NATIONAL INSTITUTE OF MENTAL HEALTH V preventing stress-related psychiatric disorders Proceedings of a Research Planning Workshop held at the Langley Porter Psychiatric Institute, University of California, San Francisco, December 10-11, 1981 Compiled and Edited by Howard H. Goldman, M.D., Ph.D. Associate Professor of Psychiatry Langley Porter Psychiatric Institute University of California, San Francisco (on leave) Assistant Director for Mental Health Financing National Institute of Mental Health and Stephen E. Goldston, Ed.D., M.S.P.H. Director, Office of Prevention National Institute of Mental Health U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Public Health Service Alcohol, Drug Abuse, and Mental Health Administration National Institute of Mental Health 5600 Fishers Lane Rockville, Maryland 20857 A» 7 / This oroceedingr odflod i the € wenn issue in the Prevention Pub- lication Series. The purpose of this series is to disseminate and ex- change information on prevention activities, especially primary prevention, in the mental health field; stimulate development of pre- vention projects in mental health, public health, and other human service facilities; encourage the training of mental health workers in prevention; and promote prevention research. The series includes scientific monographs, conference proceedings, commissioned papers, and other materials that meet a need for information about preven- tion. Publications in the series will be issued as materials and manu- scripts are developed. The views expressed in these publications are those of the author(s) and do not necessarily reflect the official posi- tion of the National Institute of Mental Health or any other part of the U.S. Department of Health and Human Services. Stephen E. Goldston, Ed.D., M.S.P.H. Director, Office of Prevention National Institute of Mental Health and Editor, Prevention Publication Series Previous Prevention Publication Series issues: Primary Prevention: An Idea Whose Time Has Come Mutual Help Groups: A Guide for Mental Health Workers Preventive Intervention in Schizophrenia: Are We Ready? New Directions in Prevention Among American Indian and Alaska Native Communities S. Preventing the Harmful Consequences of Severe and Persistent Loneliness 6. A Guide to Evaluating Prevention Programs in Mental Health BLN This document was partially compiled and edited under contract number 81MO 49264301D from the National Institute of Mental Health. Dr. Stephen E. Goldston, Director, Office of Prevention, served as the NIMH project officer. The table on pages 79 and 80 and the material on pages 112-147 are copyrighted. The chapter on pages 3-20 is adapted from copyrighted material and therefore may still be under the copyright. These materials are reproduced with permission of the copyright holders. Further reproduction of these copyrighted materials is prohibited without specific permission of the copyright holders. All other material contained in this volume except quoted passages from copyrighted sources is in the public domain and may be used or reproduced without permission from the Institute or the authors. Citation of the source is appreciated. DHHS Publication No. (ADM) 85-1366 Printed 1985 For sale by the Superintendent of Documents, U.S. Government Printing Office “Washington, D.C. 20402 FOREWORD re 5 3 [ UA pli The role of stressful life events in the pathogenesis of mental disorder has been a subject of research for over half a century. In recent years our knowledge base in this area has expanded, improving our ability to identify spe- cific mental disorders and characterize the risk associated with stressful life events. A review of theory and research is timely to guide the growing role of the National Institute of Mental Health in supporting investigation in the field of prevention. While it is abundantly clear that stressful life events have the potential for precipitating or exacerbating the course of specific mental disorders, as this proceedings document indicates, a great deal still is not known about how the magnitude of stressful life events should be as— sessed, the circumstances under which the risk of deteri- oration in mental health is increased, or the processes whereby external stress becomes internalized. Further- more, the study of stressful life events is unusually com- plex and demanding from a methodological point of view. This proceedings document reviews both substantive and methodological literature in the field of stressful life events and their relation to mental disorders to assist in- vestigators interested in entering this important research arena. The need for prospective studies is emphasized. This document discusses past, current, and planned re- search designed to explore the complex relationship between stressful life events, specific mental disorders, and discrete preventive interventions. Considerable basic epidemiology and carefully designed preventive clinical trials are necessary to delineate productive areas for de— veloping preventive intervention service programs. The potential utility of the stressful life event paradigm in prevention interventions has not been lost on creative mental health practitioners. Together, practitioners, iii service innovators, theorists, epidemiologists, and clinical investigators can advance the field of prevention. The foundation of successful interventions is well- designed research. This document is intended to inform the field of prevention research and to guide the National Institute of Mental Health in its support of this exciting field. Larry B. Silver, M.D. Acting Director National Institute of Mental Health iv PREFACE This proceedings document contains the papers prepared for the NIMH-sponsored research planning workshop on "Preventing Stress-Related Psychiatric Disorders," held at Langley Porter Psychiatric Institute, University of Cali- fornia, San Francisco, December 10-11, 1981. The work- shop brought together a group of outstanding research experts to obtain their advice and guidance on prevention research planning in this new important field. Considerable research has implicated stress in the path- ogenesis of many psychiatric and physical disorders. Major traumatic events such as natural disasters, combat, and captivity have long been associated with psychological se- quelae. Less extreme but personally distressing experiences such as physical assault, life-threatening illnesses, marital disruptions, and the death of a loved one also have been implicated in psychological and physical illness. Specific events may be significant in the etiology and exacerbation of many forms of illness and disability. Com- bat neurosis, battle fatigue, traumatic shock following natural catastrophes, and Stockholm syndrome are all ex- amples of posttraumatic stress disorders. Stress response syndromes, bereavement, depressive disorders, anxiety disorders, and schizophrenia have all been described fol- lowing stressful life events. Furthermore, stressful life events have been identified as risk factors for a variety of physical problems, including ulcers, strokes, and heart disease. The association of stressful life events and psychiatric disorders suggests an appealing paradigm for preventive intervention and prevention research. A stressful life event represents an identifiable point of exposure to risk and, as such, the likely target for a variety of intervention strat-— egies. Appropriate interventions could be made at a point prior to or just following exposure to an event to prevent subsequent disorder (primary prevention). All individuals so Vv exposed, or a subset of those at highest risk for subse- quent disorder, could be identified and administered some prophylactic intervention. Likewise, those at risk and likely to be exposed could receive some prophylaxis prior to the likely life event. Or, the entire population might undergo the preventive intervention if the life event/exposure were universal or even common. If early signs of a disorder could be identified during the otherwise latent period following a life event, early in- tervention might decrease the duration and morbidity as- sociated with the disorder (secondary prevention). Finally, if the disorder developed, interventions might prevent increased morbidity and disability during exacerbation of the disorder following stressful life events (tertiary prevention). The semantics of prevention can be confusing and prob- lematic. However, the definitions mentioned here are gen- erally recognized as correct and are restated in chapter 1 by Dr. Bloom and in chapter 4 by Drs. Klerman and Weissman. Treatment and prevention may result from the same intervention. What differs is the specification of outcome. Treatment provides relief from an established problem or disorder; prevention is measured by the future problem, disorder, or disability that is avoided. The distinctions are in the questions that are asked and the outcomes that are expected. Clarity accompanies specificity. Prevention in mental health has been criticized for its failure to specify outcomes and to identify what exactly is to be prevented. Diagnostic unreliability has permitted this lack of specificity. The Diagnostic and Statistical Manual, Third Edition provides a system for making reliable psy- chiatric diagnoses and for distinguishing discrete syn- dromes. As noted, risk factor research has identified several disorders in the diagnostic manual that may be pre— cipitated or exacerbated by stressful life events. These stress-related psychiatric disorders provide the focus of this proceedings document. Preventing stress-related psychiatric disorders implies three elements: 1. A specified psychiatric disorder or syndrome caused, precipitated, or exacerbated by 2. A life event, defined as stressful because of the risk of illness or disability from the disorder associated with the event's occurrence, and 3. An intervention that will prevent the disorder, re- duce its duration, or ameliorate associated disability from the disorder. This document discusses those issues in detail, presents examples of preventing stress-related psychiatric dis- orders, and recommends areas for preventive trials and prevention research. This workshop resulted from the collaboration of two NIMH administrative components—the Office of Preven- tion (OP), organizationally part of the Office of the Di- rector, NIMH, and the Center for Epidemiologic Studies (CES) of the Division of Biometry and Epidemiology—with the Department of Psychiatry, University of California, San Francisco, which hosted the 2-day meeting. Stephen E. Goldston, Director, OP, and Ben Z. Locke, Chief, CES, conceived the project and arranged for NIMH's sponsorship. Howard H. Goldman organized and moderated the workshop sessions. Consultation, planning, and editorial assistance were provided by Mardi J. Horowitz, Kathryn N. DeWitt, and Daniel S. Weiss of Langley Porter Psychiatric Insti- tute's Center for the Study of Neuroses and its Stress and Anxiety Clinic. The workshop had the following stated objectives: ® To review the state of the art of research knowledge on the prevention of stress-related psychiatric disorders ® To present and critique proposed research studies in this area ® To suggest preventive intervention strategies ® To recommend a research agenda to NIMH The contents of this document emanated from the third research planning workshop collaboratively organized and sponsored by the NIMH Office of Prevention. A research planning workshop serves as a key, initial step in the proc- ess of program planning and development. Operationally, the research planning workshop mechanism calls for con- vening a group of about a dozen research scientists and related professionals for a 2- to 2%-day period to discuss in depth a specific research problem or set of research vii issues. The agenda of each research planning workshop includes consideration of the following areas: ® Review of the state of the art of research knowledge with respect to the problem area on which the work- shop is focusing e Identification of the gaps in research knowledge and promising areas for further research e Indication as to the readiness of the field to pursue preventive intervention research on that specific subject e Delineation of a prioritized research agenda for the support of both basic and preventive intervention research ® Preparation of a set of recommendations to NIMH and to other groups or organizations about research con- cerns related to the specific theme of the workshop This document represents one tangible outcome from the research planning workshop. The Office of Prevention proposes to produce a monograph or proceedings document from each research planning workshop in order to make the information generated at the workshop readily available to the field, thereby promoting knowledge transfer and in- formation dissemination. Finally, the workshop mechanism may be perceived as a pipeline for program development in the sense that when the experts indicate the readiness of the field for preven- tive interventions, budget planning may proceed to set aside funds and to invite applications in that specific area relevant to preventive intervention research. All workshop participants made formal contributions either in the form of providing papers and presentations or by chairing discussions. Most papers were circulated prior to the meeting. Discussions focused on scientific criticism, clarification of definitions and opinions, and presentation of preliminary research findings, as well as on recommen- dations to NIMH. For purposes of clarity in developing the dominant themes at the workshop, the papers in this document do not follow the order of presentation at the sessions. Some papers were background essays and were not presented viii formally. In some instances, the papers differ slightly from those presented at the meeting; most authors provided revised versions that incorporated discussion points and clarifications made during the workshop. This document and the workshop for which the manu- scripts were prepared reflect growing interest in stress and other risk factors in research on mental illness. Within that context, this document is a companion piece to the NIMH conference proceedings publication entitled Risk Factor Research in the Major Mental Disorders. Several indi- viduals participated in both efforts, contributing continuity between these NIMH research planning endeavors. Selected papers from the risk factor volume that served as back- ground reading for the prevention workshop are reprinted in this document. This publication is divided into four parts. Part I, "Pre- vention, Risk, and Stress," which provides a conceptual overview of the field of concern, includes three previously published essays. Bernard Bloom's chapter, "Focal Issues in the Prevention of Mental Disorders," reviews the basic concepts of prevention and prevention research. Ernest Gruenberg discusses "Risk Factor Research Methods," and the Dohrenwends elaborate on the fundamentals of "Life Stress and Psychopathology"; the latter two chapters are reprinted from Risk Factor Research in the Major Mental Disorders. Part II, "Stress and Psychiatric Disorders," focuses on the role of stressful life events in the patho- genesis and life history of three specific syndromes. Kler— man and Weissman discuss depressive disorders in "Affective Responses to Stressful Life Events." Marks provides a comprehensive review of the anxiety disorders and mixed anxiety-depression syndrome in "Stress and Other Risk Factors in Anxiety Disorders." Horowitz details the relationship between stressful life events and a syn- drome characterized by intrusive thoughts and avoidant behavior in a chapter on "Psychological Responses to Stress: The Stress Response Syndromes." Part III takes its title from the major focus of the work- shop, "Preventing Stress-Related Psychiatric Disorders." Following an introductory chapter, "Let's Do Some Pre- ventive Trials," by Gruenberg and La Resche, three pre- ventive trials are described. DeWitt and her colleagues outline "A Research Approach for the Study of Secondary Prevention of Posttraumatic Stress Disorders." Beutler reviews a pilot study and research proposal in "Loss and Anticipated Death: Risk Factors in Depression." Falloon ix and his colleagues discuss preliminary work in "Life Events in Schizophrenia: Measuring the Effects and Modifying the Impact." Part IV contains a section on discussant comments and a summary and recommendations chapter. We anticipate that continuities between risk factor re- search and prevention research, reflected in cooperation between the Office of Prevention and the Division of Biometry and Epidemiology at NIMH and the research community, will benefit the entire mental health field. Howard H. Goldman, M.D., Ph.D. Stephen E. Goldston, Ed.D., M.S.P.H. April 1983 LIST OF WORKSHOP PARTICIPANTS AND ADDITIONAL CONTRIBUTORS Workshop Participants Larry E. Beutler, Ph.D., Professor and Chief of Clinical Psychology, Department of Psychiatry, University of Arizona, Health Sciences Center, Tucson, Arizona 85724 Bernard L. Bloom, Ph.D., Professor of Psychology, Department of Psychology, Box 345, University of Colorado, Boulder, Colorado 80309 Kathryn N. DeWitt, Ph.D., Assistant Professor of Psychiatry, Department of Psychiatry, University of California, San Francisco, Langley Porter Psychiatric Institute, 401 Parnassus Avenue, San Francisco, California 94143 Bruce P. Dohrenwend, Ph.D., Chief of Psychiatric Research, New York State Psychiatric Institute, 722 West 168th Street, New York, New York 10032 lan Falloon, M.D., Associate Professor of Psychiatry, Department of Psychiatry, University of Southern California, 1934 Hospital Place, Los Angeles, California 90033 Larry Fisher, Ph.D., Chief, Psychology Service, Veterans Administration Medical Center, 2615 East Clinton Avenue, Fresno, California 93703 Howard H. Goldman, M.D., Ph.D., Assistant Professor of Psychiatry, Department of Psychiatry, University of California, San Francisco, Langley Porter Psychiatric Institute, 401 Parnassus Avenue, San Francisco, California 94143 Michael Goldstein, Ph.D., Professor of Psychology, Uni- versity of California, Los Angeles, Department of Psy- chology, 1283 Franz Hall, Los Angeles, California 90024 Stephen E. Goldston, Ed.D., M.S.P.H., Director, Office of Prevention, National Institute of Mental Health, 5600 xi Fishers Lane, Rockville, Maryland 20857 Ernest M. Gruenberg, M.D., Dr.P.H., Professor Emeritus, Department of Mental Hygiene, School of Hygiene and Public Health, Johns Hopkins University, 615 North Wolfe Street, Baltimore, Maryland 21205 William A. Hargreaves, Ph.D., Professor of Medical Psychology, Department of Psychiatry, University of California, San Francisco, Langley Porter Psychiatric Institute, 401 Parnassus Avenue, San Francisco, California 94143 Mardi J. Horowitz, M.D., Professor of Psychiatry, Depart— ment of Psychiatry, University of California, San Francisco, Langley Porter Psychiatric Institute, 401 Parnassus Avenue, San Francisco, California 94143 Gerald L. Klerman, M.D., Professor of Psychiatry, Harvard Medical School, Director of Research, Stanley Cobb Psychiatry Research Laboratories, Massachusetts General Hospital, Boston, Massachusetts 21205 Ben Z. Locke, Chief, Center for Epidemiologic Studies, Division of Biometry and Epidemiology, National Institute of Mental Health, S600 Fishers Lane, Rockville, Maryland 20857 Isaac Marks, M.D., FRCPsych., D.P.M., Professor of Experimental Psychopathology, The Bethlem Royal Hospital, DeCrespigny Park, Denmark Hill, London, England SES 8AF Jim Mintz, Ph.D., Professor in Residence in Psychiatry, Department of Psychiatry, University of California, Los Angeles, Los Angeles, California 90024 Jon E. Rolf, Ph.D. Visiting Scientist, Division of Extramural Research Programs, National Institute of Mental Health, 5600 Fishers Lane, Rockville, Maryland 20857 Daniel S. Weiss, Ph.D., Assistant Professor of Psychiatry, Department of Psychiatry, University of California, San Francisco, Langley Porter Psychiatric Institute, 401 Parnassus Avenue, San Francisco, California 94143 Myrna M. Weissman, Ph.D., Professor of Psychiatry and Epidemiology, Director, Depression Research Unit, Yale University, 904 Howard Avenue, New Haven, Connecticut 06519 Additional Contributors Barbara Snell Dohrenwend, Ph.D., Professor and Head of the Division of Sociomedical Sciences, Columbia University, College of Physicians and Surgeons, New York, New York 10027 Linda La Resche, Dr.Sc., Research Associate, Department of Mental Hygiene, School of Hygiene and Public Health, Johns Hopkins University, 615 North Wolfe Street, Baltimore, Maryland 2120S Ronald Moss, B.S., Research Assistant, Department of Psychiatry, University of Southern California, 1934 Hospital Place, Los Angeles, California 90033 Jean Pederson, M.S., Research Assistant, Department of Psychiatry, University of Southern California, 1934 Hospital Place, Los Angeles, California 90033 Kim Shirin, M.S., Research Assistant, Department of Psychiatry, University of Southern California, 1934 Hospital Place, Los Angeles, California 90033 CONTENTS Page FOREWORD — iii PREFACE v LIST OF WORKSHOP PARTICIPANTS AND . ADDITIONAL CONTRIBUTORS —————— eee — X1 PART I - PREVENTION, RISK, AND STRESS——————- 1 Chapter 1. — Focal Issues in the Prevention of Mental Disorders Bernard L. Bloomer 3 Chapter 2. — Risk Factor Research Methods Ernest M. Gruenberg—————————— 21 Chapter 3. — Life Stress and Psychopathology Barbara Snell Dohrenwend and Bruce P. Dohrenwend-————- 37 PART II —- STRESS AND PSYCHIATRIC DISORDERS 53 Chapter 4. — Affective Responses to Stressful Life Events Gerald L. Klerman and Myrna M. Weissman——————————- SS Chapter S. — Stress and Other Risk Factors in Anxiety Disorders 15230 Marke momma 77 Chapter 6. — Psychological Responses to Stress: The Stress Response Syndromes Mardi J. Horowitz ——————————— 107 Xiv PART III - PREVENTING STRESS-RELATED PSYCHIATRIC DISORDERS——- =m meme. 149 Chapter 7. — Let's Do Some Preventive Trials Ernest M. Gruenberg and Linda La Resche——————————eee 151 Chapter 8. — A Research Approach for the Study of Secondary Prevention of Posttraumatic Stress Disorders Kathryn DeWitt, Mardi J. Horowitz, Daniel S. Weiss, and Howard H. Goldman--——————--—161 Chapter 9. — Loss and Anticipated Death: Risk Factors in Depression Larry E. Beutler—————-————————- 177 Chapter 10. — Life Events in Schizophrenia: Measuring the Effects and Modifying the Impact — A Preliminary Report Ian R. Falloon, Jean Pederson, Kim Shirin, and Ronald Moss——--195 PART IV - COMMENTARY AND CONCLUSIONS----213 Chapter 11. — Discussant Comments (edited by Daniel S. Weiss)————- 215 Chapter 12. - Summary and Recommendations-- 221 Part | PREVENTION, RISK, AND STRESS What is meant by prevention? How does one distinguish primary, secondary, and tertiary prevention? How do stressful life events figure as risk factors in the patho- genesis and prevention of disease, disorder, or dysfunction? How do we measure risk? What is a stressful life event, and what factors contribute to the life stress process? These questions are addressed in papers by Bernard Bloom, Ernest Gruenberg, and Barbara and Bruce Dohrenwend. The papers provide a foundation for understanding the prevention of stress-related psychiatric disorders. Bloom's paper elaborates on the "specific disease pre- vention paradigm, and describes a shift to a general disease prevention paradigm." This newer approach introduces stress as a risk factor in disease and as a target for pre- ventive intervention. In his paper, Gruenberg discusses risk factor research techniques, methods for measuring risk, study designs for assessing risk factors, and risk measures from preventive trials. The Dohrenwends review the re- lationship between psychopathology and stressful life events, presenting basic terms and concepts, five hypoth- eses about the life stress process, and a model for studying life stress and psychopathology. Chapter 1 FOCAL ISSUES IN THE PREVENTION OF MENTAL DISORDERS* Bernard L. Bloom, Ph.D. Until very recently, it was generally believed that at any given moment 10 percent of the population were at least partially disabled by an emotional or psychiatric disorder. Indeed, the phrase "1 out of 10" has been the long-time rallying cry of those citizens who have organized them- selves as advocates for expanded services for the mentally ill. Yet, tragically, in spite of the continually expanded capacity of the American mental health service delivery system to provide mental health care, it is now estimated that 15 percent of the American population suffer from some form of emotional disorder during the course of a year and that, of these 32 million persons, nearly 7 million receive no care of any kind (Regier et al. 1978). These facts are illustrative of the public health axiom that major disorders can be controlled only by prevention, not by providing treatment services. In the United States, at least, there seems to be no alternative other than to conclude that emotional disorders are out of control and are likely to remain so until more resources are put into *This paper appears with a somewhat different title in Schul- berg, H. C., and Killilea, M., eds. Principles and Practices of Community Mental Health. San Francisco, Calif.: Jossey-Bass, 1982; and is a revision and updating of an earlier publication (Bloom, B. L. Prevention of mental disorders: Recent advances in theory and practice. Community Mental Health Journal 15:179-191, 1979). Acknowledgment is made to both publishers for their permission to draw on those papers in the preparation of this manuscript. effective preventive intervention programs. Albee (1980) has suggested that the growing interest in primary pre- vention has many of the characteristics of a revolution, one that will challenge both the authority of the mental health establishment and the current disproportionately large allocation of mental health resources to the provision of treatment services for persons who are already emo- tionally disordered. While the rationale for expanded preventive programs seems unobjectionable, the issues that are raised as one discusses prevention can quickly become very complex and often controversial. It is important to try to clarify these issues, since part of this complexity is a function of in- adequate understanding of some basic concepts relevant to the field of prevention (Caplan 1961, 1964; DeWild 1981). Basic Concepts and Definitions With regard to the control of any disorder—emotional or physical—two types of interventions exist. The first type seeks to reduce the number of persons suffering from the disorder, that is, to decrease the prevalence of that disorder. The second type of intervention seeks to reduce the severity, discomfort, or disability associated with the disorder. Such programs are formally known as tertiary prevention but are better known as rehabilitation. With lifelong disorders, rehabilitation programs generally have little effect on prevalence. Indeed, a well-run reha- bilitation program may actually increase the prevalence of these disorders, because it may increase life expectancy. Unfortunately, within the context of our current knowl- edge, many emotional disorders appear to be lifelong or nearly lifelong illnesses. Because the prevalence of any disorder is a function of its duration and the rate at which new cases are produced, two approaches to reduce the actual prevalence of a dis- order are commonly employed. The first seeks to reduce prevalence by reducing the duration of the disorder, usually through the development of some form of early casefinding combined with the prompt application of effective treat— ment. Techniques for the control of any disorder that focus on early casefinding and prompt effective treatment, that is, techniques that seek to reduce the prevalence of a disorder by reducing its duration, are formally called 4 secondary prevention. Secondary prevention efforts are preventive only in the sense that systematic early case- finding brings with it the possibility that the duration of the disorder might be able to be reduced. Should a technique for the early identification of some disorder be developed and employed, without the concomi- tant development of more effective treatment procedures, a paradoxical increase in prevalence of that disorder would occur. Gruenberg has commented that "without an effec- tive treatment, early diagnosis only provides more work for clinicians without changing the prevalence of the disorder" (1980, p. 1323). For example, an increase in duration and prevalence has occurred in the case of diabetes because of improved techniques for its early identification. A similar increase in duration and prevalence has occurred in the case of Down's syndrome as a consequence of the medical advances that have taken place in the development of an- tibiotics. In this case, the death rate from secondary causes among persons with that syndrome has been sig- nificantly reduced (Gruenberg 1977). The alternative approach to prevalence reduction is to reduce the rate at which new cases of a disorder develop. This approach seeks to reduce prevalence by reducing incidence and is formally designated as primary pre-— vention. This concept most closely matches the lay use of the term prevention. Effective primary prevention pro- grams actually prevent disorders from occurring or reduce the likelihood that a disorder will occur in a particular population (Perlmutter et al. 1976). The concept of primary prevention requires that the in- tervention take place in a population free of the disorder being prevented. It might be appropriate to consider the development of a primary prevention program for schizo- phrenia aimed at a sample of neurotics, but as long as it is thought that acute and chronic schizophrenia are part of a single continuum, it would be incorrect to suggest that effective treatment services provided for acute schizo- phrenics constitute a primary prevention program for chronic schizophrenia. Such programs, while critically im- portant, are in fact secondary prevention. That is, they seek to reduce the duration of a disorder already present. On the other hand, physical rehabilitation and conscious- ness-raising programs provided for postmastectomy patients can quite legitimately be examined in terms of the extent to which they serve a primary prevention function for emotional disorders (Goldston 1977). 5 Disease Prevention and Health Promotion One additional important distinction should be made, that is, the distinction between disease prevention and health promotion (McPheeters 1976). Some disorders can be prevented by highly specific procedures—procedures that do not appear to be effective in preventing anything other than that specified disease. Malaria can be prevented by destroying the breeding grounds of a particular type of mosquito. There is no evidence that any other disease is thereby prevented. Drinking fluoridated water results in a dramatic reduction in the incidence of dental caries. No other disorder is reduced by fluoridated water. These are examples of specific forms of disease prevention. Although the mechanisms that give rise to a specific disease or disorder may not be completely understood, many may be prevented by the application of such procedures. As will be seen, this disease-specific prevention strategy has been very effectively employed in the case of several psychi- atric disorders. In contrast, a variety of nonspecific practices, for ex- ample, providing crisis intervention services or social sup- port during times of stress, may have a positive effect on health in general and may, in fact, prevent a variety of forms of disordered behavior. Those practices that have a generally salutory but unspecifiable effect on health are referred to as health promotion. Eisenberg similarly de- fines health promotion as those activities that "contribute to resistance to disease, even when the disease agents are not known or beyond control" (1981, p. 5). One of the most influential recent reports regarding ways of improving the general level of health of a popula- tion by both disease prevention and health promotion strategies was prepared by the Canadian Minister of Na- tional Health and Welfare (Lalonde 1974). In this report, Lalonde introduced the concept of the health field and its four components: human biology, environment, lifestyle, and health care organization. Human biology refers to aspects of health that are de- veloped within the human body and are related to the or- ganic nature of the individual, including the individual's genetics. The environment refers to those matters outside the body, over which the individual has little if any control, that can have an effect upon health. Included in this cate- gory would be food and water purity, air pollution, safe 6 disposal of sewage, noise control, and road and vehicular safety. Lifestyle refers to decisions people make about their own behavior, over which they have considerable control, that can have an effect upon their health, e.g., overeating, smoking, the abuse of alcohol and other drugs, insufficient exercise, or careless driving. Finally, the health care or- ganization refers to the quality, quantity, and distribution of health-related services in any community. Using the health field concept, it is possible to examine morbidity or mortality rates in an effort to determine to what extent these rates could be reduced. In the case of traffic deaths, for example, it has been estimated that 75 percent of them can be accounted for by pathological lifestyles, 20 percent by the environment, and 5 percent by defects in the health care organization. Similarly, it has been estimated that self-destructive lifestyles account for about half of all deaths that occur before age 70, and that 20 percent of those premature deaths may be attributed to just two lifestyle habits—smoking cigarettes and excessive use of alcohol. Since the publication of this Canadian report, there has been growing interest in examining lifestyles and their role in illness and in premature death. As part of this interest in health promotion, mental health professionals are begin- ning to examine various aspects of lifestyles in terms of their roles in predisposing people to emotional disorders or precipitating such disorders in populations that are vul- nerable but not disordered. The explicit objective in this examination is to develop programmatic strategies to prevent emotional disorders before they start. Preventable Psychiatric Disorders There is by now a well-established knowledge base re- garding the role of specific interventions in preventing specific psychiatric disorders. The American Public Health Association (1962) has identified six categories of mental disorders that are preventable, in part because they are all disorders of known etiology. While these disorders do not constitute a large proportion of all mental disorders, nearly all result in chronic brain syndromes, many are lifelong in their effects, and they represent, in total, an enormous 7 cost to society as well as to the victims and their families (Gruenberg 1980; Kornberg and Caplan 1980). The first group of preventable disorders involves those related to poisoning. These disorders are typically sub- divided into acute poisoning—disorders resulting from in- tentional or accidental ingestion of drugs, inhalants, or solvents—and chronic poisoning—disorders most commonly caused by prolonged exposure to industrial toxins or pro- longed use of medications or addicting drugs. More than 250,000 chemicals and drugs can be poisonous, including, for example, acetone, arsenic, barbiturates, carbon mon- oxide, cyanides, DDT, kerosene, morphine, and turpentine. Poisonings account for about 3,000 deaths every year and another 3,000 survivors who are left with chronic brain syndromes. Prevention requires changes in the environment, in lifestyles, and in the health care system; specifically, more prudent prescribing, dispensing, and storage of drugs, more careful control over drug prescriptions and their renewal, reduced exposure to industrial poisons, better safety standards in industrial and agricultural settings, improved labeling of industrial and household products, and prompt and accurate diagnosis and treatment. Health education can play an important role in achieving these objectives, as can changes in laws and regulations and the development of a network of readily accessible poison control centers. The second category of preventable disorders includes those resulting from infectious agents that can invade the central nervous system and leave permanent brain damage. Infections during the fetal period, such as those caused by rubella, syphilis, and toxoplasmosis, can produce ‘severe mental retardation, epilepsy, perceptual and cognitive defects, and difficulty in impulse control. Rubella can be prevented by the administration of gamma globulin to women who have not previously had the disease. Successful treatment of syphilis in the mother prevents the development of fetal syphilis. Prompt treatment of toxo- plasmosis can prevent the transmission of the disease during pregnancy. Infectious diseases during childhood, such as pertussis, influenza, measles, meningitis, mumps, and tuberculosis, can also produce permanent brain damage that can lead to defects in sensory, motor, and intellectual development. Many of these infectious processes can be prevented by immunizations, yet substantial proportions of young chil- dren are not completely immunized against these diseases (Eisenberg 1981, pp. 9-10). The third group of preventable disorders originates with genetic processes. Among these conditions are Tay-Sachs disease, phenylketonuria, galactosemia, tuberous sclerosis, and Huntington's chorea. In the case of phenylketonuria and galactosemia, both of which can produce severe mental retardation, special dietary intake can prevent the dis- order. In the case of most genetic disorders, carriers can be identified or risks estimated on the basis of family studies. Under these circumstances, genetic counseling can help reduce the numbers of vulnerable children. The fourth category of preventable mental disorders is produced by nutritional deficiencies, including beri-beri, Wernicke's encephalopathy, kwashiorkor, pellagra, and an- oxemia. These prolonged nutritional deficiencies during childhood appear to have a considerable general impact on mental development and increase the risk of mental retar- dation, epilepsy, and a variety of perceptual and cognitive disorders. In addition, there is considerable evidence that general nutritional deficiencies increase vulnerability to infections. Nutritional dietary supplementation, combined with nutritional education, can prevent these disorders. The fifth kind of preventable mental disorders result from injuries to the central nervous system, including those incurred in falls, gunshot wounds, and vehicular accidents. The incidence of such injuries can be dramatically reduced by making the sources, such as guns, less readily available, by encouraging the use of protective equipment, such as motorcycle helmets and seat belts, and by improving the safety of motor vehicles and roads. Finally, a number of preventable or treatable general systemic disorders—such as erythroblastosis fetalis, hyper— thyroidism, cretinism, intracranial masses, toxemia of pregnancy, or prematurity—can produce chronic brain syndromes. With respect to this knowledge base, additional inves- tigation is required only when it appears that an effective procedure is not being applied—for example, why, in view of their clear effectiveness in reducing automobile acci- dent injuries, drivers and passengers persist in failing to wear seat belts; or why a substantial proportion of people are not immunized against diseases that can sometimes, as in the case of measles, have profoundly harmful effects upon central nervous system functioning. The Paradigm Shift in Primary Prevention Even if every preventable specific psychiatric disorder were prevented, however, we would still be left with a very large number of psychiatric casualties in our communities. The following specific disease prevention paradigm that has worked so well for the past 200 years in the prevention of infectious and nutritional diseases no longer seems to be effective. 1. Identify a disease of sufficient importance to justify the development of a preventive intervention program. Develop reliable methods for its diagnosis, so that people can be divided with confidence into groups ac- cording to whether they do or do not have the disease. 2. By a series of epidemiological and laboratory studies, identify the most likely theories of that disease's path of development. 3. Mount and evaluate an experimental preventive in- tervention program based on the results of those research studies. This, in somewhat oversimplified form, is the paradigm one assumes when thinking about preventing a specific disorder. The paradigm has been remarkably effective for a broad array of communicable diseases—smallpox, typhus, cholera, typhoid fever, the plague, malaria, diphtheria, tuberculosis, tetanus, and, more recently, sexually trans— mitted diseases, rubella, and polio. It has worked also with an equally impressive list of what are now known to be nutritional diseases—scurvy, beriberi, pellagra, rickets, kwashiorkor, endemic goiter, and dental caries. All of these diseases have one attribute in common. For each, there is an identified, necessary (although not always sufficient) biological precondition for its appearance—lack of niacin or vitamin B, protein deficiency, invasion of a particular bacillus, lack of fluoride, and so on. This neces— sary precondition allows talk about the cause of a partic- ular disease. In keeping with this tradition, active research has been underway for some time in an effort to identify the biological bases of specific psychiatric disorders. The General Disease Prevention Paradigm In an effort to understand the development and preven- tion of currently unpreventable disorders, a new paradigm 10 has arisen within the past decade or so. This paradigm does not begin with the assumption that every specific disorder has a single or even multiple necessary preconditions. Rather, it is based on the clearly established association of stress with increased risk of illness and the assumption that all individuals are variously vulnerable to stressful life ex— periences and that "almost any disease or disability may be associated with these events" (Dohrenwend and Dohren- wend 1974, p. 314). Four vulnerable persons can face a stressful life experience, perhaps the collapse of their marriage or the loss of their job. One person may become severely depressed, the second may be subsequently in- volved in an automobile accident, the third may become clinically alcoholic, and the fourth may develop a psychotic thought disorder or coronary artery disease. In the most recent report of the Surgeon General (USDHHS 1980), one important goal articulated for promoting health and pre- venting disease is more effective control of stress. On the basis of the new paradigm, preventive interven- tion programs can be organized to reduce the incidence of particular stressful life events, whenever possible, or fa-— cilitate their mastery once they occur. In either case, one need not have undue regard for the prior specification of which forms of disability might thereby be prevented. That is, this new paradigm begins by abandoning at the outset the search for a unique cause or set of causes for each disorder. In contrast to the classic paradigm already de- scribed, the new paradigm has this sequence of steps: 1. Identify a stressful life event that appears to have undesirable consequences. Develop procedures for reliably identifying persons who have undergone or who are undergoing that stressful experience. 2. By traditional epidemiological and laboratory meth- ods, study the consequences of that event and develop hypotheses related to how one might go about reducing or eliminating the negative consequences of the event. 3. Mount and evaluate experimental preventive inter- vention programs based on these hypotheses. The general disease prevention paradigm has turned re- searchers' attention from longstanding predisposing factors in psychopathology to far more recent precipitating factors and is part of an even broader phenomenon. It has long been known that biological, psychological, and sociological factors differentially predispose persons to emotional dis- orders. With few exceptions, however, efforts to develop effective preventive services based on attempts to modify 11 these distal predisposing factors have been unsuccessful. Eisenberg has recently commented that "measurement of distant outcome places a terrible burden of proof on child- hood interventions; they must be powerful indeed to be able to show a clear effect despite the vicissitudes of sub- sequent life experience" (1981, p. 4). Furthermore, psycho- therapeutic strategies that focus on developing increased understanding and acceptance of these predisposing factors in the case of casualties have not been conspicuously suc- cessful and, even when successful, have demanded un- realistically large proportions of our resources. There is every reason to believe that preventive pro- grams linked to stressful life events can be effective, particularly when scientists set about to build on what is already known about crisis theory and crisis intervention (Bloom 1975S, pp. 134-172; Caplan 1964, pp. 34-54; Mann 1978; Parad 1965; Parad et al. 1976). The author's applied research program at the University of Colorado concerns itself with separation and divorce as stressful life events and draws very heavily on these concepts. But other crit- ical precipitating factors, to mention just a few, that are receiving increasing attention include school entrance, parenting, retirement, and widowhood. It should be noted that these events are all common, many are becoming more common, and few sustained and comprehensive services exist within our society to assist people in mas- tering any of them. As mental health professionals and social scientists be- came interested in stress, their initial approach was to examine the possibility that each type of stressful expe- rience was associated with a particular stress disease. Cassel makes a persuasive case that "there is serious doubt as to the utility or appropriateness of both of these no- tions. . . . It is most unlikely that any given psychosocial process or stressor will be chronologically specific for any disease, at least as currently classified" (1976, p. 109; Cassel 1973, 1974). Cassel's own research, as well as his review of the studies of others, led him to conclude that although psychosocial processes enhance susceptibility to disease, the clinical manifestations of this enhanced sus- ceptibility would not be functions of the particular psy- chosocial stressor. If readers then keep in mind these two developments— the growing interest in precipitating crisis events rather than in predisposing developmental variables, and the growing acceptance of general as well as disorder—specific 12 preventive interventions—they can better appreciate much of the recent research literature. Two particularly compelling concepts that have been introduced to link individual disorders with the stressful properties of the social system have been the ideas of so- cial competence and competence building, on the one hand, and social support, on the other. Much human misery ap- pears to be the result of lack of competence, that is, lack of control over one's life, of effective coping strategies, and the lowered self-esteem that accompanies these de- ficiencies (Ryan 1971; Shure 1979). A substantial body of research from a variety of domains appears to converge on social competence building as one of the most persuasive preventive strategies for dealing with individual and social issues. The second line of recent research links the availability of social support systems to vulnerability to disease. Cap- lan has provided a valuable analysis of the nature of sup- port systems, those "continuing social aggregates that provide individuals with opportunities for feedback about themselves and for validations of their expectations about others, which may offset deficiencies in these communica tions within the larger community context" (1974, pp. 4-5). Cassel's (1973) review of the research suggested that in- creased susceptibility to disease is likely to come about when people are new to an unfamiliar environment, or when there is considerable social disorganization, and that susceptibility to disease is greater in those persons who are subordinate rather than dominant in a society and persons who are deprived of meaningful social contacts and social supports. He concluded that a high level of disease in gen- eral might be anticipated under conditions of social change and social disorganization, and that "preventive action in the future should focus more directly on attempts at mod- ifying these psychosocial factors, on improving and strengthening social supports, and reducing the circum- stances which produce ambiguities between actions and their consequences" (Cassel 1973, p. 547; Berkman and Symes 1979; Gottlieb and Todd 1979; Kagey et al. 1981). The links between these areas of research and concep- tualization are striking and make a persuasive case for competence assessment and competence building at the individual and aggregate community level (Iscoe 1974) as a high-priority community mental health program objective. In summary, these studies suggest that significant progress may be made in preventing emotional disorders, first, by 13 rephrasing the primary prevention question to ask about preventing mental disorders in general, and, second, by identifying stress—producing critical precipitating events in environments and developing systematic and multi- dimensional interventions to help in the mastery of these stresses. Included among these strategies might be strengthened sources of social support and improved social competencies. While these two orientations to health promotion appear to be unusually promising, they do not exhaust the list of potential approaches. The American Public Health As- sociation has suggested a number of other strategies, in- cluding prevention of maternal deprivation, improvement of child-rearing practices, better prenatal care, reduction of radiological exposure, and genetic counseling (1962, pp. 57-61). More recently, Gruenberg (1980) has suggested ad- ditions to this list—improved nutrition, improved dental health, and improved medical care. The Immediate Future of Primary Prevention Evaluation of the status of primary prevention programs in a typical community will generally show that prevent-— able disorders continue to occur, that only a small pro- portion of the population is thriving, and that insufficient resources are allocated toward extending the frontiers of knowledge that might serve to diminish the incidence of emotional disorders. Because those disorders are rarely fatal and express their impact primarily in terms of high morbidity rather than mortality, communities face unmet needs in their efforts to treat the already ill and are often reluctant to divert scarce resources away from that im- portant humanitarian task (Lamb and Zusman 1979). Yet the public is not unwilling to seek preventive serv- ices. It has recently been estimated (Cypress 1981) that 17 percent of physicians office visits during 1977-78 took place primarily in the interest of preventive care. Clearly, a number of barriers exist that make it difficult to enlarge upon primary prevention activities in the field of emotional disorders. First, no national policy exists with regard to the en- hancement of mental health, although the President's Commission on Mental Health (1978) has proposed steps toward such a policy. This lack of national policy is re- 14 flected in a wide variety of phenomena. No one seems to be in charge of health maintenance or promotion. Respon- sibilities are fragmented even when assigned, and coordi- nation among agencies is often impossible to achieve. A second barrier relates to difficulty in extending knowledge regarding effective approaches to primary pre- vention. Excellent research is now being undertaken and reported (Albee and Joffe 1977; Bond and Rosen 1980; Forgays 1978; Kent and Rolf 1979; and Price et al. 1980), but continuing to extend this knowledge base will require increased numbers of skilled and dedicated scientists and expanded research support. Eisenberg has noted that "when one considers that the entire investment in the basic and applied research and the field trials that made the polio- myelitis vaccine possible was less than the cost of the illness burden from a single epidemic year of the disease, pursuit of the same strategy for other disorders is im- mensely appealing" (1981, p. 4). It may require a decade or longer to determine the extent to which preventive serv- ices have been effective. These longitudinal studies need not be prohibitively expensive in any one year, but there needs to be a stable and predictable support base over a relatively long period of time. Gruenberg has recently commented about the need for better research in the field of primary prevention by noting that a special group of prevention advocates share three critical assumptions about the interventions they advocate. They assume that the interventions are harmless; socially acceptable and not frightening to the subjects; and able to prevent some disorders on some reasonable basis. Whether or not they actually prevent a disorder is best found out by starting with a planned preventive trial. It is extraordinary that many of the fashionable mental health prevention programs have these characteristics, yet none of them have produced a preventive trial. Such programs as widow-to-widow counseling, to prevent extended depression; correction of low self-esteem in pregnant women to prevent neurotic symptom formation in their daughters; and enhancing the coping skills of mentally retarded parents and of their children to prevent mental retardation in the children are ex- cellent examples of hypotheses that are best tested in the first instance by a carefully prepared preventive trial. If such educational and counseling methods ac- 15 tually have large effects, they deserve to become more widespread. If they do not, other preventable causes must be found to lower the incidence of these disorders. (1980, pp. 1323-1324) A third barrier lies in our failure to accept the conse- quences of the growing realization that monoetiological theories are not viable with regard to understanding the development of most forms of psychiatric disorders (Weissman and Klerman 1978), and that the concept of health is itself multidimensional. Effective preventive ef- forts will require the active collaboration of many pro- fessions, many agencies, and many social systems. These efforts will demand the active involvement of teachers, parents, the clergy, nutritionists, economists, architects, and so on. Although this sentiment is often subscribed to in theory, it is rarely adopted in practice. Most activities in the field of primary prevention appear to equate health with the absence of disease and thus see their goal as the enhancement of health by means of the reduction of disease incidence. For more than 2,000 years, however, an alternative view of health has had its adher- ents—a view that equates health with those "natural laws which will ensure . . . a healthy mind in a healthy body" (Dubos 1959, p. 111). From this point of view, health is not a state; it is a process, a way of life. It is from this point of view that Williamson and Pearse described health as "intrinsically and distinctively a positive process not de- finable in terms of ‘'absence' of disease and/or disorder; hence not attainable through a 'preventive' approach nec- essarily directed to securing the absence of disease and disorder" (1966, pp. 293-294). This broad continuum in the concept of health brings with it an equally broad array of appropriate objectives, ranging from primary prevention activities designed to maintain a disease-free status quo to those activities designed to promote growth and wholeness and to help persons make maximum use of their potential, independently of whether a disease process might be present. Some preventive interventions can be effective if made available to individuals singly or in groups. Other inter— ventions must be directed at social systems to have an ul- timate effect on individuals. In part, this issue relates to the theory to which one subscribes regarding a particular type of stress and appropriate strategies for intervention. Just as theories and strategies may vary, so may levels of intervention differ. Some years ago, C. Wright Mills (1959) 16 proposed a distinction between troubles—when an indi- vidual's values are perceived as being threatened—and is— sues—when a society's values are perceived as being threatened. Reiff (1977) makes use of this distinction when contrasting the instability of an individual marriage—a personal trouble—with the phenomenon in which 50 percent of marriages end in divorce. In this latter case, Reiff suggests that "there is something beyond the intrapsychic and interpersonal, something beyond the situation, and something in the social structure that accounts for these personal troubles" (1977, p. 49). The clear implication of this line of reasoning is that any intervention that stops short of a thoroughgoing examination of societal conditions that are producing this much disruption may very well fail. A final barrier that should be noted lies within the men- tal health establishment itself and is reflected in the characteristics of contemporary practice and of societal values that underlie these practices. First, as Albee (1979) has noted, our society is oriented toward quick solutions to current problems, and mental health professionals are rarely enthusiastic about long-term planning aimed at long-term goals. Second, most members of the mental health establishment have been socialized into believing that treatment is a far more valued activity than preven- tion. Third, the mental health establishment is unprepared, by either training or ideology, to advocate for needed so- cial change in an effort to prevent mental disorders. Fi- nally, members of the mental health establishment are part of a larger third-party reimbursement procedure that le- gitimizes little more than the provision of treatment for diagnosed diseases. In such a fiscal environment, there is little motivation to develop preventive programs that are neither fundable nor reimbursable. It may be necessary to develop an agency solely con- cerned with primary prevention—an agency that would serve the healthy for the purpose of maintaining and en- hancing health and robustness. At one time, the public health agency played that role. Perhaps local health de- partments need to be encouraged to resume that respon- sibility in a late 20th century version of an older esteemed community service. Such an agency must deal with the entire spectrum of preventive services. It must recognize that health is a social as well as a biological phenomenon and that there are healthy ways of being sick just as there are unhealthy ways of being well. It must legitimize and encourage holistic as well as specialized views of the proc- 17 472-367 O - 85 - 2 ess of health and of a healthy life and must attend to the entire range of contemporary stress, from the biological to the psychological to the sociological and cultural. It must take a broad view of health as not only the absence of disease but also the presence of joy in living and hope in the ultimate resolution of life's conflicts. References Albee, G.W. Primary prevention. Canada's Mental Health 27:5-9, 1979. Albee, G.W. The fourth mental health revolution. Journal of Prevention 1:67-70, 1980. Albee, G.W., and Joffe, J.M., eds. The Issues: An Over— view of Primary Prevention. Hanover, N.H.: University Press of New England, 1977. American Public Health Association. Mental Disorders: A Guide to Control Methods. New York: APHA, 1962. Berkman, L.F., and Symes, S.L. Social networks, host re— sistance, and mortality: A nine-year follow-up study of Alameda County residents. American Journal of Epidemiology 109:186-204, 1979. Bloom, B.L. Community Mental Health: A General In- troduction. Monterey, Calif: Brooks/Cole, 1975. Bond, L., and Rosen, J., eds. Competence and Coping During Adulthood. Hanover, N.H.: University Press of New England, 1980. Caplan, G., ed. Prevention of Mental Disorders in Chil- dren: Initial Explorations. New York: Basic Books, 1961. Caplan, G. Principles of Preventive Psychiatry. New York: Basic Books, 1964. Caplan, G. Support Systems and Community Mental Health: Lectures on Concept Development. New York: Behavioral Publications, 1974. Cassel, J. The relation of the urban environment to health: Implications for prevention. Mount Sinai Journal of Medicine 40:539-550, 1973. Cassel, J. Psychosocial processes and 'stress': Theoretical formulation. International Journal of Health Services 4:471-482, 1974. Cassel, J. The contribution of the social environment to host resistance. American Journal of Epidemiology 104:107-123, 1976. Cypress, B.K. Office visits for preventive care, National Ambulatory Medical Care Survey: United States, 1977- 18 78. National Center for Health Statistics Advanced Data. No. 69. April 1981. DHHS Pub. No. (PHS)81-1250. Washington, D.C.: U.S. Govt. Print. Off., 1981. DeWild, D.W. Toward a clarification of primary prevention. Community Mental Health Journal 16:306-316, 1981. Dohrenwend, B.S., and Dohrenwend, B.P., eds. Stressful Life Events: Their Nature and Effects. New York: John Wiley, 1974. Dubos, R. Mirage of Health. New York: Harper & Row, 1959. Eisenberg, L. A research framework for evaluating the promotion of mental health and prevention of mental ill- ness. Public Health Reports 96:3-19, 1981. Forgays, D., ed. Environmental Influences. Hanover, N.H.: University Press of New England, 1978. Goldston, S.E. An overview of primary prevention pro- gramming. In: Klein, D.C., and Goldston, S.E., eds. Pri- mary Prevention: An Idea Whose Time Has Come. Washington, D.C.: U.S. Govt. Print. Off., 1977. Gottlieb, B.H., and Todd, D.M. Characterizing and promot- ing social support in natural settings. In: Munoz, R.F.; Snowden, L.R.; and Kelly, J.G., eds. Social and Psycho- logical Research in Community Settings. San Francisco: Jossey-Bass, 1979. Gruenberg, E.M. The failures of success. Milbank Memo— rial Fund Quarterly/Health and Society 55:3-24, 1977. Gruenberg, E.M. Mental disorders. In: Last, J.M., ed. Maxcy-Rosenau Public Health and Preventive Medi- cine. 11th ed. New York: Appleton—-Century-Crofts, 1980. Iscoe, I. Community psychology and the competent com- munity. American Psychologist 29:607-613, 1974. Kagey, J.R.; Vivace, J.; and Lutz, W. Mental health pri- mary prevention: The role of parent mutual support groups. American Journal of Public Health 71:166-167, 1981. Kent, M.W., and Rolf, M.E., eds. Promoting Social Com- petence in Children. Hanover, N.H.: University Press of New England, 1979. Kornberg, M.S., and Caplan, G. Risk factors and preventive intervention in child psychotherapy. Journal of Preven- tion 1:71-133, 1980. Lalonde, M. A New Perspective on the Health of Canadians. Ottawa: Canadian Govt. Print. Off., 1974. Lamb, H.R., and Zusman, J. Primary prevention in per— spective. American Journal of Psychiatry 136:12-17, 1979. 19 Mann, P.A. Community Psychology: Concepts and Ap- plications. New York: Free Press, 1978. McPheeters, H.L. Primary prevention and health promotion in mental health. Preventive Medicine 5:187-198, 1976. Mills, C.W. The Sociological Imagination. London: Oxford University Press, 1959. Parad, H.J., ed. Crisis Intervention: Selected Readings. New York: Family Service Association of America, 1965. Parad, H.J.; Resnik, H.L.P.; and Parad, L.G., eds. Emer- gency and Disaster Management. Bowie, Md.: Charles Press, 1976. Perlmutter, F.D.; Vayda, A.M.; and Woodburn, P.K. An instrument for differentiating programs in prevention— primary, secondary, and tertiary. American Journal of Orthopsychiatry 46:533-541, 1976. President's Commission on Mental Health. Report to the President. Vol. 1. Washington, D.C.: U.S. Govt. Print. Off., 1978. Price, R.H.; Ketterer, R.F.; Bader, B.C.; and Monahan, J. Prevention in Mental Health: Research, Policy, and Practice. Beverly Hills, Calif.: Sage Publications, 1980. Regier, D.A.; Goldberg, 1.D.; and Taube, C.A. The de facto U.S. mental health services system. Archives of General Psychiatry 35:685-693, 1978. Reiff, R. "Ya gotta believe." In: Iscoe, I.; Bloom, B.L.; and Spielberger, C.D., eds. Community Psychology in Tran- sition. Washington, D.C.: Hemisphere Publishing, 1977. Ryan, W. Blaming the Victim. New York: Random House, 1971. Shure, M.B. Training children to solve interpersonal prob- lems: A preventive mental health program. In: Munoz, R.F.; Snowden, L.R.; and Kelly, J.G., eds. Social and Psychological Research in Community Settings. San Francisco: Jossey-Bass, 1979. U.S. Department of Health and Human Services. Pro- moting Health/Preventing Disease: Objectives for the Nation. Washington, D.C.: U.S. Govt. Print. Off., 1980. Weissman, M.M., and Klerman, G.L. Epidemiology of men- tal disorders: Emerging trends in the United States. Ar-— chives of General Psychiatry 35:705-712, 1978. Williamson, G.S., and Pearse, I.H. Science, Synthesis, and Sanity. Chicago: Henry Regnery, 1966. 20 Chapter 2 RISK FACTOR RESEARCH METHODS* Ernest M. Gruenberg, M.D., Dr.P.H. Introduction This sketch of risk factor research methods is an at- tempt to show that biostatisticians and epidemiologists have accumulated techniques for determining basic risk measurement. It reviews the relevant literature as an in- troduction to, not a substitute for, the publications cited. Epidemiological research on risk factors has a tradition. One feature of that tradition is a readiness to switch levels to pursue a problem. One may get into a problem with a psychological hypothesis, but, in pursuing the problem, one may have to switch to interest in some DNA molecules or some brain tissue changes or the social network theory. When the evidence points away from the originally sus- pected antecedent, there are three choices: drop the prob- lem because it is outside the investigator's "discipline," learn more about the nature of the apparent risk factor, or (if one does not wish to keep acquiring new doctorates) learn to collaborate. Epidemiology has a tradition of col- laboration; it progresses by integrating its hypotheses and research with clinical, laboratory, and biostatistical re- search workers. Today, to facilitate interdisciplinary collaboration, we must all learn about risk research methods, whether we *Reprinted from National Institute of Mental Health. Risk Factor Research in the Major Mental Disorders. DHHS Publi- cation No. (ADM)81-1068. Washington, D.C.: Superintendent of Documents, U.S. Government Printing Office, 1981. 21 want to or not. We cannot leave it to the "riskologists." Epidemiology is a Greek word for a science. As Hip- pocrates used it, the word means the study of "upon the people." It is a study of how things fall on the people. A disease that strikes the people in one place at a particular time is an epidemic. But epidemiology is not just the study of epidemics (as Webster would have you believe). It is more generally the study of: e Who gets sick, and who doesn't get sick? e Why? (What are the risk factors?) ® What can we do to make the sickness less common? So epidemiology is the basic science of public health practice and preventive medicine. It is an action science, directed at a set of practical problems. Many useful epi- demiologic studies do not lead directly to action. But the goal of epidemiologic inquiries is to find a modifiable risk factor, a risk factor that can be manipulated through social action or individual action so as to lower the frequency of a disease, of a disability, or of death. Hence, epidemiology is pursued in the Public Health Service and in State and city health departments. One cannot help being an epidemiologist, if one is interested in disease prevention. Actuarial methods are designed to get an accurate measure of contingency risks. Epidemiologists are preoccupied with finding ways to lower the risk of people becoming ill or disabled or dead. Epidemiology is defined by its questions, not by its method of getting answers. Epidemiology does not try to go it alone. Many con- tributions to mental disorder epidemiology have been made by nonepidemiologists who insist that they were doing something else. For example, H. Warren Dunham insists he was doing social ecology research in Mental Disorders in Urban Areas (Faris and Dunham 1939). Also, Seymour Kety has been doing risk factor research in his adoption studies, although that is not his term for it. Quantification of Risk The concept of risk is older than the idea of measuring risk. People took chances; they did not say how big the 22 chance was. Being a sailor was a risky occupation, sending a ship from England to Italy was risky. Everyone knew it was dangerous, but no one put a numerical value on the risk. Risk became something to measure when people tried to insure themselves against risk. Risks were assigned nu- merical values as a result of bargaining between those who took the risk and those who would pay. The ship owner bet the ship would sink; the insurer bet it was going to survive the trip. They had to agree on odds. How much does the ship owner pay the insurer (the premium) so the insurer will pay the owner the cash value of the ship and its cargo (the benefit) if the ship sinks? They negotiated until they made a deal. The ratio of premium to benefit quantifies the size of the risk. Each side guessed. It was a betting proposition. This went on for a long time in risk measurement. The whole life insurance business is built on it. Actuarial statistics try to state how many people out of a cohort are going to die so as to guide those who sell life insurance or annuities. Until the time when the National Institutes of Health were created, almost all statistical methods of figuring odds and chances about death were based on risk of dying. Individual Risk J.N. Morris (1975) describes individual risk in his text Uses of Epidemiology. The risk of dying is higher for the very fat (or the very lean) than for the average-weight person. At different ages, people are at different risks of having automobile accidents. An outstanding example of measured individual risk is the well-established fact that the risk of a woman's newborn baby having Down's syn- drome rises exponentially with the mother's age. One way of measuring the individual risk of some par- ticular event by a particular age is a simple summation of the accumulated experience over time of age-specific in- cidence rates of that event. (The annual incidence rate is the number of new cases starting in a year divided by the average number of people observed during that year.) Measurement of the accumulated risk in time, or individual risk as a function of age, is illustrated by a study of senile dementia. The probability that a man who survives to a given age will have escaped the clinical signs of senile de- 23 Table 1. Effects of age as a risk factor in developing senile dementia, Alzheimer's type, in male subjects Age Proportion unaffected 60 1.00 70 97 75 .93 80 .83 85 71 Source: Sluss 1980. mentia (Alzheimer's type) is given in table 1.* This summary of longitudinally gathered data counted the number of individuals who have survived to each age and who have escaped the development of a particular condition. The proportion who escaped, of course, gives, by subtracting from one, the proportion who had developed it. This use of a life-table technique is particularly valuable when one is interested in the long-term risks of developing a condition. This way of organizing information tells us the effect of age as a risk factor. With such a summary, it is then possible to divide the cohort of observed individuals into two parts according to a given suspected "risk factor" and determine whether the risk for one group is higher than for the other. This type of analysis is necessary in this ex- ample for two reasons. First, age is a known powerful risk factor, and we wish to discount its effect when seeking to identify other risk factors. Second, death is a powerful outcome competitor in the population, and we do not want its determinants to confound our search for the deter— minants of SDAT. Contingency Risk A familiar example of a simple contingency risk, as presented in table 2, would be the increased risk which a *This refers to the WHO (1978) ICD category "senile dementia” (attributed to senile brain disease). It is currently believed that what was formerly referred to as senile brain disease is actually Alzheimer's disease. Paradoxically, Alzheimer described his "pre-senile dementia" as occurring only before age 60. In this study no case of senile dementia before age 60 was observed. 24 person with a history of previous mental hospital admission has of being admitted during the next 2 years to a mental hospital. The insurance companies will charge you more for your automobile insurance if you have a history of acci- dents. These are examples of the principle that a person's past experience is the best predictor of that individual's future experience in the absence of any other information. "Nothing fails like failure and nothing succeeds like success." Table 2. A simple contingency risk With C Without C With R a b Without R c d R = risk factor which is being studied. C = condition which the risk factor is suspected of favoring. Obviously, if R is in fact a risk factor for C, a/(a + b) will be larger than c/(c + d). For any single set of research data, the probability that such a finding is due to chance sampling will be measured by the "significance test." But the significance test is not a useful indicator of how strongly R affects the frequency of C; for that, something more is needed. Relative Risk How much risk does cigarette smoking produce for cor- onary deaths and how much for lung cancer deaths? Which is the greater risk? Clearer concepts of risk quantification emerged from efforts to interpret data which were not directly related to this question. Cornfield in 1951 introduced the precise concept of relative risk—the incidence of disease in an exposed group divided by the incidence of disease in a nonexposed group. It measures how much the risk factor affects incidence rates. Relative risk can also be defined in terms of conditional probabilities, which is the way in which Armitage (1973) and Fleiss (1973) discuss it. 25 Cornfield was interpreting case-control data. He imag- ined the existence of a suitably representative sample of people about whom a, b, c, and d in table 2 are all known on a prospective basis. That is, (c + d) and (a + b) had been identified in the past, and the values of a and b were observed as the cases developed. He then introduced the concept of relative risk to show how case-control data from retrospective studies could be used to estimate the value of relative risk, given certain assumptions. In a case-control retrospective study the new cases arising from a population (a + c) are identified and a group of comparable controls (b + d) are identified; then data are obtained retrospectively as to which cases and which controls had previously been exposed to R. He showed that the "odds ratio," which is the ratio of cross products, ad/bc, is a good estimate of the relative risk. His paper ended as follows: The results of even the most carefully analyzed set of such data may be open to question, and these doubts can be resolved only by methods of data collection that provide representative samples of diseased and nondiseased persons (Cornfield 1951, emphasis added). Excellent discussions of the odds ratio approach to rel- ative risk estimates and the value of the logit as developed by Haldane (1956) are to be found in Fleiss (1973) and in Lilienfeld and Lilienfeld (1980). The logit expression of relative risks makes it possible to examine the interaction of risk factors quantitatively so as to estimate the total risk resulting from clusters of risks. (Lilienfelds' second edition appeared after this paper was given and contains detailed discussions of most of the risk-measurement con- cepts in this paper.) The Prospective Design As indicated above, Cornfield (1951) imagined a pros— pective study when he defined relative risk. If the data are gathered prospectively, the relative risk is the ratio of a c a+b to c+d Thus, Cornfield's relative risk is measured directly rather 26 than estimated. If the condition being studied is not com- mon, both a and c are small while b and d are large. This is thought to be inefficient. A great deal is spent finding out the ratio of b to d without adding greatly to the findings. A prospective study is also inefficient if many years elapse between R and C. However, if the condition is fairly common, a prospec- tive design can be seriously considered. For example, it makes sense to study factors associated with rehospital- ization after leaving a psychiatric inpatient service, if the psychiatric inpatient service readmission probability is about 50 percent within a reasonable time period (6 months to a year). If one wishes to find what kinds of risk factors are associated with the readmission, a prospective design of a cohort of releases would not take an unreasonable amount of time, and the data regarding the risk factors can be gathered while the stream of patients is followed in the community. The Nonconcurrent Prospective Design The prospective study selects two populations (with R and without R) and follows them forward in time to find out who develops C. It measures the relative risk directly. It takes a long time and, unless the condition is very common, a large number of subjects. The nonconcurrent prospective study saves time. It lo- cates an old record of exposure to R in a population some time in the past, then locates the subjects at present, and sorts them into cases and noncases. Such studies are becoming more and more common in general epidemiology risk factor research but are not frequent in mental disorder risk factor research. Stein and Susser et al. (1972) tried to use the Nazi- imposed famine in certain districts of Holland during World War II as a means of determining whether a sharp reduction in maternal diet affected the mental development of children born in the succeeding months. There is significant question as to whether the "experiment" fit the needed design and whether the gathered data were appropriate to the hypothesis. But it was an attempt to do a noncon- current prospective study. 27 The Case-Control Retrospective Design The case-control study is extremely suitable as a first stage of investigation when the risk factors suspected are many and can be ascertained through histories. For a case-control study, it is necessary to systemati- cally monitor the new cases of the condition arising in a population over a specified time period. This is usually done through a medical care record system which presum- ably has an adequate casefinding and recording mechanism. The frequency of the suspected risk factor or risk factors is ascertained for each case. A group of comparable controls is simultaneously identified and the same sus— pected risk factors ascertained for them. If the condition being studied is of a low frequency, this method provides a good approximation to relative risk, as Cornfield (1951) has pointed out, in the form of the odds ratio, (a x d/b x c). The weakness of this method must be kept in mind. The identification of a comparable population, which differs from the cases only in that they do not have the condition, can rarely be accomplished with confidence. It is custom- ary to take a group of cases with another condition in the same medical care facility, but there can be a systematic bias for the other condition with respect to the use of that medical care facility by the population from which the cases have been drawn. Nonetheless, the method has great power and economy. Problems arise in ensuring that the ascertaining of the risk factors is done by the same methods in both cases and controls. For example, if the interviewer believes that cigarette smoking causes lung cancer and knows which subjects have cancer, the interviewer might probe with more vigor a history of no cigarette smoking in the cancer cases than in the controls. If the interviewers and subjects can be blind as to the way in which the sample was drawn and the reason for asking about the risk factors, much of this bias can be removed. The most influential series of case-control studies on mental disorders was initiated by Lilienfeld and Parkhurst (1951) and Lilienfeld and Pasamanick (1954; Pasamanick and Lilienfeld 1955) and pursued by them and Pasamanick and Knobloch (1960) regarding the "continuum of fetal damage" in obstetrical complications. This series of in- vestigations linked mental retardation, epilepsy, reading disability, and impulsive behavior disorders to a wide va- 28 riety of obstetrical complications. They have not been followed by appropriate prospective investigations starting with obstetrical complications and following the offspring through the years to enumerate the full range of conse- quent disorders. Many case-control studies have been done in an effort to identify antecedent factors or current factors in such conditions as schizophrenia. Most studies have had inap- propriate controls. Too often the controls have been hos- pital employees or students from a social background al- together different from the patients'. Biochemical studies have not controlled diets. It cannot be emphasized too strongly that the careful selection of comparable controls is the heart of the case-control investigation, if it is not to be misleading. The Epidemiologic Catchment Area (ECA) Project Clinical cases do not make good samples for case-control risk factor research, mainly because every clinical sample has some special selection bias which one can never be sure has been reproduced in the controls. A model data source for case-control studies is provided in the ECA surveys of prevalence of mental disorder, currently supported by the NIMH. The prevalence surveys are taken in representative samples of area populations and repeated on the same subjects 1 year later with a resurvey to detect newly arisen cases. Prevalence data from these morbidity surveys (how many people are sick at one time) can be used in case-control studies, drawing controls from the same surveyed popula- tion at the same time by the same methods. But incidence data (how many got sick during the year) are superior for detecting a risk which is associated with onset of a dis- order. Some conditions may be frequent enough in the ECA surveys to permit retrospective case-control studies based on new cases only. Prevalence rates tend to be equal to the product of in- cidence rates times the average duration of each episode. So case-control studies based on prevailing cases may give clues to risk factors without differentiating those which increase the risk of developing a condition from those which make each episode likely to last longer. For ex- 29 ample, effective medical treatment can work to either lower or raise prevalence rates without affecting incidence rates. It can shorten the duration of high blood pressure episodes or depressive episodes. Or it can lengthen the du- ration of Down's syndrome or senile dementia (Gruenberg 1977). The ECA surveys will give us the best examples anyone has ever had for mental disorder case-control retrospective studies. The Hopkins-based East Baltimore survey will be most valuable because each case and each control will have had a clinical reappraisal which will remove false positives and false negatives from the sample. Attributable Risk Morton Levin (1953) asked a different question about the importance of a risk factor: "What portion of lung cancer would disappear from the population if everybody stopped smoking cigarettes?" He used relative risk data to estimate how much of the lung cancer would disappear if you took away the cigarettes, that is, the proportion of cases one can attribute to cigarette smoking. This is an important quantity. If b is the proportion of the general population exposed tor, then S, the attributable risk is S _ b(r-1) b(r-1) +1 where r is the relative risk as defined above. Case-con- trol designs can provide estimates of S to the extent that they provide good measures of b and of r (Lilienfeld and Lilienfeld 1980). Prospective designs give more direct measurements of r, but they also provide no direct measurement of b, unless the controls are representative of the general population in their rates of exposure to R. Attributable risk can be measured directly if a preventive trial is conducted. Risk Measures From Preventive Trials: Historical Examples The prevention of pellagra among institutional inmates was a dramatic investigation started in 1913 (Goldberg et 30 al. 1925). It apparently prevented new cases of pellagra from arising by improving and supplementing diets. The report leaves something to be desired because no relative risk figures were published. They did not system- atically tabulate the risk of developing pellagra among the controls. Instead, they pointed out that the death rate in the communities surrounding the orphanages and mental hospitals continued to be high for pellagra. In such in- stances, a preventive trial related to a key risk factor can provide such dramatic evidence that subtle quantification of relative risk becomes an example of "misplaced precision." The Newburgh-Kingston Fluoridation Experiment lasted from 1945 to 1955 (Ast and Schlesinger 1956). The exper- iment was organized largely by the New York State Health Department but required a great deal of collaboration with other agencies. The survey examined children's teeth in the two cities. Fluorides were added to the water supply of Newburgh in 1945. Children's teeth were surveyed from both cities for 10 years. The results were conclusive. The frequency of cavity- free mouths in the children who grew up in Newburgh was much higher than the frequency of cavity-free mouths in the children who grew up in Kingston. It was established that adding fluoride to a city's water supply was a safe, effective way of reducing dental caries. Low-fluoride- content drinking water is a risk factor for dental decay. The preventive trial showed that most decay could be at— tributed to low fluoride concentrations in the water. Neither of these preventive trials randomly assigned individual subjects to the preventive and control groups. Whole populations were assigned because the preventive intervention was conceptualized around a community-based resource—food or water. In an era when the double-blind, randomized clinical trial is riding high, we should remind ourselves that other designs still have their use. The Dutchess County Preventive Trial of the Chronic Social Breakdown Syndrome began in 1959 (Brandon and Gruenberg 1966). Based on Hudson River State Hospital in Poughkeepsie, N.Y., the experiment reformed the care of seriously mentally ill patients who had been entering a large mental hospital admission service with later transfer to a specialized chronic service or discharged. The reform created a Dutchess County service with its own buildings and staff. All Dutchess County cases, new and old, were cared for in this service. It was a very ac- 31 tive, acute care program which released patients from hospital as early as possible following admission and con- tinued to provide care for them in the community. It was assumed that first admission rates and readmission rates would rise. Would there be a decline in the rate at which new cases of chronic deterioration in personal and social functioning were developing? A decline was anticipated because of the following pro- gram features: the systematic effort to preserve the patient's assets during the initial hospitalization, the as- siduous efforts made to preserve his social support systems and community support systems, and the speedy efforts to re-relate the individual as rapidly as possible to the ex- isting social support system and community support sys- tems. The continuing availability of the hospital-based clinical team (with easy brief readmission) was considered a means of preventing community and family rejection in cases of chronic or recurring psychosis. Hence, a rise in readmission rates was seen as a means of achieving the end: reduction of the risk of Chronic Social Breakdown Syndrome (CSBS). The part of the hospital used for this experiment had no locked services and no visiting restrictions. Compulsory hospitalization was avoided by seeking the patient's active cooperation from the very first contact and by providing precare consultation by hospital psychiatrists to anyone contemplating sending a patient to the State hospital. It was assumed that the frequency of acute episodes of psy- chotic decompensation would not decrease in the general population. Because there was fear that the new pattern of care might lead families to keep patients at home or to send them elsewhere when they became troublesome, all residents of the county between the ages of 16 and 65 who had been in any kind of inpatient psychiatric service after 1955 were surveyed annually to ascertain the frequency of CSBS. Because the program had to be introduced for the entire county or not at all, there was no way of getting a com- parable control on a random basis. A matching community, as in the Kingston/Newburgh study, was not practicable. The only control arose from the before-and-after obser- vations. Between 1960 and 1963, the annual number of new cases of severe CSBS which arose in this population of 100,000 declined from something over 45 to something under 16 per year, a drop to less than one-third of the previous rate. The rates of recovery from this chronic 32 syndrome were the same, regardless of the year of onset, so that the change could be attributed to the rate of re- covery in the first year (actually, probably in the first half year). Thus, the risk of developing a CSBS dropped from 45 to 15 per 100,000 per year; so about 30 new cases of CSBS per 100,000 per year could be attributed to the old system of care. The attributable risk was then 67 percent. The findings of this study (when attended to at all) tend to be interpreted to mean that it is justifiable to abandon chronic care patients by discharge from mental hospitals into inadequate community-care programs under the slogan of "deinstitutionalization" (Gruenberg and Archer 1979). It is rarely recognized that these services were, in fact, provided by a reorganization of existing services, tech- niques, and personnel from a State mental hospital with the continuing availability of the hospital and its staff to patients, family, and community. New Opportunities for Preventive Trials Several theories regarding risk factors with respect to mental disorders are now ripe for preventive trials. The relation of maternal malnutrition to central nervous system development before birth remains cloudy (Steichen et al. 1980). While there are various programs in existence to raise the nutritional status of marginally nourished young women in this country and elsewhere, none of them is applied with sufficient intensity to get all of the young women at risk of having a child damaged because of in- adequate nutrition of the mother. It certainly would not do such young women any harm if they were systematically educated to feed themselves better and were provided with the nutritional supplements necessary for optimal nutrition. Such experiments have been tried, but they share one deficiency. From the available evidence, it would be im- portant to start improving the nutrition of the treated group of young women long before they became pregnant in order to assure that they had excellent reserves of the es— sential nutrients at the time they became pregnant. This would require artful community organization and education programs and not just the erection of a scientific design. Widow-to-widow counseling has been fashionable as a method of improving the handling of grief and, presumably, preventing psychosomatic and depressive complications 33 (Young et al. 1963). But, so far, no program has been set up on a preventive trial design with appropriate controls, and no data are available as to how much these programs re- duce the risk of specific disorders. Meanwhile, there is pressure to expand the programs. If these programs are truly effective, they should be expanded much more quickly. If they are not effective, they should not be sup- ported in the name of prevention. Elsie Broussard (1976) identified a syndrome of poor emotional development in daughters of women who, when they were pregnant, had low self-esteem. There is an easily administered screening test to locate such women in prenatal clinics and obstetrical services, and there is a reeducation program designed for them. An objectified screening device to locate the children whose emotional development has been damaged needs to be developed. A preventive trial is indicated, even at this early stage of knowledge. It is not likely that any women so identified would be significantly damaged by these efforts to improve their self-esteem, their skills in living and mothering, and their ability to love their daughters. There are other studies suggesting useful innovations for improving coping skills in children, for improving mother- ing, and for improving teaching skills in mothers with bor- derline intelligence. These are best approached through a preventive trial design. The preventive trial design is emphasized in this review because it gives the most direct information on both rel- ative and attributable risks of undesirable outcomes. Alertness to opportunities to conduct preventive trials will speed up the acquisition of knowledge regarding prevention. Preventive trials sometimes require dangerous or anx— iety-producing interventions. These preventive trials nec— essarily represent a late stage in a sequence of epidemi- ologic studies. But preventive trials which remove a risk or counteract a risk with a harmless or desirable intervention do not need to await the accumulation of supporting data. Such preventive trials can sometimes be the first stage of investigation. References Armitage, P. Statistical Methods in Medical Research. New York: John Wiley, 1973. 34 Ast, D., and Schlesinger, E.R. The conclusion of a ten-year study of water fluoridation. American Journal of Public Health 46:265-271, 1956. Brandon, S., and Gruenberg, E.M. Measurement of the in- cidence of chronic severe social breakdown syndrome: Has the Dutchess County service been associated with a decline in incidence. Milbank Memorial Fund Quarterly 44:129- 142, 1966. Broussard, E. Neonatal prediction and outcome at 10/11 years. Child Psychiatry and Human Development 7:85- 93, 1976. Cornfield, J. A method of estimating comparative rates from clinical data: Application to cancer of the lung, breast, and cervix. Journal of the National Cancer In- stitute 11:1269-1275, 1951. Summarized in Lilienfeld and Lilienfeld 1980. Faris, R.E.L., and Dunham, H.W. Mental Disorders in Ur- ban Areas. Chicago: Phoenix Books, 1939. Fleiss, J. Statistical Methods for Rates and Proportions. New York: John Wiley, 1973. Goldberg, J.; Waring, C.H.; and Tanner, W.F. Pellagra prevention by diet among institutional inmates. Public Health Reports 38:2361-2368, 1925. Gruenberg, E.M. Failures of success. Milbank Memorial Fund Quarterly 55:3-24, 1977. Gruenberg, E.M., and Archer, J. Abandonment of respon- sibility for the seriously mentally ill. Milbank Memorial Fund Quarterly 57:485-506, 1979. Haldane, J.B.S. The estimation and significance of the logarithm of a ratio of frequencies. Annals of Human Genetics 20:309-311, 1956. Levin, M. The occurrence of lung cancer in man. Acta Unio International Contra Cancrum 9:531-541, 1953. Lilienfeld, A.M., and Lilienfeld, D. Foundations of Epi- demiology. 2d ed. New York: Oxford Press, 1980. Lilienfeld, A.M., and Parkhurst, E. A study of the associ- ation of factors of pregnancy and parturition with the development of cerebral palsy: Preliminary report. American Journal of Hygiene 53:262-282, 1951. Lilienfeld, A.M., and Pasamanick, B. Association of ma- ternal and fetal factors with the development of epilepsy: Abnormalities in the prenatal and paranatal periods. Journal of the American Medical Association 155:719- 724, 1954. Morris, J.N. Uses of Epidemiology. Edinburgh: Churchill Livingstone, 1975S. 3s Pasamanick, B., and Knobloch, H. Brain damage and re- productive casualty. American Journal of Orthopsy- chiatry 30:298-305, 1960. Pasamanick, B., and Lilienfeld, A.M. Maternal and fetal factors in the development of epilepsy. II. Relationship to some clinical features of epilepsy. Neurology 5:77-83, 1955. Sluss, T.K. "A Method for Investigating a Risk Factor for Senile Dementia—Alzheimer's Type in the Baltimore Longitudinal Study." Unpublished doctoral thesis, the Johns Hopkins University School of Hygiene and Public Health, 1980. Steichen, J.J.; Tsang, R.C.; Gratton, T.L.; Hamstra, A.; and DeLuca, H.F. Vitamin D Homeostasis in the perinatal period: 1,25-Dihydroxyvitamin D in maternal, cord, and neonatal blood. New England Journal of Medicine 302: 315-319, 1980. Stein, Z.; Susser, M.; Saenger, G.; and Marolla, F. Nutrition and mental performance. Science 178:708-713, 1972. World Health Organization. Mental Disorders: Glossary and Guide to Their Classification in Accordance with the Ninth Revision of the International Classification of Dis— eases. Geneva: the Organization, 1978. Young, M.; Benjamin, B.; and Willis, G. The mortality of widowers. Lancet 2:454-456, 1963. 36 Chapter 3 LIFE STRESS AND PSYCHOPATHOLOGY* Barbara Snell Dohrenwend, Ph.D. and Bruce P. Dohrenwend, Ph.D. Introduction This paper gives little time to what might seem to be a fundamental question: Is life stress related to psycho- pathology? When one considers the full range of evidence from war and other extreme situations (Dohrenwend and Dohrenwend 1969, pp. 116-124) to stressful life events oc— curring in everyday community settings (Dohrenwend and Dohrenwend 1974), the answer to this question seems to be affirmative though not very informative. It is, at best, the starting point from which to turn to the more interesting question: How is life stress related to psychopathology? It is this second question on which we focus. In asking it we make two assumptions. First, we assume that there is some kind of a relationship between life stress and psy- chopathology, an assumption that, as is apparent when we discuss alternative hypotheses about the nature of the re- lationship, is not necessarily a radical one. Our second as- sumption is that life stress is a complex process that can *Reprinted from National Institute of Mental Health. Risk Factor Research in the Major Mental Disorders. DHHS Publi- cation No. (ADM)81-1068. Washington, D.C.: Superintendent of Documents, U.S. Government Printing Office, 1981. 37 be conceptualized in terms of three major components: life events, personal dispositions, and ongoing social situations. What Is a Stressful Life Event? The stressful life events with which we are concerned at this point are those that are proximate to, rather than re— mote from, the onset of a disorder. For example, this cat- egory includes the recent death of a friend or relative but not the fact that an adult's father died when he or she was a child. The latter event is not irrelevant to life stress but is subsumed under personal dispositions, since we assume that the early death of a parent can affect an adult's be- havior only insofar as its impact was internalized. Although some studies have been directed at the ques- tion of what qualities of life events are related to psycho- logical problems, most have examined psychosocial symp- tom scores, whose diagnostic implications are unclear, rather than cases of diagnosed psychopathology. Further- more, many studies of life events and diagnosed psycho- pathology are case studies without controls. We have found only five studies that provide information about the kinds of events that are specific to cases of diagnosed psycho- pathology in contrast to nonpatient controls. Our review concerning the qualities of life events that are related to psychopathology focuses on these case-control studies of relations between diagnosed schizophrenia, clinical de- pression, or neurosis and life events. The findings, not surprisingly, depend to some extent on the kind of psycho- pathology being studied. Two case-control studies have shown that the number of life events reported by schizophrenics before onset of an episode was significantly greater than the number reported by controls for a comparable period. One study, by Jacobs and Myers (1976) in New Haven, focused on a survey of 62 first admissions; the other, by Brown and Birley (1968) in London, was concerned with SO patients admitted for acute episodes of schizophrenia. Although the consistent finding of these two studies was that there was a significantly higher rate of life events in their reporting periods, 1 year in the former and 3 months in the latter study, for patients compared to community controls, this excess was due mainly to events that could be dependent on the patients’ mental condition. It is of considerable interest, therefore, 38 that Brown and Birley, who dated the occurrence of events within 1-week periods, found that events that were inde- pendent of a person's psychiatric condition occurred more frequently in the 3-week period preceding episodes of schizophrenia than in a comparable period in the lives of their controls. Forty-six percent of the patients but only 12 percent of the controls experienced at least one inde- pendent event in this 3-week period. This difference was not found in earlier 3-week periods. Moreover, this finding is not an isolated one, since Leff and his colleagues (1973) found a relatively high frequency of events in the period just prior to relapse in a sample of schizophrenics being treated with phenothiazines in the community. In addition, Brown and Birley showed that their result held regardless of first admission versus readmission status of the patient, and also for patients experiencing first episodes versus those experiencing relapses. In these studies the events included were both desirable and undesirable, suggesting that any kind of change may increase a person's risk of experiencing a schizophrenic episode. It should be noted, however, that Brown and Birley conceived their events to be stressful not because they entailed change but because they "would for most people be likely to be followed by a strong negative or positive emotion" (Brown and Birley 1968, p. 66). In his case-control studies of life events and clinical de- pression, Paykel (1974) found that undesirable events, but not desirable events, were reported in excess in the 6 months prior to the onset of depression. He also identified a class of events that he labeled "exits," in which a person leaves the social field of the subject. The events which Paykel included in this category were death of close family member, marital separation, divorce, family member leaving home, child getting married, and son being drafted. Paykel found that the presence of one or more of these events was strongly associated with onsets of depression. While only S percent of controls reported one or more exits in the 6-month reporting period, 25 percent of the de- pressed patients reported at least one such event. Brown and Harris (1978) also reported a difference in recent life events between depressed women, whether treated or untreated, and nondepressed community con- trols. They found that depressed women had experienced an excess of a class of events that Brown and Harris labeled "severe" in the period, on the average, of 38 weeks before the onset of depression. Specifically, 61 percent of treated 39 cases and 68 percent of untreated cases, but only 20 per- cent of community controls, had experienced at least one such event. Unfortunately, Brown's and Harris' concept of severity embodies two components of the life-stress process, the occurrence of an event together with those aspects of the ongoing social context that are judged by raters to endow the event with a moderate or marked degree of threat that is likely to last more than 1 week. While these judgments were tested and found to be reliable across raters (Tennant et al. 1979), the criteria on which they are based are often specific to particular events and do not lend themselves to generalized description. These results cannot, therefore, be compared directly with Paykel's. In an exploratory study of the difference in recent life events experienced by 34 diagnosed acute neurotic cases found in general practices and in matched general practice patients free of psychiatric disturbance, Cooper and Sylph (1973) adopted Brown's and Harris' procedure for identi- fying severe events by means of ratings of long-term con- textual threat. They found that 47 percent of their neurotic patients but only 6 percent of their controls had experienced at least one severe event during the 3-month comparison period. However, adopting another -classifi- cation employed by Brown in his study of schizophrenics, but not in his study of depressed women, they did not find a significant difference in the number of events experienced by cases and controls that were independent of the illness. This statistical result may, however, be a function of the small number of subjects in the study, since 53 percent of the cases and only 26 percent of controls reported at least one independent event. What have we learned about the nature of stressful life events from studies of diagnosed cases of psychopathology? Studies to date suggest that any life event that involves change may be stressful for persons otherwise at risk of schizophrenia, but only certain types of life events may be stressful for persons otherwise at risk of clinical depression or neurosis. These findings imply that stressfulness is not a simple, unidimensional property of life events and that more research is needed to map associations between event properties and various kinds of psychopathology. In carrying out such studies, we must not lose sight of a crucial distinction that has been attended to only errat- ically in research to date—the distinction among life events in terms of their functional relation to psycho- 40 pathology. For example, in his research on schizophrenia, Brown analyzed separately events classified as independent and events classified as possibly independent of the sub- ject's psychopathology (Brown and Birley 1968), as did Jacobs and Myers (1976). Cooper and Sylph (1973) made the same distinction in their study of neurotics. However, neither Brown and Harris (1978) nor Paykel (1974) made this distinction in their studies of clinical depression. Given the problems of insidious onset of many types of psycho- pathology, clear inferences concerning causal relations between life events and episodes of psychopathology can- not be made unless events that are inherently independent of pathology and events that may be part of a pathological process are separated in analyses of their relations to pathology (cf. Dohrenwend 1974). What Personal Dispositions Are Related to the Life-Stress Process? Hypotheses concerning the nature of personal disposi- tions involved in relationships between life stress and psy- chopathology are of four general types. These types are not necessarily mutually exclusive. One focuses on genetic risk factors as a predisposing condition for particular types of psychopathology. We do not elaborate on this hypothesis since it will be discussed by others more competent to do so. Another hypothesis is that prior psychopathology pre- disposes the individual to subsequent episodes of psycho- pathology. While the reality of this kind of risk factor is not in doubt, the mechanisms that lead to new episodes are not well understood. We propose studies directed at testing alternative models of the life-stress process that are re— lated to this issue. Let us turn to the third type of hypothesis. We said at the outset that we would be concerned first with stressful life events that are proximate to, rather than remote from, the onset of a disorder. We noted, however, that we are not assuming that remote events are irrelevant, but only that it is more convenient to consider such events as they might influence personal dispositions later in life. Is there evi- dence, then, that stressful events early in life do, in fact, 41 render the individual more or less vulnerable to stressful events later in life? There are quite consistent findings that homes broken by divorce or separation in childhood are associated with late antisocial behavior (cf. Rutter 1974). The results of re- search on the later effects of death of one or both parents during childhood appear to be much less consistent (Gran- ville-Grossman 1968). It seems possible, however, that the apparent inconsistency has been due to insufficient atten- tion to whether bereavement took place early or late in childhood. There are now at least three studies with results suggesting that death of one or both parents before age 11 is associated with vulnerability to depression (Brown and Harris 1978) and/or severe psychological distress (Dohren— wend and de Figuerido in press; Langner and Michael 1963, pp. 167-169) among adults in samples from the general population who experience stressful life events. There is considerably less information on whether early experience with some types of events tends to decrease vulnerability to later stressful events. It is intriguing to consider whether there may be such inoculation effects (cf. Born- stein et al. 1973). The fourth type of hypothesis considers ways in which normal variations in personality may affect an individual's response to stressful life events. Much has been learned in the past two decades from clinical observations, observa- tions in natural settings, and experimental studies about effective psychological defense and coping responses to stressful events (e.g., Hamburg and Adams 1967; Horowitz 1976; Lazarus 1966). At the same time, hypotheses have been developed concerning the nature of individual dif- ferences in disposition that account for pathological out- comes related to stressful life events. As one example, an hypothesis concerning individual differences in vulnerability to stress-related psycho- pathology grows out of Rotter's social learning theory and his concept that individuals have a generalized expectancy about the extent to which they control the rewards, pun- ishments, and, in general, the events that occur in their lives. Rotter (1966) conceived this expectancy, which he labeled "locus of control," as varying on a dimension from internal to external. Persons at the extreme internal end of the locus-of-control dimension expect to be in control of their life events to a high degree. In contrast, persons at the extreme external end of the locus-of—control dimension expect that their life events will generally be controlled by 42 others or by fate. Although there is some controversy concerning the relation of locus-of-control expectancy to vulnerability to pathology, the hypothesis with the most extensive empirical support is that persons with an ex- ternal locus—of-control expectancy have greater proneness to psychopathology than those with an internal locus-of- control expectancy (e.g., Lefcourt 1976, ch. 7). In fact, in some studies, internal locus-of-control expectancy has been equated with competence, coping ability, and relative invulnerability to debilitating effects of stressful events (Campbell et al. 1976, pp. 59-60). In general, we are suggesting that, where theory or em- pirical findings suggest meaningful hypotheses, normal var— iations in personality should be studied as part of the life-stress process. We have provided one example of this kind of personality variable. What Social Conditions Are Related to the Life-Stress Process? Complementing efforts to find factors within the in- dividual that affect his risk of psychopathology related to stressful life events are more recent attempts to identify social factors in the individual's environment that affect this risk. One of the most influential hypotheses in the re— cent literature on life stress states that social support mitigates the effects of stressful events (Caplan and Kil- lilea 1976). Secondary analyses of data not specifically designed to test this hypothesis have suggested that highly stressful events combined with low social support are sig- nificantly more pathogenic than highly stressful events combined with high social support or less stressful events with high or low social support (Gore 1980). What is the structure of the support systems available to an individual? Although early efforts to conceptualize so- cial supports tended to focus on formal helping agencies, such as mental health centers (Caplan and Killilea 1976), work has also been done on analysis of the structure of varied forms of informal support systems, particularly family, neighborhood, and friendship networks (Fischer et al. 1977). What is needed now in measures to be used in stress research is objective information about social net-— works and their structures, unconfounded with the quality of an individual's participation in these networks. Without 43 this separation the relative contribution to the life-stress process of an individual's personality as against his or her social situation cannot be assessed. Another question is: What functions can be performed by the various structures in an individual's support systems? Although to date most studies of social support have, by implication, emphasized emotional support, some explor— atory studies suggest that it is equally important to in- vestigate the availability of instrumental support during life events (Carveth and Gottlieb 1979). Financial support is probably an important type of instrumental support in many life events. For example, if there is illness in the family, is there adequate insurance available, or, if not, is there someone in the person's informal network who will help out or a community agency to serve this purpose? Other types of instrumental support may be specific to particular life events. For example, in the case of serious illness, can the person obtain adequate medical care? Does the new mother have someone to turn to for advice about the care of her baby? Exploratory studies suggest that different functions may be served by different support structures (Litwak and Szelenyi 1969). For example, friends and acquaintances may be able to do more than family about helping a person find a new job, but family may be a better source of help with problems in personal relation- ships. Thus, at least in our present state of ignorance about how support systems function in the life-stress process, it is important to obtain as complete a description as possible of potentially supportive structures. Hypotheses About the Life-Stress Process The majority of studies of life stress and illness have been concerned primarily or exclusively with one or another of the three types of constructs that we have discussed: stressful life events, personal dispositions, or social conditions. Up to this point we have dealt with these constructs singly. The task now is to integrate them into hypotheses about the life-stress process, a task that has received increasing attention recently. In theoretical and empirical studies investigators have developed at least five different hypotheses, as schematized in figure 1. The first hypothesis, in figure 1A, indicates that cumu- lations of stressful life events cause adverse health 44 Figure 1. Five hypotheses about the life-stress process. Stressful \ life + Psychopathology events - Social situations Stressful ee life ty A Psychopathology events Personal dispositions Social situations | Stressful life Psychopathology events Personal dispositions Social situations ; Psychopathology Personal : dispositions Prior Stressful Exacerbation psycho- life of pathology events psychopathology 45 changes. This model was developed empirically in studies of extreme situations such as combat and concentration camps. It has been generalized to normal civil life in terms of a pathogenic triad of concomitant events and conditions that involve physical exhaustion, resulting often from severe physical illness or injury; loss of social support, as a result, for example, of geographical relocation; and the fateful negative events other than physical illness or injury over whose occurrence the individual has no control, such as the death of a loved one (Dohrenwend 1979). It is also the model that underlies the early work of Holmes and Rahe and their coworkers. This could be called the vic- timization hypothesis. The second hypothesis, in figure 1B, supposes that pre- existing personal dispositions and social conditions moder— ate the causal relation between stressful life events and psychopathology. This hypothesis underlies the literature on vulnerability involving conceptions such as coping ability and social support. It was also proposed by Zubin and Spring (1977) in their discussion of the etiology of schizophrenia and by Brown and Harris (1978) in their so- cial model of the etiology of depression. This could be called the vulnerability hypothesis. The third hypothesis, in figure 1C, contrasts with the second in that, rather than moderating the impact of stressful life events, personal dispositions and social con- ditions make an independent causal contribution to the occurrence of psychopathology. This hypothesis received empirical support in a study of psychosocial symptoms by Andrews and his colleagues (1978) and was also proposed by Tennant and Bebbington (1978) as a better fit than the second hypothesis to the data presented by Brown and Harris (1978) in their study of social factors in the etiology of depression. This could be characterized as the additive burden hypothesis. The fourth hypothesis, in figure 1D, is another and fur- ther modification of the second. It denies any role to life events, proposing instead that stable personal dispositions and social conditions alone cause the adverse health changes. This hypothesis was presented by Gersten and her colleagues (1974) as an explanation of changes in children's psychological symptom patterns. The method they use in developing the empirical basis for this hypothesis is, how- ever, open to criticism (Link 1978). This hypothesis, in contrast to the previous one, could be called the chronic burden hypothesis. 46 The final hypothesis, in figure 1E, raises the crucial issue of the direction of the causal relation between life events and psychopathology. It proposes that the presence of a disorder leads to stressful events which, in turn, exacer- bate the disorder. In empirical tests this hypothesis was supported by a study of chronic patients (Fontana et al. 1972) but not by a study of neurotics (Tennant and Andrews 1978). This might be called the proneness hypothesis, in the sense that proneness to stressful events is the central mechanism posited. These hypotheses are neither exhaustive nor mutually exclusive. They provide, nevertheless, a framework for designing research to advance our understanding of rela- tions between life stress and psychopathology. Research Design for Study of Life Stress and Psychopathology We suggest that tests of alternative hypotheses about the relation of life stress to psychopathology be done in a longitudinal case-control design with groups that vary with respect to genetic and/or early experiential risk factors. Cases should include two subgroups: those with and those without genetic and/or early experiential risk factors. Controls with no history of disorder should include the same subgroups. In addition, the inclusion of cases of different types of psychopathology within a single case-control study is es— sential for differentiating factors related to distress and other adverse outcomes in general from those that are specific to particular types of psychopathology (cf. Barrett 1979; Jacobs et al. 1974). Baseline measures should include not only standardized research diagnostic procedure with all controls as well as cases, but also measures indicating nonpathological per- sonal dispositions, particularly personality traits and cer- tain early life experiences, in cases as well as controls. Current social situations, including social and material supports and demands, should also be carefully measured for all subjects. Finally, a baseline measure of recent life events should be taken. The baseline measures of life events must be dated in relation to the onset of the episode. Although it is far easier to date onset of treatment than onset of an episode 47 (cf. Ambelas 1979), use of the former as the reference point for baseline life-event measures leaves the nature of the relationship between life events and the onset of dis- order ambiguous. In the prospective part of the study, both short-term and long-term followup measures are needed. The occurrence of life events in relation to the occurrence of episodes or exacerbations of the disorders under study should be moni- tored, with careful dating, on a relatively short-term basis, say every 2, 3, or 4 months. A longer term followup of a year or more should, minimally, repeat the diagnostic procedures and might well repeat other baseline measures to test their stability. In this research, it will be of utmost importance to pay careful attention to measurement problems that have made results from much previous research on life stress and psychopathology difficult to interpret. It is all too easy to confound measures of life events, personal dispositions, enduring social conditions, and psychopathology with one another. This kind of confounding guarantees results that, at best, are uninterpretable and, at worst, can lead to er- roneous conclusions. Enough progress has been made in the " clarification of these measurement problems so that there is no need to repeat past mistakes, for example, failure to distinguish between life events that are independent of a subject's mental condition and those that are not, or be- tween the structure of objective social networks and their effectiveness, or between these networks and personal dispositions related to having friends or not having them. In conclusion, there is now enough correlational evidence so that the interesting question is not whether life stress is associated with various types of psychopathology but what hypotheses best account for these relationships. Some of the hypotheses that we presented earlier specifically dis- count the role of recent stressful life events as etiological agents; some have a place for personality variables and genetic vulnerability as etiological factors. Taken to- gether, they hardly represent rampant environmental im- perialism. Moreover, research design to test hypotheses of this kind provides a safeguard against continuing to do studies that merely document a position. It will provide, instead, tests of alternative hypotheses about the inter- play of life stress, other risk factors, and various types of psychopathology. The testing of such hypotheses, we submit, is the next task of research on life stress and psychopathology. 48 Acknowledgments This work was supported in part by Research Grant MH30710, by Research Scientist Award KO0S MH14663, from the National Institute of Mental Health, and by the Foundations' Fund for Research in Psychiatry. References Ambelas, A. Psychologically stressful events in the pre- cipitation of manic episodes. British Journal of Psychi- atry 135:15-21, 1979. Andrews, G.; Tennant, C.; Hewson, D.M.; and Vaillant, G.E. Life event stress, social support, coping style, and risk of psychological impairment. Journal of Nervous and Mental Disease 166(5):307-317, 1978. Barrett, J.E. The relationship of life events to the onset of neurotic disorders. In: Barrett, J.E.; Rose, R.M.; and Klerman, G.L., eds. Stress and Mental Disorder. New York: Raven Press, 1979. pp. 87-109. Bornstein, P.E.; Clayton, P.J.; Halikas, J.A.; Maurice, W.L.; and Robins, E. The depression of widowhood after thirteen months. British Journal of Psychiatry 122:561- 566, 1973. Brown, G.W., and Birley, J.L.T. Crises and life changes and the onset of schizophrenia. Journal of Health and Social Behavior 9(3):203-214, 1968. Brown, G.W., and Harris, T. Social Origins of Depression: A Study of Psychiatric Disorder in Women. New York: Free Press, 1978. Campbell, A.; Converse, P.E.; and Rodgers, W.L. The Quality of American Life. New York: Russell Sage Foun- dation, 1976. Caplan, G., and Killilea, M., eds. Support Systems and Mutual Help: Multi-disciplinary Explorations. New York: Grune & Stratton, 1976. Carveth, W.B., and Gottlieb, B.H. The measurement of social support and its relation to stress. Canadian Journal of Behavioral Sciences 11(3):179-188, 1979. Cooper, B., and Sylph, J. Life events and the onset of neu- rotic illness: An investigation in general practice. Psy- chological Medicine 3:421-43S, 1973. Dohrenwend, B.P. Stressful life events and psychopathol- ogy: Some issues of theory and method. In: Barrett, J.E.; 49 472-367 O - 85 - 3 Rose, R.M.; and Klerman, G.L., eds. Stress and Mental Disorder. New York: Raven Press, 1979. pp. 1-15. Dohrenwend, B.P. Problems in defining and sampling the relevant population of stressful life events. In: Dohren- wend, B.S., and Dohrenwend, B.P., eds. Stressful Life Events: Their Nature and Effects. New York: John Wiley, 1974. Dohrenwend, B.P., and deFiguerido, J.M. Remote and re- cent life events and psychopathology. In: Ricks, D., and Dohrenwend, B.S., eds. Life History Research in Psycho- pathology. New York: Cambridge University Press, in press. Dohrenwend, B.P., and Dohrenwend, B.S. Social Status and Psychological Disorder. New York: John Wiley, 1969. Dohrenwend, B.S., and Dohrenwend, B.P., eds. Stressful Life Events: Their Nature and Effects. New York: John Wiley, 1974. Fischer, C.S.; Jackson, R.M.; Stueve, C.A.; Gerson, K.; Jones, L.Mc.; and Baldassare, M. Networks and Places: Social Relations in the Urban Setting. New York: Free Press, 1977. Fontana, A.F.; Marcus, J.L.; Noel, B.; and Rakusin, J.M. Prehospitalization coping styles of psychiatric patients: The goal-directedness of life events. Journal of Nervous and Mental Disease 155:311-321, 1972. Gersten, J.C.; Langner, T.S.; Eisenberg, J.G.; and Orzek, L. Child behavior and life events: Undesirable change or change per se? In: Dohrenwend, B.S., and Dohrenwend, B.P., eds. Stressful Life Events: Their Nature and Ef- fects. New York: John Wiley, 1974. pp. 159-170. Gore, S. Stress-buffering functions of social supports: An appraisal and clarification of research models. In: Dohrenwend, B.S., and Dohrenwend, B.P., eds. Life Stress and Illness. New York: Neale Watson Academic Publications, 1980. Granville-Grossman, K.L. The early environment in af- fective disorder. British Journal of Psychiatry Special Publication No. 2:65-70, 1968. Hamburg, D., and Adams, J. A perspective on coping behavior: Seeking and utilizing information in major transitions. Archives of General Psychiatry 17:277-284, 1967. Horowitz, M.J. Stress Response Syndromes. New York: Jason Aronson, 1976. Jacobs, S., and Myers, J. Recent life events and acute schizophrenic psychosis: A controlled study. Journal of S0 Nervous and Mental Disease 162:75-87, 1976. Jacobs, S.; Prusoff, B.; and Paykel, E.S. Recent life events in schizophrenia and depression. Psychological Medicine 4:444-453, 1974. Langner, T.S., and Michael, S.T. Life Stress and Mental Health. New York: Free Press of Glencoe, 1963. Lazarus, R.S. Psychological Stress and the Coping Proc- ess. New York: McGraw-Hill, 1966. Lefcourt, H.M. Locus of Control: Current Trends on Theory and Research. New York: Lawrence Erlbaum Associates, 1976. Leff, J.P.; Hirsch, S.R.; Gaind, R.; Rohde, P.D.; and Stevens, B.C. Life events and maintenance therapy in schizophrenic relapse. British Journal of Psychiatry 123:659-660, 1973. Link, B. On the etiological role of stressful life-change events (Comment on Gersten et al., September 1977). Journal of Health and Social Behavior 19:343-345, 1978. Litwak, E., and Szelenyi, I. Primary group structures and their functions: Kin, neighbors, and friends. American Sociological Review 34:465-481, 1969. Paykel, E.S. Life stress and psychiatric disorder: Applica- tions of the clinical approach. In: Dohrenwend, B.S., and Dohrenwend, B.P., eds. Stressful Life Events: Their Na- ture and Effects. New York: John Wiley, 1974. pp. 135-149. Rotter, J.B. Generalized expectancies of internal versus external control of reinforcement. Psychological Mono- graphs 80(609):whole issue, 1966. Rutter, M. The Qualities of Mothering: Maternal Depri- vation Reassessed. New York: Jason Aronson, 1974. Tennant, C., and Andrews, G. The cause of life events in neurosis. Journal of Psychosomatic Research 22(1):41-45, 1978. Tennant, C., and Bebbington, P. The social causation of depression: A critique of the work of Brown and his col- leagues. Psychological Medicine 8:565-575, 1978. Tennant, C.; Smith, A.; Bebbington, P.; and Hurry, J. The contextual threat of life events: The concept and its re- liability. Psychological Medicine 9:525-528, 1979. Zubin, J., and Spring, B. Vulnerability—A new view of schizophrenia. Journal of Abnormal Psychology 86:103- 126, 1977. 51 52 Part II STRESS AND PSYCHIATRIC DISORDERS What are some of the psychopathological responses to stress? Are stressful life events risk factors in the patho- genesis of identifiable psychiatric disorders? Klerman and Weissman define stress and affect and then explore af- fective adaptation to stress, as well as the role of stressful life events as risk factors for clinical affective disorders. Marks analyzes stress and its interaction with other risk factors in the pathogenesis of the anxiety disorders, ex- amines the heterogeneity of these disorders, and proposes several avenues for further investigation of the stress- related syndromes. Horowitz provides a detailed exposition on the stress response syndromes after reviewing several methodological challenges to this field of study. Each pa- per recommends areas for future research in epidemiology, descriptive psychopathology, and clinical trials for pre- venting stress-related psychiatric disorders. 53 Chapter 4 AFFECTIVE RESPONSES TO STRESSFUL LIFE EVENTS Gerald L. Klerman, M.D. and Myrna M. Weissman, Ph.D. Introduction Stress is currently a popular topic, both in scientific circles and among the lay public, where it is common to presume the cause of many bodily symptoms as stress. Stress clinics and seminars have been developed, and prac- titioners in the mental health field, particularly in in- dustrial settings, have built a lively growing industry around stress management. In scientific circles, active research is underway on the assessment of stressful life events, particularly since the development by Holmes and Rahe of a quantitative index for assessing environmental changes. Parallel with these psychometric and epidemiologic developments, studies in neurophysiology and neuroendocrinology are attempting to elucidate the central nervous system (CNS) mechanisms by which environmental changes mobilize the organism to respond. In this paper, we will discuss two aspects of affective responses to stress—acute affective responses to stressful life events and stressful life events as risk factors for clinical affective disorders. Before discussing these two substantive areas, however, the general concepts of stress 55 and affect must be addressed, since these concepts are often subjects of ambiguity and controversy. Stress, Affect, and Adaptation Modern research on stress began with Walter Cannon's (1936) discovery of epinephrine and his concept of its role in the "fight or flight" responses. Cannon's work was fol- lowed by the ideas of Hans Selye, who popularized the term stress in professional, scientific, and public arenas. Selye coined the term general adaptation syndrome for the bodily responses to environmental change, particularly those changes mediated through the adrenal cortex. The avail- ability of increased knowledge about neuroendocrine mech- anisms as well as more refined quantitative assays for measuring epinephrine, norepinephrine, other catechol- amines, adrenal steroids, and the pituitary hormones has greatly accelerated the biological investigations. At the same time, considerable ambiguity remains, which often clouds communication and makes interpre— tation of research difficult. For purposes of this paper, we distinguish among the following: ® Stressors—stressful environmental changes, partic— ularly circumstances e Stress—hypothesized psychological and physiological changes within the organism in consequence of the stressor ® Stress responses—the behavioral, psychological, and physiologic responses of the organism in the attempt to cope and adapt ® Adverse health -consequences—increased suscepti- bility to physical illness or emotional disorders Stressors In parallel with the use of the terms stress and strain in physics, stressor refers to the external change that causes a demand on the organism. Stressors can be natural events S6 (hurricanes, floods), severe adverse life circumstances (combat, wartime), changes in interpersonal relations (di- vorce, separation), or changes in economic or social con- ditions (unemployment, retirement). Physical illness of the individual also can be regarded as a stressor, external to the self. Illness may often become a major stressor. How- ever, because the illness may alter the organism's capacity to respond, particularly if it involves the central nervous system and/or the endocrine system, illness complicates understanding the responses to stressors. Stress A hypothesized change in the inner state of the organ- ism, stress is analogous to the physicist's concept of strain and can be either psychological or physiologic. For ex- ample, psychologists have described changes in cognitive functioning and predispositions that mediate between the stressor and the responses of the individual. Similarly, psychoanalysts infer changes in ego functions and the bal- ance of instincts and defenses in response to trauma or conflict. Physiologists often infer changes in CNS func- tioning, particularly at subcortical centers. This stress is inferential; it cannot be directly observed but is inferred as mediating factors because of lack of a direct one-to-one relationship between a stressor and the responses of the organism, particularly the human organism. Stress Responses Stress responses are directly observable changes that include various coping techniques people use—emotional responses often of a distressing nature (such as anxiety, tension, depression, anger), changes in behavior, and changes in physiologic functioning (most often manifested by changes in the autonomic nervous system, catechol- amine, and pituitary/adrenal/cortical changes). Adverse Health Consequences In recent years, evidence has increased that the process of responding to stressors, including stressful life events, leads to adverse health consequences. These negative ef- 57 fects appear as susceptibility to various emotional dis- orders such as depression or alcoholism, such physical illnesses as myocardial infarction and hypertension, and accidents. This paper focuses mainly on stressful life events as one subcategory of stressors that has been im- plicated in increased risk of affective disorders. Adaptive Aspects of Affects An adaptational approach views affective states, espe- cially depressions, as a component of the adaptive re- sponses of the human organism to its environments, including psychosocial stressors. In understanding this adaptation, we need to determine whether the processes involved in both initiation and perpetuation of the affec- tive responses are similar in normal emotional states and in clinical depressions. A definitive answer to this question would resolve a continuing controversy as to whether the clinical affective disorders are quantitative extensions of the normal affects of depression encountered in everyday life, or whether they represent states that are qualitatively different from that of a normal affect. The modern concept of adaptation derives from Darwin's theory of evolution. According to the strictest criterion of evolutionary theory, from the phylogenetic viewpoint, an anatomical structure, a behavior trait, or an emotion is adaptive if it promotes the survival of the species. Simi- larly, from the ontogenetic viewpoint, a trait or behavior is adaptive to the extent that it promotes the growth and survival of an individual member of the species. Darwin himself pioneered in the application of an evolutionary approach to behavior, including emotional responses. As Plutchik (1970) has described, Darwin's theory of evolution explicitly states the thesis that there existed an evolution not only of morphological structures, but also of what Darwin called "mental and expressive capacities." To es- tablish his general thesis, Darwin (1965) collected material to document the continuity of emotional expression in lower animals as well as primates and in humans. However, these ideas and observations lay dormant for many decades. Since World War II, there has been a remarkable upsurge of interest in the comparative biology of emotional states. In this upsurge, the field of ethology has emerged as a specialized area overlapping studies of animal behavior in 58 psychology and zoology. Much of this research involves a significant convergence of findings from neurobiology, ethology, and comparative psychology. Studies of human infant development, particularly those following upon psy- choanalytic theories, have paralleled this animal research. Studies of mammalian behavior, particularly of the primate infant and child development, have produced a growing body of evidence as to the nature of the mother-child bond of primates. Depression, anxiety, and other affective states have an important function in the regulation of these bonds. It is now well substantiated that all mammals inherit complex behavioral systems that promote and reinforce the mother-infant attachment and, as a consequence, facilitate the formation of social groups. In this context, depression is part of the affective response to the actual or threatened disruption of the mother-infant attachment bond. These behavioral systems and the related affective expressive behaviors have been profoundly adaptive for the species and for the individual over the millennia. They have facilitated the biological survival of infant mammals during the long period of extrauterine development before the mammals achieve self-sufficiency. Moreover, these biobehavioral systems encourage social learning, which has also been highly adaptive in the evolu- tionary survival of mammals. Mammals, unlike lower spe- cies, have a large repertoire of learned responses, which has enabled them to react appropriately to changes pro- duced by the environment, including changes in the com- position of social groups. When the maternal bond of mammalian species is broken by separation and loss, a typical behavior pattern emerges in which the initial stage is characterized by anxiety, agitation, protest, and in- creased psychomotor activity, followed by a phase of social withdrawal and decreased motor activity. It now appears certain that the animal reactions best studied in dogs and primates in response to separation and loss are so similar to human sadness that few observers doubt the continuity between the animal and the human infant experience. The adaptive view lends special importance to research on animal models that relates stressors in general and the stress of separation in particular to affective states. The adaptive function of depressive affect as a signal for social communication has been demonstrated by studies in dogs, in primates, and in human infants. As a mood, affect, feeling state, or emotion—those 59 terms are used interchangeably for the purposes of this paper—depression is a pervasive feature of normal human experience. Feelings of sadness, disappointment, frustra-— tion, discouragement, and related emotions are frequent concomitants of the vicissitudes of the human condition. The normality of depression poses multiple problems for clinical practice and theory. For clinical practice, criteria are needed to specify the boundaries between the normal mood state and those abnormal states that merit clinical intervention. For theory, it is necessary to understand the nature and the function of depression as a normal emotion and to elucidate which aspects of depression are common to both normal and pathological states, as distinguished from those features that are unique to the abnormal states. Thus, depression as an emotion in primates and human beings serves a signal function. It alerts the mother or other members of the social group that one of its helpless members, the infant, is in potential danger. That is espe- cially true during the phase of rapid central nervous system maturation and during the acquisition of cognitive, per- ceptual, motor, and social skills. Alerted protectors can rally resources for nurturance, support, and protection and thus promote biological survival. These generalizations are true for animals and for human children, but what of adults living in a modern industrial society? Is depression among civilized men and women merely the automatic perpetuation of previously developed evolutionary responses? If so, is the clinical depression of adults an adaptive response, or is it a maladaptive recur— rence of behaviors that were adaptive in an earlier devel- opmental state? Investigations into these questions involve clinical and biological research and theoretical analysis. Adaptive Functions of Affective Responses to Stressors This brief discussion of the significance of separation and loss in animals and infants allows the delineation of some general aspects of the adaptive role of affects. From an adaptational view, affects can be regarded as having multiple psychobiological functions. Four adaptive func- tions of affects are relevant to clinical psychiatry: social communication, physiological arousal, subjective aware- ness, and psychodynamic defense. 60 Social Communication The adaptive role of depressive emotion as social com- munication has been elucidated by animal studies, espe- cially in primates (Kaufman and Rosenbaum 1967; Harlow et al. 1971; Suomi et al. 1978), and by studies of human infancy (Bowlby 1969). The components of affective com- munication—crying, facial expression, posture, touch, smell—have been studied clinically and experimentally. It is now widely accepted that all mammals inherit com- plex behavioral systems that promote the mother-infant attachment and facilitate the formation of other social bonds. The importance of those bonds is increased among primates, for whom the period of extrauterine growth and development is prolonged. The helplessness and the dependence of the infant primate promote those bonds and provide the opportunity for social communication, learning, and group interactions. The development of such behavioral systems has been profoundly adaptive for the species and for the individual; these systems facilitate the biological survival of infants during the long period of extrauterine development before biological self-sufficiency is achieved. Moreover, these systems encourage social learning, which is highly significant from an evolutionary standpoint. Mammals, unlike lower species, learn responses that enable them to react appropriately and adaptively to changes produced by the environment. Physiological Arousal Clinical observation and animal experimentation have established the relation of separation-loss to infant de- pressive affect and have clarified the role of depressive affect as a social signal. Problems arise in specifying the neuroanatomical, electrophysiological, and neurochemical mechanisms by which those affective states are initiated, perpetuated, and terminated. Parallel problems have been investigated for anxiety- fear, an affect closely related clinically and developmen-— tally to depression. Following on the research of Cannon (1936), scientists accept the hypothesis that anxiety-fear arouses the organism in preparation for fight or flight. That function is carried out by complex neuroendocrine systems, especially those involving hypothalamic and ad- renergic structures and the release of epinephrine. 61 When one attempts a similar description of the mechan- isms involved in depression, less agreement is evident. Impressive research, mostly based on advances in psycho- pharmacology, strongly implicates the central nervous system biogenic amines and neuroendocrine systems in those reactions. Although the evidence derives more from neuropharmacological studies than from direct observation in humans, the patterns and trends increasingly support a role for biogenic amines in the mediation of affective re— sponses, especially depression. Another hypothesis defining the physiological adaptive function of depression was offered by Engel (1962) and Schmale (1970) and their associates. They postulated that depression involves conservation-withdrawal, with reduced psychomotor activity. That is an intriguing hypothesis, for which experimental verification is required. Although the formulation may be consistent with observations of infan- tile states, clinical depression in adults is associated with increased adrenal-cortical activity and with anxiety and tension, presumably due to heightened adrenergic activity changes inconsistent with the conservation-withdrawal formulation. One explanation for the discrepancy is to as- sume that the clinical depressions of adults involve a fail- ure of mechanisms operative in normal and infantile states. Also, different psychobiological changes probably dis— tinguish the protest phase from the despair phase after loss and separation. The increased sympathetic and neuro- endocrine activity is probably related to the protest phase, and the conservation-withdrawal processes come into operation later, in the despair phase. Subjective Awareness The subjective aspect of affect has been emphasized by most clinical research and theory and is probably unique to human experience. It is widely if not universally accepted that the subjec- tive components of human emotion, whether epiphenom- enal or real, conscious or unconscious, play important functions in setting goals and monitoring current behavior, particularly by judging personal reality against internalized values and goals. Engel (1962) stated: Intrapsychically, affects provide information con- cerning the status of the self per se and of the self in 62 relation to objects, and they indicate the level of drive tensions. In the adult, such intrapsychic activi- ties are ego functions having motivational and warn- ing properties. As signals, affects anticipate, on the basis of past experience, intrapsychic changes there- by permitting the ego to initiate psychic processes or behavior to maintain the psychic balance. There is growing convergence of, on the one hand, so- cial-psychological studies that have explored fluctuations in self-esteem, self-image, aspirations, and interpersonal relations and, on the other hand, behavioral approaches that search for sources of reinforcement in the real-life social world of normal persons related to their mood fluctuations. Beck (1969) described cognitive dysfunctions in depressed patients, particularly their impaired capacity to judge themselves and their performances realistically, and from those observations proposed a cognitive psycho- therapy of depression. The capacity of human beings to relate affective awareness to associated ideas and cogni- tive representations serves important self-regulating functions in maintaining self-esteem and in setting and modifying goals. That capacity depends on the species’ achievement of language and rational thought, evolutionary attainments of great significance. Psychodynamic Defense The fourth adaptive function of depression is intra-— psychic defense, the element most discussed in clinical practice, especially by those influenced by psychoanalytic approaches. Nevertheless, it represents an area of con- tinuing debate. Originally, Freud viewed emotions as archaic discharge syndromes. In addition to drive discharge, the specifically psychodynamic functions of affect include the initiation of defense mechanisms. After Freud's writing about anxiety in the mid-1920s, psychodynamic theorists emphasized the defensive functions of affects in the studies of ego psy- chology. Freud maintained a strong biological view that derived human emotional capacities from instinctual drives. He also emphasized the continuity between adult behavior and infant behavior. Thus, psychodynamic theory stimulated interest in the developmental aspects of emo- tions, particularly in the possible role of early childhood 63 experiences as determinants of adult psychopathology. Psychodynamic thinking also stressed the important and perhaps crucial role of emotional experience not directly within the conscious awareness of the self but potentially recoverable by reconstruction, free association, dreams, projective tests, or hypnosis. Although psychodynamic theory has reinterpreted its formulations regarding anxiety to incorporate Cannon's (1936) discovery of adrenergic mechanisms and Freud's concept, stated in Inhibitions, Symptoms and Anxiety (1926), that anxiety serves as a signal to initiate ego de- fenses, similar formulations of depression have been slow to appear. Ego psychology approaches (Hartmann 1958) have been applied in depression (Bibring 1953; Rapaport 1967). A major obstacle in developing a comprehensive ego- adaptive view of depression arises from the psychodynamic postulate that hostility is primary to depression. Until re- cently, dominant clinical teaching stressed that depression is hostility turned against itself. In many clinical settings, that oversimplification became the practical distillate of the classic psychodynamic theory of depression initially proposed by Abraham (1927) and Freud during the period of 1908 to 1921. In the classical view, the clinical features of adult depression result from retroflexion onto the self of the hostility directed at the lost object. Through incorpo- ration, the self identifies with the lost object. The self is defended against the trauma of loss and thus avoids its psychic consequences. In depression, predisposition to that mechanism is postulated to have been determined by li- bidinal fixations in the late oral or early anal stages of development. Crucial to that formulation is the central role given to aggressive drives aroused by the ambivalence and dependence characteristic of the depressed person in his or her interpersonal and object relations (Rado 1929; Cohen et al. 1954). Freud used a transformation or alchemy explanation for three clinical phenomena: conversion hysteria, anxiety, and depression. Freud's earliest conception of the conversion mechanism in hysteria, developed about 1895, postulated that, in the process of symptom formation, psychic energy that was not discharged was converted into a physical symptom of a hysterical nature, such as paralysis of limb or loss of vision. Freud proposed a similar transformation explanation of anxiety in his early toxicological theory. For many years, he held that anxiety neurosis was one of the 64 actual neuroses, arising from biological causes, rather than a psychoneurosis, arising from conflict. His initial formu- lation was that anxiety was the psychological manifesta— tion of undischarged sexual libido, and not until 1923 did he formulate his signal theory of anxiety, in which he desig- nated the role of anxiety as the initiator of defense proc- esses. Freud's concept of depression as the transformation (retroflexion) of aggressive drives deriving from oral fix- ations persists into the present time as the dominant view in many clinical settings. As Bibring (1953), Rapaport (1967), and Chodoff (1970) elucidated, the classical theory derives from an exclusively instinctual view of the genesis of emotions and from in- sufficient attention to modern views of ego functions. The classic theory of depression fails to classify depression as a primary affective disorder in its own right but, rather, rel- egates depression to a transmutation or transformation of another affect, hostility. In recent years, numerous studies have directly inves— tigated the relations between depression and hostility (Gershon et al. 1968; Klerman and Gershon 1970; Weissman et al. 1971). Findings from these empirical studies have confirmed the independence of depression from hostility. The data show alternative patterns; both high and low intensities of manifest hostility occur, depending on the patient's personality patterns and the nature of his or her social relations. These research projects support the desirability of rejecting the hostility-turned-against-itself mechanisms as universally necessary and primary in the pathogenesis of depression. The findings support the hypothesis that depression is a primary ego state that serves important ego-adaptive functions. That view places the psychodynamic aspects of depression closer to other approaches. Acute Affective Responses to Stressful Life Events Against this general background, some current issues concerning affective responses to stressful life events can now be presented. There is increasing evidence that, as part of the response to stressful life events, a medley, often a discordant med- ley, of affective responses ensues. Most stressful life events elicit various mixtures of anxiety or fear, sadness or 65 depression, surprise and arousal, anger, hostility, and rage. Different types of life events are likely to elicit more or less greater predominance of one affect or another as shown in table 1. Table 1. Stressful life situation and dominant affect Stressful life situation Dominant affect Bereavement Sadness Threat to survival Fear Frustration of needs Rage, anger Ethical or legal transgressions Shame and guilt A comprehensive program of research should elaborate the relationships among the types of stressful life events and the various affects. To a great extent, this research program is now underway. For example, during World War II, extensive studies in the military about the stress of combat elucidated the role of anxiety and fear among military personnel and clarified the relationship between exposure to combat and the clinical syndromes variously called combat fatigue or traumatic war neuroses. It was found that the rate of combat neurosis within a unit was directly proportional to the time under fire, and that more than 90 days of direct combat exposure would result in almost 100-percent casualty from psychological reasons for those not injured. The other finding from World War II was that evacuation of soldiers with traumatic combat reactions to military hospitals geographically far removed from the soldiers' unit may result in subsidence of symptoms but seldom resulted in their return to combat. Based upon these experiences, Glass and his associates in the military during the Korean War developed the policy of limited exposure of units to combat, with alternative periods of rest and relaxation. In addition, combat neurotic reactions were treated close to the front line in battalion aid stations through the use of reassurance, group techniques, and sedation. These practices resulted in a marked reduction in the rate of combat neurosis in the Korean and Vietnam wars. Similarly, considerable research has been done on psy- chological responses to natural disasters such as hurri- 66 canes, floods, and earthquakes. Among the most interesting are the studies of the Buffalo Creek disaster, indicating the extent to which initial fear was followed by feelings of helplessness and isolation depending upon the prior state of social organization of the community. Followup studies of the survivors of the Buffalo Creek disaster indicate in- creases in hypertension as a late adverse health conse- quence. In clinical psychiatry, however, affective responses generally refer to anxiety, depression, and elation. Because Marks discusses anxiety reactions in chapter 5, comments here are limited to depression. Grief and Bereavement as Prototype of Affective Responses to Stressful Life Events Probably the best understood and most powerfully stressful life event in human experience is that of grief and bereavement. Almost every human being experiences one or more episodes of bereavement at some time in his or her life as a consequence of the death of a loved one, partic— ularly a parent, child, or spouse. Grief and bereavement are universal, not only through human cultures but also throughout mammalian species. In modern times, the psychological descriptions of grief in relationship to clinical depression were elucidated in the early 1900s by Karl Abraham and Sigmund Freud. Linde- mann's (1944) description of survivors' reactions to the deaths in the Coconut Grove fire in Boston in 1942 stands as a landmark in phenomenologic description and theoret— ical inference. Lindemann's paper is important not only for its description of the grief responses but because of the theoretical concept that there is a normal "work" of "grief." Equally significant is his inference that failure to complete "the work of grieving renders the individual vul- nerable to later psychiatric or psychophysiologic reactions, particularly to psychosomatic disorder, such as ulcerative colitis." Since the Lindemann paper, clinical interest has increased in delayed grief and abnormal grief and also in the role of grief as precipitant medical and psychosomatic disorders such as colitis, thyroid disease, and arthritis. A number of unresolved questions regarding grief are of particular relevance to this presentation. These include the following: 67 eo To what extent are guilt, shame, and hostility neces— sary components of the grieving process? e Does grief in general predispose to adverse health consequences? e Is the pathophysiology of grief similar to or different from clinical depression? e What is the value of drug treatment of the acute grief reaction? e What are the short-term and long-term effects of psychological intervention? Relationship of Grief to Guilt, Shame, and Hostility Freud, in his 1910 paper "Mourning and Melancholia," postulated that normal grief does not include feelings of guilt. In melancholia, guilt was present, and, furthermore, the guilt represented turning of hostility toward the lost object against the self. Recent research, particularly that of Lindemann and others, has challenged the empirical basis of both of Freud's propositions. It is now apparent that grief is often but not universally accompanied by feelings of guilt and shame as part of the normal mourning process, particularly in Western civilization. Moreover, feelings of guilt and hostility are not universal among clinically depressed patients, questioning the universality of Freud's postulated mechanism that depression involves the turning of the hostility initially directed against the lost object back toward the self. Further, cross-cultural studies indicate that the idea- tional and cognitive components of depression and grief that are so frequent in Western society occur less often in non-Western societies. In Asian cultures, somatization is more common, and expressions of guilt or similar impair- ment are relatively infrequent. Thus, one must conclude that guilt and anger, while frequent in Western societies, are not necessarily universal concomitants of the grieving process. In fact, given the animal evidence, it is likely that the ideational components, so characteristic of human grief and depression, represent a later evolutionary addi- tion rather than the care or the capacity to respond to 68 disruption of attachment bonds that we inherit from pri- mates and other mammals. Pathophysiology of Grief All the descriptions of grief report profound bodily changes in sleep, appetite, level of energy, muscle tension, breathing, and gastrointestinal functioning. Yet there has been little direct physiological or biochemical study of the altered but reversible physiology of the grieving process. Such studies are indicated and should allow us to compare bodily function changes during grieving with those of clin- ical depression. This would be important not only to clarify whether clinical depression is qualitatively different from grief in its pathophysiology and psychodynamics but also possibly to elucidate some of the physiologic mechanisms that may predispose grieving patients to increased mor- tality, morbidity, and other adverse health consequences. Adverse Health Consequences of Grief Many clinicians and epidemiologists hypothesize that failure to adequately work through grief predisposes the individual to later health impairments. One line of inves- tigation has proposed that grieving persons are at greater risk for illnesses in general (Izen and Klerman 1977). This hypothesis has been an area of controversy in the research literature. Clayton (1978), among others, questions the evidence that grieving persons are at greater risk for ill- ness beyond the first 6 months. On the other hand, large epidemiologic studies indicate a higher mortality rate, particularly from cardiovascular disease, and more so in men than in women. Assuming that these findings are cor- roborated, the mechanisms that cause this mortality rate are worthy of study. Possible Value of Drug Treatment Surprisingly, 25 years after the introduction of imipra- mine, there exist few systematic reports and no controlled trials of the value of imipramine in treating the symptoms of grief. Epidemiologic and clinical studies indicate that a moderate percentage of grieving individuals seek medical 69 attention because of such emotionally distressing symp- toms as tension, anxiety, guilt, and insomnia. An increase in the use of alcohol, analgesics, and various sedatives and hypnotics occurs among grieving persons. Yet no system- atic study has been done of the possible value of tricyclic antidepressants or MAO inhibitors for grieving persons. Following the Lindemann (1944) hypothesis, some cli- nicians have opposed the use of medication, holding that decreasing or blocking emotional response will hinder the normal work of grief and predispose the individual to later medical, psychosomatic, or psychiatric disorders. Ulti- mately, this is a testable hypothesis, not only for its prac- tical implications but also for theory. At the practical level, it would be valuable to know whether the distressing symptoms of tension, insomnia, and dysphoria of the grieving state similar to clinical depression would be al- leviated by a tricyclic. Moreover, long-term followup would allow testing the two alternative hypotheses of late adverse health conse- quences of grieving. Some centers speculate that failure to grieve predisposes to later illness. If this were true, patients whose affect response was diminished by imipra- mine would be predisposed to later health problems. On the other hand, Engel (1962) and others have argued that the disruptive and distressing aspects of grieving increase the risk for cardiovascular disease and other illnesses. In that hypothesis, reduction of the intensity of distressing symp toms by tricyclics might reduce risk of subsequent mor- bidity and mortality. Some indirect support for this hypothesis can be found in the report by Avery and Winokur that acute clinically depressed patients treated with tri- cyclics and ECT have a reduced mortality rate, not only from suicide but also from cardiovascular disease, sug- gesting that the distress of prolonged depression has a physiologic toll as well as psychological impairment. Alternative Psychological Interventions in Grief Over the past several decades, clinicians have developed a number of forms of brief psychotherapy, crisis inter— vention, and counseling techniques for grieving patients. Some are intended for patients experiencing intense symp- toms as part of normal grief reactions. Others, such as those developed by Lazare, are attempts to deal with ab- normal or delayed grief. Additional techniques are the 70 work of Horowitz and his associates at the University of California, San Francisco-Langley Porter Psychiatric Institute (see chapter 6). At the same time, an extensive program of mutual help called the Widow to Widow Program, which was initially developed in the Boston area by Phyllis Rolfe Silverman, has gained a good deal of national attention. The time is appropriate for a controlled trial comparing some form of professional counseling with self-help groups such as Widow to Widow. Of interest in such a study would be both short-term and long-term effects. Short-term re- sults of interest would be the extent to which these in— terventions reduce distressing symptoms of grief and pro- mote the resolution of the grief reaction. In the long run, it would be interesting to see whether these interventions lower the risk of grieving patients for late-appearing adverse health consequences, such as those reported in the literature (e.g., increased alcohol consumption, cardio vascular death, and clinical depression). Grief serves as a useful prototype of emotional responses to stressful life events and particularly highlights the con- vergence of findings from clinical psychiatry, comparative biology, and epidemiology. Therefore, a coordinated series of two types of public health projects is called for: ® A prospective longitudinal cohort study that would allow some estimates of direct and attributable risk ® Control trial of interventions to assess the feasibility of both primary and secondary prevention of grief- related health consequences Stressful Life Events as Risk Factors for Affective Disorders Although loss and separation serve as the stressors precipitating grief and depressed mood in mammals and in children, the research on adult clinical depressions in- dicates that the specificity of this relationship is far from resolved. In fact, the research reported in recent years is that loss and separation are not universal events anteced- ent to all clinical depressions. Other stressors often serve as precipitants, and in a significant proportion of clinical depressive episodes, no antecedent stressor can be iden- 71 tified. At best, loss and separation account for about 25 percent of precipitating events in a sample of depressed patients studied in case control designs. Depressed patients seem more susceptible to stressful life events prior to the onset of depressive episodes as compared with normal controls or schizophrenics, but loss and separation cannot in themselves account for the morbidity. Moreover, not all individuals who are exposed to loss and separation will subsequently develop depression. Obviously, other factors are involved. As risk factors, loss and sep- aration are not specific to clinical depression but seem implicated in a wide variety of clinical conditions, not only psychiatric but also medical. Therefore, an attempt to develop a theory based on one category of environmental stressor, psychosocial loss and separation, is insufficient. This risk factor is neither nec- essary nor sufficient to account for the clinical disorder. It appears that loss and separation call for an adaptive re- sponse that most individuals are able to make. About 10 to 20 percent of individuals experiencing loss and separation of a family member have persistent depressive symptoms at the end of 1 year. This seems to be the conclusion of cohort studies of bereavement and of other loss and sep- arations as reported by Paykel (1970), Parkes and Brown (1972), Clayton et al. (1968), and others. Thus, as painful and disruptive as the depressive response to loss and sep- aration in adults may be, it is usually time-limited, seldom lasting more than 6 to 9 months. Other. psychosocial stressors less intense and personally significant than the death of a spouse or child are equally time-limited. We must come to the conclusion that stressful life events, although risk factors for acute depressions, are neither universal nor specific. We have yet to determine why certain individuals are able to cope with stressors and separation, while other individuals fail in their adaptive efforts and subsequently develop clinical symptoms. The inability to cope under stress may be genetically determined or may be due to individual differences in early life experiences. For example, early parental loss predis— poses the individual to conditioned sensitivity, loss, and dependency. This concept has been proposed by psycho- dynamic theorists and documented in part by several studies of child development. Also of note are the behav- jorists' contingencies that relate propensity to depression, loss of self-esteem, and feelings of helplessness or worth- lessness to an inadequate repertoire of learned social skills. 72 Still another theory that has been proposed deals with ab- sence of social supports to provide substitute attachments and financial, social, and religious assistance. The current formulation is that the adult depressive episode represents a failed attempt by the individual to adapt. Clinical depressions are maladaptive outcomes of partially successful efforts at adaptation. At this point, reference can be made to a parallel con- cept in physiology, that of homeostasis. The concept of homeostasis was conceived by Walter Cannon (1936), based on earlier research by Claude Bernard on the internal mi- lieu. The concept of homeostasis has been extended to psychology to apply to the internal psychological forces in personality akin to the physiologic balance. D.W. Richards, Nobel Laureate, argued that, in studying disease, it is necessary to examine the failures of homeostasis as well as those mechanisms that maintain homeostasis in normal states. He argued that there are limits to the homeostatic capabilities of the physiologic organism. Similarly, there are limits to the capability of the organism to adapt to stressors. Public Health and Epidemiologic Concerns The field of stressful life events represents an inter- esting convergence of trends from psychobiology, clinical psychiatry, epidemiology, and public health. Viewed from this context, stressful life events emerge as important risk factors for a variety of medical and psychiatric illnesses. Although this paper has focused its attention on the role of stressful life events in clinical depression and related af- fective disorders, an extensive body of research now in- dicates the probable relationship of stressful life events to other psychiatric illnesses. These include schizophrenia, anxiety states, alcoholism, and personality disorder. Medical illnesses, including cardiovascular disorder and other organic pathology, can be traced to stress as well. Most of the research thus far has been based on survey techniques in which a population of clinically ill individuals is compared with some control group. Translated into epi- demiologic methodology, this would approach the case control method. However, with the exception of papers by Paykel and the Yale group, relatively few of the investi- gators in the area of stressful life events have used epi- 73 demiologic concepts such as risk factors or quantitative measures of estimating relative risk. It is no longer suffi- cient in epidemiologic research to identify a given risk fac— tor; some estimate of the magnitude of risk is called for. Similarly, prospective cohort studies are more powerful means of demonstrating the level of risk factors and of developing quantitative measures of the magnitude of risk. What appears to be called for is a series of prospective studies of individuals exposed to specific stressful life events such as grief and bereavement, unemployment, re— tirement, and separation and divorce. Such prospective cohort studies, framed in epidemiologic methodology, would allow quantitative estimates of outcome using such techniques as life table method as well as afford more quantifiable measures of direct and attributable risk. Such studies might also allow the quantitative estimation of other associated risk factors to determine the interaction of multiple factors in multifactorial models. Assuming that further research confirms the growing evidence already available as to the role of stressful life events as risk factors, particularly for affective disorder, the possibility of preventive interventions becomes more eminent. A folklore already exists about the value of stress reduction and stress management techniques in relieving everyday distress presumably associated with stressful life events and adverse circumstances. However, relatively few of these techniques have been subjected to systematic in- vestigation using the methodology of the controlled clinical trial. In this paper, we have taken the situation of grief and bereavement as a prototype and outlined a number of pos— sible control trials investigating the role of intervention, such as individual counseling, Widow to Widow and other self-help programs, and medication. These interventions would have secondary prevention functions in relieving the distress and potential disability associated with the acute emotional response to loss and bereavement. Such studies might also demonstrate a primary prevention success in reducing the risk for subsequent adverse health conse- quences of a delayed or long-term nature. The successful design and execution of these epidemio- logic and preventive intervention studies would contribute greatly to the rapprochement between clinical psychiatry and public health. The results of these studies would con- tribute to the theoretical understanding of the relationship between stressful life events and health consequences and 74 would enhance practical efforts at therapeutic and pre- ventive intervention. References Abraham, K. Selected Papers of Karl Abraham. London: Hogarth Press, 1927. Beck, A.T. Cognition and psychotherapy. In: Beck, A.T., ed. Depression: Clinical and Theoretical Aspects. New York: Harper and Row, 1969. p. 253. Bibring, E. Mechanism of depression. In: Greenacre, P., ed. Affective Disorders. New York: International Univer- sity Press, 1953. p. 13. Bowlby, J. Attachment. Basic Books: New York, 1969. Cannon, W.B. The Wisdom of the Body. New York: W.W. Norton, 1936. Chodoff, P. The core problem in depression: Interpersonal aspects. In: Masserman, J., ed. Science and Psycho- analysis. New York: Grune & Stratton, 1970. Clayton, P.J.; Desmarais, L.; and Winokur, G. A study of normal bereavement. American Journal of Psychiatry 125:168-178, 1968. Clayton, P.J. "The Course of Depressive Symptoms Fol- lowing the Stress of Bereavement." Paper presented at the meeting of the American Psychopathological As- sociation, March 10, 1978. Cohen, M.B.; Baker, G.; Cohen, R.A.; Fromm-Reichmann, G.; and Weigert, E. An intensive study of 12 cases of manic depressive psychosis. Psychiatry 17:103, 1954. Darwin, C. The Expression of the Emotions in Man and Animals. Chicago: University of Chicago Press, 1965. Engel, G.L. Psychological Development in Health and Disease. Philadelphia: W.B. Saunders, 1962. Gershon, E.S.; Cromer, M.; and Klerman, G.L. Hostility and depression. Psychiatry 31:224, 1968. Harlow, H.F.; Harlow, M.K.; and Suomi, S.J. From thought to therapy: Lessons from a primate laboratory. Ameri- can Scientist 59:538, 1971. Hartmann, E. Ego Psychology and the Problem of Adap- tation. New York: International University Press, 1958. Holmes, R.H., and Rahe, R.H. The social readjustment rating scale. Journal of Psychosomatic Research 11:213- 218, 1967. Kaufman, J.C., and Rosenbaum, L.A. The reaction to sep- aration in infant monkeys: Anaclitic depression and 75 conservation withdrawal. Psychosomatic Medicine 29:648, 1967. Klerman, G.L., and Gershon, E.S. Imipramine effects upon hostility in depression. Journal of Nervous and Mental Disease 150:127, 1970. Klerman, G.L., and Izen, J.E. The effects of bereavement and grief on physical and health general well being. Advances in Psychosomatic Medicine 9, 1977. Lazare, A. Unresolved grief. In: Lazare, A., ed. Out- patient Psychiatry: Diagnosis and Treatment. Balti- more: Williams & Wilkins, 1979. Lindemann, E. Symptomatology and management of acute grief. American Journal of Psychiatry 101:141-146, 1944. Parkes, C.M., and Brown, R. Health after bereavement. A controlled study of young Boston widows and widowers. Psychosomatic Medicine 34:499-461, 1972. Paykel, E.S. Life events and acute depression. In: Scott, J.P., and Senay, E.C., eds. Separation and Depression, Clinical and Research Aspects, pp. 215-236. Publication 94 of the American Association for the Advancement of Science, Washington, D.C., 1970. Plutchik, R. Emotions, evolution, and adaptive processes. In: Arnold, M.B., ed. Feelings and Emotions. New York: Academic Press, 1970. pp. 3-24. Rado, S. The problem of melancholia. International Jour- nal of Psychoanalysis 9:420, 1929. Rapaport, D. Edward Bibring's theory of depression. In: Gill, M., ed. Collected Papers of David Rapaport. New York: Basic Books, 1967. p. 728. Schmale, A.J. Adaptive role of depression in health and disease. In: Scott, J.P., and Senay, E.C., eds. Separation and Depression, Clinical and Research Aspects, pp. 187-214. Publication 94 of the American Association for the Advancement of Science, Washington, D.C., 1970. Suomi, S.J.; Seaman, S.F.; Lewis, J.K.; Delizio, R.D.; and McKinney, W.T. Effects of imipramine treatment of separation induced social disorders in rhesus monkeys. Archives of General Psychiatry 35:321, 1978. Weissman, M.M.; Klerman, G.L.; and Paykel, E.S. Clinical evaluation of hostility in depression. American Journal of Psychiatry 128:261, 1971. 76 Chapter 5 STRESS AND OTHER RISK FACTORS IN ANXIETY DISORDERS Isaac Marks, M.D., FRCPsych., DPM Analysis of stressful life influences on anxiety disorders of considerable heterogeneity requires understanding of the intimate interaction of such influences with many other psychosocial and biological risk factors. This review ac- cordingly will not confine itself to the effect of stressful life events alone but will consider also a range of inter- acting variables. The ultimate quest is prevention of dis- orders, in either delaying or inhibiting their onset (primary prevention), or by early or late remedy of the developed condition (secondary and tertiary prevention, respectively). Diagnostic Issues Diagnostic imprecision is a major problem in evaluating different studies of stress and anxiety disorders. The clin- ical populations involved are often unclear, so we cannot decide whether findings about stressful events relate to symptoms or to syndromes and to what extent anxiety, depression, and related features are occurring together or separately. Better units of analysis are needed in this area than we have at the moment. The most recent edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) (APA 1980) defines anxiety disorders as those in which anxiety is experienced directly. These are a frag- ment of the earlier DSM-II (APA 1968) category of neu- 71 roses, which was renamed and dispersed among several categories in DSM-III, as shown in table 1. The neuroses had been defined as conditions in which the chief characteristics were anxiety either (1) felt or expressed directly or (2) controlled unconsciously or auto- matically by conversion, displacement, and other psycho- logical mechanisms. The DSM-III category of anxiety disorders is thus smaller than the DSM-II category of neu- rosis, corresponding only to the first of the two DSM-II definitions. Three of the DSM-II sections of neurosis were subdivided into nine DSM-III sections of anxiety disorders, and two areas were added. Some might feel that certain of the DSM-III subdivisions are too elaborate, e.g., the use of panic attacks to define separate sections and the separa- tion of atypical from generalized anxiety disorder and all of these from dythymic disorder and from atypical depres— sion. Hard research is needed to indicate whether such sub- divisions have etiological or therapeutic import. The DSM-II neuroses omitted from the DSM-III category of anxiety disorders also are seen in table 1. Some prob- lems are immediately apparent. The classification excludes disorders such as depression, depersonalization, hypochon- driasis, and even conversion hysteria, which, despite the traditional view about belle indifference, are commonly associated with much directly experienced anxiety and, to that extent, overlap with anxiety disorders. Also excluded are psychosexual dysfunctions, which are commonly as- sociated with directly experienced anxiety. Conversely, many obsessive-compulsive and phobia-like conditions are not associated with anxiety but are classified as anxiety disorders in DSM-III. Some anxiety disorders are easy to diagnose and form coherent entities, even though each can be subdivided into several variants. Examples are the phobic and the obses- sive-compulsive syndromes. This is less true for anxiety states, and questions arise about their relationship to the syndromes of dysthymic disorder (depressive neurosis), atypical depression, and posttraumatic stress disorder. The distinction between anxiety and depression syndromes is often so difficult to draw that many workers have des- paired of making it at all. In a community sample, 67 percent of subjects with psychiatric disorder had anxiety plus depression that could not be differentiated, most of these running a brief course (Tennant et al. 19815). Cooper and Sylph (1973) found that 96 percent of their primary care population with psychi- 78 6L Table 1. DSM-II neuroses and DSM-III anxiety disorders DSM-II neuroses DSM-III equivalent 300.00 300.2 300.3 300.5 300.6 Anxiety neurosis Phobic neurosis Obsessive-compulsive neurosis Not represented Not represented Neurasthenic neurosis Depersonalization neurosis 300.00 300.01 300.02 300.21 300.22 300.23 300.29 309.21 300.30 308.30 309.81 300.12 300.60 Anxiety disorders Atypical anxiety disorder Panic disorder Generalized anxiety disorder Agoraphobia with panic attacks Agoraphobia without panic attacks Social phobia Simple phobia Separation anxiety disorder (in childhood section) Obsessive—-compulsive disorder Posttraumatic stress disorder, acute Posttraumatic stress disorder, chronic Other categories Chronic depressive disorder Depersonalization disorder (dissociative disorder) 08 Table 1.—Continued DSM-II neuroses DSM-III equivalent 300.7 Hypochondriacal neurosis 300.70 Hypochondriasis Not represented 300.81 Somatization disorder (hysteria, Briquet's disorder) 300.4 Depressive neurosis 296.21 Major depressive disorder, single episode or 31 or recurrent, without melancholia 301.12 Chronic depressive disorder 309.00 Adjustment disorder with depressed mood (concept "hysteria" avoided) 300.13 Hysterical neurosis 300.1 Hysterical neurosis, 307.80 Psychogenic pain disorder (somatoform disorder) conversion type 300.11 Conversion pain disorder (Dissociative disorder) 300.14 Hysterical neurosis, 300.12 Conversion amnesia disorder dissociative type 300.13 Conversion fuque disorder 300.14 Multiple personality (in childhood section) 307.46 Somnambulism (in childhood section) 300.8 Other neurosis 300.9 Unspecified neurosis Source: American Psychiatric Association 1980 atric disorder had an anxiety-depression syndrome that was "a constellation which at present cannot be reliably sub- divided" and termed it "minor affective illness." Johnstone et al. (1980) could not reliably subdivide 240 neurotic out- patients into anxious and depressed syndromes on the basis of the clinical picture. Kendell (1974) reported that 24 per-— cent of patients with anxiety states were rediagnosed as depressive illness when readmitted S years later (the re- verse occurring in only 2 percent) and opined that "the magnitude of the diagnostic change creates obvious dif- ficulties for those who maintain that there is a fundamen- tal difference between anxiety states and depression." Forty-four percent of anxiety state cases had brief depres-— sive episodes during followup over 6 years (Noyes et al. 1980) and 60 percent over S5- to 12-year followup (Cloninger et al. 1981). Fleiss et al. (1971) noted that anx- iety states scored like depressive neuroses on a depression scale and were distinguished only by higher anxiety scores; depression and anxiety correlated .47. In patients meeting DSM-III criteria for generalized anxiety disorder or panic disorder, Dealy et al. (1981) found that one-third described a syndrome of major affective disorder in the course of their anxiety. The sample with depression was overrepresented with women and had higher ratings for depression and higher frequency of panic dis- order. Analysis of the family history of these patients failed to reveal significant differences in the morbid risk of anxiety neurosis or depression among patients separated into those with and without depression. Panic disorder and depression were highly correlated among their patients with generalized anxiety disorder, and their "evidence does not appear to support the notion of two diseases." The utility of the diagnosis of mixed anxiety—depression has been questioned. Differentiation of depressed from anxious neurotic female outpatients was achieved in a dis- criminant function analysis by Prusoff and Klerman (1974), although 35 percent of the patients could not be assigned correctly even in that study, and the authors noted cor- relations between anxiety and depression of .5-.6. Com- pared with depressed patients, anxious subjects had similar anxiety but higher depression and somatization. Even when discrimination of this kind has been achieved, it by no means proves that this has any etiological or therapeutic import. It is quite possible, for example, that patients with generalized anxiety disorder respond as well to antide- pressant drugs as do patients with dysthymia, and double- 81 472-367 O - 85 - 4 blind research of this point is important to complete. More evidence for distinguishing between anxiety state and depressive illness was claimed by Gurney et al. (1970) and Schapira et al. (1972), who differentiated between their symptoms, mood, history of previous adjustment, and prognosis; depressed patients responded better to tricyclics and to ECT and were more improved at 4-year followup. Diagnostic crossover between the two groups during fol- lowup was exceptional, although chronic anxiety states tended to acquire depression with the passage of time. A rare discrimination of antecedents of anxiety states as opposed to depressive illness was found by Finlay-Jones and Brown (1981). Of 164 young middle class women attending a London general practitioner, 78 (45 percent) showed psychiatric morbidity on the GHQ, 45 of these to a degree of "caseness" (17 depression, 13 anxiety, 15 anxiety-depression—it is chastening how many cases had this mixed diagnosis). Ratings of unpleasant life events in the previous year, made without knowledge of the psychiatric state, were significantly associated with psychiatric disorder; loss events were related to depression, danger events to anxiety, and both loss and danger events to mixed anxiety-depression. Another distinction between anxiety and depression is seen in the data of Noyes et al. (1982); compared with families of controls, families of probands with agoraphobia and panic disorder gave significantly increased histories of anxiety disorder and of alcoholism but not of affective disorder. When we turn from syndromes to symptoms, an intimate relationship is again seen between anxiety and depression. In a normal community sample, Dohrenwend et al. (1980) found eight scales to be reflecting a single dimension of nonspecific stress, which they likened to demoralization; “components involved anxiety and dread as well as sadness and poor self-esteem. These eight components were "so strongly related to each. other that they are almost cer- tainly measures of the same thing." On the SCL-90, scores for anxiety and for depression correlated .6 in emergency room patients and .7 in psychiatric outpatients (Beutler, personal communication). In posttraumatic stress disorder, Horowitz et al. (1981) found a correlation of .88 between the SCL-90 subscores of anxiety and of depression, which in turn correlated no less than .93 and .94 with the total pathology score. Ninety-seven percent of these patients complained of feeling blue, 97 percent of worrying, 95 per- cent of feeling tense and keyed up, and 94 percent of nerv- 82 ousness and shakiness inside (Horowitz et al. 1980). Dysphoria may be likened to inflammation, which comes in several forms that are regarded as nonspecifically symp- tomatic except when they coexist with other phenomena, in which case we speak of syndromes. Similarly, there are some drugs with general anti-inflammatory properties and others that are especially useful for particular syndromes. Antidepressant drugs may well be useful for dysphoria of widely different etiologies. Epidemiological Considerations Until recently, epidemiological studies of mental health have commonly employed terms like mental or psycho- logical disorder, neurosis, and the like with little distinc- tion between symptom and syndrome, let alone between anxiety and depression. Studies finding high prevalences of pathology in the community usually refer to symptoms, and a similar picture emerges from many parts of the world, with point prevalences of 10-20 percent or higher, the great majority of these problems being anxiety—depression (Cooper 1972; Marsland et al. 1976; Andrews et al. 1978; Hoeper et al. 1979; Harding et al. 1980; Henderson et al. 1980; Katan et al. 1982; Tennant et al. 1981a, b; Vasquez- Barquero et al. 1981). These figures are in line with the 9-20 percentage point prevalence of depressive symptoms reported by Boyd and Weissman (1981). Anxiety—depression is also common in pain patients (re- viewed by Katan et al. 1982), with both anxiety-depression and pain improving with tricyclic drug treatment of such patients. Specific anxiety states, and depressive, phobic, and obsessive—compulsive disorders are much rarer than cor- responding symptoms (Boyd and Weissman 1981). Preva- lence of anxiety states in the United Kingdom, United States, and Sweden was estimated at 2-4.7 percent (Marks 1981), while phobic and obsessive-compulsive disorders are substantially rarer still (Agras et al. 1969; Cooper 1972; Brown et al. 1977; Hoeper et al. 1979). Most workers who have followed up anxiety—depression in primary care have found them to be usually short-lived episodes that clear up within a few months, so there is a temptation to regard these episodes as normal distress re- sponses (Hagnell 1970; Henderson 1981; Tennant et al. 83 1981a, b). (Exceptions are Harvey-Smith and Cooper 1970 and Schwab et al. 1979.) It is not clear how these episodes differ, except in degree, from posttraumatic stress dis- order. Such a distinction requires differentiation of a stressful event from a trauma, as well as between the pic- tures of the pathology concerned. Let us now examine potential risk factors for anxiety disorders and population subgroups at risk, and consider which interventions might modify some of the identifiable risk factors. Only tentative outlines and few prescriptive suggestions can be made at this stage of knowledge. Age, Sex, and Class Effects Most forms of anxiety disorder begin in young adult life between the ages of 15 and 40. Exceptions are animal phobias, which in the great majority begin before age 6 and rarely start de novo in adult life. The female preponder- ance so evident in most phobias is far less obvious for so- cial anxieties, obsessive—compulsive disorders, and anxiety states, for reasons that are not self-evident (Marks 1969, 1981). Whether socioeconomic status is crucial remains to be seen. Brown et al. (1977) found no class differences for psychiatrically disordered women in the Scottish island of North Uist, even though in London more disorders were found in working class than in middle class women. In North Uist, those women who were most integrated cul- turally had more anxiety, while those who were least in- tegrated had more depression, but this finding requires replication. Socioenvironmental Influences It is not clear which, if any, environmental events trig- ger anxiety disorders and to what extent they are neces- sary conditions for the onset of such disorders. Certain life events seem to precipitate many problems in mental health, including neuropsychiatric admissions of soldiers, neurosis in general practice, and depression (Dohrenwend and Dohrenwend 1981). Research into the relationship between stressful life events and anxiety commonly reports only on broad cat- 84 egories like mental disorders or neurosis, which generally involve anxious-depressive symptoms, and it is not clear how these relate to syndromes. This caveat should be borne in mind throughout this paper. We might also remember that good health for most people does not mean the ab- sence of problems but rather the presence of ones that are in some way manageable or tolerable, and that events classified by individuals as illness occur against this back- ground of normal symptomatology (Demers et al. 1980). Distressing or exit life events preceded neurotic impair— ment in samples from the community (Cooper and Sylph 1973; Tennant et al. 1981a, b) and also in manual workers when these events were associated with interpersonal dis— cord (Theorell 1976). Over 3 years, life events scores cor— related .58 with SCL scores (Gerst et al. 1977). It should be emphasized, however, that a history of clearcut trauma is absent in most patients with phobic and obsessive-com- pulsive disorders, the origins of such disorders usually being shrouded in mystery. The association of social dysfunction with the persist- ence of anxiety and depression was observed by Goldberg and Blackwell (1970). Those primary care patients with depression who were still ill at 6-month followup tended to have insoluble life problems, e.g., sick relatives or long- standing personality disorder. Similarly, the primary care sample of Cooper (1972) showed a marked excess of social difficulties among chronic neurotics compared with con- trols, and the clinical severity of their neurosis was related to their degree of social dysfunction. The potential importance of social bonds for anxiety disorders is attested by the partially protective effect of a network of acquaintances against the development of symptoms after the occurrence of stressful life events (Miller and Ingham 1976). Among air traffic controllers studied over 2% years, psychological problems tended to develop in those who initially were anomic and poor in so- cial coping resources (Jenkins 1979). Conversely, patients with anxiety and depression tend to perceive that they have defective social bonds with insufficient support from attachment figures and spend more time in unpleasant so- cial interaction (Henderson et al. 1978a, b; Silberfeld 1978). In keeping with these findings, neurotic inpatients engage in less self-disclosure to others than do normals (Mayo 1968), and women with neurotic depression have few confidants (Harris 1977). Perceived inadequacy of social attachment predicted subsequent increased anxiety- 85 depression more than did adverse life events (Henderson 1981). During their study, Henderson et al. (1980) noted that in a subsample visited four times over a year, several respondents began to confide in interviewers and to find them helpful. Emotional disorder in medical outpatients was associated with adverse life events and decreased so— cial support (Bird 1981). Jenkins (1979) and Eaton (1981) also reviewed evidence for impoverished social networks of mentally disordered persons compared with the general population and found that impoverished networks poten- tiated the adverse effect of stressful life events. Eaton further noted evidence that mental health impairment oc— curs more in disorganized than in integrated communities. Subjects who lacked social and community ties were much more likely to have died at 9-year followup, irre- spective of self-reported physical health status 9 years earlier, or of their socioeconomic status, smoking, alco- holism, obesity, physical activity, or utilization of pre- ventive health services (Berkman and Syme 1979). On this same theme, Sims and Prior (1978) found that neurotics had an increased risk of dying earlier from natural causes, even excluding suicide, single-vehicle accidents, and alcoholism. Vaillant (1979), following up a cohort that was fit at age 20, found that those with poor mental health in their thirties had by their fifties a ninefold greater risk of new chronic illness or death. In line with this, Eastwood and Trevelyan (1972) noted that primary care patients with psychological conditions had a sevenfold greater risk of major physical illness. Neurotic symptoms like anxiety, depression, and hypo- chondriasis are significantly more common in primary care patients who were classified by their general practitioners as physically ill (Vasquez-Barquero et al. 1981). Conse- quences also can be transmitted to the sufferer's children, as in the mothers with psychiatric disorder studied by Brown and Davidson (1978), whose children had a fourfold greater risk of accidents—the greatest killer of British children over age 1. Neurotics also behave in a more life- threatening manner. Compared with control inpatients suf- fering with varicose veins, neurotic inpatients smoked more cigarettes, inhaled more deeply, and had begun smoking younger (Salmons and Sims 1981); they also had a higher mortality rate from respiratory and cardiovascular diseases. Psychosocial and pharmacological factors can interact. Relevant to the role of life events and social support 86 systems in anxiety disorders are Leff's findings in chronic schizophrenics that high emotionality in the family pre- disposed the patient to relapse, and that patients on pheno- thiazines could tolerate more stressful life events before relapsing than patients on placebo. Appropriate medication thus raised the threshold at which breakdown occurred after environmental stress. Unfortunately, this protective effect of phenothiazines seems to continue only so long as the drug is given, and the relapse rate increases once the drug is withdrawn, even after several years. This pattern of relapse on drug withdrawal regrettably also seems to be emerging with tricyclics and MAOIs given for phobias, rituals, and anxiety—depression. Whether such drugs protect against the emergence of anxiety-depression during life stress remains to be seen and would be inter- esting to examine. One is all too aware of the potential snags of long-term drug administration and also of ac- cusations that drugs are being used to paper over cracks in society, but only longitudinal studies of the cost-effec- tiveness of drugs and psychological treatments in various anxiety disorders can inform us about the relevant trade- offs. Preliminary work suggests a time scale of at least 3-year followup (Marks 1981), and simple ways of meas— uring life events and social supports are critical to the effort. Childhood and Premorbid Personality Childhood events are associated with the subsequent development of some emotional disorders. Loss of a mother before age 11 predisposes women to develop depression 20 years later. The relationship of emotional disorders in childhood to later anxiety disorders in adult life was re- viewed by Cox (1976). He concluded that most children with emotional disorders do not become neurotic adults, and most neurotic adults were not neurotic as children, but that there is some evidence of continuity. Some have failed to find a relationship between records of childhood behavior and later adult neurosis. Lewine et al. (1978) examined teachers' comments in the childhood school records at grades 11-12 of 143 psychiatric patients and their matched controls. No differences were apparent in comments on those who later became neurotics com- pared with those who later became controls, even though 87 significant differences appeared for schizophrenics and personality disorders. Granted that temperament and personality have both genetic and environmental components, some tempera- ments may facilitate the development of certain anxiety disorders. The cliche that timidity is the soil out of which phobias necessarily grow is often disproved by clinical ex- perience that phobias can develop in the most courageous of people. Nevertheless, timidity and perhaps overprotec— tion in childhood might sensitize individuals to develop later phobias. Agoraphobics with a history of school phobia developed their agoraphobia significantly earlier than ago-— raphobics without such a history. This could occur, how- ever, because the two clinical states are two sides of the same syndrome at different ages, or the one state might predispose to the other, or both might reflect a common third problem (Marks 1969). Another type of personality— meticulous personality—is often said to be associated with the development of subsequent obsessive-compulsive disorders, even though compulsive rituals can develop in the most slipshod and untidy individuals. Associations have been found between the presence of hardworking, time- conscious personalities and coronary heart disease. By that token, most of us here today are at risk. Whether this is relevant to anxiety disorders is problematic. Prospective longitudinal studies, which can shed light on these questions, would be less expensive if appropriate data were collected as part of a longitudinal study of wider is— sues. This area is important to explore to identify high-risk populations for future preventive programs. An important question is whether stoicism can illuminate anxiety-proneness. Can we learn anything about anxiety disorders from the background of stoical people? As an example, Moshe Dayan had unusual detachment, which helped him weather virtually any criticism or tension. He reported that when Opposition Members of Parliament screamed abuse at him while he was speaking at the Knes- set, "Though it was not at all pleasant, I went on with my speech, and felt no psychological stress. I withdrew myself, a sensation familiar to me from the battlefield, when I would cut myself off emotionally from reality. What was happening in the Knesset was in a fog and unreal to me, like the burst of shells when crossing a field of fire." (Newsweek, Oct. 26, 1981, p. 59) Of 105 superior jet pilots aged 20-40, first-born children with unusually close father-son relationships numbered 67 88 (Reinhardt 1970). The pilots were candid, self-confident, desired challenge and success, were not introspective, and tended toward interpersonal and emotional distance. They showed no evidence of compulsive risk-taking or counter— phobic activity. These personalities are similar to those found in astronauts. Bomb disposal technicians in Northern Ireland who per— formed well were "toughminded" and "forthright," i.e., un- sophisticated, easily pleased, natural, sometimes crude and awkward (Hallam and Rachman 1980). Those who were decorated for exceptional bravery were especially stable and well-adjusted, with extremely low hypochondriasis scores ("My mind works quickly and well these days," "I feel fit and happy."). It is noteworthy that several of these men with steel nerves nevertheless reported being fearful of heights, snakes, dentists, and injections. A final intriguing anecdote from Girodo (personal com- munication) highlights the kind of phenomenon deserving investigation. He wished to train nuns to endure keeping their hands in a bucket full of ice until the pain became intolerable, but found that, before any training, the nuns, unlike other people, kept their hands in the ice bucket in- definitely. A host of interesting questions arise whose answers could illuminate the mechanisms of stress immunization. Biochemical and Other Physical Factors So far no reliable biochemical correlates of anxiety have been found in humans. Traditional indicators of stress may concern not so much anxiety or even unpleasant affect but rather orientation or arousal, regardless of hedonic tone. Catecholamine production increases after comic as well as after frightening films (Levi 1972), and an indirect psycho- logical indicator of stress such as intrusive and repetitive thoughts occurred equally often after erotic and unpleasant films (Horowitz 1976). Immediate cortisol production can be an orienting response to a new situation rather than an indicator of severe anxiety, e.g., phobics showed temporary increase in plasma cortisol while getting used to a lab- oratory, but during subsequent treatment there by flooding in vivo their plasma cortisol did not rise, despite intense subjective and behavioral anxiety (Curtis et al. 1976). 89 There is one report of a greater chemical increase after anxiety than after another emotion. Free fatty acids rose in normals, within 15 minutes, more after an anxiety-in- ducing than after a hostility-inducing situation; free fatty acids also increased during REM sleep proportionally to the anxiety content of dreams reported during those periods (Cleghorn et al. 1967). This increase was thought to result from mobilization from fat, the inverse depending upon catecholamine production in the short term but in the longer term on other mechanisms, and was decreased by propanolol. An intriguing preliminary finding by Curtis et al. (1976) well deserves attempts at replication. These researchers found that treatment of phobics by exposure in vivo took twice the time to reach criterion of improvement in the morning, when cortisol secretion was at its maximum, than in the late afternoon when cortisol secretion was lowest. If this finding is replicated, the question will arise whether this has to do with the biological clocks of the subjects or of the therapist. Such findings could not only influence clinical practice but also raise questions about the dif- ferential sensitivities of individuals to stressful stimuli according to the phases of their biological clocks at the time of stress. There is increasing evidence that peptides affect avoid- ance responses in mammals (Gispin et al, 1976; Nemeroff and Prange 1978), different peptides overlapping in their behavioral profiles, yet retaining areas of specificity. The time courses of behavioral actions of peptides are inter- mediate between those of very rapidly acting neurotrans— mitters and those of slowly acting hormones. The interaction between stressors and the response of the organism involves multiple systems which, in mice, include interplay between, at least, corticosteroids, T, B, and NK cells, and the thymus, among others. There are subtle neuroendocrine influences on immunocompetence, which has major effects on the growth rate of tumors in experimental animals (Riley 1981). In human volunteers experimentally inoculated with cold viruses, the magnitude of infection (estimated from the amount of virus present in nasal washings) was not related to antibody level, but rather to the number of preceding stressful life events, and introverts developed worse symp— toms and infections. How such influences affect anxiety disorders in the human clearly requires investigation. Med- ical outpatients have a high rate of emotional disorder 90 (Bird 1981). Actual physical disturbance is commonly ac- companied by anxiety-depression. In patients attending a hearing clinic, the increased GHQ score correlated with the amount of objective hearing loss and also with the presence of tinnitus and vertigo (Singerman et al. 1980). Individual Genetic Influences In animals, selective breeding experiments have demon- strated that emotional behavior, such as aggression and fearfulness, is strongly influenced by multiple genetic factors and is associated with selective biochemical changes. Anxiety-related features that have been selec- tively bred in animals include, in rats, general level of fearfulness, escape avoidance, and tonic immobility; in dogs, passive defensive behavior and general fearfulness. Tonic immobility, which can be regarded as an analog of freezing during fear, was increased four times in duration by selective breeding over just one generation of chickens (Gallup et al. 1977). Similarly, stress-induced fatal hyper- tension has been selectively bred in rats. The potential relationship of genetic factors to stressful life events is highlighted by the experiment of Morley (1977); in rats, the incubation of avoidance responses after a single shock occurred only in emotionally reactive, not nonreactive, strains. In rhesus monkeys, anxious responses to stressful events showed a strong interaction between genetic and experiential factors (Suomi et al. 1981). There is a high incidence of anxiety states in the fam- ilies of patients with anxiety disorder, but none of the studies has excluded the effect of familial environment (Lader and Marks 1971; Shields 1976). Genetic influences might act by predisposing individuals to have timid or other temperaments leading to later anxiety disorders. Twin research in the area is suggestive both for person- ality and for illness. Fear of strangers in twins during the first year of life is more concordant among monozygotic (MZ) twins than dizygotic (DZ) twins, and adult twins show more concordance for neuroticism among MZ than DZ, even when the MZ pairs have been reared apart. Similarly, physiological responses such as skin conductance, EEG, pulse, and respiration rate are more alike among MZ than DZ twins (Lader 1976). For phobic and obsessive-compulsive illness, 43 Maudsley 91 twin pairs were studied where one or both had the diagnosis (Carey and Gottesman 1980). Carey found a higher con- cordance among MZ than DZ twins for symptoms than for diagnosis of phobic and obsessive-compulsive phenomena. All the concordant phobics were polyphobic (including ago- raphobia), not monophobic, and the concordance was for presence of phobic symptoms rather than for the detailed nature of the phobic situation. Another twin study, of 99 same-sex twin pairs (Torgersen 1979), also found that MZ twins were more similar than were DZ twins in strength of phobia and in type of situation feared, but in this study the effect was restricted to the nonagoraphobic fears. Studies of genetic risk factors for anxiety disorder yield better evidence if undertaken in MZ twins reared apart compared with those reared together, so that the en- vironmental contribution can be teased out. Because such tiny numbers are available, such research is viable only when undertaken across several centers. Carey's 43 twin pairs constituted the entire relevant Maudsley sample over many decades. On the other hand, co-twin control studies can illuminate environmental issues by matching the discordant characteristics of MZ twins with discordant elements in their upbringing. Phyletic Influences (Species Genetics) The design characteristics of the species and its wiring diagram clearly affect certain aspects of anxiety. What we fear and when we show fear are not random. There are numerous natural fears and rituals, and maturational fac- tors affect the timing of their emergence. Organisms, in- cluding humans, appear to have selective stimulus-response valencies or facilitatory connections. Some of these may be related to adult syndromes and need to be taken into account in programs of treatment and prevention (Marks 1981). Treatment of Developed Disorders (Secondary and Tertiary Prevention) For phobic and obsessive-compulsive disorders, the be- havioral approach of exposure in vivo is currently the treatment of choice, producing lasting improvement in up 92 to 10 sessions, gains in phobias and rituals persisting up to 7-year followup in at least six countries (Marks 1981). An increasing trend is for most of the exposure treatment to be carried out by the patient himself or herself as self- exposure homework, with the therapist fading early from the scene. The principles of exposure in vivo have been worked out so well that not only can the treatment be given effectively by nurse-therapists, but a self-help man- ual (Marks 1978) has been as effective as the same treat- ment given by a well-trained psychiatrist. The exposure program can be given equally well by a computer, through which the patient interacts, although this technology con- fers no advantage over a manual. Exposure in vivo does not lastingly affect the minor af- fective episodes to which many of these patients are prone; for these, antidepressant drugs are useful, but there is a high relapse rate on drug withdrawal, even after a year's treatment, so that the drugs do not seem to be altering the underlying pathology. In the absence of dysphoria, anti- depressant drugs are of limited value in these syndromes. For anxiety states without situational anxiety, there are no clear guidelines to treatment of lasting value. Neither relaxation nor biofeedback is yet based on well-controlled long-term outcome studies, nor are behavioral claims for anxiety-management methods. Some studies indicate the short-term value of MAOIs (Paykel et al. 1982) and tri- cyclics (Johnstone et al. 1980), but long-term followups remain to be done. Anxiolytic drugs are merely palliative and carry the risk of addiction. Many drug studies to date have been too short term in nature to gauge the efficacy of drugs in the long term. Given (a) the overlap between anx-— iety and dysthymic disorders, (b) the results of antidepres— sants obtained by the New Haven group with nonpsychotic depression, and (c) the effects of antidepressants in minor depression complicating phobic and compulsive disorders, it would be surprising if antidepressants had any long-term effect in anxiety states after drug withdrawal. The control study of Sloane et al. (1975) found no dif- ference in outcome between behavioral and psychodynamic methods for a patient population suffering largely from anxiety states and personality disorders, and neither treat— ment was substantially better than being on a waiting list for 4 weeks, even though the last group had far less ther— apist contact. Whether cognitive therapies will prove any more effective remains to be seen. In nonpsychotic depres— sion, only one controlled study so far has found cognitive 93 therapy to be significantly superior (to imipramine) at 1- year followup (Kovacs et al. 1981). In that study, the cog- nitive therapy group had had 350 percent more therapist contact than the drug group, so the role of the cognitive component of cognitive therapy per se remains in some doubt. Another study (Harpin et al. 1982) failed to find significant superiority of cognitive therapy over a wait-list control group in chronic depression. Cognitive therapy also has failed to enhance the therapeutic value of behavioral treatment in agoraphobia and of social skills problems (Stravynski et al. 1982). The helpfulness for anxiety states of nonbehavioral psy- chological treatment is also unclear. With students who had high depression anxiety on the MMPI, there was no advan- tage to using experienced therapists as opposed to untrained supportive figures (Strupp and Hadley 1979). Although some form of treatment appeared superior to no treatment . . . no support (was given) to the major hypothesis that, given a benign human rela- tionship, the technical skills of professional psycho- therapists produce measurably greater therapeutic change. In line with this is the thoughtful review of Havik (1980). He summarized 25 U.S. studies of the effects of patient information and education on physically ill patients under— going a variety of procedures. On average, patients having psychological treatment were better off than 76 percent of control hospital patients and spent significantly less time in hospitals. Interventions of less than 30 minutes were as good as longer interventions. Havik's review found that information and education from nurses are as effective as that given by doctors or psychologists. This finding is widely supported in the plentiful literature showing the value of paraprofessionals and less trained people in health care as a whole and in psychiatry in particular (Marks 1981, chapter 9). Of relevance here, Cooper et al. (1975) found that chronic neurotics in primary care who received brief help from a social worker were better adjusted after a year than a similar control group that did not receive such as— sistance, although the assisted group's total consultation rate with general practitioners did not go down. We can summarize the evidence as indicating exposure modalities to be the best approach for phobic-compulsive disorders, with antidepressants for affective complications, 94 and brief counseling for anxiety states, with or without antidepressants; ongoing research here will give further guidelines about drugs. Clearly, less lengthily trained workers can often afford as much relief as M.D.s and Ph.Ds. In fact, most health problems are dealt with in the home by sufferers and their relatives (Kleinman 1981), and professional therapists see only the tip of the iceberg, when self-help has failed. More research is needed into how sufferers and their relatives can be taught to help themselves. Feasible Longitudinal Epidemiological Studies Well-planned prospective longitudinal studies can yield much definitive evidence and answer several questions at the same time (see tables 2 and 3). Given our ignorance of the long-term outcome of anx- iety—depression in primary care populations (the com- monest of all mental health referrals), there seems merit in undertaking systematic 5-year followup of such patients to see what happens to them—whether sufferers are pre- disposed to develop other problems later and whether bet- ter diagnostic categories can be developed for prognosis and treatment. Such a followup study could include a con-— trolled trial of brief supportive and counseling therapies given by less highly trained workers familiar with the prin- ciples of the minimum care necessary for maximum out- come. At this stage, it would be premature to test elaborate therapeutic technologies when the evidence is that simpler procedures might commonly do just as well. Therapists would probably have to make themselves briefly but repeatedly available over long periods, although their total therapeutic time invested would remain small. Anti depressant medication is also worth testing as an additional factor (versus placebo) in a 2x2 design. The converging evidence that supportive social networks protect against adverse effects of stressful life events suggests the need for increased attention to long-term support of vulnerable anxious patients. Such support would be readily available but brief on each occasion over long periods and would include practical advice and emotional reinforcement. Teaching patients to develop their own sup- portive social networks is more problematic. The gradually increasing potency of social skills training might eventually prove helpful, although at the moment it is not impressive 95 Table 2. Issues for research in anxiety disorders 1. Psychosocial and environmental Longitudinal study of anxiety-depression in primary care, identification of long-term outcome patterns and predispositions Simple brief therapies and possibilities for support over long periods Provision of protective social and pharmacological support 2. Personality types Relationships to subsequent adult syndromes Identification of high-risk individuals for prevention Studies of stoical populations 3. Biochemical Do biological clocks and states affect acquisition and subsidence of anxiety? Matching different emotions with different biological markers 4. Genetic Can we identify temperament and physiological fea— tures predicting later morbidity? Possibility for multicenter research on MZ twins reared apart Matching discordant features of MZ twins reared together with discordant elements in their upbringing (Gotwin control method) S. Phyletic (design characteristics of species, wiring diagram) What are "natural" fears and rituals, and when do they develop? Can we map S-R valencies? Are these related to adult syndromes? Do they affect treatment and prevention? 96 Table 3. Possibilities for prevention and management 1. General stress immunization and education, plus followup: Will this decrease anxiety states and posttraumatic stress disorders? 2. Specific stress immunization: Identification of high-risk individuals and situations, management, and followup of their problems: Possibly applicable to separation, social, sexual, and dental anxieties, morbid grief, social skills problems and isolation, phobias (dental, school, agoraphobia), depressive neurosis, sexual dysfunction. in helping patients achieve intimacy as opposed to super- ficial relationships. Many options are possible for the prevention and man- agement of anxiety disorders in children and adults, and the strategies and cost-effectiveness of each approach require careful examination. A tantalizing vista, for ex-— ample, is general stress immunization of children to be carried out both in schools and at home. One would not propose the education given in ancient Sparta but rather encouragement to deal with stresses rather than to avoid them. No general prescription for such encouragement is yet possible, and the optimum approach may vary with different individuals, e.g., those who readily monitor their feelings may respond better to confrontation than will people who usually avoid or blunt the experience of intense emotion. Relationships between early personality and the development of subsequent syndromes need to be examined longitudinally, e.g., whether timidity predisposes to phobias and to anxiety states, and meticulousness to compulsive rituals. This would aid identification of high-risk indi- viduals as target populations for prevention programs. Most people have trivial fears and rituals at some time or another. Certain individuals, however, allow these minor fixations to grow to invade their life space. Maybe we should not think in terms of why phobias and rituals are acquired in the first place, but rather why, once they are acquired, some persons fail to extinguish them. This view 97 resembles that prevalent about the spread of cancer. Cancer cells are regularly produced in the average organ- ism, but they spread only when immunological and other defense systems fail. Maybe stress immunization could strengthen psychological antibodies against stress, if given in the appropriate manner at the right time. What then is the appropriate manner, and what is the right time? It is worth educating parents and teachers about the exposure principle of fear reduction, which has proved so effective in behavioral psychotherapy. Perhaps children can be taught early to manage some of their own discom- forts. One is reminded of the anecdotes about the child- rearing practices of the Senoi, who employed dream interpretation as a special form of exposure therapy at the breakfast table, so to speak. While it is possible to reduce anxiety about dental procedures and surgical operations, it is by no means clear how far improving such specific psy- chological areas leads to a more general tolerance of stress. And it is a major leap from that to producing a de- crease in the incidence of anxiety disorders of various kinds e.g., anxiety states and posttraumatic stress disorder. Spinoff might ultimately come from behavioral treat- ments by exposure and anxiety management. These treat- ments produce lasting relief from phobias and rituals in most patients, but we cannot yet specify the critical fea- tures that separate the sensitizing exposure such patients report they have had during acquisition of their pathology from the habituating exposure they undergo during therapy. Identification of the critical differentiating factors could help us not only refine the treatments by eliminating re- dundant elements, but also broaden applicability of the exposure approach to stress tolerance in general, i.e., de- velop the exposure paradigm into a wider and even more useful coping theory. At the moment this seems a distant possibility, and specific stress immunization appears to be more realistic. It is easier to identify certain high-risk populations, to an- ticipate their problems, and to prevent them than to man- age such problems after they have developed. Much work has already been done on reducing separation anxiety in children admitted to hospitals. Children with dental anx- iety and with school phobia can easily be identified and treated so that these problems are minimized. Widows at high risk for morbid grief after bereavement have been picked out and given bereavement counseling (Raphael 1977), which reduced their morbidity as compared with a 98 control procedure. Another strategy worth considering is identification of children who are social isolates at school for social skills training at an early age. Work on this effort to develop social contacts and support that might reduce later anxiety and depression is already in progress. To lower the incidence of adult sexual dysfunction, we might consider widespread early sexual education given in a matter-of-fact manner, including films of coitus, and counseling about the emotional as well as the physical problems of sex and love. Although such education might prevent much sexual anxiety, it is unlikely to make an im- pact on sexual deviation (paraphilia). An experiment in sexual education of this kind would require cooperation of educational and parental institutions in those areas where such an undertaking would be socially acceptable. The work involved in a controlled study of this kind would be no greater than that in the successful studies of the preven- tion of cardiovascular disease in California towns. If such an educational program began at age 10, a followup of 15 years would be needed before meaningful data about a re- duction in sexual anxieties could be collected. The same long vista is needed for programs to reduce phobias, social anxieties, skills deficits, and depressive neurosis. In health care as a whole, many advances have been made ahead of precise knowledge. Jenner's inoculation against smallpox, Semelweiss's measures against childbirth fever, and Snow's reduction of cholera in the Broad Street area resulted from actions taken before the exact etio- logical agents had been identified. They were short-term experiments that were highly successful, and we tend to forget the many that failed. The risk with longitudinal studies is that, when they are ineffective, failure takes a long time to be revealed. Nevertheless, informed guesses can be made about potentially fruitful areas in which to start. We can already successfully treat several anxiety syndromes by behavioral means, even when the pathology is far advanced. Prevention of such pathology is a logical next step where it is possible to modify probable risk factors. References Agras, W.S.; Sylvester, D.; and Oliveau, D.C. The epidemiology of common fears and phobias. Compre- hensive Psychiatry 10:151-156, 1969. 99 American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 2d ed. Wash- ington, D.C.: the Association, 1968. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 3d ed. Wash- ington, D.C.: the Association, 1980. Andrews, G.; Tennant, C.; Herson, D.M.; and Vaillant, G.E. Life event stress, social support, coping style and risk of psychological impairment. Journal of Nervous and Men— tal Disease 166:307-316, 1978. Berkman, L.F., and Syme, S.L. Social networks, host-resistance and mortality: 9-year followup of Alameda County residents. American Journal of Epi- demiology 109:186-204, 1979. Bird, J. "Management of Psychological Social Problems by General Physicians." Talk to Royal College of Psy- chiatrists, 1981. Boyd, J.H., and Weissman, M.M. Epidemiology of affective disorders. Archives of General Psychiatry 38:1039-1046, 1981. Brown, G.W., and Davidson, S. Social class, psychiatric disorders of mothers, and accidents to children. Lancet 1:378-380, 1978. Brown, G.W.; Davidson, S.; Harris, T.; Maclean, U.; Pollock, S.; and Prudo, R. Psychiatric disorder in London and North Uist. Social Science and Medicine 2:367-3717, 1977. Carey, G., and Gottesman, L.I. Twin and family studies of anxiety, phobic and obsessive disorders. In: Rabkin, J., and Klein, D., eds. Anxiety Revisited. New York: Raven Press, 1980. Cleghorn, J.M.; Peterfly, E.J.; Pinter, E.J.; and Pattee, C.J. Effect of stress on levels of free fatty acid. Cana- dian Psychiatric Association Journal 12:539-548, 1967. Cloninger, C.R.; Martin, R.L.; Clayton, P.; and Guze, S.B. Blind followup and family study of anxiety neurosis: Preliminary analysis of St. Louis 500. In: Klein, D., and Rabkin, J., eds. Anxiety, New Research and Changing Concepts. New York: Raven Press, 1981. Cooper, B. Clinical and social aspects of chronic neurosis. Proceedings of the Royal Society of Medicine 65: 509-512, 1972. Cooper, B., and Sylph, J. Life events in the onset of neurotic illness: An investigation in general practice. Psychological Medicine 3:421-435, 1973. Cooper, B.; Harwin, B.G.; Depla, C.; and Shepherd, M. 100 Mental health care in the community: An evaluative study. Psychological Medicine 5:372-380, 1975. Cox, A. The association between emotional disorder in childhood and neurosis in adult life. In: Van Praag, H.M., ed. Research in Neurosis. Utrecht: Bohn, Scheltema and Holkema, 1976. Curtis, G.; Buxton, M.; Lippman, D.; Nesse, R.; and Wright, J. Flooding in vivo during the circadian phase of minimal cortisol secretion. Comprehensive Psychiatry 17:153- 160, 1976. Dealy, R.S.; Ishiki, D.M.; Avery, D.H.; Wilson, L.G.; and Dunner, D.L. Secondary depression in anxiety disorders. Comprehensive Psychiatry 22:612-618, 1981. Demers, R.Y.; Altamore, R.; Mustin, H.; Kleinman, A.; and Leonardi, D. Exploration of dimensions of illness behav- ior. Journal of Family Practice 11:1085-1092, 1980. Dohrenwend, B.S., and Dohrenwend, B.P. Life stress and psychopathology. In: Regier, D.A., and Allen, G., eds. Risk Factor Research in the Major Mental Disorders. National Institute of Mental Health. DHHS Pub. No. (ADM)81-1068. Washington, D.C.: Supt. of Docs., U.S. Govt. Print. Off., 1981. Dohrenwend, B.P.; Shrov, P.E.; and Mendelsohn, F.S. Non-specific psychological distress and other dimensions of psychopathology. Archives of General Psycho- pathology 37:1229-1236, 1980. Eastwood, M.R., and Trevelyan, M.H. Relationship between physical and psychiatric disorder. Psychological Medi- cine 2:363-372, 1972. Eaton, W.W. Demography and social ecology as risk factors for mental disorders. In: Regier, D.A., and Allen, G., eds. Risk Factor Research in the Major Mental Disorders. National Institute of Mental Health. DHHS Pub. No. (ADM)81-1068. Washington, D.C.: Supt. of Docs., U.S. Govt. Print. Off., 1981. Endicott, J.; Spitzer, R.L.; and Fleiss, J.L. Mental status examination record (MSER): Reliability and validity. Comprehensive Psychiatry 16:285-301, 1975. Finlay-Jones, R., and Brown, G.W. Types of stressful life event and the onset of anxiety and depressive disorders. Psychological Medicine 11:803-815, 1981. Fleiss, J.L.; Gurland, B.G.; and Cooper, J.E. Some contributions to the measurement of psychopathology. British Journal of Psychiatry 119:647-656, 1971. Gallup, G.G., Jr.; Maser, J.D.; Wallnau, L.B.; Boren, J.D; Gagliardi, G.J.; and Edson, P.D. Tryptophan and tonic 101 immobility in chickens. Effects of dietary and systemic manipulations. Journal of Comparative and Physio logical Psychology 91:642-648, 1977. Gerst, M.S.; Grant, I.; and Yager, J. Computer generated self report symptom checklist. Abstract 529 in Pro- ceedings of World Psychiatric Association, Honolulu, September 1977. Gispin, W.H., et al. Neuropeptides, brain and behavior. In: Van Praag, H.M. Research in Neurosis. Utrecht: Bohn, Scheltema and Holkema, 1976. Goldberg, D.P., and Blackwell, B. Psychiatric illness in general practice. A detailed study using a new method of case identification. British Medical Journal 2:439-443, 1970. Gurney, C.; Roth, M.; Kerr, T.A.; and Schapira, K. Bearing of treatment on classification of affective disorders. British Journal of Psychiatry 117:251-255, 1970. Hagnell, L.O. Incidence and duration of episodes of mental illness in the total population. In: Hare, E.H., and Wing, J.K., eds. Psychiatric Epidemiology. London: OUP, 1970. pp. 213-214. Hallam, R.S., and Rachman, S.J. Courageous acts or cour— ageous actors? Personality and Individual Differences 1:341-346, 1980. Harding, T.W.; De Arango, M.V.; Baltazar, J.; Ibrahim, H.H.A.; Ladrido-Ignacio, L.; Srinivasa Murthy, R.; and Wig, N.N. Mental disorders in primary health care: A study of their frequency and diagnosis in four developing countries. Psychological Medicine 10:231-241, 1980. Harpin, H.E.; Liberman, R.P.; Marks, I.; Stern, R.; and Bohannon, W.E. Cognitive-behavior therapy for chroni- cally depressed patients: A controlled pilot study. Journal of Nervous and Mental Disease 170:295-301, 1982. Harris, T. Social factors in neurosis with special reference to depression. In: Van Praag, H.M., ed. Research in Neu- rosis. Utrecht: Bohn, Scheltema and Holkema, 1977. Harvey-Smith, E.A., and Cooper, B. Patterns of neurotic illness in the community. Journal of the Royal College of General Practitioners 19:132-139, 1970. Havik, O.E. Patient information and education as an intervention in physical illness. Review of outcome re- search. Chapter in proceedings of conference on Cost-— Efficiency Analysis in Clinical Psychology, Geila, Nor-— way, December 1979. Henderson, S. Social relationships, adversity and neurosis: 102 Prospective observations. British Journal of Psychiatry 138:391-398, 1981. Henderson, S.; Byrne, D.E.; Duncan-Jones, P.; Adcock, S.; Scott, S.; and Steele, G.P. Social bonds in the epidemi- ology of neurosis: A preliminary consideration. British Journal of Psychiatry 132:463-466, 1978a. Henderson, S.; Duncan-Jones, P.; McAuley, H.; and Ritchie, K. The patient's primary group. British Journal of Psychiatry 132:74-86, 1978b. Henderson, S.; Byrne, D.E.: Duncan-Jones, P.; Scott, R.; and Adcock, S. Social relationships, adversity and neu- rosis. British Journal of Psychiatry 136:574-583, 1980. Hoeper, E.W.; Nycz, G.; Regier, D.; and Goldberg, I. Prevalence of mental disorder in primary care. Abstract 37A, p. 181, American Psychiatric Association, Chicago, May 1979. Horowitz, M.J. Stress Response Syndromes. New York: Jason Aronson, 1976. Horowitz, M.J.; Wilner, N.; Kaltreider, N.; and Alvarez, W. Signs and symptoms of posttraumatic stress disorder. Archives of General Psychiatry 37:85-92, 1980. Horowitz, M.J.; Krupnick, J.; Kaltreider, N.; Wilner, N.; Leong, A.; and Marmar, C. Initial psychological response to parental death. Archives of General Psychiatry 38: 316-323, 1981. Jenkins, C.D. Psychosocial modifiers of response to stress. Journal of Human Stress 5:3-15, 1979. Johnstone, E.C.; Cunningham-Owens, D.G.; Frith, C.D.; McPherson, K.; Dowie, C.; Riley, G.; and Gold, A. Neu- rotic illness and its response to anxiolytic and anti- depressant prescription. Psychological Medicine 10: 321-328, 1980. Katan, W.; Kleinman, A.; and Rosen, G. Depression and somatization: A review. Part 1. American Journal of Medicine. 72:127-135, 1982. Kendell, R.E. The stability of psychiatric diagnoses. British Journal of Psychiatry 124:352-356, 1974. Kovacs, M.; Rush, A.J.; Beck, A.T.; and Hollon, S.D. Depressed outpatients treated with cognitive therapy or pharmacotherapy: A one-year follow-up. Archives of General Psychiatry 38:33-39, 1981. Kleinman, A. Patients and Healers in the Context of Culture. Berkeley: University of California Press, 1981. Lader, M.H. Physiological research in anxiety. In: Van Praag, H.M., Ed. Research in Neurosis. Utrecht: Bohn, Scheltema and Holkema, 1976. 103 Lader, M.H., and Marks, I.M. Clinical Anxiety. London: Heinemann Medical and Academic Press, 1971. Levi, L. Stress and distress in response to psychosocial stimuli. Acta Medica Scandinavica 191, Supplement No. 528, 1972. Lewine, R.J.; Watt, N.F.; Prentky, R.A.; and Fryer, J.H. Childhood behavior in schizophrenia, personality dis- order, depression and neurosis. British Journal of Psy- chiatry 132:347-357, 1978. Marks, IM. Fears and Phobias. London: Heinemann Medical and Academic Press, 1969. Marks, 1.M. Living with Fear. New York: McGraw-Hill, 1978. Marks, I.M. Care and Cure of Neuroses. New York: John Wiley, 1981. Marsland, D.W.; Wood, M.; and Mayo, F. Content of family practice: Data bank for patient care, curriculum and research in family practice. Journal of Family Practice 3:25-68, 1976. Mayo, P.R. Self-disclosure and neurosis. British Journal of Social and Clinical Psychology 7:140-148, 1968. Mawson, D.; Marks, I.M.; and Ramm, I. Clomipramine and exposure for chronic obsessive-compulsive rituals: IIL Two year follow-up and further findings. British Journal of Psychiatry 140:11-18, 1982. Miller, P.M., and Ingham, J.G. Friends, confidants and symptoms. Social Psychiatry 11:51-58, 1976. Morley, S. Incubation of avoidance behaviour, strain dif- ferences in susceptibility. Behaviour Research and Therapy 15:365-367, 1977. Nemeroff, C.B., and Prange, A.J. Peptides and psycho- neuroendocrinology. Archives of General Psychiatry 35:999-1010, 1978. Noyes, R.; Clancy, J.; Hoenk, P.R.; and Slymen, D.J. Prognosis of anxiety neurosis. Archives of General Psy- chiatry 37:173-178, 1980. Noyes, R. Paper to NIMH workshop on Psychopharmacology of Anxiety Disorders. Washington, D.C.: 1982. Paykel, E.S.; Rowan, P.R.; Parker, R.R.; and Bhat, A.V. Response to phenelzine and amitriptyline in subtypes of outpatient depression. Archives of General Psychiatry 39:1041-1049, 1982. Prusoff, B., and Klerman, G. Differentiating depressed from anxious neurotic outpatients. Archives of General Psychiatry 30:302-309, 1974. Raphael, B. Preventive intervention with the recently be- 104 reaved. Archives of General Psychiatry 34:1450-1454, 1977. Reinhardt, R.F. The outstanding jet pilot. American Journal of Psychiatry 127:732-736, 1970. Riley, V. Psychoneuroendocrine influences on immuno- competence and neoplasia. Science 212:11009-11019, 1981. Salmons, P., and Sims, A. Smoking profiles of patients ad- mitted for neurosis. British Journal of Psychiatry 139:43-46, 1981. Schapira, K.; Roth, M.; Kent, T.A.; and Gurney, C. Dif- ferentiation of anxiety states from depressive illnesses. British Journal of Psychiatry 121:175-181, 1972. Schwab, J.J.; Bell, R.A.; Warheit, G.J.; and Schwab, W.B. Social Order and Mental Health. New York: Brunner- Mazel, 1979. Stravynski, A.; Marks, LLM.; and Yule, W. Social skills problems in neurotic outpatients. Archives of General Psychiatry 39:1378-1385, 1982. Shields, J. Genetic factors in neurosis. In: Van Praag, H.M., ed. Research in Neurosis. Utrecht: Bohn, Scheltema and Holkema, 1976. Silberfeld, M. Psychological significances and social sup- ports. Social Psychiatry 13:11-17, 1978. Sims, A., and Prior, P. The pattern of mortality in severe neurosis. British Journal of Psychiatry 133:299-305, 1978. Singerman, B.A; Riedner, E.; and Folstein, J. Emotional disturbance in hearing clinic patients. British Journal of Psychiatry 137:58-62, 1980. Sloane, R.B.; Staples, F.R.; Cristol, A.H.; Yorkston, N.J.; and Whipple, K. Psychotherapy Versus Behavior Therapy. London: Harvard University Press, 1975S. Strupp, H.H., and Hadley, S.W. Specific vs. nonspecific factors in psychotherapy. Archives of General Psy- chiatry 36:1125-1136, 1979. Suomi, S.J.; Kraemer, G.W.; Baysinger, C.M.; and Elizio, R.D. Inherited and experiential factors associated with individual differences in anxious behaviors displayed by Rhesus monkeys. In: Klein, D.F., and Rabkin, J., eds. Anxiety: New Research and Changing Concepts. New York: Raven Press, 1981. Tennant, C.; Bebbington, P.; and Hurry, J. Short-term evidence of neurotic disorders in the community: Re- mission clinical factors and ‘'neutralizing' life events. British Journal of Psychiatry 139:213-220, 1981a. Tennant, C.; Hurry, J.; and Bebbington, P. Short-term 105 evidence of neurotic disorders in the community: Demographic clinical predictors of remission. Australian and New Zealand Journal of Psychiatry 15:111-116, 1981b. Theorell, T. Selected illnesses and somatic factors in relation to two psychosocial stress indices: A prospective study on middle-aged construction building workers. Journal of Psychosomatic Research 20:7-20, 1976. Torgersen, S. The nature and origin of common phobic fears. British Journal of Psychiatry 134:343-351, 1979. Vaillant, G.E. Natural history of male psychologic health: Effects of mental health on physical health. New England Journal of Medicine 301:1249-1254, 1979. Vasquez-Barquero, J.L.; Munoz, P.E.; and Madoz-Jarregui, V. Interaction between physical illness and neurotic mor— bidity in the community. British Journal of Psychiatry 139:328-335, 1981. 106 Chapter 6 PSYCHOLOGICAL RESPONSES TO STRESS: THE STRESS RESPONSE SYNDROMES Mardi J. Horowitz, M.D. Introduction Our collective aim is to help prevent stress-related disorders. A series of investigations within the University of California, San Francisco, Center for the Study of Neu- roses, an NIMH-funded Clinical Research Center, has aimed at secondary and tertiary prevention of neurotic level responses to serious life events. To achieve these prevention and treatment goals, we have joined in the widespread effort to understand human responses to psy- chological stress. Our strategy is to understand the processes of working through serious life events well enough to inhibit the for- mation of symptoms as a route of response and to facili- tate healing and adaptation. This requires combined efforts to study what happens and to gauge the effects of inter- ventions. A feasibility study on secondary prevention of posttraumatic stress disorder has been designed (see chap- ter 8). For some years, our center has studied the signs and symptoms of stress response syndromes in clinical, field, and experimental contexts. In a research clinic, we had developed a brief psychoanalytic psychotherapy with cer— tain special features for such disorders. When the outcome of such treatment showed it to be useful and safe, we 107 planned a study of early interventions based on these sa- lient techniques. The design of that study involved the cooperation of a large corporation, an insurance company that agreed to allow up to 4,000 employees to participate during working hours. We aimed to include a unique feature, the collection of baseline data before the occurrence of the serious life event. This would be a vital new addition to our archives of data on patients and field subjects. Those employees who gave written consent to be sub-— jects within the study would, at prescribed intervals, com- plete specialized rating scales indicating life events, stress—specific symptoms anchored to those events, general symptoms, and social functioning. From these data, we would select individuals who had specific kinds of serious life events, identify those who had symptomatic levels of usual responses (intrusive and avoidant), and invite them to come for a clinical interview. Through the application of our evaluation procedures, we would select a group whose distress from the serious life event was confirmed by clinical judgment. These individuals would then be assigned randomly into a treatment group and an untreated sample whose members received usual community care. Evaluation of both groups over time would assess morbidity and well-being, including work production and medical health care utilization. The study design is discussed in chapter 8, along with brief comments on a few design issues that we confronted. The first issue is how to study stress-related disorders—by degree of cumulative stress from multiple life events (along with personal dispositional features and social con- textual variables), by experience of discrete events, or both. If discrete life events are selected, how catastrophic should they be? The second concerns the issue of how to classify normal and abnormal responses to a specific life event. When can normal stress responses contribute to morbidity and even mortality, and when can pathological reactions have additional ill effects? Cumulative and Specific Stressor Events Cumulative stress is the concept of strain imposed on an individual by multiple factors. We are all most familiar with it in terms of the concept of defining presumptive 108 levels of distress that might be caused by particular life events and evaluating these levels through the use of life events questionnaires. The problem with the cumulative approach is its generality. This leads to a diffuse inter- vention effort aimed vaguely at improving coping and does not produce a well-defined theoretical construct. We therefore decided to select discrete life events as the bases for choice of subjects. This had the great ad- vantage of allowing the response to a specific event to be tracked over time and led to the question of which specific life events to choose. The more catastrophic the event, the more one would have to study only small groups of homog- eneous subjects or the less likely one would be to obtain prospective data. We thus selected more typical serious life events. These diverse events would then be examined in terms of specific response levels, using our Impact of Event Scale for preliminary subject self-ratings. Previous research on such sequelae as signals of intrusive thinking has indicated, through measures of cumulative presumptive stress, that a higher level of response to the specified discrete life event is likely if the individual has experienced many other life events. Such data are presented here in detail. They in- dicate the need to examine the complete individual, even when the context is his or her response to specific life events. Our pilot study also indicated the prevalence of displacement of emphasis from one event to another, with a tendency to explain distress by the most recent in a series of life experiences. Normal and Pathological Responses to Stressful Life Events A second thread to weave into consideration is the range of responses and the ways in which they are made mani- fest. Due to the unitary nature of human psychology, there is no particularly clear dividing point between normal and abnormal signs of adaptive or maladaptive responses. In planning prevention studies, therefore, we would be wise to use threshold concepts in order to detect those who would be helped by interventions to assist their intrinsic coping processes. This leads to the question of the kind of psychiatric dis— orders that might be precipitated by stressful life events or 109 possibly prevented by interventions. Posttraumatic stress disorders (PTSD) are defined in DSM-III and, because of their clear linkage to life events, would be central to any such effort. But while intrusive and numbing phases are the cardinal manifestations for PTSD, the whole range of anxiety and affective disorders and possibly even schizo- phreniform psychotic reactions (e.g., in borderline per- sonality disorders) may be found to occur more frequently or intensely in persons strained by psychological stress factors. Also, intrusive thinking and feeling are prominent in diverse disorders; having disorders is itself a stressful event! These disorders are precipitated by such life changes as serious separations from loved ones, work, school, and home. Nonetheless, posttraumatic stress disorders may lie at the heart of the task, and our aim may be defined by considering how to prevent them from becoming chronic and how to determine the presence of other DSM-III diagnoses. Studying any population will require systematic formulation of the range of stress response syndromes, PTSD in particular, as well as other disorders, especially anxiety, depression, and brief psychotic reactions. It is the consensus of investigators that a structured clinical inter- view is the vehicle of choice. The Schedule for Affective Disorders and Schizophrenia (SADS), one such approach, is favored by our own group. Rendering it appropriate to our focus on prevention involved developing our own research diagnostic criteria for PTSD to apply to previously defined disorders. In doing so, we have noted a problem in the wording of DSM-III for these disorders. It concerns not the nature of the syndrome (defined in table 10 in the accompanying article) but the situational context: "The stressor producing this syndrome would evoke significant symptoms of distress in most people, and is generally outside the range of such common experiences as simple bereavement, chronic illness, business losses, or marital conflict." This wording allows two interpretations. In one inter- pretation, posttraumatic stress disorders might follow only an unusually catastrophic stress event, such as the death of a spouse by flood (rather than a heart attack). In another reading, one would understand the wording to mean that just because most people have periods of intrusion and numbing after a serious loss such as the death of a spouse, all individuals should not necessarily be given this diagnosis. The PTSD diagnosis would be made only for 110 persons for whom the intrusion and avoidant experiences became extremely intense, overwhelming, or persisted for long periods of time without resolution. We favor the latter position. An attendant diagnostic problem concerns how the patients present their complaints. This will depend, in part, on how subjects are assembled and how a structured in- terview is given. A widow who lost her husband after a heart attack would probably be diagnosed as having a PTSD, an anxiety disorder, and a depressive disorder during different phases of her response. (This may not be irrele- vant to treatment considerations, since interventions might vary according to the specific pathology at the moment or might also relate to the underlying psychological processes that led to these varied symptomatic states of mind.) But it should be noted that the criteria for treatment are not necessarily presenting signs and symptoms—all that is included in DSM-III—but underlying causal factors. For background reading, it seems wise to provide a de- tailed examination of intrusive and avoidant episodes, the cardinal symptoms of posttraumatic stress disorder as de- scribed in DSM-III. These episodes occur in the other di- agnoses, as phobias, obsessions, despairing ruminations, or as overwhelming pangs of rage, remorse, sadness, and fear. The following material, which has been published as a book chapter, is the collective work of many collaborators. It is personal, in that it covers a scientific journey that began outside the domain of stress research according to its present boundaries, with the clinical investigation of unbidden images. These subjective episodes of intrusive experiences formed a pathway this investigator chose to follow into the study of normal responses to discrete stress events and eventually into the study of the treatment of persons with stress disorders. The text that follows, then, deals with the theme of in- trusive and avoidant experiences. These are the two car- dinal symptoms that define the diagnostic criteria for posttraumatic stress disorders in DSM-III (listed in detail in table 10). 111 STRESS RESPONSE SYNDROMES AND THEIR TREATMENT* Mardi J. Horowitz, M.D. The compulsive repetition of traumatic perceptions is an observed phenomenon with a long history. Breuer and Freud (1895) greatly advanced the modern conceptualization of this pattern. Freud (1953) reemphasized and generalized their observation that "hysterics suffer from reminis— cences," a form of compulsive repetition, and Schur (1976) gave a more contemporary summary. With an awareness of this literature, I began a clinical investigation of the phenomenon of unbidden images. But, as I made my own observations, I was surprised by the intensity of the phe- nomenon. It seemed possible that the intrusive quality of ideas related to a previous life event constituted a specific index of distress. Unbidden Images: Clinical Observations Before undertaking a systematic series of investigations of stress response syndromes, [I had been interested in certain abnormalities of thought as they were made mani- fest by visual images. The central phenomenon in this mode *This line of work on stress, especially on the phenomenon of intrusion, has been supported by a research scientist develop- ment award from NIMH (K3-22-573), a stress research grant from NIMH (MH 17373, 01-07), a stress and coronary heart dis- ease grant from NHLBI (HL 17580, 01-06), a program project grant from NICHD (AGOQOO2), and funds granted to the Center for the Study of Neuroses as a Clinical Research Center of NIMH (MH 30899). 1 would also like to thank the following colleagues for their assistance: Robert Wallerstein, Nancy Wilner, Nancy Kaltreider, Charles Marmar, Janice Krupnick, Phyllis Cameron, William Alvarez, Tony Leong, Matthew Holden, Kathryn DeWitt, and Daniel Weiss. Reprinted with permission from Goldberger, L., and Breznitz, S., eds. Handbook of Stress: Theoretical and Clinical Aspects. New York: The Free Press, 1982. Copyright 1982. 112 of representation seemed to be the loss of voluntary con- trol over the flow of information. Intrusiveness was of special interest because it was a quality that ranged across phenomenology and diagnosis, hallucinations, illusions, nightmares, obsessive thought images, flashbacks, and al- tered body images. Careful explanation of such unbidden images would shed some light on visual thinking, regulation of thinking, and failures of defensive processes. A series of clinical investigations were begun that ranged from psychoanalysis of persons who complained of unbidden images to study of psychiatric patients and to people who had bizarre experiences associated with tem- poral lobe epilepsy. The contents, genesis, changes, and derivatives of recurrent unbidden imagery experiences were examined, with special interest in change: cessation of a recurrent image, alteration of its contents, and appearance of greater volitional control. The details are published elsewhere (Horowitz 1978). What is important here is the conviction, based on obser— vations, that the contents of unbidden images are quite frequently based upon previous perceptual realities. This conviction was startling because I expected, given the patient populations selected and my training in theory, that these contents would emerge mainly as fantasy elabora- tions based on current intrapsychic conflicts. The contents did, of course, contain fantasy elaborations and served var— ious impulsive and defensive purposes, but the frequency of contents repeating a traumatic perception of a serious life event that had actually occurred was impressive. Experimental Work on Intrusive Thinking After Perception of Stress-Inducing Stimuli It seemed that the connection between perception of stress—-inducing stimuli and subsequent experiences of in- trusive thinking, especially in the same modality as the perception, might be strong enough to be called a general human response tendency. If this were so, the association between these phenomena could serve as an index of how stressful a set of people found a set of life experiences. Change in degree of distress could be monitored by fol- lowing this index and noting the factors that mediated the process of working through a reaction to a serious life 113 472-367 0 - 85 - 5 event. However, review of the voluminous experimental work on laboratory induced stress revealed an emphasis on biological response variables. While there were many data on ratings of emotional responses, such as degree of fear or anger after a stress—inducing stimulus, there were few re- ports on subjective experiences. These were considered by some investigators as too fragile for scientific measure- ment and by others as too epiphenomenal to be important; yet, Antrobus and Singer (1969) had done very relevant work with such variables, and Lazarus (1966) and Janis (1969) had placed strong theoretical emphasis on cognitive variables. In order to test the hypothesis that intrusive and repet- itive thought is a general stress response tendency, I de- signed an experimental investigation that used films, a well-studied and replicable laboratory device for observing visual stress events. Intrusive and repetitive thoughts were defined in operational terms and then quantified by two methods: subjective self-report and content analysis of reported episodes of consciousness (Horowitz 1975). A series of investigations was conducted. In a prototypical experiment, groups of subjects saw a stress film and a neutral contrast film in counterbalanced order. Measure- ments were taken at baseline, poststress, and postneutral periods. Over 300 subjects were selected by systematic sampling of various populations. The mental set given to subjects varied in terms of specific and general instruc- tions on how to report episodes of consciousness, normal and abnormal slants as to the meaning of having intrusive images, and the context of attention deployment. The re— sults confirmed the hypothesis. For example, a cross-experimental data analysis was performed using the Finn (1970) Multivariate Analysis of Variance program for nonorthogonal designs. The change in conditions (neutral or stress) had a significant effect on the levels of reported quality of intrusiveness and repeti- tiveness of original stimuli. These two factors, one of form and one of content, correlated significantly and positively (r =.51, p<.001, n = 133). Table 1 reports adjusted means for the effects of all factors and conditions on intrusions and film references. The population consisted of 133 subjects in various ex- periments who each had data on the variable in all condi- tions (baseline, neutral, and stress). Positive scores for the adjusted means in table 1 indicate the number of intrusions per subject above expectation, while negative scores in- 114 SIT Table 1. Combined means as adjusted according to report length Intrusions Film references Factor n Base Neutral Stress Base Neutral Stress Population Military inpatients 23 -.44 -.16 41 -1.69 -.88 -.08 Civilian students 82 -.83 -.40 1.38 -2.14 -.23 1.51 Health students 28 -.50 -.11 .39 -1.53 .08 1.61 Sex Men 99 -.66 -.33 +35 -1.81 -.13 -.98 Women 34 -.80 -.20 2.33 -2.29 -.69 2.05 Instructional demand Normal 24 -.72 -.30 1.73 -2.26 -.53 2.44 Abnormal 25 -.63 .02 1.79 -1.96 —25 1.65 Specific 16 -.49 .01 .93 -1.70 .07 0.86 General 68 -.76 -.49 .48 -1.86 -.28 .78 Order Stress 1st S1 -.67 -.20 57 -1.82 .16 .86 Neutral 1st 82 -.71 -.36 1.27 -2.00 -.55 1.50 Film "Subincision" 110 - 13 -.33 1.13 -1.98 -15 1.53 Woodshop 23 —.44 -.16 41 -1.69 -.88 -.08 Overall 133 -.69 -.30 1.00 -1.23 -.29 1.23 Source: Horowitz (1975). dicate lower than expected levels. At the p < .0S cutoff level, only the change in conditions, from neutral to stress— ful, exerted a significant effect on both intrusions (MS = 57.5; df = 2; F = 20.7; p < .001) and film references (MS = 146.5; df = 2; F = 49.1; p < .001). Population differences (health sciences students, college students, and military health personnel), film order (stress or neutral first), sex of subjects, and instructional demands (general and specific; abnormal or normal slants) did not exert a significant effect. The significant condition effect for intrusions can be accounted for by the stress film. Baseline film references were, of course, especially low because no film had yet been seen, and only occasional anticipatory remarks were scored. The neutral and stress condition film reference levels were significantly different. While the large-scale analysis of variance found condi- tion to be the only variable to exert a significant effect on levels of intrusion and film reference, the variance asso-— ciated with instructional demand and sex approached sig- nificance (p < .09). Analyses of variance and covariance were performed on film references and intrusions only as scored in the stress condition to observe more closely the effect of all other factors except condition on these two variables. The entire group of 302 subjects had yielded data in the stress condi- tion and was, therefore, used for these analyses. Table 2 presents the results of the two analyses for in- trusions. The analysis of variance used intrusion means both adjusted and unadjusted for report length, while the analysis of covariance used number of words in a subject's report as the covariate of intrusions. This analysis and an additionally reduced analysis of covariance found sex and population to have a significant effect on intrusions. Female subjects scored the highest levels of intrusions in the stress condition; civilian men scored higher than mili- tary nonpatients (health personnel) (see table 3). Film ref- erence data did not show a significant difference between the sexes or across population groups, although military psychiatric patients reported more film repetitions after the stress film than did military nonpatients. These data can be interpreted as either a disinclination of military nonpatients and, to a lesser extent, civilian men to disclose mental contents they might construe as weak or a greater tendency on the part of women and psychiatric patients to develop such response to stress. 116 LTT Table 2. Intrusions for all factors in the stress condition (n = 302) Using unadjusted means Using adjusted means Source? af MS F P MS F P ANOVAb Population 3 7.22 2:55 .06 4.96 1.85 <.14 Sex 1 37.34 13.18 <.001 35.10 13.09 <.001 Error 27 2.83 2.68 ANOC (reduced model) Population 3 13.97 5.14 <.002 Sex 1 48.03 17.69 <.001 Error 294 2.72 Source: Horowitz (1975). 2 Analysis of variance is indicated by ANOVA; analysis of covariance, by ANOC. bInstructions, order, film, and residual were not significant. 811 Table 3. Average level of intrusions per subject in stress condition? Military Military Civilian Health inpatients nonpatients students students Estimated Population means Men 41 -.11 .64 44 27 Women 31 2.06 .78 1.64 Estimated 41 -.09 1.24 37 n Men S1 92 64 27 234 Women 0 4 47 17 68 Total 51 96 111 44 302 Source: Horowitz (1975). 2Adjusted combined means. Intrusions Tend to Correlate with Degree of Reported Stress In the stress condition, persons who rated themselves high on negative emotions tended also to report high levels of intrusions. The adjusted intrusion scores in 133 subjects who did the same affect report measure during the stress condition correlated significantly and positively with a composite of negative effects (r = .27, p < .001) and sig- nificantly and negatively with a composite of positive affects (r = -.16, p. < .05). The highest correlation with specific individual affects was with pain (r = .38; p < .001) and surprise (r = .35, p < .001). These correlations and others are found in table 4. Of course, with large numbers of subjects, some low levels of correlation (e.g., r = -.16) may reach statistical significance but indicate only a small effect. Table 4. Correlation of intrusions with affective variables in stress condition (n = 133) Correlations with Affect adjusted intrusions Interest -.12 Pleasantness -.202 Happiness -.02 Positive composite ~.16® Sadness .202 Anger .05 Disgust .14b Contempt .07 Surprise .35¢ Fear .13 Pain .38€ Nervousness 12 Negative compositive 27¢ Source: Horowitz (1975). ap < .01 bp <.05 Cp < .001 A group of 77 subjects also rated themselves after the stress film on 1-100 "thermometer" type scales for emo- 119 tional and physical stress, following the method of Stevens (1966). Both scales correlated significantly and positively with intrusion levels. For emotional stress the correlation was r = .39, p < .01; for physical stress, r = .34, p < .01. A subsequent experiment contrasted the effects of dif- ferent types of stressful films (Horowitz and Wilner 1976). The earlier series had involved two films, both depicting events of bodily injury, which tended to induce feelings of fear. In this later experiment, a film was included that showed no bodily injury, but rather the abandonment of a young boy by his father. This tended to induce feelings of anger and sadness. Both stress films, and an arousing erotic film, led to significantly higher levels of intrusive and re- petitive thinking than did a neutral control film. Of in- terest here, this experiment included an inventory of life events for each subject (Horowitz et al. 1974). Thus, pre- experiment experiences could be compared with reactions to witnessing the stressful stimuli in the study. There was no correlation between response to the films and an overall presumptive stress score obtained by sum- ming weights assigned to each experienced item on our life events questionnaire. People who had had the highest fre- quency of deaths and separations in the past experienced the most sadness and fear after the separation film (r = .44, p < .0S for sadness; r = .41,, p < .0S for fear). Field Studies Since the laboratory investigations supported the im- portance of intrusive thinking as one index of the degree of stress after a recent life event, it seemed valuable to de- sign a self-rating instrument that would allow for repeated measurement of such variables (Horowitz et al. 1979). Open-ended interviews were conducted with people who had experienced diverse stress-inducing life events. We hoped to add to the clinical experiences already assembled in the study of unbidden images and in everyday clinical practice. The sample included patients who had sought help after a recent loss or injury or who had sustained such events during the course of extended psychoanalysis or psychoanalytic psychotherapy (allowing for some prospec- tive understanding of the mix between character and cir- cumstances). Items for a self-report instrument were assembled from written reports and the clearest wording of 120 redundant experiences was selected. The list that evolved in this way contained experiences of a particular quality, such as intrusiveness, worded so that they might apply to any event. To anchor the qualities of experience to a par- ticular context, the life event specific to each person was entered at the top of the form and served as a referent for each of the statements on the list. Clinical experience suggested that the various items be divided into two subgroups, intrusion and avoidance. The goal was a scale that would provide subscores for these response sets, as well as a total subjective stress score. Over a period of several years various forms of this item list were given both to psychotherapy patients with stress response syndromes and to nonpatient volunteers exposed to serious life events. The wording and the format were revised. Through examination of the data it became evident that although we had asked for separate ratings of how often during the past week a particular experience had occurred, and what the peak intensity of that episode had been, the scores for frequency and intensity of episodes were rela- tively similar. In other words, the experience of a single, very intense intrusive image was similar in effect to mul- tiple experiences of less intense and only mildly intrusive images. Therefore, although subjects were asked to respond to both variables in a study of the refined impact of event scale, only the higher score of either frequency or intensity was used as the indicator of the magnitude of the item (Horowitz et al. 1979). In its form after pilot studies, the scale consisted of 20 items: 9 items described episodes of intrusion; 11 described episodes of avoidance. This scale was given to 66 adults who had sought psychotherapy at the University of Cali- fornia outpatient department in response to a serious life event. These patients were referred to a clinic for spe- cialized treatment of stress response syndromes. This clinic accepted patients with neurotic levels of psycho- pathology; those persons referred elsewhere either had overt psychoses, drug dependence, or alcoholism or were unwilling to give written consent to research review of their records. The serious life event that had preceded the develop- ment of a stress response syndrome varied in this group. About half the patients (n = 34) had experienced bereave- ment, the remainder had suffered personal injuries re- sulting from accidents, violence, illness, or surgery. When 121 scales completed by bereaved subjects were compared with those of subjects who had incurred personal injury, en- dorsements of items were found to be relatively similar. The subjects were 16 men and 50 women between 20 and 75 years of age, with a mean age of 34. All items were endorsed frequently. Those most often endorsed, "Things I saw or heard suddenly reminded me of it" and "I got waves or pangs of intense or deep feelings about it," were acknowledged by 85 percent of the subject sample (see table S). Even the item with the lowest en— dorsement, "It seemed to me that I was reacting less than would be expected," was acknowledged by 38 percent of the 66 subjects. All six of the most frequently reported items had a mean weighted score of 3 or more, indicating that as a group these subjects experienced such episodes at a high level of frequency or intensity. This occurred even though the particular stressful event usually had been ex- perienced several months earlier (mean of 5.5 months), while the time frame for endorsement of any scale item was limited to the past week. The mean scores for each item were calculated sep- arately for men and women, as well as for the total group. The sexes differed significantly in frequency or intensity on only 3 of 20 items. In each instance, women indicated a higher level of endorsement. As shown in table 5S, the primary and secondary clusters included 15 of the 20 items, with two smaller residual clusters containing S items. The primary cluster contained items from the clinically derived intrusion subset, while the second cluster was composed of clinically derived avoidance items. This finding, added to the logic with which items were usually selected, supported the use of the intrusion and avoidance classifications. Subsequently, the number of items was reduced by se- lecting only those that empirically clustered and had sig- nificant item to subscale correlations beyond the .01 level of significance. Next, comparison of scores derived by frequency and by intensity indicated a degree of similarity that made dual response for each item unnecessary. We retained only the frequency variable since subjects seemed able to score it more accurately than the intensity vari- able. Finally, wording of a few items was slightly modified further to reduce ambiguity. After reduction of the scale to the 15 most powerful items, the split half reliability of the total scale was high (r = 0.86). Internal consistency of the subscales, calculated 122 Zl Table 5. Impact of event scale: Frequency and mean of positive report and cluster analysis of items (n=66) Item to % subscale Logical Item Endorsement Mean (SD) r cluster? Cluster 1 I had waves of strong feelings about it. 88 3.8 (1.9) .57P I Things I saw or heard suddenly reminded me of it. 85 3.7 (1.9) 570 I I thought about it when I didn't mean to. 76 3.3 (2.2) .45D I Images related to it popped into my mind. 76 3.2 (2.2) .39b I Any reminder brought back emotions related to it. 76 3.0 (2.1) .62b I I have difficulty falling asleep because of images or thoughts related to the event. 64 2.6 (2.4) .35b I I had bad dreams related to the event. 44 1.7 (2.2) .41b I Cluster 2 I knew that a lot of unresolved feelings were still there, but I kept them under wraps. 71 3.0 (2.2) .46D A I avoided letting myself get emotional when I thought about it or was reminded of it. 70 2.8 (2.1) .62 b A I wish to banish it from my store of memories. 65 2.8 (2.3) .49b A I made an effort to avoid talking about it. 61 2.2 (2.0) .59b A My emotions related to it were kind of numb. S59 2.1 (2.1) .42b A I felt unreleased about it, as if it hadn't happened or as if it wasn't real. 58 2.2 (2.3) .48b A vii Table 5.— Continued Item to % subscale Logical Item Endorsement Mean (SD) r cluster? I stayed away from things or situations that might remind me of it. 53 2.2 (2.3) .57° A I didn't let myself have thoughts related to it. 50 1.8 (2.2) .65P A Cluster 3 I kept wondering why it had to happen to me or to persons near me and not someone else. S9 2.6 (2.3) .18 I I used alcohol, drugs, or a lot of activity to help me forget. 54 2.0 (2.2) .29 It seemed to me that I was reacting less than would be expected. 38 1.3 (2.0) .20 Cluster 4 I found myself almost waiting for something like that to happen again. 44 1.6 (2.1) .29 I I found myself making plans and decisions which were inappropriate in light of the event. 41 1.5 (2.0) .08 A 2l=intrusion item; A=avoidance item. p .01. Source: Horowitz et al. (1979). Copyright 1979 by The American Psychosomatic Society, Inc. Reprinted with permission. using Cronbach's alpha, was also high (intrusion = .78, avoidance = .82). A correlation of .42 (p < 0.0002) between intrusion and avoidance subscale scores indicated that the two subsets were associated but did not measure identical dimensions. In a later sample, test-retest reliability and sensitivity were found to be good (Horowitz et al. 1979). In a subsequent series of studies, people with diverse life events were asked to fill out the revised impact of event scale.* Some were prospective patients at a special clinic for the treatment of individuals with posttraumatic stress disorders or adjustment reactions to serious life events. For these patients, the specific event was used as the referent for their impact of event scale. Other field populations were selected because of their participation in a preselected event such as death of a parent; we also studied medical and physical therapy students shortly after they had participated for the first time in cadaver dis- section. Other respondents were women who had received news of the possibility of cancer and were about to undergo a breast biopsy. Each event for each group was the refer— ent for their impact of event scales. Still other groups consisted of patients with head or neck cancer who, when they came for treatment, related the scale to the diagnosis of their disease some months earlier, and the people closest to these cancer patients, who filled out the scale as related to their recent experiences concerning the news that their relative had cancer (see table 6). The comparative impact of diverse events could be as- sessed on this kind of scale because the instrument focuses on qualities of experience that allow it to be anchored for reference to any specific event. As might be expected, the least stressful of the events was cadaver dissection. The experience of possibly having cancer was much more dis— tressing, even though the first news of the disease in these people was more remote in time than the experience of the cadaver dissection for the health students. Respondents with the highest levels of distress as evidenced by intrusive and avoidant experiences were people seeking help for psychological stress associated with a serious life event. The levels of intrusion and avoidance were significantly different across the groups by ANOVA and Scheffe con- trast statistical examinations. The means ranged from a *These studies were carried out in collaboration with Frances Cohen, Barbara Kent, Nancy Wilner, Anthony Leong, Nancy Horowitz, and Michael Paglarouli and will be presented in forth- coming publications. 125 9Z1 Table 6. Populations n Time since event (weeks) Age Groups Life event Total Women Men mean Mean (SD) Stress clinic patients Loss to self or others 77 67 10 35.3 24.2 (33.6) Bereavement field Death of a parent 36 18 18 37.9 7.6 (2.4) subjects Surgical patients Breast biopsy for 68 68 0 41.8 1.0 (0) possible cancer Medical patients Diagnosis of cancer of S54 17 37 56.5 4.8 (3.1) head or neck beyond simple surgical cure Collateral of medical Above in related 36 22 14 53.9 4.9 (3.6) patients person Medical students First exposure to 69 23 46 23.6 2.0 (0) and dissection of cadaver Physical therapy First exposure to 29 20 S 25.3 2.0 (0) students and dissection of cadaver high of 23 for intrusion levels among the stress clinic help-seeking patients, the majority of whom had post- traumatic stress disorders, to a low of 2.4 for the physical therapy students after cadaver dissection. Equivalent avoidance levels ranged from 19.4 in the stress clinic patients to 2.1 for the physical therapy students. But these are mean levels, and a clinically more relevant consider— ation has to do with how many people have high, medium, or low distress. In order to derive such approximations, we establish clinical criteria for symptom levels on the impact of event scale. A low level was established below the indifference point on the ascending scores according to these criteria for low and medium levels of symptoms reported on the impact of event scale: Low level: As a clinician, you would not be concerned about persons who fall within this range because, while they might have some complaints or signs from time to time, the outlook is for ready improvement. There is no indication for further diagnostic, evaluative, or treat- ment procedures. Medium level: As a clinician, you would be concerned with this level of complaints and/or signs because they may give a global indication of conditions that warrant further evaluation. In a similar manner, the indifference level between med- ium and high ratings was established. High levels of symp- toms on the scale were based on the following criteria: High level: As a clinician, you feel that this level of signs and symptoms merits concern; that diagnostic, evaluative, or treatment procedures are clearly war- ranted; and that the person is more likely to be in a problem or pathological category. Four clinicians arrived at these indifference points, and their independent judgments (within one gradation of each other) were averaged to serve as a threshold (Horowitz et al. 1981). For the impact of event scale (IES), low scores were those below 8.5, high scores were those over 19, and medium scores ranged from 9 to 19. Tables 7 and 8 indicate the percentage of subjects from each group at these levels of distress. These percentages allowed comparison of the 127 Table 7. Percentage of persons at three levels of distress on the intrusion subscale of the impact of event scale Level of distress Low Medium High Stress clinic patients 10 16 74 (on entry to therapy) m=77 Bereavement field subjects 28 39 33 (2 months after death of a parent) (n = 36) Surgical patients 66 22 12 (within 1 week of breast biopsy) (n = 68) Medical patients 59 22 19 (1 month after cancer diagnosis) (n = 54) Collaterals of medical patients 42 31 28 (1 month after diagnosis) (n = 36) Medical students 86 13 1 (2 weeks after beginning cadaver dissection) (n = 69) Physical therapy students 92 8 0 (2 weeks after beginning cadaver dissection) (n = 25) degree of distress within diverse groups, experiencing di- verse life events, for their given situations and contexts. The majority of stress clinic patients, on seeking help, were at high distress (74 percent). About one-fifth of the cancer patients and their relatives were at high distress by these criteria. Only one of the students was at an equiva- lent level of distress on the first confrontation with death through cadaver dissection. These variations indicated the sensitivity of even this self-report measure for intrusion. Another investigation involved a large group of men who were given the usually unexpected news that they were at high risk for coronary heart disease. Half of this group was then randomly assigned to a special intervention program 128 Table 8. Percentage of persons at three levels of distress on the avoidance subscale of the impact of event scale Level of distress Low Medium High Stress clinic patients 21 27 S52 (on entry to therapy) (n=177) Bereavement field subjects 61 17 22 (2 months after death of a parent) (n = 36) Surgical patients 65 24 12 (within 1 week of breast biopsy) (n = 68) Medical patients S56 32 13 (1 month after cancer diagnosis) (n = 54) Collaterals of medical patients 53 25 22 (1 month after diagnosis) (n = 36) Medical students 75 20 4 (2 weeks after beginning cadaver dissection) (n = 69) Physical therapy students 96 4 0 (2 weeks after beginning cadaver dissection) (n = 25) designed to reduce their risk factors, while the other half continued with their usual community sources of health care, knowing that they were at risk because of any com- bination of elevated serum cholesterol, excessive smoking, or high blood pressure. At yearly intervals after receiving this news, the men filled out a special derivative of the impact of event scale. The group that received the special intervention attended special meetings for risk reduction, which served to remind them of their risk and tended to reduce denial. Probably as a consequence, this group had significant elevations in the levels of intrusive ideas and 129 feelings related to the news of their high risk. Of interest, this significantly higher level of intrusive experience per- sisted over each of the 3 years of followup. As reported by Horowitz and associates (1980a), the levels for intrusion on the impact of study stress scale are shown in table 9. On ANOVA, the group effects were significantly different (MS = 480.0; df = 1; F = 8.4; p < .004). The decline over time was also significant (MS = 307.0; df = Z; F = 2.8; p < .00001). The group by year interaction was not significant. These data confirmed the importance of intrusive thinking as an indicator of the degree of current stress imposed by a past life event. Table 9. Intrusion levels of impact of study stress scale Special intervention Usual care group (n = 226) \group (n = 260) Year 1 Mean 4.5 3.6 (SD) (5.6) (5.1) Year 2 Mean 4.1 2.6 (SD) (6.3) (5.1) Year 3 Mean 3.0 1.9 (SD) (5.1) (4.9) Source: Horowitz et al. (1983a). Clinical Investigations A group at the University of California, Center for the Study of Neuroses, studied patients who presented with problems precipitated by a recent serious life event. In- trusive experiences were common chief complaints in these patients, and such symptoms represent an important de- fining characteristic of the new category of posttraumatic stress disorder in the American Psychiatric Association's DSM-III (1980) (see table 10). In 66 cases of stress response syndromes, intrusive symptoms were virtually universal according to both self- report and clinician's rating (Horowitz et al. 19800). 130 Table 10. Diagnostic criteria for posttraumatic stress disorder A. A recognizable stressor that would be expected to evoke significant symptoms of distress in almost all individuals B. Reexperiencing the traumatic event either by (1) Recurrent and intrusive recollections of the event; or (2) Recurrent dreams of the event; or (3) Suddenly acting or feeling as if the traumatic event were occurring because of an association with an environmental or ideational stimulus C. Numbing of responsiveness to, or involvement with, the external world, beginning some time after the trau- matic event(s) as shown by either (1) Markedly diminished interest in one or more sig- nificant activities; or (2) Feelings of detachment or estrangement from others; or (3) Marked constriction of affective responses D. At least two of the following (not present prior to the traumatic event) (1) Hyperalertness or exaggerated startle response; (2) Initial, middle, or terminal sleep disturbance; (3) Guilt about surviving when others have not, or about behavior required to achieve survival; (4) Memory impairment or trouble concentrating; (5) Avoidance of activities that arouse recollection of the traumatic event; (6) Intensification of symptoms by exposure to events that symbolize or resemble the traumatic event Source: American Psychiatric Association (1980). Reprinted with permission. 131 Differences were not noted between patients who had ex- perienced the loss of someone close to them and those who had sustained a bodily injury or had had a near-fatal experience. This study produced very little data on the differences among persons according to dispositional factors such as personality. Clinical judgment of person- ality style was included in the study, although the reli- ability of such assessments was not fully examined. It will be necessary to accumulate a larger consecu- tive sample and to evolve a reliable method before any conclusions can be drawn on the interaction of information—processing styles and the experiential qualities of stress response states. Only the most tentative results can be reported herein, but they are worthy of mention in order to underscore the im- portance of the eventual inclusion of such variables. Among the 66 patients in the sample, 14 had a good to excellent fit to the obsessional type and 13 had a good to excellent fit to the hysterical type. As would be expected from clinical theory, the hysterical group (12 women and one man) reported more symp- toms than the obsessional group (11 women and three men). On the intrusion subset of the Impact of Event Scale, the hysterical group produced a mean score of 23 + 8.0, as compared with a mean score of 17 + 11.0 for the obsessional group. On the avoidance subset, the mean score for the hysterical group was 20 + 10.0, as compared with 15 + 13.0 for the obsessional group. Three of the 15 items of the Impact of Event Scale significantly differentiated the groups with levels of at least p < .0S, with the hysterical group always higher. On responses to the Symptom Checklist and other variables, the groups were similar. Two of the items on the Impact of Event Scale were in the intrusion subset. They were: (1) "Things I saw or heard suddenly reminded me of it," with a mean score for the hysterical group of 4.4 + 1.0, as compared with the obsessional group, who had mean scores of 2.9 + 2.1 (t = 2.5, p < .05); and (2) "Images related to it popped into my mind," with a mean score for the hysterical group of 4.2 + 1.5, as com- pared with the obsessional group who had mean scores of 2.0 + 2.2 (t = 2.9, p < .05). These data can be accounted for, perhaps, on the increased readiness of the hysterical personality to admit to emotional 132 qualities of experience." (Horowitz et al. 198005, p.91) The clinical nature of intrusive experiences can be con- veyed with a case vignette. A young man was called by an acquaintance, who asked him for help, and described himself as feeling desperate. He said no, because he didn't want to get too entangled with this person, who was not a very close friend, and was seen as "too demanding." Later, the body of this acquaintance was found with a gun beside it. He had shot himself later that night. The young man, who became a patient in our clinic, didn't go to the funeral because he did not want to get upset. A few weeks later, after keeping the event out of mind (which we, in retrospect, would call a denial and numbing period), he had an intrusive image while lying down to go to sleep. He lay in his bed, closed his eyes, and had a pseudo-hallucination of a felt presence of the dead acquaintance standing at the bedside looking calmly down at him. He felt very guilty and anxious. From then on he developed a syndrome in which he was unable to sleep except with the lights on and when he was exhausted. He was afraid to lie in bed and close his eyes. He did not wash his face because he was afraid that if he closed his eyes he would once again see the intrusive image of the face of the person who committed suicide. People often go through a period of latency after a trau- matic experience. Five or 6 months later, they may un- expectedly, with very minor triggers, begin to have intrusive thoughts or images with intense pangs of feeling. The word "pang" is particularly effective because it ex- presses the sharp, overwhelming intensity of these feelings. For example, people do not think they can tolerate the experience of bereavement if it were to continue. The feelings experienced at this time usually come in a wave- like form; the unbearable emotional quality passes, some- times in seconds. The person is able to bear the feeling because it seems to occur in doses. One can also gain an idea of the intensity of intrusive experiences by comparing different groups who have shared the same life event but who respond to it differently. In another cohort studied at the center, the particular life event common to all adult subjects was the recent death of a parent. One group con- sisted of people who had sought treatment for a neurotic level of reaction to this life event. A nonequivalent field study contrast group was selected by a review of hospital 133 death records. In a tactful way, the controls were asked to participate in the same evaluation procedures used with the patient sample. This allowed us systematically to con- trast persons who had sought psychotherapy with people who had experienced the same event but did not seek therapy (Horowitz et al. 1981). As might be expected, the group seeking psychotherapy had significantly higher levels of intrusive thinking on the impact of event scale at the first evaluation. Their mean was 21.52, compared with a mean of 13.83 for the field subjects (SD = 7.99 and 9.05, respectively; t = 3.66; p < .001). Clinician rating of intrusion on a separate instrument was also significantly higher for the group seeking therapy (mean = 18.53 for patients versus 8.36 for field subjects; SD = 11.05 and 10.72, respectively; t = 3.79, p < .0001). These two ratings for the total sample also correlated significantly (r = .69). In terms of the thresholds mentioned earlier, only 3 percent of the patients were at a low or nonsymptomatic level for intrusive experiences; only 10 percent were at this level for the avoidance report portion of the impact of event scale. This compared with 28 per- cent and 61 percent, respectively, for the field subjects (chi square = 8.98, p < .01). Meanwhile, on clinician's rat- ings for intrusion, 17 percent of the patients were at a low or nonsymptomatic level compared with 67 percent of the field subjects. The remainder of both groups were at medium or high symptomatic levels. As in prior studies, female subjects tended to have higher ratings on both intrusion and avoidance than did males (r = .39 for the correlation between level and sex for intrusion, .35 for avoidance, both significant at p < .01). Age did not correlate, nor did marital status or number of children. Thematic Contents of Intrusive Thoughts and Feelings Prior perceptions of stress-inducing events are often combined with reactive thoughts, forming a constellation of external and internal bits of information that are the elements of intrusive experiences. These constellations of ideas and feelings have a thematic valence, often reflected in the quality of the felt emotions. The main emotional ingredients are familiar from clinical experience: anger, 134 sadness, fear, and guilt about the occurrence of serious life events and the possibility of their repetition. A large clinical sample of persons intensely interviewed after similar life events allows one to determine the generality of these common tendencies. Among the latter is one that is sometimes neglected because it is hard to deal with: rage that the event has taken place at all. Most clinicians are quite ready to help a person work through his or her fear that a painful event in the past may repeat itself. They are also quite ready to help patients work on their fear of merger with others who are victims, their guilt over survival, and the shame and rage they feel that they, or a loved one, were vulnerable to harm. Rage at the source of an event is especially difficult, however, be— cause it often involves an unjust or irrational component that is avoided in communication and recognition but is a common human response. Krupnick and I (1981) worked on developing operational definitions for the themes most commonly noted in our case material on posttraumatic stress disorder. We ex- amined 15 cases in which the stress event was the death of a loved one and 15 in which it was a personal injury such as amputation, an assault, or a near-fatal experience. The frequency of conflicts involving eight common themes in stress response syndromes, as averaged for two judges following a manual of operational definitions for each theme in reviewing written case material, is reported in table 11. As expected, all themes, especially discomfort over vulnerability and rage at the source, were quite com- mon. The main differences between the groups involved sadness over loss (more common in bereavement cases) and fear of repetition (more common after personal injury). Explanation of Intrusive Ideas and Feelings The data presented so far have suggested that a quality of thinking—intrusive entry into consciousness—is an important psychological index of stress. Intrusive thoughts are a general tendency in wide populations, as noted in experimental, field, and clinical studies. Some of the themes that provide the content are not only perceptual residues of the experience but also major affectively colored attitudes. As table 12 shows, these intrusive epi- sodes (as measured by self-report) decline with time and 135 9¢1 Table 11. Frequency of eight major themes Personal Bereavement cases injury cases Combined Theme {n = 15) (n = 19) (n = 30) Discomfort over vulnerability 13 (87%) 11 (73%) 24 (80%) Rage at the source 12 (80%) 11 (73%) 23 (77%) Feelings of responsibility 9 (60%) 12 (80%) 21 (70%) Fear of repetition 7 (47%) 13 (87%) 20 (67%) Sadness over loss 14 (93%) 4 (27%) 18 (60%) Discomfort over aggressive impulses 7 (47%) 9 (60%) 16 (53%) Fear of loss of control over aggressive impulses 6 (40%) 7 (47%) 13 (43%) Rage at those exempted 6 (40%) 4 (27%) 10 (33%) ae Source: Krupnick and Horowitz (1981). Copyright 1981, American Medical Association. Reprinted with permission. Table 12. Impact of event scale data on intrusion over time and treatment Intrusion items Impact of event scale mean score Entry (n = 45) 20 Therapy hour 4 (n = 25) 16 Therapy hour 8 (n = 23) 12 Therapy hour 12 (end of therapy) (n = 45) 11 Followup (n = 45) 10 Source: Horowitz and Wilner (1980). working through (Horowitz and Wilner 1980). What is it that eventually alters mental states from those characterized by either intrusion or denial to those in which a serious life event can be contemplated or put out of mind in a volitional manner? Apparently, there is a gradual integration of both memories and associations ac- tivated by the incident. Until that time, the event and ac- tivated and reactive attitudes are maintained in an active memory system that seems to have an intrinsic property of repetition. Because each repetition evokes painful expe- riences, defensive processes are instituted. The result is an unstable dynamic conflict, so that after repeated repre- sentation of the active memories, the experience is re- garded as intrusive. The detailed exposition of this theory has been presented elsewhere (Horowitz 1976) and has led to some revision of the classical theory of the differentiation between normal and pathological grief (Horowitz et al. 1980¢). A synopsis of one aspect of this theory appeared in Human Stress and Cognition (Horowitz 1980). Serious life events such as a loss or injury present news that will eventually change inner models. But change is slow; time is essential for review of the 137 implications of the news and available options for response. The mind continues to process important new information until the situation or the models change, and reality and models of reality reach ac- cord. This important tendency to integrate reality and schemata can be called a completion tendency. Until completion occurs, the new information and reactions to it are stored in active memory. Accord- ing to this theory, active memory contents will be transformed into representations wherever that proc- ess is not actively inhibited (Horowitz and Becker 1972). This tendency for repeated representations will end only when these are no longer stored in ac- tive memory. In the instance of very important con- tents, termination in active memory will not occur with decay but only when information processing is complete. At that point, the news will be part of long-term models and revised inner schemata. As ideas related to the stress event are repre- sented, there will be a very natural comparison of the news with relevant schemata. Because a stress event is, by definition, a significant change, there will be a discrepancy between the implications of the news, and these schemata. This discrepancy evokes emo- tion. Serious life events, and the repetition of in- formation related to them, are so different from inner models of attachment that very painful emo- tional responses occur; emotional states of such power that controls are activated to prevent the threat of unendurable anguish or flooding. . . . News and immediate responses to serious life events remain stored in active memory because, on first encounter, the meanings seem to have great personal importance. Because the contents are strongly coded in active memory they tend to be rep- resented intensely and frequently. With each re- currence of the information, comparisons are made again, and emotional activation increases. Emotional responses are also represented, and so become part of the constellation stored in active memory. When other tasks are more immediately relevant, or when emotional responses such as fear, guilt, rage, or sorrow are a threat, controls are initiated. This feed- back modulates the flow of information and reduces emotional response. Excessive controls interrupt the process, change 138 the state of the person to some form of denial and may prevent complete processing of the event. Failures of control lead to excessive levels of emo- tion, flooding and retraumatization, causing entry into intrusive states. Optimal controls slow down recognition processes and so provide tolerable doses of new information and emotional responses. They lead to working states or less intense oscillations between denial and intrusive states. In this optimal condition some intrusiveness will occur with repeated representation. Some denial will occur when controls operate more pervasively, but the overall result will be adaptive, in that completion will eventually occur. Inner models will eventually conform to the new re-— ality, as in the process of completion of mourning. When this happens, information storage in active memory will terminate. At any given time, different sets of meanings of a stress event will exist in different stages. For ex- ample, fear of repetition, or fear of merger with a victim might be a recurrent intrusive experience, while survivor-guilt themes might be completely in- hibited and for the moment avoided in experience. Later, in situations of greater safety, this theme might be allowed to enter awareness, and be repre- sented as an intrusive experience. This model accounts for compulsive repetition, phasic states of intrusion and denial or numbing, variation in the level of experience between different constellations of response, and eventual resolution of stress response syndromes. While general stress re- sponse tendencies can be abstracted in this way, persons also respond uniquely. This is due in part to how their developmental history colours personal meanings of an event and to how their current life tasks and environment are affected by the event. Individual variation in response to the same type of life event is also partly due to variations in the ha- bitual style and capacity for control. (pp. 248-251. Reprinted by permission of John Wiley & Sons, Ltd.) Treatment of Stress Response Syndromes Treatment can be conceptualized as an effort to assist 139 patients in their own natural completion process. This will usually involve efforts to work through conflicts that have stymied the patient's own attempts toward this goal. It will also involve examination of latent conflicts that have been activated by association to this recent event and its pres— ent implications. This process has been reported in detail elsewhere (Horowitz 1976) but can be summarized by a citation from Horowitz and Kaltreider (1979). The thera- pist who sees a patient can organize treatment around three goals: 1. The patient needs to retain a sense of his com- petence and self-worth. This necessitates his ac- cepting whatever unalterable limitations are placed on his life plans by the loss or injury. This should be done without loss of hope or a sense of meaning in life. 2. The person should continue realistic and adaptive actions, including maintenance of available re- lationships and development of new, adaptively useful ones. 3. In working through the reactions to such a serious life event, the situation should be used for ad- ditional growth and maturation. The Pattern of Treatment After a serious life event, persons usually recon- sider the meanings and plans for response to that event in a manner that is systematic, step by step, and dosed. When emotional responses become exces— sive, or threaten flooding, the person initiates control operations. The recollection of the unfinished proc- essing of sets of meanings will tend to counteract these controls. When the person cannot handle both the repetition compulsion and the defensive counters, he seeks help. The therapist establishes a working alliance through which he assists the patient in working through his natural responses to the event and to the overall situation. In addition, efforts may be directed at modification of pre-existing conflicts, developmental difficulties, and defensive styles that made the person unusually vulnerable to traumati- zation by this particular experience. Therapy depends, in part, on establishing a safe relationship. Once this is done, work within the 140 therapy alters the status of the patient's controls. With a safe relationship and gradual modification of controls, the patient can then proceed to reappraise the serious life event, and the meanings associated with it, and make the necessary revisions of his inner models of himself and the world. As this reappraisal and revision takes place, the person is in a position to make new decisions and to engage in adaptive actions. He can practice the altered models until they gradually become automatic. Overlapping with these processes is the necessity for working through reactions to the approaching loss of the therapist and the therapy. As the person is able to accommodate to new levels of awareness, this process repeats itself. When he can relate in a still more mutual and intimate manner, he can examine himself more deeply, and controls can be modified further. Additional work of this sort may modify aspects of character structure. Within the time limits of a brief psychotherapy, the therapist works to establish conditions which will be helpful to the processing of the painful event. There is an early testing by the patient of both the safety of the relationship and the therapist's ability to help him cope with symptoms. These symptoms can seem less overwhelming when the therapist pro- vides support, suggests some immediate structuring of time and events, prescribes medication if anxiety or insomnia is too disruptive, and gives "permission" for the patient to dose his feelings rather than to doggedly attempt to work them through as quickly as possible. Such interventions are often called for even during the period of acute medical hospitalization after trauma. Patients who are more handicapped by their symp- toms of avoidance can be helped by encouragement from the therapist to recollect the stress event with associations and abreactions, while working toward changing attitudes that made the controls necessary. Frequently, symptoms subside rapidly with the es— tablishment of a good working alliance. Then, the relationship of the stress event to the patient's various selfconcepts can become a focus. Intro— duction of the plans for termination of therapy sev- eral sessions before the final one leads to a re- experience of loss, often with a return of symp- 141 toms. But this time, loss can be faced gradually, ac- tively rather than passively, and in a communicative helping relationship. Specific interpretations of the link of the termination experience to the stress event are made and the final hours center on this theme. At termination, the patient will usually still have symp- toms, in part because of the time needed to process a major loss and because of anxiety about the loss of the relationship with the therapist. Followup evalu- ations suggest to us that the therapy serves as a catalyst for both symptomatic and structural change over the ensuing year or more. This is a very global and generalized overview for a model 12 hour therapy as it has been applied in our center (see table 13). Patients sometimes become aware, in the course of these brief therapies, of a particular style they have for not thinking about events, and they are able to deliberately alter that situation. It may be possible for them, by continued work on their own after therapy, to live out changes that may gradually be incremental in altering habitual controls. When a person experiences the impact of a serious life event, such as a loss or injury, his most advanced, adaptive role relationships can be threatened. He may regress to earlier role relationships, or the meaning of the event itself may tend to create some new role relationship, perhaps with unattractive, dangerous, or undesirable characteristics. The person may then enter a series of painful, strongly affective states based on altered self-images and the changed role relationship. As a consequence of therapeutic facilitation of normal processes, the disturbing role relationships or selfimages can once again be sub- ordinated to more adaptive, mature self-images and role relationships. Intensive work using a brief ther- apy model may both alter the symptomatic response to a stressful life event and facilitate further prog- ress along developmental lines. (pp. 374-376. Re- printed with permission of W.B. Saunders Company, Publishers) Conclusion Intrusive thinking can be used as an index of the degree of distress resulting from a serious life event when the 142 evi Session Table 13. An example of timing in brief psychotherapy Relationship issues Initial positive feeling for helper Lull as sense of pressure is reduced Patient tests therapist Therapeutic alliance deepened Patient activity Patient tells story of event Event related to life of patient Patient adds associations Work on what has been avoided Therapist activity Preliminary focus discussed Takes psychiatric history Realignment of focus; interpretation of resistances with empathic recognition of reasons why they are currently reasonable based on past relationships Further interpretation of defenses and warded off contents linking of latter to stress event and responses Time of termination discussed yl Table 13.—Continued Session Relationship issues Patient activity Therapist activity 7-11 Transference reactions interpreted Continued working through when seen and indicated, and of central conflicts and issues linked to other configurations of termination as related to the life event and reactions as related to the life event and reactions to it 12 Saying goodbye Realization of work to continue on own Clarification and interpretation related to central conflicts and termination; clarifi- cation of unfinished issues and recommen- dations Acknowledgment of real gains and real future work in continued mourning and formation of new schemata Source: Horowitz and Kaltreider (1979). Reprinted with permission of W.B. Saunders Company, Publishers. thought is linked by content to themes and emotions related to that event. Such an approach may give false positive results when distress related to some other area of stress and conflict is displaced onto a traumatic constel- lation of memories. Such an approach may also give false negatives when unconscious maneuvers of denial and dis- avowal work sufficiently well to prevent intrusive ideas and feelings. By accepting and understanding these limi- tations, researchers can include intrusive thinking in ex- perimental, field, and clinical investigations of stress as a useful variable, amenable to time series sampling. The manner of inquiry can range from self-reports, such as the impact of event scale, through indepth clinical interviews. In treatment, the contents of mental life that have an in- trusive quality may provide a guide for fruitful areas of investigation. References American Psychiatric Association. Diagnostic and Statis— tical Manual of Mental Disorders (DSM III). 3d ed. Wash— ington, D.C.: the Association, 1980. Antrobus, J.S., and Singer, J.L. "Mind Wandering and Cog- nitive Structure." Paper presented to the New York Academy of Science, October 20, 1969. Antrobus, J.S.; Singer, J.L.; and Greenberg, S. Studies in the stream of consciousness: Experimental enhancement and suppression of spontaneous cognitive processes. Per— ceptual and Motor Skills 23:399-417, 1966. Breuer, J., and Freud, S. Studies on hysteria. In: Strachey, J., ed. The Standard Edition of the Complete Psychological Works. Vol. 2. London: Hogarth, 1957. Finn, R.H. A note on estimating the reliability of cate- gorical data. Educational and Psychological Measurement 3:71-76, 1970. Freud, S. Beyond the pleasure principle. In: Strachey, J., ed. The Standard Edition of the Complete Psychological Works, Vol. 18. London: Hogarth, 1953. Horowitz, M.J. Intrusive and repetitive thoughts after experimental stress: A summary. Achives of General Psychiatry 32:1457-1463, 1975. Horowitz, M.J. Stress Response Syndromes. New York: Aronson, 1976. Horowitz, M.J. Image Formation and Cognition. 2d ed. New York: Appleton, 1978. 145 472-367 0 - 85 - 6 Horowitz, M.J. Psychological response to serious life events. In: Hamilton, V., and Warburton, D., eds. Human Stress and Cognition. New York: Wiley, 1980. Horowitz, M.J., and Becker, S. Cognitive response to stress: Experimental studies of a "compulsion to repeat trauma." In: Holt, R., and Peterfreund, E., eds. Psycho- analysis and Contemporary Science. New York: Macmillan, 1972. Horowitz, M.J.; Hulley, S.; Alvarez, W.; Billings, J.; Ben- fari, R.; Blair, S.; Borhani, N.; and Simon, N. News of risk for early heart disease as a stressful event. Psychosomatic Medicine 42:37-46, 1980a. Horowitz, M.J.; Krupnick, J.; Kaltreider, K.; Wilner, N.; Leong, A.; and Marmar, C. Initial psychological response to death of a parent. Archives of General Psychiatry 38:316-323, 1981. Horowitz, M.J., and Kaltreider, N. Brief therapy of stress response syndromes. Psychiatric Clinics of North America 2:365-378, 1979. Horowitz, M.J.; Schaefer, C.; and Cooney, P. Life event scaling for recency of experience. In: Gundersen, E., and Rahe, R., eds. Life Stress and Illness. Springfield, Ill.: Thomas, 1974. Horowitz, M.J., and Wilner, N. Stress films, emotion, and cognitive response. Archives of General Psychiatry 30: 1339-1344, 1976. Horowitz, M.J., and Wilner, N., Life events, stress, and coping. In: Poon, L., ed. Aging in the 80's. Washington, D.C.: American Psychiatric Association, 1980. Horowitz, M.J.; Wilner, N.; and Alvarez, W. Impact of event scale: A measure of subjective stress. Psychosomatic Medicine 41:209-218, 1979. Horowitz, M.J.; Wilner, N.; Kaltreider, K.; and Alvarez, W. Signs and symptoms of post-traumatic stress disorder. Archives of General Psychiatry 37:85-92, 1980b. Horowitz, M.J.; Wilner, N.; Marmar, C.; and Krupnick, J. Pathological grief and the activation of latent self-images. American Journal of Psychiatry 137: 1157-1162, 1980c. Janis, I. Stress and Frustration. New York: Harcourt, 1969. Krupnick, J., and Horowitz, M.J. Stress response syn- dromes: Recurrent themes. Archives of General Psychiatry 38:428-435, 1981. Lazarus, R. Psychological Stress and the Coping Process. New York: McGraw-Hill, 1966. 146 Malan, D. Frontier of Brief Therapy. New York: Plenum, 1976. Schur, M. The Id and the Regulatory Process of the Ego. New York: International Universities, 1966. Sifneos, P.E. Short-term Psychotherapy and Emotional Crises. Cambridge: Harvard University Press, 1972. Stevens, S.S. A metric for the social consensus. Science 151:530-541, 1966. 147 148 Part II PREVENTING STRESS- RELATED PSYCHIATRIC DISORDERS This section addresses key issues concerned with the design of studies to prevent stress-related psychiatric disorders. Gruenberg and La Resche suggest a range of preventive trials, offering examples from the past and prospects for the future. DeWitt and Horowitz present a model for the study of secondary prevention of posttrau- matic stress disorder, including data from a pilot study. Beutler proposes an approach to studying bereavement, loss, and anticipated death as "a precursor to the devel- opment of a preventive trial." Falloon and his colleagues present preliminary results from their study on the impact of life events on schizophrenia and their efforts to reduce the risk of stress. Each paper suggests several potential elements of a comprehensive research program in pre- venting stress-related psychiatric disorders. 149 Chapter 7 LET'S DO SOME PREVENTIVE TRIALS Ernest M. Gruenberg, M.D., Dr.P.H. and Linda La Resche, Dr.Sc. When the famous English biologist Sir Peter Medawar (1967) said, "If politics is the art of the possible, science is the art of the soluble," he was making a general statement indicating that the solutions to scientific problems come through a process analogous to art, in which a scientist recognizes that a question about some aspect of nature has reached the point of solubility by a possible method of research. We suggest that we look at the art of epidemiological inquiry from this point of view. Epidemiology is to a large extent an observational science, which seeks natural ex- periments and systematically organizes information so as to get a picture of the dynamics of disease occurrence in populations. Its ultimate intent is to reduce the prevalence of disease and the frequency of premature death from disease and injury. When knowledge reaches the point of active intervention in the natural course of disease occur-— rence, a sudden step forward in understanding the dynamics of disease occurs. The most recent dramatic occurrence in our minds is the campaign to eradicate smallpox, which substituted the search and surround strategy for mass vaccination. Methods for tracing chains of possible exposure were de- veloped. Our knowledge about the absence of a nonhuman host for the virus was reinforced. Much more was learned about the viruses in the material used for inoculations than had ever been known before. And there were other new 151 bits of information. We now know more about the epidemiology of smallpox than we ever knew before, and the beauty of our current knowledge is that we hope never to have to use it again. Examples of Preventive Trials The idea of preventive interventions is as old as epi- demiology. Draining swamps to avoid malaria, quarantine to avoid infections, removing oneself from the locus of an epidemic as in Boccaccio's famous writings, are all ex- amples of preventive interventions. When antiscurvy diets were introduced on boats, the results were dramatic. Fly extermination and other sanitary measures are very old. In the past century, Sandwith experimented with modifying diets in Egypt to prevent pellagra (Roe 1973). Goldberger in the 1920s experimented with various ways to prevent pellagra in orphanages and hospitals (Goldberger et al. 1915; Goldberger and Tanner 1922, 1925). The prevention of pellagra and its associated psychoses among institutional inmates was a dramatic investigation started in 1913 (Goldberger et al. 1915). It apparently prevented new cases of pellagra by improving and sup- plementing diets. The report leaves something to be de- sired, because no relative risk figures were published. They did not systematically tabulate the risk of developing pel- lagra among the controls. Instead, they pointed out that the death rate in the communities surrounding the orphan- ages and hospitals continued to be high for pellagra. In such instances, a preventive trial related to a key risk factor can provide such dramatic evidence that subtle quantifi- cation of relative risk becomes an example of misplaced precision. After the Second World War, we witnessed Thomas Francis' elegant test of the Salk vaccine (Francis et al. 1955) and the dramatic experiment with fluoridation of the water supplies in the Newburgh-Kingston trial for the prevention of dental caries (Ast et al. 1956). The Newburgh-Kingston fluoridation experiment lasted from 1945 to 1955. The experiment was organized largely by the New York State Health Department but required a great deal of collaboration with other agencies. The survey first examined children's teeth in the two cities. Fluorides were added to the water supply of Newburgh in 1945, and 152 children's teeth were monitored in both cities for the next 10 years. The results were conclusive. The frequency of cavity- free mouths in the children who grew up in Newburgh was much higher than the frequency of cavity-free mouths in the children who grew up in Kingston. It was established that adding fluoride to a city's water supply was a safe, effective way of reducing dental caries. Low fluoride con- tent drinking water is a risk factor for dental decay. The preventive trial showed that most decay could be attrib- uted to low fluoride concentrations in the water. Neither of these preventive trials randomly assigned individual subjects to the preventive and control groups. Whole populations were assigned, because the preventive intervention was conceptualized around a community- based resource—food or water. In an era when the double- blind, randomized clinical trial is riding high, we should remind ourselves that other designs still have their use. Of course, during the past century the art of systematic intervention trials has developed most around clinical trials. The clinical trial has become increasingly powerful as investigators have learned to carefully select similar cases and randomly assign them to different treatments. This random assignment in the clinical trial has become something of a hallmark of systematic trials, but of the preventive trials referred to, none had randomly assigned study subjects except for the Salk vaccine trials. Those trials were exceptional also, because it was unusual for preventive interventions to be applied only to individuals rather than to populations. In the other preventive inter- ventions referred to, whole populations have been sub- jected to an intervention, and comparative populations had to be used for evaluation of effects. As far as we know, the Newburgh-Kingston fluoridation studies were the most carefully executed intercommunity comparisons that have been done in the realm of preventive trials. In Newburgh, where fluoride was added to the water supply to bring the concentration of fluoride ion up to 1.0-1.2 parts per million (ppm), the rate of dental caries was 57 percent less than in Kingston, where fluoride ion concentration was 0.5 ppm. Another important difference among the preventive trials mentioned is that those that gave nutritional sup- plements were based on rather weak evidence that the nu- tritional supplement would, in fact, have an effect. For instance, Goldberger, in a State mental hospital, took groups of patients and gave different groups different 153 supplements. Some were given soybeans, others were given dried milk or casein or cowpeas, and some were given brewer's yeast. In the five groups that he used, only the group given brewer's yeast showed a preventive effect (Goldberger and Tanner 1925). The important principle illustrated here is that a safe intervention need not be based on extremely well-devel- oped knowledge regarding the likelihood that the inter- vention will work. In contrast, the Newburgh and Kingston studies and the Salk vaccine studies were dealing with ma- terials for which good grounds exist to suspect possible dangers. The investigators had to proceed with consider- able caution. Preventive Trials Based on Weak Hypotheses The enormous increase in the number of people con- ducting observational epidemiological research contrasts with the slow growth of preventive trials. It is our sug- gestion that attention be focused more sharply on the conditions under which the preventive trial is suitable at an early stage in our knowledge. First, the intervention must be seen as safe. Second, it is advantageous if the intervention has char- acteristics people see as desirable in general, regardless of whether it has the hoped-for preventive effect. Third, the preventive trial cannot be used if the inter— vention is seen as very desirable and is available for everyone who wants it, because then controls cannot be found. IHlustrative Preventive Trials Two examples of trials for preventing stress-related psychiatric disorders include (1) new cases of chronic social breakdown syndrome associated with hospitalizations that excessively disrupt the patients' relationships to their so- cial support systems and community supports and their own responsibility for their behavior, care, and future, and (2) bereavement in recent widows. Each demonstrates the po- tential for research in preventive intervention and the as- sessment of risk. 154 Preventing the Social Breakdown Syndrome One interesting preventive trial began in 1959, at a time when mental hospitals in both Britain and the United States were going through rapid transformations. These facilities had tended to become places of long-term care, with a relatively small part of their facilities devoted to short- term treatment, which then was regarded as a matter of S or 6 months. Some remarkable demonstrations starting after the Second World War, however, had transformed a few hospitals into locations where new admissions were treated on an informal basis in an atmosphere similar to that of an acute general hospital (MacMillan 1958). Pains were taken to remove as much of the stigma of institu- tionalization as possible from the acute episodes of care, and the family and community support systems of the patient were encouraged to expect the patient back after a week or two of intensive attention. The hospitals adopted a policy of releasing patients before recovery was complete and standing ready to readmit over an indefinite period if the need should arise. This pattern of care, called com- munity care of the severely mentally ill, was introduced initially as an efficient way of using available staff and a humane and kindly way of treating seriously mentally ill patients. Observations led to the hypothesis that, in fact, chronic severe deterioration in personal and social functioning had become much less common in the new system of care. When a hospital in upstate New York wished to copy these innovations, it was suggested that the effectiveness of the reformed clinical services in preventing long-term de- terioration should be evaluated. At that time, the first- named author of this paper was in favor of the reform but against the allocation of personnel and resources to the evaluation effort, on the grounds that the reform had plenty of precedents and would succeed in its immediate goals. The reformers were trying to do what had been done before by others. The evaluation of effects in reducing the frequency of chronic deterioration would start with no es- tablished research devices or research design and had little promise of producing any clear indication of whether this desired outcome had been achieved. Others argued that it was worth the trouble and time, even if no definite sci- entific results could be obtained. Their views prevailed, and the effort was made. The part of the hospital used for this experiment had no 155 locked services and no visiting restrictions. Compulsory hospitalization was avoided by seeking the patient's active cooperation from the very first contact and by providing precare consultation by hospital psychiatrists to anyone contemplating sending a patient to the State hospital. It was assumed that the frequency of new episodes of psy- chotic decompensation would not decrease in the general population. Because there was fear that the new pattern of care might lead families to keep patients at home or to send them elsewhere when they became troublesome, all residents of the county between the ages of 16 and 65 who had been in any kind of inpatient psychiatric service after 1955 were surveyed annually to determine the frequency of an operationally defined chronic social breakdown syn- drome (CSBS), an objectively ascertainable measure of deterioration. The SBS data form ascertains some two dozen elements of behavior reflecting self-neglect and withdrawal from social and work activities on the one hand, and troublesome behavior such as fighting, destructiveness, and shouting on the other hand. The SBS frequently brings the patient to the hospital or to crisis intervention. The SBS episode was labeled chronic when these behaviors were present every week for 52 consecutive weeks. It was hoped that the reformed services would reduce the annual number of new Dutchess County CSBS cases; they were not expected to have any effect on the total number of new SBS cases. Because the new program of services had to be introduced for the entire county or not at all, there was no way of assigning individuals to it ran- domly. A matching community, as in the Kingston-New- burgh study, was not practicable. The only control arose from the before-and-after observations. Uneasiness about being able to innovate research tech- nology fast enough to measure the intended effects was justified, but so was the courage of those who insisted that an effort had to be made. The operationally defined cri- teria had to be developed, a data-gathering technique de- signed and pretested, a register established of the county's entire population who had had inpatient treatment during the previous S years (mostly long since out of clinical care), and a large-scale plan assembled to use recruited, trained, and supervised interviewers to conduct many thousands of interviews. The only way to address the di- lemma required an absolutely original design. The results were dramatic. Benefits were concentrated 156 in the proportion of new SBS episodes that recovered in the first 6 months. A rapid, relevant, full-service response to the crisis, with continuing availability of the full spectrum of services (including brief rehospitalizations), appeared to increase the probability of quick recovery from the SBS episode. Hundreds of acute onsets of SBS occur each year among these former patients (and some new patients, of course), but most of them end the SBS behaviors within 2 to 3 weeks. There is a remarkably smooth decline in recovery rates as the number of weeks increases; it is very high dur- ing the first week, but in the 10th .week is only one one- hundredth as great, and by the end of a year (when we ar— bitrarily had decided to call the SBS condition chronic) it was extremely slow. During the first 3 years of the demon- stration, the number of people with SBS who reached the 1-year anniversary point dropped year by year. In the first year, more than 45 CSBS cases had developed; by the third year, fewer than 16 had. The population at risk remained steady at about 100,000 persons (county residents) aged 16 to 6S. The subsequent rate of recovery was very low and did not vary between the cohorts. Apparently, in terms of recovery from SBS, nothing succeeds like success, and nothing fails like failure. Thus, the risk of developing chronic social breakdown syndrome dropped from 45 to 15 per 100,000 per year; so about 30 new cases of chronic social breakdown syndrome per 100,000 per year could be attributed to the old system of care. The attributable risk of new cases was then 67 percent. Patients who had already become chronically deterio- rated recovered only slowly and no more rapidly than com- parable chronically deteriorated patients in the more conventional services. This confirmed the observations made in the British community care hospitals (Gruenberg 1983). The findings of this study (when attended to at all) tend to be interpreted to mean that abandoning chronically dis- abled patients by discharging them from mental hospitals to no care or to inadequate community care programs is justifiable under the slogan of deinstitutionalization (Gru- enberg 1974). It is rarely recognized that these preventive interventions were, in fact, provided by a reorganization of existing services, techniques, and personnel from a State mental hospital with the continuing availability of the hospital and its treatment team to patients, family, and 157 community. The release of patients before complete re— covery and with continuing care in the community was coupled with a readiness to provide readmissions when indicated. While this preventive trial did not avoid all possible sources of error and bias, it seemed convincing that many cases of chronic severe deterioration, as measured by the instruments developed in the study, were sociogenic sec— ondary manifestations of severe mental disorders and were preventable through a social reorganization of the treat- ment system that emphasized the maintenance of the patients' community support systems. This lesson stands as the only example of a change in the rate of illness of any type in a population as a result of a reorganization in the delivery of medical care services without the introduction of new techniques or new personnel. These lessons have gone largely unnoticed in the scramble to create community mental health centers and to shift the economic burden from one level of government to another and to various other third-party payers under the guise of deinstitutionalization. This development has led to shocking forms of community neglect of seriously ill mental patients. Nevertheless, abuses do not diminish the advance in knowledge these studies provide. There would appear to be merit in looking at some of the untested preventive hypotheses currently being advocated for action to see if one can imagine introducing some of them on a preventive trial design. Widow-to-Widow Counseling for the Prevention of Depression The widow-to-widow program, begun at the Laboratory of Community Psychiatry at Harvard in the late 1960s (Silverman 1970), has been described as a demonstration program in primary prevention. The program's effect on the incidence of mental or physical disorders, however, has not yet been demonstrated. The goal of the program is to help recently widowed women under age 60 adjust to widowhood through the use of lay counselors who are themselves widows. The counselors are mostly high-school-educated women in their mid-forties who have been widowed for about 3 years. Initially, the counselor's role is to share her own experiences concerning financial and emotional problems 158 and to help the new widow cope with being alone, making decisions, and developing new social relationships. Later, the counselor encourages and supports the new widow's independence. In many ways, the situation of the widow-to-widow pro— gram parallels that of the early programs in community care of the seriously mentally ill. The program is generally thought to provide a humane and socially acceptable intervention. Although its advocates do not specify the condition they hope to prevent, we suspect that this in- tervention also lowers the incidence of clinical depression by preventing undue extension of bereavement in recently widowed middle-aged women. The time is ripe for a pre- ventive trial to confirm or refute this suspicion. Because the widow-to-widow program finds its clients by scanning newspaper death notices, and because there are never enough counselors to extend the service to every newly widowed woman, the service could be offered to a random half of the widows in a given city. The clinical diagnosis of depression can be made quite accurately, so the incidence of this condition can be measured fairly easily. Thus, no major obstacles exist to initiating such a preventive trial. If the results were positive, we would have learned how to prevent a sizable number of depressions. Yet negative re— sults would not mean that the widow-to-widow program is no good; they would simply indicate that it does not pre- vent depression, and its advocates would have to be more specific about what it does prevent. Conclusion These proposals suggest the potential of preventive trials research in stress-related disorders. Even when our hy- potheses are weak and not based on a complete under— standing of the pathogenesis of specific disorders, we may proceed with preventive trials, provided they are safe, ethical, and acceptable to the public. Let's do some preventive trials! References Ast, D.B.; Smith, D.J.; Wachs, B.; and Cantwell, K.T. New- burgh-Kingston caries—fluoride study XIV. Combined 159 clinical and roentgenographic dental findings after ten years of fluoride experience. Journal of the American Dental Association 52:314-325, 1956. Francis, T.J.; Korns, R.F.; Voight, R.B.; Boisen, M.; Hemp- hill, F.M.; Napier, J.A.; and Tolchinsky, E. An evaluation of the 1954 poliomyelitis vaccine trials. American Journal of Public Health 45:Part 2, 1955S. Goldberger, J., and Tanner, W.F. Amino-acid deficiency probably the primary etiological factor in pellagra. Public Health Reports 37:462-486, 1922. Goldberger, J., and Tanner, W.F. A study of the pellagra- preventive action of dried beans, casein, dried milk, and brewer's yeast, with a consideration of the essential pre- ventive factors involved. Public Health Reports 40:54-80, 1925. Goldberger, J.; Waring, C.H.; and Willets, D.G. The pre- vention of pellagra. A test of diet among institutional inmates. Public Health Reports 30:3117-3131, 1915. Gruenberg, E.M. Mission to Britain. International Journal of Mental Health 11(4):24-47, 1983. Gruenberg, E.M. Benefits of brief hospitalization. In: Can- cro, R.; Fox, N.; and Shapiro, E., eds. Strategic Interven- tion in Schizophrenia: Current Developments in Treatment. New York: Behavioral Publications, 1974. pp. 251-259. MacMillan, D. Hospital-community relationships. In: An Approach to the Prevention of Disability from Chronic Psychoses. New York: Milbank Memorial Fund, 1958. pp. 29-50. Medawar, P.B. The Art of the Soluble. London: Methuen, 1967. Roe, D.A. A Plague of Corn: The Social History of Pel- lagra. Ithaca: Cornell University Press, 1973. Silverman, P.R. The widow as caregiver in a program of preventive intervention with other widows. Mental Hygiene 54:540-547, 1970. 160 Chapter 8 A RESEARCH APPROACH FOR THE STUDY OF SECONDARY PREVENTION OF POSTTRAUMATIC STRESS DISORDERS Kathryn N. DeWitt, Ph.D. Mardi J. Horowitz, M.D. Daniel S. Weiss, Ph.D. Howard H. Goldman, M.D., Ph.D. Introduction The Research Group at the Center for the Study of Neu- roses, Langley Porter Psychiatric Institute, has been en- gaged in research on stress-related disorders for some time. We have developed a brief, time-limited psycho- therapy found to be effective for many persons who seek help for these problems and have come to recognize the need for basic information about the incidence and natural course of posttraumatic stress disorders (PTSD). We have also been impressed by the number of persons who seek treatment only as a last resort, after their distress has significant negative effects on their natural support sys- tem— family, friends, and work. We feel a need to find an avenue for preventive treatment that could use early case- finding methods to offer an effective intervention before these disorders have taken their toll. While these goals seem straightforward, the practical hurdles in implementing them are considerable. Prospec- tive research designs are very costly. Since the incidence 161 of PTSD is unknown and may be relatively small, it would be necessary to follow large numbers to obtain information on a few. Because the risk factors for PTSD are currently unknown, it would not be possible to limit the sample to persons with risk-related characteristics. Since traumatic events occur in a random or quasi-random period over time, the population would need to be monitored frequently to ensure that the disorder had been caught at a time when preventive intervention would be possible. Because many persons with early signs of PTSD may not be receptive to professional help, methods would have to be found to min- imize barriers to treatment. Faced with these dilemmas, the group has come up with one possible solution— using a large private corporation as a sample. This approach would solve many of the problems of a prospective research design. Use of an intact group would greatly reduce project costs. Since the corporate sample meets together nearly every day, subject recruitment and data collection expenses would be greatly reduced. In addition, most corporations have systems for recording such employee information as attendance and work productivity and could facilitate access to medical care utilization via insurance records. The proper procedures and subject consent could provide access to these hard data. Finally, a corporation with good employee relations could foster an atmosphere in which it is acceptable to seek professional help. As with all solutions, many of these advantages are matched by disadvantages. A major concern is with the representativeness of the sample. In the case of a large corporation, the results might not generalize to nonworking populations or to distinctly different occupational groups. However, this strategy could be developed on one group and replicated on others. In addition, if the secondary prevention strategy were to become practical for general application, it would be necessary to use some vehicle like an intact work group for case identification and referral. A second disadvantage presents more knotty problems. The ideal research strategy would be to complete a study to identify risk factors for the development of PTSD first and then to institute an intervention trial. Unfortunately, few corporations would be willing to cooperate for the number of years necessary to achieve the basic research objective before receiving the benefits of some treatment. Services to employees act as an incentive to secure cor- porate participation. This means that we would have to 162 make an educated guess as to probable early signs of PTSD and offer treatment early in the research program before results as to the actual risk factors were known. This so- lution can be a practical success in securing corporate cooperation; it has research costs in that the timing of the intervention phase is premature and the need to treat some subjects early reduces the numbers available for charting the natural course of the disorder. Our purpose in bringing this research to public attention is to reconsider the pros and cons of this approach and ex- plore possible practical solutions to the drawbacks. The material presented here has been excerpted from a longer project description that included additional sections des- cribing the suggested treatment, the therapists, measures, issues involved in confidentiality and collaboration, logis tics, and data management. A pilot study indicated that corporate cooperation could be secured and employce participation would be sufficient to meet the sample needs of the study. Specific Aims This study aims to assess the efficacy and safety of time-limited, brief dynamic psychotherapy (BDP) as a secondary preventive treatment for posttraumatic stress disorders. Periodic collection of data from volunteers at a large corporation would provide information about persons before and after they experience traumatic events. This prospective design would permit examination of subgroups who have and have not had serious life events, as well as those who do or do not develop distress. Those who develop distress would be randomly assigned to usual community care (UCC) or brief dynamic psychotherapy. These groups would be compared for ensuing years to appraise their symptoms, social functioning, and quality of life. We predict that BDP would be shown to be a safe, effective, and inexpensive way of preventing chronic stress disorders. In specific terms, the following would occur: 1. We would determine the incidence of PTSD as related to specific serious life events over a 2-year period in a sample of 1,500-2,000 employees of a large corpora- tion. These observations would be based on self-reports of subjects, obtained at fixed 4-month intervals. The self-reports would include signs and symptoms that meet an operational definition of the DSM-III diagnostic criteria for PTSD. 163 2. With these data, we would test the hypothesis that persons who are young and female, and have low levels of perceived social support, would have not only more distress after serious life events but also more frequent life events and hence more cumulative stress. 3. We would also test the hypothesis that persons who share a high level of distress 1 to 4 months following a serious life event would have a higher rate of subse- quent chronic stress disorders than persons with lower levels of distress. 4. Finally, we would test the hypothesis that a brief psy- chotherapy focused on the event and its personal im- plications, when offered to 40 persons with a high level of distress at a period from 2 to 5S months following the event, would significantly reduce the incidence of chronic PTSD and levels of anxiety and depression in general, when compared with 40 equivalent persons who serve as a usual care control. Significance We would propose the study for several reasons. 1. Current information on the impact of stress events is based largely on retrospective self-reports of pre- and postevent functioning. Studies that collect information only at points following the stress event are not able to determine which portions of postevent functioning were preexistent and which were precipitated by the event. This points to the need for a prospective research design. 2. A preventive intervention promises savings in both suf- fering and health care expenses if it proves to be ef- fective and safe. Since PTSD is marked by a definite and recent life event, it provides an excellent candidate for early identification and treatment of persons at risk. 3. The measures and treatment procedures needed in this study are now available. Reliable measures of distress after serious life events, and of the signs and symptoms of PTSD, have been developed, and a brief time-limited psychodynamic psychotherapy is available. The concept of intervention during a crisis is threaded through a long history of clinical investigation. Primary prevention is an effort to protect persons from harmful agents. This may not be appropriate to some life events, 164 such as bereavement, but can apply to the prevention of disasters. Secondary prevention is appropriate to the focus of the present study, since it is an effort to use early detection and treatment to alter a downward course. The reviews of Lamb and Zusman (1979) and Adler et al. (1978) suggest that secondary prevention of mental disorders may be the most economical approach. Models of effective secondary prevention research in- clude those recently published by Raphael (1977) and Gruen (1975), who focused on groups of persons who had ex- perienced specific life events. For Raphael, the event was bereavement. Sixty-four women considered to be at risk for later morbidity were selected from a larger group of widows. Raphael offered brief individual psychotherapy, consisting of three home visits, each 2 hours long, and found that a treated group fared better than an untreated group. Gruen studied persons at risk for cardiac neurosis following myocardial infarction. He selected persons at risk by eliminating those who showed signs of coping in initial interviews. Sixty male patients received from 10 to 12 half-hour hospital room visits. The outcome was better in the treated subsample than in an untreated control group. Careful identification of persons at risk is an important aspect of the secondary prevention model. Studies by Williams and Polak (1979) and by Polak et al. (1975) show the problems that can result from an overinclusive selec— tion of the target population. These authors offered six home visits to a randomly selected sample of 82 families who had experienced the sudden death of a family member. Results at 16 and 18 months postevent showed that treated families fared worse than the untreated comparison fam- ilies, leading the authors to conclude that "the current re— sults suggest that such techniques (crisis intervention) are less applicable to the treatment of normal persons or fam- ilies who did not actively seek help and who are experi- encing a normal life crisis." Without proper identification of appropriate at-risk persons, treatment interventions can do more harm than good. Such findings also indicate the need to study both the efficacy and the safety of treatment. Many previous investigators have examined the shift from transient stress reactions to persistent psychopath- ology to determine what variables might serve as risk factors. Chiriboga and Dean (1978) found that age (younger) and sex (female) are related to event-associated distress. Various studies have shown that a perceived lack 165 of social support during a crisis may be associated with more distress in the ensuing period (e.g., Madison and Walker 1967; Gore 1974; Brown et al. 1975; Nuckolls et al. 1976; and Kaplan et al. 1977). Investigators such as Gersten et al. (1974) have found that behavioral pathology prior to a stress event may be associated with postevent symptoms. Both neurotic predispositions and limitations on coping capacities have also been connected (Lazarus 1966, 1974; Hinkle 1974; Horowitz 1976). Dohrenwend (1978) has sum- marized this as follows: The results of this research indicate that intrapsychic predispositions and processes play an important role in determining the outcome of exposure to stressful events, but the ways in which they interact with a stress reaction to determine the ultimate outcome are too complex to lend themselves to summary in one or two hypotheses. (p. 5) Based on this review, we would propose a prospective study of the incidence of serious life events, the development of distress, and the efficacy of treatment in preventing chronic posttraumatic stress disorders and in preventing other possible stress-related physical and mental disorders as manifested by health care utilization. Methods Design Three interrelated studies would be incorporated within the research design. They would progressively analyze, in increasing detail, smaller samples of consenting employees of a large corporation. The studies would be: ® A longitudinal study of the incidence and impact of stress events and of individual factors related to their occurrence, in which data would be collected on a sample of 1,500-2,000 persons at six 4-month intervals over a 2-year period ® A contrast groups study of differences in characteristics and response patterns of two groups of 175 persons each, selected because they report serious losses or injuries and show high or low levels of postevent distress when compared with the sample as a whole 166 ® A comparative treatment study of differences in long- term (up to 33 months postevent) adjustment patterns of two groups of 40 persons each, who show high levels of acute distress after serious life events and who are ran- domly assigned to receive a brief preventive treatment (BDP) or to rely on their usual care resources (UCC) Subjects The 3,000-4,000 employees of a large local corporation would serve as the subject pool. The corporation would be chosen to provide a large multiracial sample of privately employed persons at various levels of skill and remunera- tion. Sample Selection and Procedures Samples would be selected in a progression of steps be- ginning with the longitudinal study, as follows. Longitudinal study. The longitudinal sample would include all employees who volunteer and give written informed consent to complete self-report instruments at six data collection points scheduled at 4-month intervals over a 2-year period, and to evaluation interviews if they enter a high distress zone after a serious life event. Data from the longitudinal sample would be used to provide information on: ® The incidence of stress events and of PTSD ® Preevent factors related to the occurrence of stress events ® The expected range of responses to stress events Contrast groups study. Subjects for the contrast groups study would be selected from among those in the longitudinal sample on the basis of their responses to the self-report measures. Persons who had experienced a major stress event between 1 and 4 months prior to that data collection point would be identified first. Selection of these subjects would begin with the second collection point (4 months after the study starts), so that it will be possible 167 to obtain baseline data from at least one data collection point prior to the stress event. Events qualifying as major stresses would include bereavement, injury to self or prop- erty, and divorce completed to final decree. Once the sample of persons with recent stress events had been identified, their level of stress-related symptoms would be examined. A stress threshold established in a previous study at the Center (Horowitz et al. 1981) would be used to divide the sample into distressed and nondis- tressed groups. Subjects with scores above a high symptom threshold would be designated as at risk. This division would be made on the basis of symptom reports at 1 to 4 months postevent, so that treatment could begin prior to the point 6 months postevent that is used as a diagnostic cutoff for identification of chronic PTSD by DSM-III. Once these subjects have been selected, samples of 175 persons with the highest stress scores (the event stressed sample) and 175 persons with the lowest stress scores (the event nonstressed sample) would be selected from among them. One-fourth of the sample would be selected at each of four data collection points. Data from the two groups would be contrasted to provide information on factors related to the development of early signs of PTSD. Data from the event nonstressed groups would be contrasted with data from the usual care group (described below) to provide information on the incidence of PTSD among those who do and do not show high levels of initial distress. Em- ployee record measures of work productivity and medical usage would be used in these analyses along with self- report data. Comparative treatments study. The event stressed sample would serve as a subject pool for the comparative treatments study. All subjects in this sample would be asked to participate in an evaluation interview, which would be a semistructured inquiry into the stress event and the nature of the person's response to it. Interviews would be audio- or videotaped and would be conducted according to a manual. Following the interview, subjects would be asked to complete the additional self-report measures. Information from the evaluation interview would be used to confirm the presence of early signs of PTSD. Subjects also would be screened for exclusion criteria, including having had psychotherapy since the event occurred, chronic misuse of drugs or alcohol, symptoms of psychosis, in- volvement with litigation concerning the event, and any 168 reservations about giving written consent for participation in the research procedures. Persons with cancer or other serious continuing illnesses also would be excluded, along with those with severe characterological disturbances, chronically chaotic lives, or complex and mixed syndromes. All excluded subjects would be referred elsewhere. Once identified, subjects would be randomly assigned to the stressed usual care group and the stressed treated group. The stressed treated group would be offered a program of stress counseling consisting of the brief, time- limited (12 hourly sessions) psychotherapy and of three followup interviews. The treatment would be used as a form of secondary prevention in this case, as it is being administered experimentally to persons with early signs of PTSD. Followup interviews would take place at 13 months post- stress—event (an average of 6 months posttreatment), 25 months post-stress—event, and 2 years after termination of therapy (about 33 months after the life event). All ther- apies and followup evaluations would be audio or video re- corded with the informed consent of the subject. All subjects would be asked to participate in followup inter- views, even if they refused treatment or dropped out prior to the full program of 12 sessions. The stressed usual care group would not be offered treatment and would use their own resources in obtaining any help they may decide upon. They might seek help from friends, religious groups, community mental health facil- ities, or family physicians. Subjects in this group would be asked to return for the same three additional clinical in- terviews at points that coincide with the followup inter- views for the stressed treated group. Information on the formal and informal sources of assistance used by this group would be collected via a questionnaire that accom- panies the evaluation interviews. Data from the stressed usual care group and the stressed treated group would be contrasted to provide information on the impact of the preventive treatment on the devel- opment of PTSD. Schedule of Main Study Operations The study would take 4 years and 2 months to complete, with data analyses extending through the fifth year. Data for the longitudinal study would be collected at six points, 169 spaced at 4-month intervals over 2 years. Selection and interview of subjects in the event stressed sample would be done in four subgroups, beginning after the second data collection point and extending through the fifth data col- lection point. The sixth data collection point would be used to select additional subjects for this sample if fewer than the expected sample had agreed to be interviewed prior to that point. Subjects in the stressed treated group would be selected and offered psychotherapy in four subgroups, be- ginning after the second data collection point. If the par- ticipation rate is as expected, the last psychotherapy would be completed 1 year and 4 months later. Followup inter- views of participating subjects from the event stressed sample, including subjects from both the stressed treated and stressed usual care groups, would begin in the seventh month of the study and extend through the second month of the fifth year. The three followup interviews would be scheduled at 13 months, 25 months, and 33 months post- stress-event. In all, the study would call for six data col- lection periods, supplying six sets of data on 1,500-2,000 subjects, along with 420 clinical interviews of 140 subjects, and 40 psychotherapies, each of 12 sessions. Measures This study would use three sets of measures. One would be the standard battery of rating scales used at every evaluation at our Center. A second, shorter battery would be designed to obtain pre-stress-event prospective data within the necessary time constraints. These data would provide the selection criteria needed to pick the at-risk sample to be seen for evaluations. The third set of data would be obtained from work records at the cooperating company and at the company's employee benefits insurance carriers, meeting safeguards to protect the confidentiality of the subject participants. Data Analysis The data analysis would be divided into primary and sec- ondary analyses to reduce the number of tests done in the first examination of results. The primary data analyses would be used to provide tests of specific prestated hy- potheses. Results of the primary analyses would not nec- 170 essarily compromise the most important results of the study. Planned analyses and prestated hypotheses for each of the three studies are described here. Longitudinal study. Information on four main questions would be collected from the longitudinal study. 1. What is the incidence of major stress events and of PTSD in the sample? The analysis would begin by identifying persons who have experienced major stress events (bereavement, injury to self or property, and divorce). The cumulative and noncumulative frequency and percentage of persons experiencing each event would be charted for each of the six data collection periods. Then, cases of PTSD would be identified from among those reporting events. The number and per- centage of persons reporting symptoms of PTSD at 8 and 12 months following each event would be charted. 2. What factors are related to the incidence of major stress events? Specific hypotheses include: that younger age, female gender, prior pathology, and low levels of perceived social support would be related to increased incidence of major stress events. The primary data analysis would be a multiple regression with age, sex, total self-reported pathology score, total quality of life "met needs" score, and the two-way interactions as independent variables. The number of recent events and a calculated presumptive stress score (for recent events only) would serve as dependent variables. The pathology and "met needs" scores would be taken from the first data collection point, while the recent events and pre- sumptive stress score (from these events) would be taken from the second data collection point. This analysis would be performed twice on each half of the available sample and cross-validated on the. other half, in a double-split design. The secondary data analysis would use a similar ap- proach, with the remaining demographic characteristics (race, education, occupation, marital status, and num- ber of children) and symptom and personality trait subscale scores as the independent variables. Here, too, the analysis would follow the double—split cross—vali- dation format. 3. What is the pattern of response to specific major stress events, including bereavement, injury to self or property, and divorce? The subgroups of persons ex- periencing each event would be used as a sample. The 171 mean and standard deviation of their total pathology scores and intrusion and avoidance scores would be charted for three 4-month periods following the event. 4. What factors are related to the development of PTSD? Specific hypotheses include that younger age, female gender, prior pathology, and low levels of perceived social support are related to the development of PTSD. The sample used in the primary data analysis would consist of those persons who report signs of PTSD at 8 months following a major stress event and an equal number of randomly chosen persons who have experi- enced a stress event and do not report signs of PTSD. A discriminant function analysis would be used to examine between-group differences, with age, sex, total patho- logy score prior to the stress event, and needs met scores prior to the stress event as the independent variables. We would also examine the two-way inter— actions as independent variables. The secondary data analysis would use a similar sample and approach, with independent variables to include presumptive stress score prior to the event, type of stress event, the remaining demographic char- acteristics, and the symptomatic and personality trait subscale scores from the period prior to the stress event. Contrast groups study. Data on two additional ques- tions would be compiled during the contrast groups study. 1. What is the pattern of stress response in persons who do or do not show high initial postevent distress? Data reduction techniques would be used to construct a dis— turbance index from relevant self-report scales and hard data. The mean and standard deviation of the dis— turbance index scores would be charted separately for the 175 subjects in the event nonstressed group and the 13S untreated subjects in the event stressed group (175 minus 40 who receive treatment) for three 4-month periods following the event. A slope would be fitted for each subject's disturbance score at 4, 8, and 12 months postevent and a between-group t-test used to assess the difference between the slopes of the subjects in the two groups. 2. What is the relationship between a high level of initial postevent distress and the development of PTSD? It is hypothesized that persons with a high level of distress 1-4 months following a major stress event would have a 172 great probability of meeting the criteria for diagnosis of a chronic PTSD at 8-12 months postevent. Data analysis would consist of a 2x2 chi-square analysis of the association between presence or absence of high initial distress and presence or absence of reported signs of chronic PTSD. Comparative treatments study. A final question is considered in the comparative treatments study. What is the impact of a brief preventive psychotherapy on the incidence of PTSD? It is hypothesized that, in con- trast to the usual care group, the stressed treated group would show significantly lower incidence of PTSD and of general disturbance at 12, 25, and about 33 months fol- lowing the stress event. The first primary data analysis would consist of three 2x2 chi-square analyses of the association between group membership and presence or absence of PTSD at each of the three followup points. For the second primary data analysis, a slope would be fitted to each subject's values on the constructed disturbance index scores at the first, second, and third followups. Analysis of covariance would be used to assess the between-group difference in change patterns with group membership as the independent vari- able and disturbance score at intake and variables identi- fied as differing between groups in the randomization check as covariables. The secondary data analysis would be used to identify variables related to treatment success. For the stressed treated groups alone, a hierarchical regression would be conducted with the subjects' fitted disturbance slopes as the dependent variable and demographic characteristics, preevent presumptive stress score, type of stress event, symptomatic and personality trait subscale scores, pre- treatment disturbance index, sex of therapist, individual therapist, and the two-way interactions between these variables as independent variables. Closing Remarks We have presented one approach to research on the preventive treatment of stress-related disorders. Our ob- jective in presenting this research has been to stimulate interest in this area. We believe that the preventive 173 treatment of stress-related disorders offers a potentially cost-effective opportunity to reduce human suffering. Much more information is needed, however. The goal of research in this area is to determine whether and how to intervene effectively. In our opinion, the first answers to this two-phase question call for large sample research using a longitudinal, prospective design. Such a study would provide information on: (1) stress—specific contributions to problems in functioning, (2) the incidence and prevalence of stress-related disorders, and (3) specific disorder-re- lated risk factors. Appropriate contrast-groups designs could be used to isolate optimally timed, effective inter- ventions. We look forward to developments in this exciting new field of prevention research. References Adler, D.A.; Levinson, D.J.; and Astrachan, B.M. The con- cept of prevention in psychiatry: A reexamination. Ar- chives of General Psychiatry 35:786-789, 1978. Brown, G.W.; Bhrolehain, M.N.; and Harris, T. Social class and psychiatric disturbance among women in an urban population. Sociology 9:225-254, 1975. Caplan, G. Principles of Preventive Psychiatry. New York: Basic Books, 1964. Chiriboga, D.A., and Dean, H. Dimensions of stress: Per— spectives from a longitudinal study. Journal of Psycho- somatic Research 22:47-55, 1978. Dohrenwend, B.S. Social stress and community psychology. American Journal of Community Psychology 6:2-14, 1978. Gersten, J.C.; Langer, T.S.; Eisenberg, J.G.; and Orzech, L. Child behavior and life events: Undesirable change or change per se? In: Dohrenwend, B.S., and Dohrenwend, B.P., eds. Stressful Life Events. New York: John Wiley, 1974. Gore, S. The influence of social support and related vari- ables in ameliorating the consequences of job loss. Dis- sertation Abstracts International 34:5330-5331, 1974. Gruen, W. Effects of brief psychotherapy during the hos- pitalization period on the recovery process in heart at- tacks. Journal of Consulting and Clinical Psychology 43: 223-232, 1975. Hinkle, L.E., Jr. The effect of exposure to culture change, social change, and changes in interpersonal relationships 174 on health. In: Dohrenwend, B.S., and Dohrenwend, B.P., eds. Stressful Life Events. New York: John Wiley, 1974. Horowitz, M.J. Stress Response Syndromes. New York: Jason Aronson, 1976. Horowitz, M.J.; Krupnick, J.; Kaltreider, N.; Wilner, N.; Leong, A.; and Marmar, C. Initial psychological response to parental death. Archives of General Psychiatry 38:316-323, 1981. Kaplan, B.H.; Cassel, J.C.; and Gore, S. Social support and health. Medical Care 15:47-58, 1977. Lamb, H.R., and Zusman, J. Primary prevention in per-— spective. American Journal of ‘Psychiatry 136:12-17, 1979. Lazarus, R.S. Psychological Stress and the Coping Process. New York: McGraw-Hill, 1966. Lazarus, R.S. The psychology of coping; Issues of research and assessment. In Coelho: G.V., ed. Coping and Adapta- tion. New York: Basic Books, 1974. Madison, D.C., and Walker, W.L. Factors affecting the out- come of conjugal bereavement. American Journal of Psychiatry 113:1057-1067, 1967. Nuckolls, K.B.; Cassell, J.C.; and Kaplan, B.H. Psycho- social assets, life crisis and the prognosis of pregnancy. American Journal of Psychiatry 134:90-101, 1976. Polak, P.R.; Egan, D.; Vandenbergh, R.; and Williams, W.V. Prevention in mental health: A controlled study. American Journal of Psychiatry 132:146-149, 1975. Raphael, B. Preventive intervention with the recently be- reaved. Archives of General Psychiatry 34:1450-1454, 1977. Williams, W.V., and Polak, B.R. Follow-up research in pri- mary prevention: A model of adjustment in acute grief. Journal of Clinical Psychology 35:35-45, 1979. 175 176 Chapter 9 LOSS AND ANTICIPATED DEATH: RISK FACTORS IN DEPRESSION Larry E. Beutler, Ph.D. Introduction Clinical conceptualizations of depression have often im- plicated the role of loss or anticipated loss. The clearest and potentially most significant loss may be that incurred by the death of a loved one. Indeed, contemporary be- reavement literature has documented that spouse survivors are at a greatly increased risk of developing clinical de— pression. Among such individuals, pronounced grief pat- terns have been noted that, at least within the first year of bereavement, are essentially indistinguishable from major depressive disorders. Moreover, ‘the incidence of other mental and physical disorders is escalated from four to seven times over base rate during the bereavement period. Parkes (1972), for example, examined the records of over 3,000 hospitalized patients and found that the number who were admitted for psychiatric reasons following the death of a spouse was six times greater than expected. Likewise, evidence suggests that survivors tend to undergo an in- creased and sustained intake of hypnotic medication, alcohol, and tranquilizers (Parkes 1972; Clayton 1973). Sedative ingestion increases threefold during the first 6 months following loss and sevenfold by the end of 18 months. This general pattern of escalating frequency and severity of symptoms over the course of time following a loss is a fairly typical observation. Stein and Susser (1969), for ex— 177 472-367 0 - 85 - 7 ample, found that the rates of hospitalization increase substantially during the second and third years. This pat- tern is apparently exacerbated among groups under the age of 65 (Parkes 1972) and among those who lack a family support system. There is also an apparent sex effect in the rate at which clinical symptoms and syndromes develop during the bereavement period. Males seem to be partic— ularly prone to a variety of disturbances within the first 6 months following the death of a spouse (Young et al. 1963), while female survivors become increasingly symptomatic for a period extending up to 3 years (Cox and Ford 1964). Such observations underline the suggestion by Holmes and Masuda (1973) that the death of a loved one is the most significant of psychological stress events. Yet much re- mains to be understood about the nature of risk in the be- reaved population. Because of the difficulty distinguishing between a normal grief response and clinical depression, incidence rates of depression following loss are very dif- ficult to ascertain. While it seems clear that a large num- ber of individuals who engage in depression-like behavior (e.g., suicide attempts) have suffered a significant loss (Bunch 1972), efforts to apply formal diagnostic criteria (Feighner et al. 1972) have found diagnosable depression to be less prevalent as time since the loss increases. In com- parisons of bereaved samples with hospitalized controls who have primary affective disorders, Clayton et al. (1974) noted few specific differences in clinical manifestations. Indeed, it appears that the only significant feature on which clinical depression can be distinguished from grief is its chronicity. Raphael (1978), in observing the com- monalities between depression and bereavement, suggests that the distinguishing feature is that depressed individuals cannot rally from feelings of despair and hopelessness as is typically done in the course of bereavement. While the bereavement period has long been seen as a prototypic model for studying depression, the methodolog- ical difficulties encountered in this area of research have precluded the development of clear indicators of risk. While major depressive disorders may occur and linger in a significant percentage of the bereaved population, the factors distinguishing this group from those who will not develop a lingering depression have not been clearly de- termined. Virtually all studies, heretofore, have been of the bereaved after the death of their close relative (Parkes 1972; Clayton 1973; Epstein et al. 1975; Jacobs and Ostfeld 1977; Raphael 1978). Unless clear indicators can be found 178 through intensive, longitudinal studies of individuals subjected to bereavement or anticipated loss, efforts either to refine indicators of risk following loss or to de- velop preventive treatment programs will, of necessity, lack much specificity and focus. Comprehensive procedures have not yet been developed for predicting who will be at risk during the bereavement or anticipatory grief period, and little energy has been devoted to developing preventive programs to ameliorate or reduce this risk. The current research program represents an effort to understand the phenomenon of depression as it occurs in the course of adjusting to loss and to develop a clear un— derstanding of factors that place one at risk for incurring such depression. We consider such understanding to be a necessary precursor to the development of preventive treatment programs. While our hope is to eventually de- velop a preventive trial, our current focus is on-under— standing the factors that produce depression in the course of both loss and anticipated loss. We aim also to develop improved means for differentiating between those who will pass through the grief cycle and those who will not, de— veloping instead a major affective disturbance. In accomplishing the dual objectives of understanding the nature of depression and developing predictive indicators for its occurrence following a significant loss, the current research program focuses on three major populations: (1) individuals in major crises, anticipating their own death; (2) spouse survivors; and (3) adult children of patients who are anticipating or who have succumbed to death. One may rightly question the importance of studying individuals who are anticipating their own death within the context of an effort to develop a preventive treatment protocol. However, it is our assumption that the develop- ment of depression in response to anticipated loss can be most clearly seen in these individuals. We have reasoned that the evolution of depression might be magnified in these people, since no greater anticipated loss can be con- ceived. If we can understand the development of depression in these individuals, we may then be able to determine the significance of certain observed relationships in a popula- tion wherein the predictor variables themselves are less obvious. Our focus on the surviving spouses and adult children of dying patients is more logical. Here, we anticipate that the stability of social support systems and the closeness of personal attachments will have major significance for the 179 development of depressive disorders. Yet, these popula- tions must be studied in concert if we wish to determine both the ameliorating effects of social supports on the im- pact of major life events, such as loss, and the recipro- cating impact of those events on social support systems. This ever-reciprocating influence between life event and emotional-social consequence must be considered as in- teracting units across time. This presentation will outline briefly a two-pronged re— search program consisting of two major studies. The first study is an investigation of patients with a life trajectory of under 2 years. This study was initiated, as indicated, to clarify the nature of depression as one anticipates a major life event and to target certain types of variables that may be significant in an effort to develop a prevention program. This study was undertaken as part of a multicenter clinical trial investigation of laetrile in the treatment of patients with cancer. It was a small sample investigation, which sacrificed some of the benefits of a large-scale investi- gation to the depth of focus allowed by intensive interviews. The second study presented here is currently under re- view and development. It is an effort to specifically in- vestigate the affects of anticipated loss and loss of a significant family member on spouses and surviving chil- dren. This planned study is designed to serve as a precursor to the development of a prevention trial, while developing prognostic indicators that will allow a clear delineation of that part of the bereaved population most likely to be at risk for incurring a major depressive disorder. Study | Purpose and Procedures As a participating center in the federally funded laetrile clinical trial, our organization has undertaken to assess social and psychological data on matched groups of patients seeking either laetrile or other phase II agents. This study was conducted in response to the observation that many patients, when given a poor diagnosis for their cancer, seek unproven and ineffective alternatives to con- ventional treatments. This search for an alternative in the face of hopelessness has been suggested as a protective 180 mechanism erected against depression. In the current study, we undertook to investigate the psychological states of patients who elected laetrile over a more conventional but still experimental treatment. We also investigated these patients' attitudes toward health, their experience with previous physicians, their attitudes toward other so- cial support systems, and their relationships with the phy- sicians who administered the experimental treatments. Patients from a wide geographic area applied to the University of Arizona to be treated with laetrile. Upon arrival for treatment, patients were given the option of enrolling in the laetrile treatment program or participating in some other phase II treatment project. Nineteen patients, when presented with this choice, elected to re- ceive laetrile. Ultimately, two of these patients were ex— cluded from participation in the current project, because their medical conditions precluded their meeting the cri- teria for the laetrile clinical trial. The 17 patients who both elected laetrile over other treatments and were em- pirically suited for inclusion were group-matched with 17 patients who expressed preference for and received other investigational phase II agents. Matching was based on sex, age, and geographic distance from the treatment site. Table 1 is modified from data originally presented else- where (Redding et al. 1981) and describes the two samples in terms of the matching variables. In order to further assess the comparativeness of the two samples, particularly along psychological dimensions, an SCL-90R (Derogatis et al. 1976) was administered to both groups. No significant differences were obtained, and, in fact, no mean differences exceeded one-half of a standard deviation. The primary data for the current study were obtained through a series of questionnaires and structured inter— views specifically developed for this project. Three popu- lations were of concern in the course of gathering data. The primary source of data came through interviewing and evaluating the patients themselves. Secondary, focus was directed to the patients' physicians—those directing the experimental treatments to which the patients were as— signed. Third, a sample of 23 family practitioners was evaluated to compare their attitudinal responses to those of the treating oncologists. Because of the nature of the study and the nature of the patients' conditions, no effort was made to maintain a therapeutic blind among either the physicians or the 181 Table 1. Matching variables Laetrile Phase II Variable n=17 n=17 Sig. Sex n.s. Males 9 9 Females 8 8 Age n.s. Mean 55 55 Range 18-72 18-72 Annual income n.s. Mean $15,000 $18,000 Range $10,000-50,000 $10,000-50,000 Education n.s. Mean 12 years 12 years Range 8-17 years 8-17 years Marital status n.s. Married 13 14 Other 4 3 Geographic distribution n.s. County 6 7 State 5 6 Out of State 6 4 patients. Investigation of the matched samples was undertaken in three steps. First, we systematically interviewed all patients, obtaining quantifiable data on dimensions related to support systems, beliefs about health care systems, previous treatment experiences, and coping strategies. We also obtained their impressions and expectancies about the value of the treatment they were about to receive. These interviews were cross—checked periodically for reliability by a second interviewer. In the second step, we systematically evaluated all 182 patients, using standardized assessment devices that in- cluded the SCL-90R, the Health Locus of Control Index (Wallston et al. 1976), and the Life Experiences Scale (Holmes and Masuda 1973). As a third step in this study, we investigated the atti- tudes of the physicians who treated the patients. Physi cians were evaluated relative to their attitudes toward laetrile and their beliefs about patients who sought such treatment. In addition, they were asked to make ratings of study patients under their care. Six physicians maintained patients who were being treated in both groups. These latter ratings reflected psychiatric symptoms and were based on an analog form of the SCL-90 in order to yield scores comparable to those obtained from the patients themselves. As observed in table 2, patients in the two groups did not differ in the type of intensity of symptoms assessed by these means. In an effort to determine if phy- sicians perceived patients in the two groups differently, we looked first at the entire group of patients for any dif- ferences that may have accrued in doctors' perceptions. We then performed a separate analysis of the attitudes expressed by the six physicians who treated patients re- ceiving both types of treatment. Finally, we contrasted the attitudes of doctors in our study conditions with those of an independent group of 23 family practitioners, in terms of their attitudes both toward laetrile as a treatment for cancer and toward patients who sought such treatment. Results and Discussion Because of the absence of untreated controls, double— blind, and the small sample size, all results of this study must be considered very cautiously. Keeping in mind our original objective of determining some characteristic pat— terns that might precipitate depression, however, the re— sults are nonetheless interesting. Examination of patients' attitudes revealed no signifi- cant differences in the expectations of benefit as a func- tion of the particular treatment they were receiving. Patients in both groups remained surprisingly optimistic about the treatment program selected, and most indicated that they would recommend their particular treatment to others. No significant differences were shown in the two groups of patients' previous life experiences or attitudes toward health, the medical profession, chemotherapy, or 183 Table 2. Comparison between ratings of scores on dependent variables Oncologists and family Family practitioners — (t-tests)* Oncologists practitioners P n=12 n=23 1. Ratings of expected laetrile effectiveness Mean score (high is positive) 2.2 2.5 N.S. S.D. .66 44 2. Attitudes toward patients who seek laetrile Mean score (high is positive) 3.3 3.6 N.S. S.D. 1.9 .62 3. Composite attitude score Mean score (high is positive) 3.0 3.1 N.S. S.D. .26 .46 Physicians' perceptions of laetrile patients (t-tests) SCL-90R analog Laetrile Phase II Sig. Somatic 28.2 + 28.48 29.5 + 26.3 N.S Obsessive— compulsive 24.0 + 26.1 32.6 + 29.6 N.S. Interpersonal sensitivity 29.3 + 22.6 27.9 + 25.7 N.S. Depression 53.1 + 30.5 28.9 + 22.7 N.S. Anxiety 60.2 + 29.2 49.5 + 25.1 N.S. Hostility 23.4 + 26.2 18.8 + 18.19 N.S. Phobic anxiety 25.6 + 27.2 19.8 + 19.4 N.S. Paranoia 22.6 + 18.5 12.5+12.8 N.S. Psychoticism 19.6 + 26.7 14.1 + 14.2 N.S. Dr.'s confidence in treatment 1.2+1.13 3.2+1.08 .0S *Ratings were based on five-point Likert scales. unproven methods of cancer treatment. A common concern expressed by both groups was their dependency on others. Indeed, over half the patients in each group reported feeling less attractive and more unhappy about having to depend on others for their care. 184 Approximately 18 percent of the patients in both groups had moderate to severe depressive symptoms and anxiety, as indicated by levels of symptomatic expression on the SCL-90R. Nine percent of the patients in both groups re- ported scores on the Holmes Life Events Scale (Holmes and Masuda 1973) that suggested a high incidence of change in the past year. Types of life change did not discriminate among the groups. The major difference between patients who sought la- etrile and those who selected other phase II agents was in terms of their perceptions of available social support sys— tems. Considerably fewer patients in the laetrile treatment group, as compared with those in the phase II treatment group, indicated that religious systems had provided a basis of emotional support for them in coping with cancer. Sim- ilarly, only two of the laetrile patients felt that they had received emotional support from their previous physicians, whereas most patients in the phase II group felt that their physicians had been supportive. Laetrile patients were more than four times as likely to report that their previous physicians had told them there was no hope for their condition. This finding led us to com- pare the patients' attitudes toward their current physi- cians—those applying the experimental treatments. In this regard, table 2 attests to relatively low levels of felt support in both groups but high levels of perceived com- petence on the part of their doctors. In a further effort to determine if patients' perceptions of their previous physicians may have reflected their own dispositions, we both evaluated the attitudes of the phy- sicians toward the patients and compared the attitudes of oncologists and family practitioners toward laetrile and laetrile patients. In this analysis, we ascertained that phy- sicians providing treatment to both groups had a con- siderably more negative attitude toward laetrile than they did toward phase II agents. This pattern was replicated among family practitioners. However, this negativism did not generalize in any overt way to their global views of patients who sought such treatment. When physicians were asked to assess patients in both groups utilizing the symp- tom checklist analog, however, some suggestive differ— ences were observed. While the samples were too small to obtain levels of significance, table 2 reveals that physi- cians' ratings of patients' depression, anxiety, and paranoia were higher among those patients who were being treated with laetrile as compared with those receiving a phase II 185 agent. In each of these cases, a moderate level of effect was obtained with differences exceeding one-half of a standard deviation. Moreover, these differences were dis— crepant from the very consistent scores obtained between patient groups when self-administered the SCL-90R. The current findings suggest that a major contributor to individuals' seeking unproven treatments when facing their own death resides in the response of health care and social systems. If one utilizes the pattern of seeking unproven treatment as an indicator of helplessness among the patients who do so, this helplessness can probably be at- tributed in part both to the absence of religious supports and to the prevalence of previous physicians who had des— cribed the patients' condition as hopeless. Low levels of felt support were especially prevalent in this group in re- lation to their previous physicians but, in fact, were low in both groups even when describing the physicians who maintained their current treatment programs. The absence of emotionally supportive systems and the negativistic views of physicians toward both patients and their future might be very significant in the development of depression in those who are anticipating major life changes. Study II Purpose Building on previous findings, the current study proposes to more clearly extend our understanding of depression among those who are facing major losses. In this proposal, we are attempting to compensate for some of the lack of methodological sophistication in the foregoing study and to test an assessment procedure for evaluating spouse/sur- vivors' risk of incurring depressive symptoms of clinical significance during the bereavement period, both before and after a spouse's death. We also plan to apply the same procedures among adult children through the course of a 2-year period of adjustment following loss, during which time most postloss morbidity occurs. We intend to clarify patterns of interrelationship among the identified patient, his spouse, and his children in the course of this adjustment period. Marital adjustment patterns will be assessed, and variables thought to be significant to the successful coping of individual family members will be measured. Ultimately, 186 the plan is to develop and implement a preventive treat— ment regimen for spouses and family members, which will be initiated prior to the patient's death and will be de- signed to prevent the occurrence of depression among those who are at highest risk. The current proposal, how- ever, is for an initial funding period designed to develop and refine an assessment procedure for determining an individual's risk and not for the development of a preven- tion program itself. We have reasoned that if we are successful in developing a reliable assessment battery that can adequately both predict risk and acquire a further understanding of the na- ture of postloss depression, we will then be in a position to undertake an extended study that will include a preventive clinical trial. This latter phase will be a prospective re- search program to include spouses and adult children of terminally ill cancer patients. For example, one might randomly assign high-risk spouses either to a supportive treatment or to a minimal clinical management condition. A similar group of low-risk spouses might be maintained on a minimal treatment regimen. We could then assess and follow the spouses and adult children through the course of the patient's illness and for several years subsequent to the patient's death. This process would allow us to evaluate psychological adjustment and family interactional patterns to determine both the effects of the supportive, psycho- social treatment and the natural course of psychological adjustment among those who are at both high and low risk. Method To meet the needs of predicting survivor risk by infor- mation available prior to the death of the significant other, this project will be carried out using a prospective research model. We will concentrate on developing an assessment package that will predict the likelihood of survivors' de- veloping significant depressive and grief symptoms. Indeed, it is our intent to clarify the entire spectrum of grief symptoms. The differentiation between normal grief and clinical depression will be based on both duration of symptoms be- yond a 2-year period and development of clinical syn- dromes. Table 3 reveals the proposed evaluation schedule for patients themselves, their spouses, and their adult children. 187 Table 3. Assessment schedule Data Instrument/concept End of source Pretest Monthly Quarterly year Screening procedures IQ test S Laboratory tests S Memory scales S Predictor variables Demographic Questionnaire (to be developed) Family dynamics Evaluation (to be developed) . Affect sharing . Equitability of decisionmaking . Dependency . Attachment . Equanamity of communication Health locus of control (attitudes toward illness) S,C,P Personality inventory S,C S.C S.C.P oe oan Dependent variables Health status a. Cornell Medical Index S.C b. Physical exam S c. Medication/Dr. visit log S Status of depression a. Depression inventory S.C b. SADS/RDC S General psychological status a. SCL-90R b. Interview/ unstructured S.C S.C.P X X X (otherwise only if a ques- tion of dementia is raised) xx xx xX xx xX xx x S = Spouse C = Adult child P = Patient 188 As originally conceived, subjects for this project were designated to be 50 spouses and identified patients, along with all of their adult children. We have designed to in- clude approximately equal numbers of male and female patients. All designated individuals are to be postparental couples (no children under the age of 18), patients in whom terminal cancer has been diagnosed and whose life tra— jectory is less than 1 year, and those who do not evidence confounding medical disorders. At this point, we do not intend to rule out individuals with preexisting illnesses, as long as they do not require ongoing medication that might be expected to affect their emotional and mental status. Likewise, we have no intention of excluding individuals above or below a given age, designing instead to investi- gate age as a potential risk factor. Recently we have considered redesigning the study to include a second group of spouses and adult children who have lost a family member through sudden, traumatic in- jury. The inclusion of such a control group would allow us to investigate the effect of anticipatory grief in the de- velopment of depressive symptoms and syndromes. As referenced in table 3, a combination of structured interviews, paper and pencil assessment procedures, rec— ords of psychological and medical status, medical evalu- ation, and self-reports will provide primary data for this study. In our effort to differentiate between depressive symptoms and depressive syndromes, we will monitor var— iations in depressed mood through self-report measures. These SADS and RDC will be periodically utilized to de- termine the presence of a major depressive syndrome. This study plan calls for intensive evaluation at frequent points, in the hope of unraveling some of the complex interactions among the various symptoms themselves and between de-— pressive symptoms and syndromes. We are aware of the possiblity that syndromes of major depressive disorders will manifest themselves at different times for individuals with different clusters of predisposing factors. Hence, intensive evaluation will be required to differentiate appropriately between transitory manifestations and enduring symptom patterns. By combining formal assessment procedures with de- mographic information and intensive interview data, we hope to compose a risk assessment battery. Collectively, this battery will combine information about marriage, de- mographic background, available support systems, previous medical and psychiatric history, current emotional states, 189 current coping patterns and abilities, other life stresses, and family dynamics. The evaluation battery, when admin- istered to the S50 families, will be regressed against mor- bidity rates over a 2-year postloss period to distill the most consistent predictors. The dependent variables will include medical usage, medical symptoms, psychological symptoms, clinical de- pression, and shifts in family supports or family dynamics. One will note that family patterns will be assessed both as a predictor variable for major affective disorders and as a consequence of the development of these disorders. Methodological Issues When exploring risk factors in depression or other types of stress-related psychopathology, using populations who are in acute distress raises many important methodological and ethical issues. While loss and the anticipation of death appear, both clinically and empirically, to place individuals at significant risk for a multitude of physical and psychi- atric disorders, the very act of studying such populations may place stress on an already stressed system. One must give ample consideration to the provision of support sys— tems and referral sources to such individuals. Yet, by so doing, these persons' vulnerability to the symptomatology under study may be altered. By the same token, only by studying individuals in the absence of such treatment pro- grams can one hope to find a baseline measure against which prevention programs can be assessed. Methods must be developed to more adequately study relevant social support systems within the health care enterprise and within society more broadly (e.g., religious and family systems). Likewise, assuming the importance of these factors, systems must be developed for altering and fa- cilitating such supports within existing social-religious and health care systems. A notable lack in the current literature revolves around the absence of available instrumentation to assess family roles, dynamics, and other aspects of support. While elab- orate systems for assessing families are available, they are infrequently usable within the context of risk research on large populations. The current literature documents the role of loss in the development of subsequent depression. Yet, much conflict 190 still exists about the variables that precipitate major de— pressive disorders among those who have suffered loss. We do not know the strengths nor the vulnerabilities that lead some individuals to cope with loss and others to succumb to it. Likewise, we do not know the characteristics that fa— cilitate some people's successful adjustment to the antici- pation of their own death, while others become despondent, despairing, and distressed and seek unproven and perhaps dangerous remedies. In order to understand these factors, close scrutiny of case material is helpful and should be followed by intensive studies of relatively small popula— tions of individuals who are at high risk. Broad-ranging but relatively superficial evaluations of large populations have resulted in the accumulation of much contradictory data. It is the author's judgment that, until such determinations are made to more clearly specify the risk that is present to those who undergo loss, prevention programs will be un-— necessarily broad based and nonspecific in their focus and effectiveness. Nonetheless, intensive studies of both individuals and small groups of high-risk subjects raise important questions of how much one can intervene in the normal course of events to obtain usable data without altering the natural course of the phenomenon under investigation. Similarly, important questions are raised in the course of such studies as to the nature of assessment and the means by which important variables might be evaluated. Not only is there inadequate instrumentation in the area of family assess— ment, but objective data on broader social support systems are also unavailable through efficient assessment pro-— cedures. Neither is there instrumentation readily available by which subtle changes in physical and psychological symptomatology can be assessed on frequently repeated measurement occasions. While self-report devices are reasonably effective in determining the presence of symp- toms, they are still ineffective in determining the presence of syndromes. Because depression and anxiety seem to be common ingredients in a wide variety of syndromes, both medical and psychological, the presence of these symptoms cannot be equated with the presence of major depressive disorders. A wider variety of devices must be made avail- avle for research to proceed efficiently. While such in- struments as SADS/RDC are reliable and effective in de- termining syndromes, they are also difficult to utilize, and it is not clear that they will be effective in efforts to dif- ferentiate grief reactions from enduring depressive 191 disorders. Finally, with any intensive study, a significant problem rests in condensing massive amounts of data to usable di- mensions. A combination of factor analytic and clustering strategies has been suggested by which data might be com- bined along statistically defined dimensions for further analysis. Collecting multiple measures into usable units before outcome analysis is undertaken does offer the hope of applying statistical procedures to even relatively small groups of subjects. However, deciding between intensive and extensive studies is still difficult, and there may be value in combining both approaches sequentially in a re- search program. To some degree, this procedure has been undertaken in the research program described in the pre- vious pages. In a related way, a need still exists to develop improved ways for assessing the reliability of applying treatment strategies. While the current research protocol has not outlined a treatment or prevention strategy, that is clearly the aim of such research activities. In doing so, however, one should be aware that psychological treatments are particularly difficult to evaluate in this way. While efforts are being developed to ensure that treatments are applied in a reliable fashion, methods for both differentiating among treatments and ascertaining the active ingredients in each are still relatively primitive. Once we are able to determine the variables that predict the development of clinical depression in loss or anticipated death, much still needs to be done. These variables must be related to spe- cific treatment strategies and applied in a reliable fashion to the specific risk characteristics of the population and across both occasions and therapists. Acknowledgments Special acknowledgment is extended to my research colleagues on Study I: Ms. Karen Redding, Dr. Stephen E. Jones, Dr. Thomas E. Moon, and Mr. Peter Schur, all from the University of Arizona. Ms. Redding and I were co- principal investigators for this project. Study II was initially developed at the instigation of Dr. Stephen B. Shanfield. Drs. Shanfield, Beutler, and Shisslak have served as coprincipal investigators for the proposal. 192 References Bunch, J. Recent bereavement in relation to suicide. Journal of Psychosomatic Research 16:361-366, 1972. Clayton, P.J. The clinical morbidity of the first year of bereavement: A review. Comprehensive Psychiatry 14:151-157, 1973. Clayton, P.J. Mortality and morbidity in the first year of widowhood. Archives of General Psychiatry 30:747-750, 1974. Clayton, P. The sequelae and nonsequelae of conjugal be- reavement. American Journal of Psychiatry 12:136, 1979. Clayton, P.J.; Herjanic, M.; Murphy, G.E.; and Woodruff, R., Jr. Mourning and depression: Their similarities and differences. Canadian Psychiatric Association Journal 19:309-312, 1974. Cox, P.R., and Ford, J.R. The mortality of widows shortly after widowhood. Lancet 1:163-164, 1964. Derogatis, L.R.; Rickles, K.; and Rock, A.F. The SCL-90 and the MMPI: A step in the validation of a new self report scale. British Journal of Psychiatry 128:280-289, 1976. Epstein, G.; Weitz, L.; Roback, H.; and McKee, E. Re- search on bereavement: A selective and critical review. Comprehensive Psychiatry 16:537-546, 1975. Feighner, J.P.; Robins, E.; and Guze, S.B. Diagnostic cri- teria for use in psychiatric research. Archives of General Psychiatry 26:57-63, 1972. Holmes, T.H., and Masuda, M. Life change and illness sus- ceptibility. In: Scott, J.P., and Senay, E.C., eds. Separation and Disease. Washington, D.C.: American Association for the Advancement of Science, 1973. Jacobs, S., and Ostfeld, A. An epidemiological review of the mortality of bereavement. Psychosomatic Medicine 39:344-357, 1977. Parkes, C.M. Bereavement: Studies of Adult Grief. New York: International Universities Press, 1972. Raphael, B. Mourning and the prevention of melancholia. British Journal of Medical Psychology 51:303-310, 1978. Redding, K.; Beutler, L.E.; Jones, S.; Meyskens, F.; and Moon, T. "Psychosocial Attitudes of Cancer Patients Treated with Laetrile or Other Phase II Agents." Paper presented at the annual meeting of the American Society of Clinical Oncology, New York, April 1981. Stein, Z., and Susser, M. Widowhood and mental illness. British Journal of Preventive Social Medicine 23:106- 193 472-367 0 - 85 - 8 110, 1969. Wallston, E.D.; Kaplan, G.D.; and Maides, S.A. Develop- ment and validation of the health locus of control (HLC) scale. Journal of Consulting and Clinical Psychology 44:580-585, 1976. Young, M.; Benjamin, B.; and Wallis, C. The mortality of widowers. Lancet 2:454-456, 1963. 194 Chapter 10 LIFE EVENTS IN SCHIZOPHRENIA: MEASURING THE EFFECTS AND MODIFYING THE IMPACT—A PRELIMINARY REPORT Ian R.H. Falloon, M.D. Jean Pederson, M.S. Kim Shirin, M.S. Ronald Moss, B.S. Introduction The task of clarifying the role of life events in the onset and course of schizophrenia has been hampered by method- ological problems, related mostly to the inherent limita— tions of the retrospective research design. Only one prospective study of life events and schizophrenia has been reported (Michaux et al. 1967). This study suggests a re- lationship between life stressors and relapse, but due to weak methodology in other aspects of the design, the gen- eralizations that can be drawn from the findings are quite limited. The two most definitive studies in the areas of life events and schizophrenia have been retrospective in design. Brown and Birley (1968) looked at four 3-week periods prior to schizophrenia onset. They developed their own methodology for scoring life events, judging life events as "events which on common sense are likely to produce emo- tional disturbance in many people." The contextual fea— tures of the event, which might make it more or less disturbing for the average person in that situation, were taken into account in the scoring procedure. Brown and 195 Birley also judged events on independence, that is, whether the events occurred as a result of factors clearly inde- pendent of the subject's illness or as only possibly inde- pendent of the illness. Three hundred twenty-five normal subjects selected at random from the local working community and 50 schizo- phrenic patients were interviewed about life events in the preceding 3 months. Only schizophrenics whose illness on- set had occurred within 13 weeks of admission and could be dated to within 1 week were included in the patient sam- ple. The authors found that the patient group had expe- rienced nearly double the number of events per person compared with the controls, an average of 1.74 and 0.96 per person per 13-week period, respectively (p < .001). The largest difference was in those events judged to be possibly independent. The difference for the rates of the inde- pendent events did not quite reach statistical significance. Furthermore, there was a dramatic clustering of life events for the patients in the 3 weeks immediately pre- ceding illness onset. During the other weeks, the frequency of life events in the patient group was comparable with the frequency observed in the control group. Jacobs and Myers (1976) also reported a relationship be- tween the frequency of life events and the onset of schizophrenia. Sixty-two first admission schizophrenics were interviewed 2 to 4 weeks following hospital admis- sion. Life events history over the year preceding onset, defined for the schizophrenics as a combination of the oc-— currence of symptoms and change in social functioning, was compared with life events in the preceding year for a group of control subjects selected from the local com- munity. Life events were tallied according to a 58-item scale containing items assembled from several other life event schedules. Jacobs and Myers found that the schizophrenics had ap- proximately twice as many life events as compared with the control group (3.2 versus 2.1, p < .01). The schizo- phrenic group also reported more undesirable and upsetting events and more events involving family, housing, or legal disturbances. However, consistent with the findings of Brown and Birley (1968), Jacobs and Myers reported that the magnitude of the difference did not hold when "events classified as independent of a person's ability to influence them" were excluded from the analysis. The Brown and Birley (1968) and Jacobs and Myers (1976) studies both lend credence to the hypothesis that life 196 stressors are related to the symptomatology of schizo- phrenia. Unfortunately, design problems limit the extent to which influences can be made regarding the nature of this relationship. A more powerful experimental design would be one that is prospective and allows for more exacting observations of the flux in schizophrenic symptoms and life stressors. The present pilot study has been designed as a step toward this goal. We have collected an ongoing record of life events and life changes in persons who have suffered an episode of schizophrenia and their families, over a 1- to 2-year period in conjunction with longitudinal data cover- ing a wide range of the patients' social, physiological, and psychological functioning. We are interested in testing several hypotheses: ® Do life stressors contribute to exacerbations of schizophrenia? ® What is the temporal relationship between life events and exacerbation? ® Does coping behavior modify the impact of life event stress? e® Do life events contribute to depressive symptoms in schizophrenia? What is the temporal relationship? What is the relationship between depressive episodes and schizophrenic symptomatology? ® Are there identifiable warning signals preceding schizophrenic exacerbations? These may include phys iological parameters, such as heart rate and blood pressure, as well as behavioral variables, such as social withdrawal, tension, anxiety, and somatic symptoms. ® Do psychosocial interventions modify the effects of stress on the course of the illness? ® Does the degree of pleasantness of the event affect its impact as a stressor? Methods This study of life stressors in schizophrenia is being conducted within the larger context of a controlled out- 197 come study on the relative effectiveness of individual and family interventions on the course of schizophrenia. A detailed description of this project is available elsewhere (Boyd et al. 1981). Subject Selection Subjects were selected for this study on the basis of a recent episode of florid schizophrenia with evidence of family stress at the time of the episode. The criteria included: ® Age: 17-45 years ® Diagnosis confirmed by Present State Examina- tion/CATEGO and DSM-III criteria ® Living with English-speaking parent(s) (and siblings) ® Clear evidence of ongoing stress in the family living situation as evident from high levels of criticism, hostility, or emotional overinvolvement (high EE) on the Camberwell Family Interview (Vaughn and Leff 1976), or other evidence of major family difficulty such as serious communication or problem-solving deficits. No patients were excluded on the basis of race, sex, edu- cation, chronicity, associated behavioral disturbances, or drug-taking habits. Subjects included a third with first episodes and several who had never been hospitalized. Medication and Stabilization All patients were carefully maintained on minimally sufficient doses of neuroleptic drugs, chiefly chlorproma- zine and fluphenazine (oral and depot), throughout the 12- to 24-month followup period. Medication was monitored by monthly plasma levels of the drugs and compliance main- tained through the judicious use of intramuscular medication. 198 Psychiatric Assessment All psychiatric assessments were conducted at monthly intervals and at times of emergency visits or hospital ad- mission by the prescribing physicians, who were trained to a high degree of interrater reliability on the instruments used (r = 0.85). They were independent of the patient's so— cial status and of any life events data throughout the study. Each patient had a primary therapist, who managed his or her case throughout and who delivered specific psy- chosocial treatment according to random assignment. The following rating scales were employed: Target Symptom Rating (TSR): Two or three florid symptoms of schizophrenia that were most prominent during past episodes were chosen as target symptoms for each patient. Care was taken to specify clearly the symptom as it affected each subject and to avoid choosing behavioral disturbance that might recur in the presentation of nonschizophrenic episodes (e.g., social withdrawal in a depressive episode). Persistent symptoms that showed little variation with the course of the illness were similarly avoided, so that change in the mean TSR would prove a highly sensitive indicator of severity of florid schizophrenic symptomatology. Brief Psychiatric Rating Scale (BPRS): This 18-item rating scale was employed. The factor analytic scales of Goldstein and his colleagues (1978) were used. These were employed in the analysis: I. Hostile/ suspicious; II. Depression; III. Withdrawal; IV. Schizo- phrenic thought. Individual items also were examined, in particular, item #1: Somatic concern. Clinical Global Impression: Two global scales of se- verity of illness were employed. Side Effects Ratings: Ratings of significant side ef- fects and their severity were conducted. Plasma levels of drugs and an index of compliance were assessed at each monthly medication visit. Physiological measures of heart rate and blood pres— sure were conducted carefully by an experienced nurse. 199 ® Family tension levels were monitored regularly. Life Event Assessment Life event data were collected over a 12- to 24-month period. Three research assistants were trained to conduct semistructured interviews with a patient once every 2 weeks. These interviews, which lasted from 15 minutes to over an hour, consisted of a careful inquiry concerning any change or event (change/event) that had occurred in the family during the past 2 weeks. Positive, negative, and neutral change/events were elicited. Some change/events were ongoing rather than discrete, and detailed reports of these ongoing difficulties also were recorded. Change events reported from the previous weeks were given fur— ther inquiry to record subsequent developments. When any change/event was elicited, the research as- sistant then gathered information about each family mem- ber's contingent responses. These responses were recorded under the following headings: ® Behavioral responses: What each household member did following the change/event. Both immediate re- sponses and later actions were recorded. Note was made when no perceptible behavioral change was ob- served by the informant. At times, behavior did not appear to be contingent upon the change/event, but the research assistants attempted to avoid making such judgments and simply recorded what happened. ® Cognitive appraisal: The informant was asked to describe the thoughts and feelings each family mem- ber had in response to the event. In cases where family members had not openly dis— cussed their cognitions with the informant, he or she was invited to infer the nature of the thoughts of others. In addition, the informant was asked to check the accuracy of these inferences and to provide veri- fication at the subsequent interview. With the excep- tion of families where communication was extremely limited, this task was less difficult than expected, and at least gross parameters of cognitive appraisal could be gleaned from the data gathered at present. 200 ® Social network involvement: The informant was asked which persons in the social network of the family had been involved subsequent to the change/event. This included contact with professional agencies. Every person who had been informed about the event was listed, regardless of the nature of his or her involvement. Reliability of Life Change Interview This method of gathering data concerning life changes required a careful examination of several aspects of its reliability. These included: ® Agreement between different family members on description of the event. Sampling from one family informant might be expected to bias reporting of changes involving that person more than other family members, such as the patient. To examine this, each therapist independently gathered information on a typical sample of changes reported to him or her by the patient. The nature of these changes was not di- vulged to the research assistant who interviewed the informant. T..: nature of changes, as well as the re— port dates of onset, often were compared by an inde- pendent rater. The same event was reported by parent and patient on 44 of SO events (88 percent). The re- ported onset was within 1 day in 61 percent and within 1 week in 95 percent. ® Comparability of transcribed interview reports be- tween different interviewers. Because three inter- viewers were employed, there was the possibility of considerable variation in the information each tran- scribed concerning change/events. Fifteen interviews were audiotaped (five by each interviewer) and tran- scribed independently by each interviewer. A blind rater then rated the agreement between interviewers in terms of (a) change/events transcribed; (b) severity of stress potential; and (c) global coping behavior, a measure derived from reported behavioral, cognitive, and social network responses to each change/event. The reliability data were as follows: 201 ® Four out of 30 (13 percent) change/events were con- sidered too minor to be recorded by one of the inter— viewers. These were all treated by the blind rater as 1 on the 4-point blind stress potential scale (1 = low stress potential, 4 = extremely high stress potential). ® The blind rater gave identical stress potential ratings to 63 percent of jointly transcribed interviews and was within 1 scale point on the remaining cases. ® On the 10-point coping scale, the blind rater had exact agreement in 50 percent of the jointly transcribed reports and was within 1 scale point in 92 percent. The levels of agreement on the life change transcripts suggested that the interviewers obtained comparable re- ports from relatives, who, for the most part, were ex- tremely cooperative and who even, in several cases, regarded the biweekly interviews to be a form of nondi- rective counseling. At times, problems of confidentiality arose, usually when patients became resentful of their parents' "talking about them secretly." An explanation of the nature and purpose of these interviews was usually sufficient to ease such tension. Stress Potential Two directions have been taken in scoring the life events data. The first has been to apply the Holmes and Rahe (1967) standardized weightings to the change/events. Each subject's life event record was screened for events that qualified as stressors and the appropriate weightings ap- plied. The second method, now in process, was to apply the scoring methodology developed by Brown and Birley (1968). (One of our research associates was recently trained to use this scoring method by Brown and his colleagues.) Coping Each stressful change/event was globally assessed on a 10-point measurement scale as to the adequacy of the 202 family coping response. Coping has been defined as "any behavioral response (behavior, cognition, affect) by the patient, family member and/or other person that is con- tingent upon a life change/event." It may be either bene- ficial and lead to stress reduction or detrimental and lead to increased stress in the short term (up to 1 week) or the long term (1 week or more). Our ratings were performed by a rater blind to information on symptom exacerbation or relapse. A second rater used the same rating scale, and an interrater reliability of r = 0.78 was obtained. Advantages of the Prospective and Longitudinal Design The present prospective and longitudinal design has many distinct advantages. Several of these will be commented upon. ® The ongoing collection of events (every 2 weeks) re— duces the probability of error in reporting. With a longer delay between the event and the report of the event, there is a greater chance of omission of detail and lack of precision in dating events. Probability also increases that the recall of the event will have been "edited" according to information known to the person some time later but not at the time the event occurred. ® We have elicited the events primarily from a selected family member rather than from the patient. This has been done to avoid distortion in event recall due to illness-related thought and communication disturb- ances. The degree of disturbance may have varied both between patients and within a patient over a given period of time. ® Rather than looking at life events in reference to an approximated point of illness onset, we have focused on the life events and their association with subse— quent fluctuations in schizophrenic symptoms. Our primary goal is to determine whether life stressors contribute to future symptom exacerbation. As such, we are less concerned with the formidable tasks of dating illness onset and of determining whether the 203 life events are dependent or independent of the illness. We would postulate that all stressful life events, re— gardless of their origins, have the potential to cause exacerbations. Furthermore, our subjects were not selected from a population of hospitalized schizophrenics. The decision whether to hospitalize a patient and when to do so is influenced by a complex array of factors: behavioral problems, symptom severity, the availability of a hospital bed, etc. Social factors related to life events may also play a role. For example, families may have a lower threshold for dealing with difficulties arising from the patient's behavior at times of high family stress, e.g., mother ill, father out of work, etc. Se- lecting subjects from a population of hospitalized schizophrenics may introduce a confound, a confound which is avoided in this study by referencing life events to the ongoing fluctuations in symptoms rather than to a criterion of hospitalization. This design allows subjects to serve as their own con— trols. Statistically, this is a very powerful design, which affords a close scrutiny of the relationship be- tween stressors and symptoms. Analysis of the rela-— tionship can be approached in two ways. First, one can identify the occurrence of a stressful event and com-— pare the levels of symptomatology before and after an event. Second, one may observe periods of high and low levels of symptomatology and determine if there are differences in the number or severity of life events during a preceding time interval (see figure 1). Dohrenwend and Dohrenwend (1974) make several recommendations for continued research in life events and schizophrenia that we believe are addressed in this design. One of these is related to the above point, namely, to structure the opportunity to focus on specific life events, or types of life events, and then to assess the magnitude of the risk that the severity of symptoms will increase. Of particular interest to us is the impact of positive and negative events. It has been postulated that all life events/changes, positive or negative, have a detrimental effect on schizo- phrenia. The clinical implication of this position is to advocate low-input, static environments as therapeu- tically optimal for schizophrenic patients. On the 204 Figure 1 Methods used in the assessment of the relationship between life events and symptom severity. Measure symptom level over a time interval before and after an event. POST-EVENT PRE-EVENT EVENT Measure life events over an interval preceding a period of minimum and a period of maximum symptom levels. MAXIMUM A MINIMUM 8 N —_— Ng TT—CEVENTS =— Measure life events over a specified time interval preceding the estimated point of illness onset. (Brown & Birley = 13 weeks before onset; Jacobs and Myers, 1976 = 12 months before onset). ONSET 12 MONTHS 13 weeks 4 A fossa nde —— — INTERVIEW 1 T | J | TT Nevens= zo ADMISSION 205 other hand, others (Tennant et al. 1981) have argued that positive and negative events may interact to neutralize one another. The positive effect of starting a new relationship shortly after losing one's job may reduce the probability of symptom exacerbation. The assessment of whether an event is positive or negative may best be made in the same contextual way that Brown and Birley (1968) developed to assess the threat of an event. Being fired from a job that one enjoys may be far more negative than being fired from a job that is unpleasant and lacks recognition and re-— ward. The positive and negative quality of life events needs to be assessed contextually and then related to observed patterns of schizophrenic symptomatology. We are interested in seeing if this dimension of life events is critical in predicting symptom exacerbation in schizophrenics. ® Dohrenwend and Dohrenwend (1974) also recommend studying factors that may mediate the impact of life events. To this end, we have constructed a model of potential modifiers of life events and have collected data from which we hope to analyze the relative con-— tribution of these modifiers on the course of illness (see figure 2). Life Stressors and Symptom Exacerbation: Preliminary Findings An analysis of the Social Readjustment Rating Scale scores (Holmes and Rahe 1967) in the 3 months prior to the highest Target Symptom Rating (TSR) of each of 30 subjects was compared with the scores in each of the 3 months preceding the lowest TSR. Seven subjects who showed the same level of symptoms throughout, usually symptom-free, could not be included in this comparison. The results (table 1) indicate a trend that supports the hypothesis that life events contribute to symptom exacerbation of schizophrenia. The higher ratings were noted in the month closest to the exacerbation, supporting the earlier findings of Brown and Birley. However, higher stress was evident in each of the 3 months preceding the higher TSR. Care was taken to choose the high TSR point at the earliest point of increment to the peak rating in 206 Loz Figure 2. Therapeutic modifiers of life event stress mEsssssssssse ng Coping mechanisms e.g. problem solving Life event e.g. loss of job, physical illness Stress Nonspecific symptoms e.g. sleep disturbance Relapse of schizophrenia Neuroleptic medication Table 1. Mean social readjustment scores in the 3 months preceding high versus low target ratings of schizophrenic symptomatology (n = 30) mean readjustment score in months before target rating 3rd month 2nd month 1st month 10.0 13.7 15.0 High Target symptoms 7.4 10.9 11.6 Low order to avoid confounding the stress associated with increased symptomatology. This very preliminary examination that employed our least sophisticated methodology provides encouragement that detailed analysis may shed further light on this fas— cinating and confusing subject. Psychosocial Interventions and Stress Reduction The intervention study in which we are concurrently en— gaged employs random assignment of 40 subjects to two treatment conditions: individual supportive psychotherapy and behavioral family therapy (Boyd et al. 1981). Subjects are all living with their parents in stressful family envi- ronments and meet the other selection criteria previously outlined. The aim of both treatment modalities is to reduce 208 everyday life stress through improving problem-solving behavior. However, in one condition, the major target of intervention is the family system; in the other, it is the individual patient. Family Therapy Intervention: Preliminary Findings After 9 months of controlled treatment, the results ap— pear to strongly favor the family intervention in terms of clinical exacerbations (major and minor) and hospitali- zation. While the individually treated patients have, for the most part, shown a substantially improved course, family therapy group members have been almost free from major exacerbations and hospital care. Undoubtedly, some of these differences may be attrib- uted to improved medication compliance and earlier de- tection of impending crises as a result of family management. How much of this added effectiveness could be attrib- uted to changes in the stressful impact of life events? We have found that the incidence of life events was very sim- ilar in the two conditions—-the family therapy group showed a nonsignificant excess of family-related events, which may have reflected an increased focus on family conflicts contingent on family therapy. A preliminary examination of changes in family problem-solving behavior on a standardized direct observation of a naturalistic problem-solving task suggested that major changes in the ability to employ effective problem-solving methods had occurred after 3 months of treatment in a large proportion of families receiving family therapy. Furthermore, the coping abilities of patients and family members apppeared substantially better in their handling of life events, thereby reducing the risk potential of this source of stress. This report is based on a cursory analysis but appears striking, and it is suspected that sophisticated analysis will confirm these early findings. Conclusions This report provides an overview of the NIMH-funded 209 research program that is being conducted at the University of Southern California Department of Psychiatry. The major projects involve (1) a prospective, longitudinal study of life stressors and their impact on the course of estab lished cases of schizophrenia, and (2) a controlled clinical outcome study of impact of family- versus individual- based problem-solving interventions on the course of the illness in the same population. Preliminary results suggest an association between life events and schizophrenic symptom severity that may be muted by effective coping behavior (see figure 2). The family therapy approach ap- pears somewhat more effective than individual therapy in enhancing coping skills. The methodology appears adequate to support rigorous data-based analysis of these tentative findings. Despite its relatively small sample size, this study, it is hoped, will contribute in some small way to an improved understanding of the association between stress and schizophrenia, as well as to the development of in-— terventions to prevent the morbidity attributed to these environmental factors. References Boyd, J.L.; McGill, C.W.; and Falloon, I.R.H. Family par- ticipation in the community rehabilitation of schizo— phrenics. Hospital and Community Psychiatry 32:629- 632, 1981. Brown, G., and Birley, J.L. Crises and life changes and the onset of schizophrenia. Journal of Health and Social Be- havior 9:203-214, 1968. Dohrenwend, B.S., and Dohrenwend, B.P., eds. Stressful Life Events: Their Nature and Effects. New York: John Wiley, 1974. Goldstein, M.J.; Rodnick, E.H.; Evans, J.R.; May, P.R.A.; and Steinberg, M.R. Drug and family therapy in the aftercare of acute schizophrenics. Archives of General Psychiatry 35:1169-1177, 1978. Holmes, T., and Rahe, R. The social readjustment rating scale. Journal of Psychosomatic Research 11:213-218, 1967. Jacobs, S., and Myers, J. Recent life events and acute schizophrenic psychosis: A controlled study. Journal of Nervous and Mental Disease 162:75-87, 1976. Michaux, W.; Gansereit, K.; McCabe, O.; and Kurland, A. The psychopathology and measurement of environmental 210 stress. Community Mental Health Journal 3:358-371, 1967. Tennant, C.; Bebbington, P.; and Hurry, J. The short-term outcome of neurotic disorders in the community: The re- lation of remission to clinical factors and to "neutralizing" life events. British Journal of Psychiatry 139:213-220, 1981. Vaughn, C.E., and Leff, J.P. The influence of family and social factors on the cause of schizophrenic illness: A comparison of schizophrenic and depressed neurotic patients. British Journal of Psychiatry 129:125-137, 1976. 211 212 Part IV COMMENTARY AND CONCLUSIONS Some workshop participants served as discussants or commentators. Their comments are summarized in chapter 11, edited by Daniel S. Weiss. Chapter 12 summarizes the conference proceedings and presents the recommendations reached by consensus during the workshop. This section provides a sense of the future for preventing stress-related psychiatric disorders. 213 Chapter 11 DISCUSSANT COMMENTS edited by Daniel S. Weiss, Ph.D. Comment William Hargreaves, Ph.D. The concept of attributable risk, as discussed by Dr. Gruenberg, is a useful and interesting notion. It has been limited, however, in that it has been applied only to binary endpoints (present vs. absent). It seems possible that at- tributable risk could be applied to quantitative endpoints over some period of time. This would increase the scope of the concept while preserving its meaning. In the same vein, one could ask about a "monetized" attributable risk. That is, what are the cost consequences of exposure to an event? Following that line of reasoming, in an intervention trial one might flip the idea on its head and ask about attrib- utable benefit. What we would be asking in an intervention trial is this: What is the benefit of being exposed to an ex- perimentally derived intervention over some period of time? For prevention research, the question of short term vs. long term in part would determine the kind of preven- tive effort being undertaken. It is also critical to deter- mine the outcome to be examined. Taken together, these two decisions define the research focus. Figure 1 illus- trates the issues: In the first situation, we see "pure" attributable benefit. After exposure to the presumptive risk, random assignment to the preventive intervention allows us to examine the 215 91¢ Figure 1. Models for assessing "attributable benefit". Presumptive | risk No performance decrement | Random Primary assignment prevention Presumptive Performance | risk decrement Disorder as Presumptive |p performance Treatment decrement assignment » risk Random ~~ Attributable benefit Secondary prevention Primary prevention Attributable benefit on performance decrement ot] Attributable benefit on prevention of X benefit of primary prevention. The second situation differs in that individuals already show effects of exposure to the risk (labeled for convenience, performance decrement), are randomly assigned to an intervention, and any attributable benefit on the performance decrement can be understood as secondary prevention. Alternatively, this model can be interpreted to demonstrate a primary prevention of the disorder. The last situation shows the common treatment or tertiary prevention model for completeness. As the second situation makes clear, in the real world, it may be unclear whether a trial is a secondary or a primary effort. The answer depends on what is seen as the endpoint, and whether the performance decrement is equivalent to the disorder. Thus, the key issues are how long we are looking and at what outcome we are looking. The inter— vention trial design clearly does not tell us this and, in fact, applies equally well to all three situations. Comment Jim Mintz, Ph.D. Are depressive symptoms following stressful life events pathological or adaptive? Is there continuity between pathological and normal emotional states? Are the dis- tinctions only arbitrarily quantitative, or are they quali- tative? If depression and bereavement are different dis— orders, then how do those who experience the symptoms of major depression following a loss differ from those who do not? If these phenomena are continuous, we ought to re- focus our research on the perpetuating causes of depressive disorder, on the relationship between personality and stress response syndromes, on coping, and on recovery from the effects of stressful life events. The following listing illustrates these issues: Adaptive, "normal" Maladaptive, "pathological" emotional state emotional state Protest Despair Agitation Retardation Adrenergic activity Withdrawal/conservation Time-limited Extended 217 Distress, anxiety Vegetative signs Depressed mood Emotional depletion *Exogenous *Endogenous *Nonfamilial *Familial *Subjects for further study A preliminary study of very brief depression (VBD) in college students at UCLA illustrates one attempt to eluci- date the continuities and discontinuities between normal and abnormal depressive symptomatology. In a recent in- vestigation of self-reported symptoms, using DSM-III cri- teria for depressive disorder, 50-60 percent of college students report four or more categories of depressive symptoms required for the diagnosis of major disorder. Only approximately 30 percent have these symptoms for the required 2 weeks, and only 8 percent have functional impairment. The modal period of the VBD is 2-3 days, and 40 percent of the students feel the episode is completely endogenous. In this population, major disorder demon- strates the expected familial association in females. Data on familial association for VBD is less clear, suggesting that persistence of the normal depressive symptomatol- ogy into a disorder may be the familial trait linked to major depression. These observations demand further study, suggesting multiple implications for prevention research: Is early de- pressive symptomatology following a stressful life event an early warning sign or a normal response in an individual without family history of depression? Should preventive interventions focus on the recovery process following losses and other stressful life events? What perpetuates symptoms? What promotes recovery? Comment Lawrence Fisher, Ph.D. The family as a social network is a critical and often- neglected focus of investigation and intervention following tThis preliminary work was conducted by Jim Mintz, Laurette Schiff, and Howard Goldman. 218 a stressful life event. Recent experience with political crises and natural disasters seems to indicate that family concerns predominate in the minds of victims of stress. The ability of the family, as a critical element of an in- dividual's natural support system, to mediate distress emerges as a significant, yet poorly understood, variable. Yet, before we can improve our understanding of family coping, we need to accept the conceptual shift that makes the family—and not the family member—the unit of intervention and the unit of analysis. The proposed new field, preventing adverse consequences for families following stressful life events, requires con- siderable research. Needed are more studies of family types, coping mechanisms, and tools for measuring family transactions and interactions. Family assessment is in its infancy; current methods are primitive and often very expensive. Dr. Weissman suggests that we have the tools for measuring individual distress and need only add measures of distressed family interaction. Dr. Gruenberg agrees that the family unit ought to become a new unit of observation in prevention research. He adds, "We do not even know the difference between family interaction in a (‘stable’) family of three versus a ('dynamic') family of four which has just lost a family member." Dr. Hargreaves concludes that we need normative data on a variety of family units. Comment Michael Goldstein, Ph.D. In discussing preventing stress-related psychiatric disorders, one must distinguish between prevention re- search and prevention trials. To initiate a clinical trial, investigators need to know how something works. This is different from knowing that something works. Further- more, once one knows that the investigation is possible, the next question is whether it is worth doing. Prevention trials do not answer the cost/benefit question—rather, they address the mechanism by which the intervention operates. Throughout this process, the investigator must strive to continue honing the intervention down to its minimum essential ingredients for effective results. Of course the criteria that specify effective results also are not fixed. For example, they may be any one or a combination of 219 variables, such as suppression of a disorder, social role en— hancement, help in social functioning, or improvement in level of functioning. There is also some understanding to be gained from com- paring prevention efforts in psychiatric disorders with similar efforts in somatic disorders. The successes in medi- cine have a short timeframe of action and a time-limited "criterion period." This has been an important factor in their success, because both interest and funding tend to wane over long periods. Primary prevention research has encountered such difficulty. General mental health educa- tion programs have a long wait for results, and their po- tential outcomes are not measured by the reduction of a specific disorder. For secondary intervention, it may be possible to consider a preemptive intervention that allows us to act now in order to preempt further morbidity in the future. This notion of a preemptive intervention is not a shatteringly new insight, but its recognition helps the in- vestigator to ask salient questions about the potential outcome signals and timeliness of a proposed intervention. 220 Chapter 12 SUMMARY AND RECOMMENDATIONS The various papers in this document provide numerous ideas for further research. This chapter identifies some areas of consensus that emerged during the closing session of the workshop. Discussion focused on four general topic areas: (1) a critique of proposed research, (2) deficiencies in the state of the art, (3) proposed areas for preventive interventions, and (4) recommendations for research. Each area is addressed briefly here, along with some specific recommendations to the Office of Prevention, NIMH. Critique of Proposed Research The participants agreed that most preventive trials must be based on solid epidemiological evidence. Disorders or undesirable health consequences of stressful life events must be specified and reliably quantified. The desired out— comes must be clarified; their preventive nature must be elucidated, distinguishing preventive outcomes from treat- ment outcomes. The incidence, prevalence, and morbidity associated with these disorders/consequences should be known in order to assess the magnitude of the burden on the public health. It was also agreed that a complete epidemiological re- search database might not be essential if the potential pre- ventive intervention had some likelihood of success and the intervention would be associated with a minimum of risk. Preventive trials research need not wait until all related basic research is complete. Neither should preventive trials preclude detailed investigation of the epidemiology, basic science, and natural history of stress-related psychiatric disorders. 22) Deficiencies in the State of the Art Considerable agreement existed on areas in need of further study to correct deficiencies in the state of the art to prevent stress-related psychiatric disorders. Longitu- dinal and prospective studies of the natural history of stressful life events are to be encouraged. Sophisticated techniques to analyze retrospective data have been de- veloped and could be applied to this research area. The circumstances surrounding stressful life events often pre- clude prospective research. The suddenness and unexpected onset of catastrophes, as well as the confusion and dis— equilibrium of their aftermath, make it difficult to initiate careful research or collect baseline data. It was suggested that research teams could be prepared to await disasters, but the logistics seemed extremely difficult. Perhaps the most intriguing area in need of further in- vestigation is the role of social support in preventing or ameliorating stress-related disorders. Many investigators cite this understudied phenomenon as a critical variable in predicting responses both to stressful life events and to preventive interventions. Social support research would include investigating the role of families and other ele- ments of natural support networks, as well as the effect of artificial social supports and community support inter— ventions. Proposed Areas for Preventive Interventions As already noted, preventive trials are to be encouraged. Four topic areas were discussed as potentially fruitful areas for intervention: 1. Bereavements, especially widowhood. This area is sig- nificant because of the prevalence of the phenomenon and the established risk of untoward consequences. Suggested studies include clinical trials of antide- pressant medications, brief psychotherapy, and widow- to-widow peer counseling. 2. Recurrent, unipolar depressions. 3. Posttraumatic stress disorders, following both ex- treme situations and more prevalent, but stressful, life events, such as assaults, rapes, and prolonged or sudden, accidental deaths. 222 4. Adjustment disorders, such as the syndrome of psycho- pathology seen following marital separation and divorce. Recommendations for Research Consensus was shown to encourage collaboration among the various research programs at NIMH. The Epidemiology Catchment Area (ECA) program was cited as having great potential as a focus of research on preventing stress-re— lated disorders, particularly since the ECA sites already possess a critical mass of epidemiologic data and research expertise. Other sites, such as clinical research centers, epidemiology study groups, and prevention researchers, should be encouraged to develop and submit research pro- posals. Multiyear financing will be essential for the support of prospective studies on the natural history of stressful life events and their consequences. Postscript At the time this workshop was convened in December 1981, NIMH was developing plans for a major reorganiza— tion of the Institute. As a result of that reorganization, the Center for Prevention Research was established within the newly renamed Division of Prevention and Special Mental Health Programs. Responsibility for the development and conduct of NIMH's preventive intervention research grants program was transferred from the Office of Prevention, which was established in the fall of 1979, to the Center for Prevention Research. As a function of the reorganization, NIMH has two pri- mary contact points for prevention: 1. The Office of Prevention, administratively located within the Office of the Institute Director, has the following responsibilities: ® To plan and coordinate the programs required to carry out the provisions of Section 455(d) of the Public Health Service Act, which include: designing national goals and establishing national priorities for the prevention of mental illness and the promotion of mental health; encouraging local entities and State agencies to achieve these goals and priorities; and developing and coordinating Federal prevention pol- icies and programs and ensuring increased focus on 223 the prevention of mental illness and the promotion of mental health To serve as the NIMH lead for activities related to the prevention of mental illness and the promotion of mental health To coordinate and develop Institute-wide policies for prevention goals, priorities, and programs To identify, stimulate, develop, implement, and mon-- itor activities to achieve prevention goals and pri-- orities For further information, contact: Stephen E. Goldston, Ed.D., M.S.P.H. Director, Office of Prevention National Institute of Mental Health 5600 Fishers Lane Rockville, MD 20857 (301) 443-6130 . The Center for Prevention Research, administratively located within the Division of Prevention and Special Mental Health Programs, has the following mission: To serve as the focal point for NIMH activities on prevention research, which aim to develop and assess early preventive interventions to diminish or reduce emotional disorders, serious behavioral dysfunctions, and significant emotional distress of persons and populations at risk To identify, analyze, and evaluate research and re- lated program developments To stimulate, develop, and support programs of pre-— vention research and research training For further information, contact: Morton M. Silverman, M.D. Chief, Center for Prevention Research Division of Prevention and Special Mental Health Programs, NIMH 5600 Fishers Lane Rockville, MD 20857 (301) 443-4284 - 224 U.S. GOVERNMENT PRINTING OFFICE : 1985 O - 472-367 GENERAL LIBRARY - U.C. BERKELEY MURR BO00S520069 stirs vane sees ce sissreecronse arse ies ieee aes arses es em ren secon