Smokeless Tobacco Use , in-the United States — A___. —— vM-r'ffw For sale ONLY by the Superintendent of Documents, US. Government Printing Office, Washington, DC. 20402. Payment is required in advance, and check or money order should be made payable to the Superintendent of Docu- ments. ,NCI Monographs National Cancer Institute, Samuel Broder, Director International Cancer Information Center, Susan Molloy Hubbard, Director Editorial Board Daniel C. lhde, Editor-in-Chief William J. Blot Peter Greenwald Joost J. Oppenheim Peter M. Blumberg Robert N. Hoover John R. Ortaldo John D. Boice, Jr. Susan M. Hubbard David G. Poplack Michael R. Boyd Steven M. Larson Alan S. Rabson Bruce A. Chabner Brian R. Leyland-Jones Jeffrey Schlom Richard L. Cysyk Lance A. Liotta Richard M. Simon Charles H. Evans Marc E. Lippman Michael B. Sporn Eli J. Glatstein Dan L. Longo Snorri S. Thorgeirsson Michael M. Gottesman Douglas R. Lowy J. Paul Van Nevel Publications Branch, Robin A. Atkiss, Chief Florence l. Gregoric, Monograph Editor Editorial Policy Manuscripts from key conferences dealing with cancer and closely related research fields, or a related group of papers on specific subjects of importance to cancer research, are considered for publication, with the understanding that they have not been published previously and are submitted exclusively to NCI Monographs. All material submitted for consideration will be subject to review, when appropriate, by at least two outside reviewers and one member of the Editorial Board of the Journal of the National Cancer Institute. Opinions expressed by the authors are not necessarily those of the publisher or the editors. Symposia or related papers should be submitted to the Editor-in-Chief, Journal of the National Cancer Institute, R. A. Bloch International Cancer Information Center, Building 82, Room 211, National Cancer Institute, Bethesda, MD 20892. James O. Armitage Bruce C. Baguley Charles M. Balch June L. Biedler Clara D. Bloomfield Benjamin Bonavida Ernest C. Borden George J. Bosl Murray F. Brennan Edward Bresnick Ting-Chao Chou Harvey J. Cohen C. Norman Coleman 0. Michael Colvin Thomas H. Corbett Rik Derynck Leila Diamond Johanna T. Dwyer John D. Earle Merrill J. Egorin Virginia L. Ernster Nelson Fausto James S. Felton EDITORIAL ADVISORY BOARD Isaiah J. Fidler Thomas R. Fleming Gstein Fodstad Emil Frei 111 Richard D. Gelber David W. Golde Harvey M. Golomb E. Robert Greenberg Philip D. Greenberg Kenneth R. Harrap Ingegerd Hellstrom Maureen M. Henderson Gloria H. Heppner E. Carmack Holmes William J. Hoskins Robert C. Jackson Michael E. Johns Randall K. Johnson William W. Johnston Charles M. King Stanley J. Korsmeyer John S. Kovach Margaret L. Kripke Donald W. Kufe John S. Lazo Brigid G. Leventhal Bernard Levin Arnold J. Levine Allen S. Lichter David M. Livingston Virginia A. LiVolsi Jean—Pierre Mach Louis Malspeis Donald B. McCormick Richard S. Metzgar George K. Michalopoulos Anthony B. Miller Malcolm S. Mitchell Alan S. Morrison Lee M. Nadler C. Kent Osborne David F. Paulson Sidney Pestka Lester Peters Theodore L. Phillips Henry C. Pilot Janet S. Rasey Ralph A. Reisfeld Igor B. Roninson Ruth Sager Shigeako Sato David Schottenfeld William R. Shapiro Roy E. Shore Kendall A. Smith Paul M. Sondel Osias Stutman Herman D. Suit Raymond Taetle Ian Tannock Joel E. Tepper Cox Terhorst J. Tate Thigpen Giorgio Trinchieri Peter R. Twentyman I. Bernard Weinstein Sharon W. Weiss Raymond M. Welsh Richard E. Wilson Gerald N. Wogen Smokeless Tobacco Use in the United States A Compilation of Papers on Recent Research and Discussion of Directions for Future Research Scientific Editors: Gayle M. Boyd Charles A. Darby Smoking, Tobacco, and Cancer Program Division of Cancer Prevention and Control National Cancer Institute Bethesda, Maryland TABLE OF CONTENTS / Foreword C. Everett Koop Preface Joseph W. Cullen and Gayle M Boyd Smokeless Tobacco Use Among Adolescents: A Theoretical Overview Mario A. Orlandi and Gayle M Boyd Smokeless Tobacco: Association With Increased Cancer Risk Margaret E Mattson and Deborah Winn Use of Smokeless Tobacco in the United States: Recent Estimates From the Current Population Survey Alfred C Marcus, Lori A. Crane, Donald R Shopland, and William R Lynn Smokeless Tobacco Use in the United States: The Adult Use of Tobacco Surveys Thomas E Novotny, John P. Pierce, Michael C. Fiore, and Ronald M Davis Epidemiology of Smokeless Tobacco Use: A National Study Beatrice A. Rouse Use of Smokeless Tobacco by Age, Race, and Gender in Ten Standard Metropolitan Statistical Areas of the Southeast United States Karl E Baumart, Gary G. Koch, Lynn A. Fisher, and Elizabeth S. Bryan Native Youth and Smokeless Tobacco: Prevalence Rates, Gender Differences, and Descriptive Characteristics Steven P. Schinke, Robert F. Schilling II, Lewayne D. Gilchrist, Marianne Rolland Ashby, and Eifi Kitajima Parent Characteristics, Perceived Health Risk, and Smokeless Tobacco Use Among White Adolescent Males Karl E Bauman, Gary G. Koch, and G. Michael Lentz Use of Smokeless Tobacco Among Male Adolescents: Concurrent and Prospective Relationships Dennis V. Ary Activity Involvement, Risk-taking, Demographic. Variables, and Other Drug Use: Prediction of " Trying Smokeless Tobacco Steve Sussmart, Liana Holt; Clyde W. Dent, Brian R Flay, John W. Graham, William B. Hansen, and C. Anderson Johnson Initiation and Use of Smokeless Tobacco in Relation to Smoking Arthur V. Peterson, Patrick M Marek, and Sue L Mann Prevalence and Predictors of Smokeless Tobacco Use: Iowa’s Program Against Smoking Julia A. Burke, Rebecca Arbogast, Samuel L Becker, Michelle Naughton, and Ronald M Lauer Marketing Smokeless Tobacco in California Communities: Implications for Health Education Marc T Braverrnan, Carol N. D’Onofrio, and Joel M Moskowitz Page 13 17 25 29 35 39 43 49 57 63 71 79 Advertising and Promotion of Smokeless Tobacco Products Virginia L Emster Development and Evaluation of a Smokeless Tobacco Cessation Program: A Pilot Study Elizabeth Eakin, Herbert Severson and Russell E Glasgow Adolescent Smokeless Tobacco Use: Future Research Needs Laurie Chassin, Clark C. Pressom Steven J. Sherman, and Lynne Steinberg Page 87 95 101 Foreword In 1986, the Public Health Service completed the first comprehensive, in-depth review of the relation- ship between smokeless tobacco use and health, The Health Consequences of Using Smokeless Tobacco: A Report to the Surgeon General It identified three major health risks associated with the use of smokeless tobacco: Qcancer, the develflment of leukoplakias and other oral conditions, and nicotine addiction. In the intervening years, these findings have been confirmed and strengthened ibycontinued research. The 1986 Report described an alarming pattern of increasing prevalence of smokeless tobacco use in the United States, especially among youth, and data presented in this monograph indicate little change in that pattern. Smokeless tobacco use has a long history in this country but declined markedly during the first half of this century due, in part, to health concerns about spitting. The reemergence of snuff and chewing tobacco use when the attendant health hazards are so well documented is unacceptable. Over 350,000 lives are prematurely lost each year due to cigarette smoking. It is imperative that smokeless tobacco not be allowed to add to this unnecessary toll. Cigarette smoking has declined over the past twenty years, due in great measure to the combined efforts of biomedical scientists, public health experts, and organizations in the private sector, particularly the American Cancer Society, the American Lung Association, and the American Heart Association. It is important that a comparable effort now be directed toward reducing the use of smokeless tobacco. Indeed, the public must be made aware that there is no safe form of tobacco use. The papers in this monograph present the most current national survey data on smokeless tobacco prevalence and explore the nature of tobacco use by youth. These papers represent a valuable contribution to our understanding of smokeless tobacco practices by youth; they will be invaluable in the development of effective interventions. Intervention efforts are urgently needed to educate the public regarding the hazards of using any form of tobacco, to prevent youth from initiating tobacco use, and to encourage and assist attempts at quitting. I congratulate the National Cancer Institute and other health agencies for their efforts, and I strongly encourage those interested to make the elimination of tobacco use in all its forms a M4; C. Everett Koop, M.D. Surgeon General number one priority for health promotion and disease prevention. Preface The National Cancer Institute has set a goal of reducing cancer morbidity and mortality by 50% by the year 2000. As part of the Institute’s strategy to attain that goal, the Smoking, Tobacco, and Cancer Program, within the Division of Cancer Prevention and Control, is funding a number of research trials to develop and evaluate interventions designed to prevent and reduce tobacco use among children and adolescents. Although many of the intervention programs focus primarily on cigarette smoking, valuable data are also collected on smokeless tobacco use. The Smoking, Tobacco, and Cancer Program sponsors regular meetings each year to help investigators profit from each other’s experience and scientific perspective. In addition, these interactions created the collaborative opportunity wherein the articles included in this monograph were developed. Additionally, several authors were especially invited to contribute papers that were considered critical to a full under- standing of the current problem of smokeless tobacco use in the United States. These papers include a review of the cancer risk associated with smokeless tobacco, reports from recent national probability surveys regarding its use, and a summary discussion of future directions for research. Some of the contributors to this monograph have received support from, or are affiliated with, the American Cancer Society; the National Heart, Lung, and Blood Institute; the National Institute on Drug Abuse; and the Centers for Disease Control, including the Office on Smoking and Health and the National Center for Health Statistics. Joseph W. Cullen, Ph.D. Gayle M. Boyd, Ph.D. Smokeless Tobacco Use Among Adolescents: A Theoretical Overview Mario A. Orlandi1 and Gayle Boyd2 ABSTRACT—Prevalence data that are currently available indi- cate that smokeless tobacco use among youths in various parts of the United States is a significant and growing problem. Although relatively little is known about the factors that contribute to the initiation and maintenance of this behavior, previous research fo- cusing on other substance use provides a valuable framework for both a study of smokeless tobacco use and the design of effec- tive interventions for its prevention. The papers presented in this monograph address many of the gaps in our current knowledge and provide an accurate overview of research efforts in this area to date.—NCI Monogr 8:5-12, 1989. During most of this century, the use of chewing tobacco and oral snuff, collectively referred to as SLT, was con- fined primarily to rural areas and particular occupational groups, such as miners and agricultural workers (1, 2), and prevalence was highest among persons over the age of 50 years (3). Recently, however, the numbers of reports from schools about SLT use by children and adolescents have been increasing. These observations have coincided with new biomedical and epidemiologic evidence implicating its use in the development of oral cancer. In response to such reports and to a request from the Fed- eral Trade Commission, the Surgeon General appointed a special Advisory Committee to review the available sci- entific literature on health risks associated with SLT. The report from the committee, “The Health Consequences of Using Smokeless Tobacco” (3), concluded that its use can cause cancer in humans and that this association is strongest for cancers of the oral cavity. Addressing noncancerous ef- fects, the report concluded that SLT can cause oral leuko- plakia and gingival recession and that its use can lead to nicotine dependence or addiction. The International Agency for Research on Cancer (4) and the National Institutes of Health Consensus Development Conference on Health Implications of Smokeless Tobacco Use (5) independently reached similar conclusions. Evidence relating to the can— cer health risk associated with SLT use has been reviewed elsewhere (6) and is also summarized in this monograph by Mattson and Winn (7). Because large-scale use of SLT by youths is such a re- cent phenomenon, relatively little is known about the pre- ABBREVIATIONS: SLT = smokeless tobacco; CPS = Current Pop- ulation Survey; NHIS = National Health Interview Survey; NIDA = National Institute on Drug Abuse. ‘Division of Health Promotion Research, American Health Foundation, 320 East 43rd St, New York, NY 10017. 2Smoking, Tobacco, and Cancer Program, Division of Cancer Preven- tion and Control, National Cancer Institute, National Institutes of Health, Department of Health and Human Services. dictors of onset, the factors influencing use, the patterns developed, and the relationship of its use to that of other substances. Such information is essential for the develop- ment of effective intervention strategies. This introduction to the studies that follow provides an overview of what is currently known regarding the nature and the magnitude of this growing public health problem. PREVALENCE OF SMOKELESS TOBACCO USE Description of Product Of the two major forms of SLT, oral snuff and chewing tobacco (3, 8), snuff is finely ground or shredded tobacco to which various sweeteners and flavorings have been added. It is available in both dry and moist forms, but moist snuff is by far the more popular product. A user places a small amount or “pinch” in the mouth, and holds it in place between the lip and gum. Moist snuff is available in small sachets similar to tea bags or may be packaged loosely. Dry snuff is a fine powder and may be taken nasally as well as orally, although this practice is rare in the United States. Chewing tobacco is much coarser than snuff and is available in three forms: Loose leaf, by far the most popular, consists of small strips of shredded tobacco packaged loosely. Chewing tobacco may also be compressed into “plugs” or “bricks” or twisted into ropelike strands. All forms may be chewed or held in place in the mouth. Both chewing tobacco and snuff are used orally and nico— tine is absorbed through the oral mucosa. The resulting blood nicotine levels are comparable to those achieved by smokers, although the time courses differ. The SLT pro- duces prolonged, sustained levels of nicotine in the blood that contrast to the rapid rise, peaks, and troughs that char- acterize cigarette smoking (9—12). The amount of nicotine absorbed is a function of the brand used, the amount of to- bacco taken at one time, the length of time it is held in the mouth, and other individual variables. National Surveys Three recent surveys provide data on SLT use by adults. 1) In 1985, the Office on Smoking and Health sponsored a supplement to the CPS that included questions on SLT use. Due to sample structure and the large sample size of 114,000 individuals aged 16 years and over, estimates of the prevalence of its use can be calculated individually for states. These data are presented in this monograph by Marcus et al. (13). 2) In 1986, the same office included a number of questions on SLT in the Adult Use of Tobacco Survey. This telephone survey collected information from 13,031 persons aged 17 and older. Data are reported by 5 6 ORLANDI AND BOYD Novotny et al. (14). 3) The Division of Cancer Prevention and Control of the National Cancer Institute in 1987 in- cluded questions on tobacco use in their Cancer Control Supplement to the NHIS, conducted by the National Center for Health Statistics. Preliminary data from the first quar- ter, representing 4,735 males aged 18 and over, have been reported elsewhere (15). ' The NIDA included a question on SLT on the National Household Interview Survey conducted in 1985. This peri- odic household interview of persons aged 12 years and over provides the only national probability data currently avail- able on SLT use by persons under 17 years of age. There are no national data on use by children under 12. Estimates from the NIDA survey indicate that 20% of males aged 12 through 17 used it sometime in 1985. A complete analysis of these data is presented by Rouse (16). Due to differences in question format, estimates from the NIDA survey cannot be compared directly with the other national surveys. Estimates for prevalence of SLT use derived from the first 3 surveys are shown in table 1. Data from the 1970 NHIS are shown for comparison purposes. Variation among estimates from the 3 recent surveys may be attributed to differences in methodology. The 1985 CPS and the 1970 NHIS included approximately 45% proxy data, which are likely to result in underestimates, especially for snuff use. The 1986 Adult Use of Tobacco Survey did not use proxy data, but the relatively low response rate (74.3%) and the use of telephone interviews may have resulted in underes- timates (17). Although the 1987 NHIS was a household interview survey without the use of proxy respondents, data represent the first quarter of 1987 only and the resulting estimates must be considered preliminary. 4 Given the above caveats, it is clear that use of SLT, es— pecially snuff, has increased among the adult male popula- tion. Additionally, all the recent national surveys discussed in this monograph indicate that the highest rates of use oc- cur among adolescent and young adult males (13—15, I 7). Regional and Local Studies Regional and local prevalences are often much higher than national estimates (3 ). Bauman and co-workers (18) present data on all age groups from a large random sample TABLE 1.—Prevalence (%) of SLT use by males in the United States: Four national surveys SLT NHIS, 1970a CPS, 1985b 1986‘ NHIS, 1987d Snuff 1.4 1.9 2.4 3.2 Chewing 3.8 3.9 3.3 4.1 tobacco Any form 9 5.5 5.2 6.2 “ Males were 217 yr old. Survey was by household interview with ex- tensive use of proxy respondents (3). b Males were 2 16 yr old. Survey was by household interview with exten— sive use of proxy respondents (13). 9 Adult Use of Tobacco Survey (AUTS) was done of males 217 yr old by telephone survey (14 ). d Males were 2 18 yr old. Survey was done by household interview (15 ). e No information was available for this category. of 10 Standard Metropolitan Statistical Areas in southeast- ern United States. Because information on SLT use by youth was collected by adult proxy respondents, the resulting es— timates are likely to underrepresent actual use by children. Despite this, it should be noted that there were reports of current use by children under the age of 4 years and that the estimates obtained for adult use were comparable to the 1985 CPS figures for this region of the country (13). Considerable regional variation has been reported in the use of snuff and chewing tobacco by youth and adults. For example, use appears to be more common in rural areas, small communities, and in areas where regular use is traditional, although high rates have been reported in some metropolitan areas as well (3, 16, 19).3 Marcus et a1. (13) found prevalence among adults highest in states with larger rural populations. The studies reported in this monograph cover geographic areas ranging from rural to suburban and metropolitan regions in the West, Northwest, North Central, Midwest, Southeast, and Northeast regions of the country. Prevalence estimates among adolescent males for “having ever used” SLT were consistently high, over 62% in most areas. Rates for more frequent use were lower and ranged from 11% to 18%. Within all parts of the country, regular use is confined al- most exclusively to males, although substantial numbers of females may experiment with SLT (3, 20, 21 ).3 Often, use by females is not reported because so few are involved. This trend toward increased use among adolescent males rela- tive to females appears to be the reverse of that observed for cigarette smoking (22, 23). One marked exception to this pattern occurs among Native Americans, in whom sub— stantial use of SLT tobacco by females has been observed [(19, 24, 25); Batliner T, Kaltenbach R, Bothwell E: Un- published manuscript]. Native American males have higher rates than contemporary youths of other ethnic groups, al- though this pattern is not always observed (19) and may reflect the strength of traditional cultural practices. Bat- liner et al. (unpublished observations) observed differences in use between Native American youths living on and off reservations, and this was particularly marked among fe- males. Schinke and associates (26) discuss its use among young Native Americans living on reservations in Wash- ington State. Generally, blacks and Asians are less likely to use SLT than are whites and Hispanics (I9, 27).3 Similar trends have been observed in those studies reporting data by ethnic group (18), although in Los Angeles, the observed differ- ences in use between black and white youths were less than in some other areas (28). National studies of adult use con- sistently find highest prevalence among whites and lowest among blacks (13—15, 1 7). Racial differences in use of SLT vary regionally, especially Hispanic prevalence relative to other groups (13). Data reported by Bauman et a1. (18) on adults in the Southeast indicate that racial differences in use vary considerably across age groups. Because no national data on use by adolescents prior to 1985 are available, researchers cannot track changes 3McCarty D, Krakow M: Unpublished manuscript. NCI MONOGRAPHS, NUMBER 8, 1989 A THEORETICAL OVERVIEW 7 in consumption by youths nationwide. Anecdotal reports suggest a steady increase in prevalence over the past 10 years. Adult surveys indicate that over the past 15 years the increase in SLT use by young adults (under age 30) has been marked, whereas prevalence among older adults has either remained constant or diminished (3, 13—15, I 7). Most of the available data from local or regional sur— veys are cross-sectional. Generally, these data have shown consistent increases in use with age, although in some ar- eas, use appears to peak in the ninth or tenth grade and to decline subsequently (27, 29);3 only limited longitudinal data are available. In Louisiana, cross—sectional surveys of students ages 8 through 18 conducted in 1976 and 1981 showed increases in use of snuff and chewing tobacco over time within age categories, within cohorts, and across age categories (27). In an Oregon study, use of SLT at baseline was the best predictor of use 9 months later (30). Longi- tudinal data are presented in this issue by Ary (22), Burke et al. (23), and Sussman and co-workers (28). Increases in prevalence were consistently observed within cohorts, with the exception of some older student cohorts. For example, Ary (22) provides follow-up data that indicate considerable stability in use practices. The age of initiation is considerably younger than would have been predicted from studies of other substances that may be abused (table 2). In other studies, substantial rates of use have been re- ported among fourth-grade students in several areas of the country (19). Similarly, Peterson et al. (34) provide retro- spective data on the initiation of SLT use, which indicate 15.1% of tenth graders had tried it by the age of 10 and 4.2% were using it on a weekly 'basis by this age. CORRELATES OF ADOLESCENT SUBSTANCE USE Correlates of health—compromising behaviors such as cigarette smoking, alcohol consumption, and other drug use have been extensively studied during the past decade. An understanding of these correlates, especially with regard to the onset of such behaviors, has provided the principal ratio- nale for intervention development efforts targeting primary prevention (35, 36). Although far less is currently known regarding the correlates of SLT use, previous work related TABLE 2.—Age of initiation of regular SLT use among representative groups SLT users Reported age at Area Reference initiation, yr Student group Percent Arkansas (31) <4—5 Kindergarten 21 Texas (32) <12 J unior/ senior 5 5 high school Oklahoma (21) <10 Kinderganen— 33—50 high school Rhode Island ‘1 ~9 High school 18 Atlanta, (33) 12 Junior/senior 50 Georgia high school Southeastern (I 8) <10 Kindergarten— 0.3 United States fourth grade ‘1 Mariciano LA: Personal communication. SMOKELESS TOBACCO USE IN THE UNITED STATES to other substances will undoubtedly serve as a guide to current and future research in this area. Despite differences among the various substance use be— haviors that have been studied, important similarities have emerged. Although no real consensus exists with respect to etiology, some agreement has developed regarding a com- mon underlying theoretical framework and a common set of underlying factors that predispose adolescents to substance use (37). This framework has borrowed considerably from the perspectives of the social learning model of Bandura (38), the problem behavior theory of Jessor and Jessor (39), the attitude change model of Ajzen and Fishbein (40), and the peer influence concepts of Festinger (41). Within this context, health-compromising behaviors are conceptualized as socially learned, purposeful, and functional activities that are determined through the interplay of a variety of exter- nal and intrapersonal factors. These factors serve to shape behavioral responses through combinations of positive and negative reinforcement (38). The onset of substance use typically occurs during ado- lescence, and the initial use of tobacco, alcohol, or other drugs tends to be confined to social situations (42). Of the many variables that have been studied as possible correlates of initial substance use, several have shown more consistent significant relationships. Because of the role that these fac- tors are thought to play in the acquisition of other substance use behaviors, each should be considered as a potential cor- relate of SLT use. Social Influences The influence that others can exert upon an individual in the context of a social interaction has been shown to be one of the most salient predictors of substance use onset among adolescents (43). The primary factors of this type include family influences and peer influences. However, the importance of these two categories of influences is believed to vary across substances. Adolescents whose parents drink alcohol are more likely to use hard liquor and certain illicit drugs other than marijuana. For cigarettes and marijuana, on the other hand, the major social influence predicting iii- dividual use is having a friend who uses the same substance (44, 45). Adolescents’ perceptions of the attitudes of parents and peers toward substance use, specifically with regard to approval or disapproval, also has an effect on the acquisi- tion of such behaviors (46, 47). As children pass through childhood into adolescence, parental influences are thought to change, exerting their greatest effect during the preado- lescent years. This is a time when health-related knowledge, beliefs, attitudes, values, and early intentions are formu- lated as part of what has been termed a behavioral “prepa- ration” stage (48). As children approach adolescence, this progressive decline in the importance of parental influence accompanies a corresponding increase of salience of peer influences and other factors related to extrafamiliar social— ization (49). Although these factors are highly correlated with the onset of substance use, the interrelationships be- tween familial and extrafamiliar social influences are highly complex, and the relative importance of these variables varies across substances and at different points during ado- lescent development (50). 8 ORLANDI AND BOYD Self-image and Personality Traits Self-image variables also have been studied as poten- tial influences on the development of youths’ intentions to experiment with health-compromising behaviors (51, 52). Early adolescence is a period of development during which individuals gradually establish a sense of autonomy through a process that typically involves a considerable amount of behavioral experimentation, and, for certain individuals, this experimentation includes substance use. In fact, Jessor (46) has suggested that experimentation with tobacco, alcohol, and other drugs provides a major focus for some children in the process of developing a sense of personal identity. For these children, many of whom perceive themselves as unsuccessful in meeting adult expectations, self-image fac- tors may exert a greater effect on their behavior than peer pressure or other social influences (43). Other variables that have been studied in the con— text of self-image and its relationship to substance use include external locus of control, low self-esteem, low self-satisfaction, a need for greater social approval, low self-confidence, high anxiety, low assertiveness, impulsive- ness, and rebelliousness (35, 38, 53). The evidence sup- porting some of these constructs as predictors of substance use, however, is difficult for some (54—56) to interpret and, at times, conflicting. This is most likely due to a tendency for these traits to vary to some extent across subgroups of individuals and across the time span of adolescent devel- opment. Cognitive and Affective Factors Researchers’ attempts to understand the antecedents of substance use behaviors have resulted in the identification of a variety of psychologic factors that have been exten- sively studied to date. The major impetus for this research came from the realization that prevention programs that focus only on educating adolescents regarding the legal, pharmacologic, and medical consequences of substance use were virtually ineffective in deterring onset (57, 58). Con— sistent with conceptual approaches to behavior that have developed concurrently (38, 40), different categories of variables have been studied as potential correlates of sub— stance use behaviors. These categories, which include fac- tors such as knowledge, beliefs, values, attitudes, and in- tentions, have been combined in various conceptual models by behavioral scientists attempting to rationalize the behav- ior change process as it relates to substance use (48, 59). These different factors are likely to be related to behavior, but their interrelationships are highly complex and poorly understood. Multiple Substance Use and Other Behaviors The use of specific substances has been shown to be correlated with the use of other substances (60) and with other health-related behaviors (61). For example, individ- uals who smoke cigarettes are more likely to drink alcohol or to smoke marijuana than are nonsmokers of tobacco (62). In fact, individuals who experiment with more than one substance have been reported to do so in a predictable sequence that has been referred to as a substance use hi- erarchy (47). This sequence typically begins with the use of cigarettes and alcohol followed by marijuana. The use of other drugs such as stimulants and depressants generally follows later and precedes the use of “harder” drugs like opiates, cocaine, and their derivatives (63 ). Although the causal relationships between the use of one substance and that of another are poorly understood, it is clear that the use of health-compromising substances that are perceived as “less harmful” could potentially facilitate experimenta- tion with more dangerous substances (63). It also has been shown that the use of tobacco, alco- hol, and other drugs is correlated with a variety of other health-compromising behaviors. These behaviors include poor school performance as measured by low grades and absenteeism (39), premature sexuality and psychosocial risk-taking (46), various acts of delinquency and rebellion (47), and antisocial behaviors such as lying, cheating, and stealing (64). CORRELATES OF SMOKELESS TOBACCO USE Although SLT use has been far less extensively studied than the use of other substances, preliminary research sug- gests that, in general, the process of becoming a user is simi- lar in certain respects to the process of becoming a cigarette smoker. For example, social influences appear to be impor- tant factors in the onset of its use. A number of studies have reported that users are more likely than nonusers to have friends who also use (30, 65—67). This relationship also has been reported for current nonusers who intend to use (68). In addition to these re- ported peer influences, Gritz et a1. (65) noted that users were more likely than nonusers to have fathers who also use, which suggests the potential importance of parental influence as well. None of these investigators, however, attempted to define the specific roles that peer or family influences play in determining the initiation of SLT use nor did they report these data in correlations. In one study that did report results in percentage of explained variance, the social influences variables found to explain use significantly were: peer use, father’s approval, and mother’s approval (69). Young and Williamson (31) have reported data that also suggest that early family influences could have a significant effect on the onset of use. In a sample of 112 kindergarten children interviewed, 21.4% reported that they had already used SLT and an additional 35.7% expected to do so in the future. The variables related to social influences that were significantly correlated with use and expected use in this study were knowing someone who uses and having seen the product used at home. In a retrospective study of adults attending a dental clinic in Ohio, an interaction was noted between age of initiation and social influences. Those who began before age 15 were more likely to have family members who also used SLT than were those who began after age 15 (70). Additional data on the relationships. between family and peer influences and use are discussed by several investigators (22, 23, 71). _ Previous investigations have provided relatively little in- formation regarding the relationships between self-image or personality variables and SLT use. In a study by Chassin NCI MONOGRAPHS, NUMBER 8, 1989 A THEORETICAL OVERVIEW 9 et a1. (72 ), the image that adolescents associate with its use was analyzed by their using semantic differential ratings of three types of stereotyped teenage models: anflathlete, a cowboy, and an average teenager. The social image as— sociated with such use was reported to be more positive than that associated with cigarette smoking. The authors suggested that the adoption of the snuff and chewing to- bacco habit has perceived social image benefits for adoles- cent males. However, this study did not provide a prospec- tive assessment of the value of self-image measures as predictors of SLT use. Schaefer and co-workers (32) reported that in a survey of 5,392 Texas school children, 9% reported regular use of SLT. Of these, 27% indicated that they used it because friends influenced them and 20% because they wanted to look grown-up. Overall, 47% believed that there were pos- itive social reasons for taking up the habit. In addition, this survey asked respondents to indicate whether advertise- ments for SLT influenced teenagers to initiate use, and 42% indicated that they did believe that advertisements were in- fluential. Other aspects of advertising practices and product availability are discussed in this monograph (62, 73). Cognitive factors and their relationship to snuff and chewing tobacco use have been described by a number of studies. For example, Glover (21 ) reported that only 44% of a sample of 5,392 school children in Texas believed that SLT is harmful to one’s health, whereas 77% of these chil- dren believed that cigarette smoking is harmful. This gen- eral belief that its use is less harmful than cigarettes has been reported in other studies as well (72). With regard to knowledge of specific harmful effects, Schaefer et a1. (32) reported that 67% of their sample believed that dipping and chewing of tobacco could cause cancer. Other studies have reported conflicting results. For exam- ple, Marty and co-workers (74) reported that neither users nor nonusers could consistently identify the health risks as- sociated with it, although the risks associated with smoking cigarettes were accurately delineated by both groups. Fe- males in this study were more likely than males to believe that SLT had at least a moderate effect on one’s health. In a similar study by Marty, McDermott, and Williams (75), only 31% of the respondents could identify the specific health consequences associated with SLT use from among a list of six options presented. In this monograph, Bauman et a1. (70) examine the relationship between adolescent use and perceived risk. The relationship between SLT and other substance use has been described in a number of studies. In a survey of Arkansas high school students (74), only a modest level (28.2%) of individuals reported using cigarettes and SLT, but 70.7% reported using both alcohol and SLT. Other studies have reported significant correlations between use of SLT and alcohol, marijuana, and other drugs (20, 30, 67, 76). Similarly, in this volume, several authors report that the use of other substances, when assessed, was consistently observed to correlate with SLT use (22, 23, 28, 34). Because of the significant health risks associated with cigarette smoking, correlations between the use of SLT and smoked tobacco should be of special concern to health professionals. Various correlations of this type have been observed. Some have suggested, for example, that SLT is SMOKELESS TOBACCO USE IN THE UNITED STATES used by adolescent males as a perceived “safe” alterna- tive to cigarettes (72). Chassin et al. (68) observed higher rates of its use among adolescent ex-smokers than among any other group. Glover and associates (77) found three times as many ex-smokers using SLT as ex—SLT users who smoked cigarettes in a North Carolina college student pop- ulation. Perceived health risk of cigarette smoking was the reason most often cited for changing to snuff and chewing tobacco. In addition to its role as one of the sequelae of cigarette smoking, SLT use is also a potential precursor to, or a risk factor for, cigarette smoking. Several investigators suggest that SLT experimentation precedes that of cigarettes among some children (21), and SLT may be preferred to cigarettes by the younger age groups (29). Considerable variability, however, can be observed among studies in the degree of observed correlation be— tween cigarette smoking and SLT use. For example, in Louisiana, Hunter et al. (27) found little overlap between the two practices and few adolescents who used both prod— ucts. Others (22, 23, 34, 76) have found considerable over— lap with as many as 70% of the smokers also reporting SLT use (21, 68).3 This finding may reflect regional differ- ences, or, as data from Oregon suggest, an age effect (66). Lichtenstein and co-workers reported a strong association between use of cigarettes and SLT among seventh graders, but the relationship was weaker among ninth graders and was reversed among children in the tenth grade. Peterson et a1. (34) found the rates of onset for regular use of SLT and cigarettes to be identical before age 11 and to diverge subsequently. Longitudinal studies indicate a reciprocal relationship between smoking and SLT use. Hunter and associates (2 7) found that, over a 5-year period, decreases in its use were accompanied by increases in cigarette smok- ing. In 1981—1982, the prevalence of smoking among l6—year—olds was twice that of 14-year-olds, but use of snuff was only one-half as prevalent. In a 9-month follow-up study (78), cessation of SLT use was associated with a high cigarette smoking rate and intention to smoke cigarettes. The authors cautioned that their sample size was small, however, and that the relationship was suggestive at best. In a separate 9-month follow-up study by the same investigators, use of SLT was found to be a risk factor for the onset and/or increased use of cigarettes, as well as for marijuana and alcohol use (30). Burke et a1. (23) reported that between 1980 and 1985 a decrease in weekly smoking by seventh-grade boys was correlated with an even greater increase in the use of SLT. The Surgeon General has concluded that nicotine present in cigarettes and other forms of tobacco causes addiction and that the underlying processes of this addiction are sim- ilar to those that determine it to other drugs such as heroin and cocaine (79). Because users of SLT sustain significant blood nicotine levels (9—12), nicotine addiction must be considered one potential adverse health consequence from its use. Users often report subjective dependence (I4, 80, 81) and have been found in laboratory settings to expe- rience withdrawal symptoms during abstinence similar to those of abstinent cigarette smokers (3, 82). In this mono- graph, a unique study reporting data from a cessation clinic 10 ORLANDI AND BOYD for SLT dependence found that 20% of the adolescents enrolled in the intervention program were abstinent at 6 months (83). This success rate is comparable to that ob- served in smoking cessation clinic programs and suggests that the difficulty a person experiences in attempting to quit using SLT may also be comparable. REFERENCES (I) MAXWELL JC JR: Chewing snuff is growth segment. To- bacco Rep 31, 1980 (2) SHELTON A: Smokeless sales continue to climb. Tobacco Rep 42, 1982 (3) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco. A Report of the Advisory Committee to the Surgeon General. DHHS (NIH) Publ No. 86-2874. Bethesda, MD: Natl Cancer Inst, 1986 (4) INTERNATIONAL AGENCY FOR RESEARCH ON CANCER WORK- ING GROUP: Tobacco Habits Other Than Smoking; Betel—Quid and Areca-Nut Chewing; and Some Related Nitrosamines. Monographs on the Evaluation of the Car- cinogenic Risk of Chemicals to Humans, vol 37. Lyon: IARC, 1985 (5) OFFICE OF MEDICAL APPLICATIONS OF RESEARCH, NATIONAL INSTITUTES OF HEALTH: Health implications of smokeless tobacco use. JAMA 255:1045—1048, 1986 (6) HOFFMANN D, HECHT SS: Smokeless Tobacco and Cancer, ISI Atlas of Science: Pharmacology, vol 2. Philadelphia: ISI, 1988, pp 46—52 (7) MATTSON ME, WINN DM: Smokeless tobacco: Association with increased cancer risk. NCI Monogr 8:13—16, 1989 (8) AGRICULTURAL MARKET SERVICE, DEPARTMENT OF AGRI- CULTURE: Tobacco in the United States. Misc ARS Publ No. 867. Washington, DC: ARS, Dept Agr, 1979 (9) GRITZ ER, BAER-WEISS V, BENOWITz NL, ET AL: Plasma nicotine and cotinine concentrations in habitual smokeless tobacco users. Clin Pharmacol Ther 30:201—209, 1981 (10) RUSSELL MAH, JARVIS MJ, DEVITT G, ET AL: Nicotine intake by snuff users. Br Med J 283:814—816, 1981 (II ) RUSSELL MAH, JARVIS MJ, WEST RJ, ET AL: Buccal absorp- tion of nicotine from smokeless tobacco sachets. Lancet 2:1370, 1985 (12) BENOWITZ NL, PORCHET H, SHEINER L, ET AL: Nicotine ab- sorption and cardiovascular effects with smokeless to- bacco use: Comparison with cigarettes and nicotine gum. Clin Pharmacol Ther 44:23—28, 1988 (I3) MARCUS AC, CRANE LA, SHOPLAND DR, ET AL: Use of smokeless tobacco in the United States: Recent estimates from the Current Population Survey. NCI Monogr 8:17— 23, 1989 (I4) NOVOTNY TE, PIERCE JP, FIORE MC, ET AL: Smokeless tobacco use in the United States: The Adult Use of To- bacco Surveys. NCI Monogr 8:25—28, 1989 (15) BOYD GM, PECHACEK TF, MATTSON ME: Tobacco use in the U.S.——preliminary estimates from the 1987 HIS, first quarter data. Presented at a meeting of the American Public Health Association, New Orleans, Oct 18—22, 1987 (I6) ROUSE BA: Epidemiology of smokeless tobacco use: A na— tional study. NCI Monogr 8:29—33, 1989 (I 7) OFFICE ON SMOKING AND HEALTH, CENTER FOR HEALTH PROMOTION AND EDUCATION, CDC: Cigarette smoking in the United States 1986. MMWR 3621-5, 1987 (I8) BAUMAN KE, KOCH GG, FISHER LA, ET AL: Use of smoke- less tobacco by age, race, and gender in ten Standard Metropolitan Statistical Areas of the Southeast United States. NCI Monogr 8:35—37, 1989 (19) BOYD GM, ARY DV, WIRT R, ET AL: Use of smokeless tobacco among children and adolescents in the United States. Prev Med 16:402—421, 1987 (20) DENT CW, SUSSMAN S, FLAY BR: Adolescent smokeless to- bacco incidence: Relations with cigarette smoking tran- sitions, other drug use and psychosocial variables. Prev Med 16:422-431, 1987 (21) GLOVER ED: Regional prevalence and patterns of smoke- less tobacco use. Presented at the NIH Consensus Devel- opment Conference on Health Implications of Smokeless Tobacco Use, Bethesda, MD, January 1986 (22) ARY DV: Use of smokeless tobacco among male adolescents: Concurrent and prospective relationships. NCI Monogr 8:49—55, 1989 (23) BURKE JA, ARBOGAST R, BECKER SL, ET AL: Prevalence and predictors of smokeless tobacco use: Iowa’s program against smoking. NCI Monogr 8:71—77, 1989 (24) MILLAR WJ, VAN RENSBERG S: Use of chewing tobacco and snuff by students in the Northwest Territories of 1982. Chronic Dis Canada 4:54—56, 1984 (25 ) SCHINKE SP, GILCHRIST LD, SCHILLING RF, ET AL: Smokeless tobacco use among Native American adolescents. N Engl J Med 314:1051,1986 (26) SCHINKE SP, SCHILLING RF II, GILCHRIST LD, ET AL: Native youth and smokeless tobacco: Prevalence rates, gender differences, and descriptive characteristics. NCI Monogr 8:39—42, 1989 (27) HUNTER SM, CROFT JB, BURKE GL, ET AL: Longitudinal patterns of cigarette smoking and smokeless tobacco use in youth. Am J Public Health 76:193—195, 1986 (28) SUSSMAN S, HOLT L, DENT CW, ET AL: Activity involve- ment, risk-taking, demographic variables, and other drug use: Prediction of trying smokeless tobacco. NCI Monogr 8:57—62, 1989 (29) GUGGENHEIMER J, ZULLE TG, KRUPPEE DC, ET AL: Chang- ing trends of tobacco use in teenage population in Western Pennsylvania. Am J Public Health 76:196—197, 1986 (30) ARY DV, LICHTENSTEIN E, SEVERSON HH: Smokeless to- bacco use among male adolescents: Patterns, corre- lates, predictors, and the use of other drugs. Prev Med 16:385—401, 1987 (31) YOUNG M, WILLIAMSON D: Correlates of use and expected use of smokeless tobacco among kindergarten children. Psychol Rep 56:63—66, 1985 (32) SCHAEFER SD, HENDERSON AH, GLOVER ED, ET AL: Patterns of use and incidence of smokeless tobacco consumption in school-age children. Arch Otolaryngol 111:639—642, 1985 (33) OFFENBACHER S, WEATHERS DR: Effects of smokeless to- bacco on the peridontal, mucosal and caries status of ado- lescent males. J Oral Pathol 14:169—181, 1985 (34) PETERSON AV, MAREK PM, MANN SL: Initiation and use of smokeless tobacco in relation to smoking. NCI Monogr 8:63—69, 1989 (35) BOTVIN GJ: Prevention of adolescent substance abuse through the development of personal and social compe- tence. In Preventing Adolescent Drug Abuse: Intervention Strategies. NIDA Res Monogr No. 47, DHHS (ADM) Publ No. 83-1280. Washington, DC: Natl Inst Drug Abuse, 1983 (36) FLAY BR: What we know about the social influences ap- proach to smoking prevention: Review and recommen- dations. In Drug Abuse Among Children and Adoles- cents. NIDA Res Monogr No. 63. DHHS (ADM) Publ No. 85-1334. Washington, DC: Natl Inst Drug Abuse, 1985 (37) BOTVIN GJ, WILLIS TA: Personal and social skills train- ing: Cognitive behavioral approaches to substance abuse SMOKELESS TOBACCO USE IN THE UNITED STATES A THEORETICAL OVERVIEW 1 1 prevention. In Drug Abuse Among Children and Adoles- cents. NIDA Res Monogr No. 63. DHHS (ADM) Pub] No. 85-1334. Washington, DC: Natl Inst Drug Abuse, 1985 (38) BANDURA A: Social Learning Theory. Englewood Cliffs, NJ: Prentice-Hall, 1977 (39) JESSOR R, JESSOR SL: Problem Behavior and Psychosocial Development: A Longitudinal Study of Youth. New York: Academic Press, 1977 (40) AJZEN I, FISHBEIN M: Attitudinal and normative variables as predictors of specific behaviors. J Pers Soc Psychol 27:41—57, 1973 (41) FESTINGER L: Theory of social comparison processes. Hum Relations 7:117—140, 1954 (42) BOTVIN GJ, MCALISTER A: Cigarette smoking among chil- dren and adolescents: Causes and prevention. In Annual Review of Disease Prevention (Arnold CB, ed). New York: Springer-Verlag, 1981, pp 222—249 (43) BROOK JS, WHITEMAN M, GORDON AS: Stages of drug use in adolescence: Personality, peer, and family correlates. Dev Psychol 19:269—277, 1983 (44) HUBA GJ, WINGARD JA, BENTLER PM: A longitudinal anal- ysis of peer support, adult models, and peer subcultures in beginning adolescent substance use. Multivariate Behav Res 15:259—279, 1980 (45) ROONEY JF, WRIGHT TL: An extension of Jessor and Jessor’s problem behavior theory from marijuana to cigarette use. Int J Addict 17:1273—1287, 1982 (46) JESSOR R: Predicting time of onset of marijuana use: A developmental study of high school youth. J Consult Clin Psychol 44:125—134, 1976 (4 7) KANDEL DB: Drug and drinking behavior among youth. Annu Rev Sociol 6:235—285, 1980 (48) FLAY BR, D’AVERNAs JR, BEST JA, ET AL: Cigarette smok- ing: Why young people do it and ways of preventing it: The Waterloo Study. In Pediatric and Adolescent Behav— ioral Medicine (McGrath P, Firestone P, eds). New York: Springer-Verlag, 1983, pp 132—183 (49) UTECH D, HOVING KL: Parents and peers as competing influences in the decisions of children of differing ages. J Soc Psychol 78:267—274, 1969 (50) FLOYD BR, SOUTH DR: Dilemma of youth: The choice of parents or peers as a frame of reference for behavior. J Marriage Fam 34:627—634, 1972 (51) BECKER HS: History, culture, and subjective experience: An exploration of the social basis of drug-induced experi- ences. J Health Soc Behav 82163—176, 1967 (52) OFFER D, OSTROV E, HOWARD KI: Body image, self- perception, and chronic illness in adolescence. In Chronic Illness and Disabilities in Childhood and Adolescence (Blum RW, ed). Philadelphia: Grune & Stratton, 1984, pp 59—73 (53) ROTTER JB: Generalized expectancies for internal versus external control of reinforcement. In Applications of a Social Learning Theory of Personality (Rotter JB, Chance JE, Phares EJ, eds). New York: Holt, Rinehart & Winston, 1972, pp 260—295 (54) GERSICK KE, GRADY K, SEXTON E, ET AL: Personality and sociodemographic factors in adolescent drug use. In Drug Abuse and the American Adolescent (Lettieri DJ, Ludford JP, eds). NIDA Monogr No. 38. Rockville, MD: Natl Inst Drug Abuse, 1981, pp 81—116 (55) GOLDSTEIN JW, SAPPINGTON JT: Personality characteristics Of students who became heavy drug users: An MMPI study of an avant-garde. Am J Drug Alcohol Abuse 4:401—402, 1977 (56) ORIVE R, GERAND HB: Personality, attitudinal and social correlates of drug use. Int J Addict 15:869—881, 1980 SMOKELESS TOBACCO USE IN THE UNITED STATES (57) GOODSTADT MS: Alcohol and drug education: Models and outcomes. Health Educ Monogr 6:263—279, 1978 (58') THOMPSON EL: Smoking education programs: 1960—1976. Am J Public Health 68:250—257, 1978 (59) LEVANTHAL H, ZIMMERMAN R, GUTTMAN M: Compliance: A self-regulation perspective. In Handbook of Behavioral Medicine (Gentry WD, ed). New York: Guilford Press, 1984, pp 369—478 (60) BLUM R, RICHARDS L: Youthful drug use. In. Handbook on Drug Abuse (DuPont R1, Goldstein A, O’Donnell J, eds). Rockville, MD: Natl Inst Dmg Abuse, 1979, pp 257—269 (61) DEMONE HW: The nonuse and abuse of alcohol by the male adolescent. In Proceedings of the Second Annual Alcoholism Conference (Chafetz M, ed). DHHS (HSM) Pub] No. 73-9083. Washington, DC: Natl Inst Alcoholism and Alcohol Abuse, 1973, pp 24—32 (62) BRAVERMAN MT, D’ONOFORIO CN, MOSKOWITZ JM: Mar- keting smokeless tobacco in California communities: Im- plications for health education. NCI Monogr 8:79—85, 1989 (63) HAMBURG BA, BRAEMER HC, JAHNKE WA: Hierarchy of drug use in adolescence: Behavioral and attitudinal corre- lates of substantial drug use. Am J Psychiatry 132:1155— 1167, 1975 (64) WESCHLER H, THUM D: Teen-age drinking, drug use, and social correlates. Q J Stud Alcohol 34:1220—1227, 1973 (65) GRITZ ER, KSIR C, MCCARTHY WJ: Smokeless tobacco use in the United States: Past and future trends. Ann Behav Med 7:24—27, 1985 . (66) LICHTENSTEIN E, SEVERSON HH, FRIEDMAN LS, ET AL: Chew- ing tobacco use by adolescents: Prevalence and relation to cigarette smoking. Addict Behav 9:351—355, 1984 (67) SEVERSON H, LICHTENSTEIN E, GALLISON C: A pinch or a pouch instead of a puff? Implications of chewing tobacco for addictive processes. Bull Psychol Addict Behav 4:85— 92, 1985 (68) CHASSIN L, PRESSON CC, SHERMAN SJ, ET AL: Psychological correlates of adolescent smokeless tobacco use. Addict Behav 10:431—435, 1985 (69) BONAGURO JA, PUGY M, BONAGURO EW: Multivariate anal— ysis of smokeless tobacco use by adolescents in grades four through twelve. Health Educ 17:4—7, 1986 (70) BAUMAN KE, KOCH GG, LENTz GM: Parent characteristics, perceived health risk, and smokeless tobacco use among white adolescent males. NCI Monogr 8:43—48, 1989 (71) SCHROEDER KL, IADEROSA GB, CHEN MS, ET AL: Bimodal initiation of smokeless tobacco usage: Implications for cancer education. J Cancer Ed 2:15—21, 1987 (72) CHASSIN L, PRESSON CC, SHERMAN SJ: Stepping backward in order to step forward: An acquisition oriented approach to primary prevention. J Consult Clin Psychol. In press (73) ERNSTER VL: Advertising and promotion of smokeless to- bacco products. NCI Monogr 8:87—94, 1989 (74) MARTY PJ, MCDERMOTT RJ, YOUNG M, ET AL: Prevalence and psychosocial correlates of dipping and chewing in a group of rural high school students. Health Educ. In press (75) MARTY PJ, MCDERMOTT RJ, WILLIAMS T: Patterns of smokeless tobacco use in a population of high school stu- dents. Am J Public Health 76:190—l92, 1986 (76) JONES RB, MOBERG DP: Correlates of smokeless tobacco use in a male adolescent population. Am J Public Health 78:61—63, 1988 (77) GLOVER ED, JOHNSON R, LAFLIN M, ET AL: Smokeless to- bacco use trends among college students in the United States. World Smoking Health. In press (78) ARY DV, SEVERSON HH, LICHTENSTEIN E: Concurrent cor- relates and prospective factors associated with smokeless 12 ORLANDI AND BOYD tobacco use among male adolescents. Presented at the An- nual Convention of the American Psychological Associa- tion, Los Angeles, CA, August 1985 (79) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Smoking: Nicotine Addiction. A Report of the Surgeon General. DHHS (PHS) Publ No. 88-8406, Washington, DC: DHHS, May 1988 (80) OFFICE OF THE INSPECTOR GENERAL, DEPARTMENT OF HEALTH AND HUMAN SERVICES: Youth use of smokeless tobacco: More than a pinch of trouble. Washington, DC: DHHS, January 1986 (81) GLOVER ED: Conducting smokeless tobacco cessation clin- ics. Am J Public Health 76207, 1986 (82) HATSUKAMI DK, GUST SW, KEENAN RM: Physiologic and subjective changes from smokeless tobacco withdrawal. Clin Pharmacol Ther 412103—107, 1987 (83) EAKIN E, SEVERSON H, GLASGOW RE: Development and evaluation of a smokeless tobacco cessation program: A pilot study. NCI Monogr 8:95—100, 1989 Smokeless Tobacco: Association With Increased Cancer Risk Margaret E. Mattsonl and Deborah M. Winnzy 3 ABSTRACT—Smokeless tobacco (chewing tobacco and snuff) contains known carcinogens shown to increase the risk for oral cancer. The effect of snuff has been more fully documented than other forms of smokeless tobacco, although the carcinogenic po- tential of all such products is acknowledged. Risk increases with increasing length of exposure, with risks greatest for anatomic sites where the product has been held in contact the longest time. In some studies, other organs, such as the esophagus, larynx, and stomach, have been shown to be at increased risk for can- cer from the use of smokeless tobacco, although at present the data are insufficient to substantiate fully a causal association. Nu- merous reports have shown an association between snuff use and leukoplakia, with less evidence at present linking chewing tobacco use with leukoplakia. The documented early onset of the smoke- less tobacco habit and reports of increases in certain oral cancers among young men raise serious concerns of an impending oral cancer epidemic in this population. In addition, synergistic in- teractions with other oral cancer risk factors, e.g., smoking and alcohol, and a high rate for second primaries observed for these cancers add to the concern. Unless the tide of its use is stemmed, long-term use can be expected to produce an increase in oral can- cers, and perhaps cancers of other sites, as youthful users ma- ture and accumulate exposure to this carcinogenic agent.—NCI Monogr 8:13—16, 1989. Increased risk for oral cancer has been associated with the use of chewing tobacco and snuff or SLT. The recent rise in its use, especially among youth, prompted four con— sensus conferences between 1984 and 1987 in the United States and Europe to evaluate, critically and systemati- cally, the data supporting this relationship (1—4). The re- mainder of this monograph documents this upsurge in the use of SLT. In this paper, we discuss the implications of these trends in the light of the body of research evidence on SLT and cancer. The conclusions from these four con- sensus meetings are summarized, additional evidence pub- lished since these reports were presented are cited, and this evidence is related to features of the new epidemic of SLT use that may have a bearing on the incidence of oral cancer in the coming decades. The focus of this summary will be on cancers, particu- larly oral cancers, and potentially precancerous conditions, ABBREVIATION: SLT = smokeless tobacco. 1Division of Cancer Prevention and Control, National Cancer Institute, Executive Plaza North, Room 330, National Institutes of Health, Depart- ment of Health and Human Services, Bethesda, MD 20892. Address reprint requests to Margaret E. Mattson, Ph.D. 2Division of Health Interview Statistics, National Center for Health Statistics, Hyattsville, MD. 3We thank Dr. William Blot for his advice and for the review of the manuscript and Ms. Vanessa Hooker for her typing of the manuscript and assistance with preparation of the references. such as leukoplakia. The consensus reports haVe also pre- sented substantiations of an association of SLT use with gingival recession, both local and generalized. Evidence re- lated to other diseases such as periodontal disease and tooth decay was examined as well. However, although these con- ditions are also important potential sequelae of SLT use, the emphasis in this review is on epidemiologic evidence for increased cancer risk in humans. The general conclusion reached by the consensus confer- ence groups is that the risk for oral cancer is increased as a result of SLT use. The effect of snuff has been more fully documented than other forms, although the carcinogenic potential of all SLT products is acknowledged. Risk in- creases with increasing length of exposure, with risks great- est for anatomic sites where the product has been held in contact the longest time. In some studies, other organs, e.g., esophagus, larynx, and stomach, have been shown to be at increased risk for cancer from the use of SLT, although at present the data are insufficient to substantiate fully a causal association. Numerous reports have shown an association between snuff use and leukoplakia, with less evidence at present linking chewing tobacco use with leukoplakia. Cigarette smoking and alcohol use are well-documented potent risk factors for oral cancer (5, 6); together they ac- count for about 74% of the oral and pharynx cancers in the United States (6). The proportion of cancer cases at- tributable to a risk factor is a function of the magnitude of the association and the extent of exposure to the risk factor, and so the estimate will be sensitive to current and future SLT use patterns. There is no general acceptance of any one estimate of the proportion of all oral cancer cases attributable to its use. However, overall, approxi- mately 29,800 new cases of oral cancer, or about 3% of total new cancers, were expected for 1987, with the major- ity occurring in males. The estimated number of deaths for 1987 was 9,400 or about 2% of total cancer deaths. The age-adjusted incidence rate (1981—1985) is 11.5 for whites and 14.5 for blacks, with the male differential pronounced across races, i.e., 16.9 for whites and 23.6 for blacks (7). Clinicians agree that cancer stage at detection is the most influential factor determining length of survival and that “no lethal disease is easier to cure than oral cancer of less than 1 cm in diameter” (8). Hence the frustration that, despite the relative accessibility of the mouth for examination com- pared with other sites, at diagnosis most oral cancers are large, symptomatic stage III and IV lesions, with 50% of the patients having metastatic lymphadenopathy (8). De- spite surgery, radiation therapy, and chemotherapy, overall survival rates for cancers of the oral cavity and pharynx are 53% for white males and 26% for black males (7). Even with modern reconstructive techniques, the impact upon 13 14 MATI‘SON AND WIN‘N quality of life for patients with advanced and disfiguring head and neck cancer may be devastating. Those who sur- vive oral or pharynx cancer have an exceptionally high risk of developing subsequent cancers. As many as 9% of the patients with cancer of the tongue, 14% of those with other mouth cancers, and 8% of the patients with pharynx cancer develop a second primary cancer. CARCINOGENS IN SMOKELESS TOBACCO Chemical analyses of SLT indicate that three types of known carcinogenic agents are present: N—nitrosamines, polycyclic aromatic hydrocarbons, and polonium 210, a radioactive alpha-emitter. Nitrosamines are present in SLT at levels 100 times higher than the levels of these substances in food, such as bacon, that are regulated by law. The nitrosamines are metabolized in vivo to products shown to produce genetic changes in cells and also produce tumors, benign and malignant, in animals when applied in amounts comparable to a human lifetime exposure dose for regular SLT users (1—3). The nitrosamine content of snuff may be 10—100 times greater than the level received by a smoker of one cigarette (9). Because of the high levels present and their ability to in- duce cancerous lesions, the nitrosamines are probably the major contributor to the carcinogenic potential of SLT. At present, cancer has been induced in animals by application of the carcinogenic chemical but not by exposure to ac— tual SLT products. Deficiencies in design or execution of many of the studies may account for the absence of an ob- servable effect. Several recent studies indicate that infection with herpes simplex virus may play an interactive role in the carcinogenic process, because animals infected with the virus do develop malignant and premalignant changes when exposed to SLT products. This issue deserves more research attention, because 20%—40% of the United States popula— tion has episodes of labial herpes and even more carry the virus without expressing it (10). SNUFF AND CANCER The strongest data associating the use of SLT with cancer are provided by studies that link “snuff dipping” or oral use of snuff to cancers of the oral cavity. Many of the early studies on this association made no distinction between snuff and chewing tobacco and some did not control for cigarette use (11). However, the case—control studies and case series that make clear the distinction between snuff and chewing tobacco provide sufficient evidence that oral use of snuff is carcinogenic (2). Across several studies conducted in the South, where snuff dipping has been popular among adults, the percentage using snuff varied from 11 to 40, with even higher rates in some subgroups, compared with only a few percent in disease-free controls (I I ). Winn et al. (12) undertook a landmark study in North Carolina in which they overcame many previous method- ologic problems and substantiated earlier conclusions. These authors found that, among female nonsmokers, the risk of oral cancer was 4.2 times greater for those who used snuff than for those‘who did not, with the cancer risk greatest in the parts of the mouth where the snuff was ha— bitually placed. The risk increased with longer duration of use, reaching a relative risk of 48 for cancer of the gum and buccal mucosa among users of snuff for over 50 years. The authors calculated that 87% of the cancers of the gingival and oral mucosa in the southeastern United States popu- lation were due to snuff use. A more recent case—control study conducted by the National Cancer Institute with four population-based cancer registries also found a substantial oral cancer risk associated with snuff or snuff and chew- ing tobacco combined; however, this risk could be quan- titated only in the few women study subjects who did not smoke (6). CHEWING TOBACCO AND ORAL CANCER The epidemiologic evidence for the carcinogenicity of chewing tobacco is presently less convincing primarily because of methodologic shortcomings, i.e., questionable validation of use, failure to control for smoking, and failure to provide dose—response data. Evidence for a pos- . itive association was provided by two of the five available case—control reports (13, 14), although no control was pro- vided for cigarette smoking in either study. One study (15) that did control for smoking found an odds ratio of 12 for oral cancer among chewers versus nonchewers; however, these results were not confirmed in the study by Wynder and Stellman (16) that also controlled for cigarette smok- ing. Because chewing tobacco also contains the same car- cinogenic N-nitrosamines as snuff (I ), Winn speculates that the reason for the lack of agreement among findings may occur because most forms of chewing tobacco are, in fact, chewed. Because of movement in the mouth and increased salivation, physical contact between the tobacco and oral mucosa is not maintained as intensely as for snuff, which is usually maintained in one spot (I I ). Thus the proximity of tissues to the tobacco and duration of direct contact may be factors of some significance. ' SMOKELESS TOBACCO AND OTHER CANCERS Data on humans are insufficient for researchers to con- clude that SLT causes cancer at nonoral sites. However, limited but suggestive data exist to raise the possibility of the carcinogenic potential of snuff at other aerodigestive sites where exposure is direct and prolonged, such as the esophagus, supraglottic larynx, and stomach. Nasal use of snuff has been associated with increased nasal cancer in studies in the United States, the United Kingdom, other Eu— ropean countries, and Africa (I, 10). In the United States, a 50% excess risk associated with snuff use was found in a case—control study of nasal cancer with increased risks ob- served for adenoid and squamous carcinomas (I 7). A case report described squamous cell carcinoma that developed in the ear of a man who placed snuff in his ear for 42 years (18). It appears that the risk for bladder cancer in users is not increased; results for kidney cancer are inconsistent. At present, insufficient epidemiologic evidence exists for one to evaluate conclusively the relationships, although several studies have shown an increased risk (10). Investigators need to conduct studies of sufficient size to detect small effects and to control for potentially confounding variables. NCI MONOGRAPHS, NUMBER 8, 1989 SLT: ASSOCIATION WITH INCREASED RISK 15 ASIAN SMOKELESS TOBACCO PRODUCTS The highest rates Of oral cancer in the world are found in India, where oral cancer is the most common tumor (1, 2). Tobacco use is widespread in this country with be- tween 47% and 73% of the population using some form of tobacco including cigarettes, cigars, and chewing (19). The inclusion of lime, areca nut, and other ingredients in many of the SLT-containing products Of India hinders the evaluation of the contribution of tobacco per se to the in- creased risk of oral tumors found in this country. How- , ever, it is believed that none of these additives has more than a small direct carcinogenic effect on oral tissues and that the high rate of malignant transformation in groups who chew these products is due to the presence of to- bacco. Indeed, an evaluation of 5 studies of oral cancer among chewers of betel quid with or without tobacco bears this out (I ). Users of tobacco-containing products had sub— stantially higher oral cancer risks compared with users of non-tobacco-containing quid. Confounding from cigarette smoking was also ruled out in these studies. The report con- cludes that the habit of quid chewing accounts for most of, the oral cancers in the diverse populations studied, which included Calcutta, Madras, Karachi, Bombay, and several other areas of India and Sri Lanka. SMOKELESS TOBACCO AND LEUKOPLAKIA In studies in the United States and Scandinavia, between 8% and 59% of the SLT users were found to have oral leukoplakia (10). Lesions are commonly found at the ha— ' bitual site of tobacco placement, and long-term dura- tion of the habit and daily intensity of use are determinants Of the severity of the leukoplakias. Quantitative estimates Of the risk for malignant transformation are difficult .for re- searchers to obtain, but an apparent'con’sistent finding is that tobacco-induced leukoplakias can develop into cancer over a period of years (10). In a large series of over 23,000 white Americans over 35 years of age, 3.4% on examination had white kera- totic changes of the mucosal surface, 86% of whichwere leukoplakias (20). Twenty-five percent of the ‘individu: als with leukoplakias underwent biopsies; of these, about one-quarter of them were demonstrated to have'either early invasive carcinoma or severe epithelial dysplasia. If the overall rate of cancerous leukoplakias was the same as the rate in the biopsy specimens, 168 cases or 0.7% of the pop- ulation had cancers. . COMMENT AND CONCLUSION The data on carcinogenic effects of SLT, particularly snuff used orally, are convincing and sufficient to arouse public health concern over the growing use Of this prod- uct by youth. Additional features of this growing use are especially disturbing. 1) Hospital case reports and registries (21, 22) and national data (23, 24) show an increase in tongue cancer among younger males under 40. Although differing opinions exist (21, 25), this increase may be related to the rise in SLT use among young men and may be the harbinger of an impending oral SMOKELESS TOBACCO USE IN THE UNITED STATES cancer epidemic in this population. It reinforces the need for both close monitoring of the oral condition of young current users as well as former users and active education and prevention programs. 2) If this epidemic of oral cancer among the young materializes, the high rate of second primaries often observed in later life for these types of cancers (26) becomes a matter of serious concern in the survival of this population. 3) The potential for synergistic interactions with other oral cancer risk factors may be enhanced by the tendency of youthful SLT users also to smoke cigarettes and drink alcohol and for users without these other habits to adopt them with greater fre- quency than nonusers (27, 28). 4) The strong relationship between oral cancer risk and duration of SLT use seen in studies of older persons with oral cancer implies that the early onset of the SLT habit in today’s youth, and thus the opportunity for sustained exposure, is a matter of serious concern. Its use in a kindergarten population has been reported (29), and use before junior high school is common (30). 5) Most of the epidemiologic research on SLT was done with users of “dry” snuff, whereas the most popular form among young users today is moist snuff. Chemical analysis reveals that both types have the same carcinogenic potential (31). 6) Studies of youthful users show that soft tissue lesions and leukoplakias are not infrequent (I ). Long-term follow-up studies are not available, and the magnitude of malignant transformation is un- known, although regression of lesions with discon- tinuance of use has been observed (1 ). The present evidence is cause for a serious public health alarm that, unless the tide Of SLT use is stemmed, long-term use can be expected to produce an increase in oral cancers, and perhaps cancers of other sites, as youthful users mature and accumulate exposure to this carcinogenic agent. REFERENCES (1) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco: A Report of the Advisory Committee to the Surgeon General. DHHS (NIH) Publ No. 86-2874. Bethesda, MD: Natl Cancer Inst, 1986 (2) INTERNATIONAL AGENCY FOR RESEARCH ON CANCER: To- bacco Habits Other Than Smoking; Betel-Quid and Areca-Nut Chewing; and Some Related Nitrosamines. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, vol 37. Lyon: IARC, 1985 (3) DEPARTMENT OF HEALTH AND HUMAN SERVICES: Health Im- plications of Smokeless Tobacco Use. National Institutes of Health Consensus Development Conference, vol 6, No. 1. Bethesda, MD: Natl Inst Health Off Med Applications Res, 1986 (4) WORLD HEALTH ORGANIZATION. Smokeless Tobacco Con- trol: Report of a WHO Study Group. Geneva: WHO, 1987 (5) OFFICE OF SMOKING AND HEALTH: Health Consequences of Smoking: Cancer. A Report of the Surgeon General. DHHS (PHS) No. 82-50179. Washington, DC: US Govt Print Off, 1982 16 MATTSON AND WINN (6) BLOT WJ, McLAUGHLIN JK, WINN DM, ET AL: Smoking and drinking in relation to oral and pharyngeal cancer. Cancer Res 48:3282—3287, 1988 (7) DEPARTMENT OF HEALTH AND HUMAN SERVICES: 1987 An— nual Cancer Statistics Review. DHHS (NIH) Publ No. 88-2789. Bethesda, MD: Natl Cancer Inst, 1988 (8) MASHBERG A, BARSA P: Screening for oral and oropharyn- geal squamous carcinomas. CA 34:262—267, 1984 (9) CONNOLLY GN, WINN DM, HECHT SS, ET AL'. Science, public policy, and the maemergence of smokeless tobacco. N Engl J Med 314:1020—1027, 1986 (10) CULLEN JW, BLOT W, HENNINGFIELD J, ET AL: Health con- sequences of using smokeless tobacco: Summary of the advisory committee’s report to the Surgeon General. Pub— lic Health Rep 1012355—373, 1986 (I I) WINN DM: Tobacco Chewing and Snuff Dipping: An Asso— ciation with Human Cancer. IARC Sci Publ No. 57. Lyon: IARC, 1984, pp 837—849 (12) WINN DM, BLOT WJ, SHY CM, ET AL: Snuff dipping and oral cancer among women in the southern United States. N Engl J Med 304:745—749, 1981 (I3) VOGLER WR, LLOYD JW, MILMORE BK: A retrospective study of etiological factors in cancer of the mouth, phar- ynx, and larynx. Cancer 15:246—258, 1962 (I 4) WYNDER EL, BROSS IJ, FELDMAN RM: A study of the etiolog- ical factors in cancer of the mouth. Cancer 10:1300—1323, 1957 (I5) MARTINEZ 1: Factors associated with cancer of the esopha- gus, mouth, and pharynx in Puerto Rico. J Natl Cancer Inst 42:1069—1094, 1969 (I6) WYNDER EL, STELLMAN SD: Comparative epidemiology of tobacco—related cancers Cancer Res 37:4608—4622, 1977 (I 7) BRINTON LA, BLOT WJ, BECKER JA, ET AL: A case—control study of cancers of the nasal cavity and paranasal sinuses. Am J Epidemiol 119:896—906, 1984 (I8) ROOT HD, AUST JB, SULLIVAN A: Snuff and cancer of the ear. N Engl J Med 262:819—820, 1960 (I9) SQUIER CA: Smokeless tobacco and oral cancer: A cause for concern? CA 34:242—247, 1984 (20) BOUQUOT JE, GORLIN RJ: Leukoplakia, lichen planus, and other oral keratoses in 23,616 white Americans over the age of 35 years. Oral Surg 61:373—381, 1986 (21) SCHANTZ SP, BYERS RM, GOEPFERT H: Tobacco and cancer of the tongue in young adults. JAMA 259:1943—1944, 1988 (22) SHEMEN LI, KLOTZ J, SCHOTTENFELD D, ET AL: Increase of tongue cancer in young men. JAMA 252:1857, 1984 (23) DEPUE RH: Rising mortality from cancer of the tongue in young white males. N Engl J Med 3152647, 1983 (24) DAVIS S, SEVERSON RK: Increasing incidence of cancer of the tongue in the United States among young adults. Lancet 2:910—911,1987 (25) MACFARLANE GJ, BOYLE P, SCULLY C: Rising mortality from cancer of the tongue in young Scottish males. Lancet 2:912, 1987 (26) WINN DM, BLOT W]: Second cancer following cancers of the buccal cavity and pharynx in Connecticut. Natl Cancer Inst Monogr 68:25—48, 1985 (2 7) DENT CW, SUSSMAN S, JOHNSON A, ET AL: Adolescent smokeless tobacco incidence: Relations with other drugs and psychosocial variables. Prev Med 16:422—431, 1987 (28) ARY DV, LICHTENSTEIN E, SEVERSON HE: Smokeless to- bacco use among male adolescents: Patterns, corre- lates, predictors, and the use of other drugs. Prev Med 16:385—401, 1987 (29) YOUNG M, WILLIAMSON D: Correlates of use and expected use of smokeless tobacco among kindergarten children. Psychol Rep 56:63—66, 1985 (30) BOYD G, ARY DV, WIRT R, ET AL: Use of smokeless tobacco among children and adolescents in the United States. Prev Med 16:402—421, 1987 (31) HOFFMANN D, ADAMS JD, LISK D, ET AL: Toxic and carcino- genic agents in moist and dry snuff. JNCI 79:1281—1286, 1987 Use of Smokeless Tobacco in the United States: Recent “Estimates From the Current Population Survey1 Alfred C. Marcus,2 Lori A. Crane,2 Donald R. Shopland,3 and William R. Lynn3 ABSTRACT—Chewing tobacco, snuff, and total smokeless to- bacco use from the 1985 Current Population Survey (CPS) are reported. The CPS is the only survey capable of providing na- tional, regional, and individual state tobacco use estimates for all 50 states and the District of Columbia. The prevalence of smoke- less tobacco use varies considerably among and within regions of the country, by division and state. Smokeless tobacco use is high- est in the South and lowest in the Northeast. Individual states with the highest smokeless tobacco use among males are West Vir- ginia (23.1%), Mississippi (16.5%), Wyoming (15.8%), Arkansas (14.7%), and Kentucky (13.6%). In all regions of the country, use of smokeless tobacco among women is considerably less than men. Nationally, male use of such products was 5.5%; less than 1% of women use them. Snuff consumption is predominantly a behavior characteristic of white males; less than 1% of black or Hispanic males consume this product. Higher percentages of blue-collar and service workers use it compared with white-collar workers. Snuff and chewing tobacco use among teenage boys in the United States increased dramatically between 1970 and 1985, a time when their use of cigarettes was declining. The significance of in- dividual state level estimates is discussed.—NCI Monogr 8:17-23, 1989. Periodic national surveys for tobacco use assessment have been sponsored by numerous agencies and institutes within the United States Public Health Service. Most of these national surveys have focused on smoking behav- ior, with only limited attention given to the use of SLT. The first large-scale survey of tobacco use was conducted by the National Cancer Institute in 1955, when detailed questions about cigarette and pipe and cigar smoking were added to the CPS. This survey did not include questions on the use of chewing tobacco or snuff. Similarly, although the National Center for Health Statistics has collected limited data on cigarette smoking as part of the ongoing National Health Interview Survey, these surveys have not collected information routinely on other forms of tobacco use. When the National Clearinghouse for Smoking and ABBREVIATIONS: SLT = smokeless tobacco; CPS = Current Pop- ulation Survey; CI = confidence interval(s). 1Supported in part by a contract with the Office on Smoking and Health (Program Consultant Agreement No. 5651), Public Health Service, Department of Health and Human Services. 2Division of Cancer Control, Jonsson Comprehensive Cancer Center, 1100 Glendon Ave., Suite 711, Los Angeles, California 90024. Address reprint requests to Alfred C. Marcus, PhD. 3 Smoking, Tobacco, and Cancer Program, Division of Cancer Preven- tion and Control, National Cancer Institute, Bethesda, Maryland. Health (now the Office on Smoking and Health) was es- tablished after the release of the first report of the Surgeon General on smoking (1 ), several surveys on adult use of tobacco were conducted that included questions on other forms of tobacco use, including patterns of chewing to- bacco and snuff use. However between 1966 and 1986, only 4 such surveys were conducted. Beginning in the early 1980s, the attention of the public health community began to focus on the use of SLT as a significant public health problem. At that time, it became apparent that both national and regional data on its use were urgently needed. In 1985, one of the authors (DRS), while serving as Acting Director of the Office on Smoking and Health, approved funding for a short series of ques— tions about cigarette smoking and use of other tobacco products (snuff, chewing tobacco, pipe and cigar smoking) to be added to the September 1985 CPS. Information on tobacco use was collected from 114,342 individuals. The large sample size of the CPS provides a unique opportunity for researchers to examine national, regional, divisional, and state estimates for all forms of tobacco use. The focus of this paper is on SLT use in the United States METHODOLOGY The CPS is a continuous survey conducted monthly by the Census Bureau primarily as a means to assess changes in the labor force for the civilian noninstitutionalized pop- ulation. The 1985 CPS sample was initially selected from the 1980 census files with coverage in all 50 states and the District of Columbia. The 1985 sample was located in 729 areas comprising 1,973 counties, independent cities, and minor civil divisions in the nation. 0f the approximately 62,500 occupied households that were eligible for inter- view, Visits to about 2,500 occupied units (4%) resulted in no completed interviews because the occupants were not found at home (after repeated contacts), they refused to be interviewed or were otherwise unavailable for interview. Although all household members were enumerated, a to— bacco use history was obtained only for those individuals 16 years of age or older at the time of interview. With the use of a special weighting algorithm developed by the Census Bureau, the CPS household sample estimates are considered to be representative of the United States. However, one potential drawback with the CPS is the effect of proxy reports on sample estimates that is presumed to re- sult in a modest underreporting bias. Overall, approximately 55% of the total CPS sample consists of self-respondents, whereas the remaining 45% of the data represents proxy re- 17 18 MARCUS ET AL. sponses. The suspected underreporting bias should be most evident for the younger age groups (e.g., 16—19 yr), be— cause they have a higher proportion of proxy responses. In addition, it is conceivable that the proxies for younger re- spondents included in the CPS (e.g., parents) may be less knowledgeable or aware of SLT use, especially if they do not approve of such use among adolescents and younger adults living in the household. Two questions were asked in the CPS so that current use of SLT could be determined. For those individuals for whom “yes” was recorded for the question, “Does (name) presently use any other form of tobacco, such as snuff or chewing tobacco?”, a follow-up question asked, “What other form(s) of tobacco does (name) presently use?” The categories of snuff, chewing tobacco, cigars, pipe tobacco, or other were coded in response to this follow-up question. Given the substantial sample size of the CPS (~114,000 records), rates of SLT use are reported by type of product (snuff or chewing tobacco) separately by region, division, and state. Using formulas provided by the Census Bureau (which take into account the multistage sample design of the CPS), we calculated the 95% CI for each estimate and include them here. Detailed data are only presented for males; less than 1% of females use these tobacco products (0.5% and 0.2% of the women reported using snuff and chewing tobacco nationally, respectively). RESULTS Differences by Region, Division, and State As indicated in table 1, the overall prevalence rate for males is estimated at 1.9% for snuff and 3.9% for chew- ing tobacco. Use of SLT among males was lowest in the Northeast and highest in the South, with the North Cen- tral and West ranking intermediate. Within the Northeast region, both the New England and the Mid-Atlantic divi— sions had relatively low reported use, but prevalence rates among males in the North Central region approximated the national average for snuff (2.1%) and chewing tobacco (3.4%). In the South, use of snuff (2.7%) and chewing tobacco (6.0%) exceeded the national average among males. For TABLE l.——Percent current users of snuff and chewing tobacco by region, division, and state: 216-yr-old males” Chewing Any Chewing Any Snuff tobacco smokeless Snuff tobacco smokeless Area Percent CI Percent CI Percent CI Area Percent CI Percent CI Percent C1 United States 1.9 0.2 3.9 0.3 5.5 0.4 Washington, DC 0.0 0.0 0.4 1.0 0.4 [.0 Northeast 1.0 0.3 1.4 0.4 2.3 0.5 Virginia 2: 1-9 125 i; 2;? :3 West irginia l . 4.4 . . . . N E l 0.4 .4 0.8 0.6 1.2 . main? and 0 9 (1)3 1 5 1 7 2 3 (2)? North Carolina 1.8 1.0 8.6 2.1 9.8 2.2 New Hampshire 1'2 1‘7 1'5 1'9 2‘7 2'5 South Carolina 0.7 1.1 5.3 3.0 6.1 3.2 Vermont 0.9 1.4 4.7 3.2 5.5 3.5 06°93” 1'4 1'6 7'3 3‘4 8‘7 3'7 Massachusetts 0.2 0.3 0.4 0.5 0.5 0.5 F'°“da 1" 0'7 1‘9 0‘9 2'9 1‘1 Rhode Island 0.5 1.1 0.6 1.2 0.9 15 East South Central 2.7 1.1 9.4 1.9 11.6 2.1 Connecticut 0.3 2.3 0.5 2.9 0.8 1.2 Kentucky 3-2 2.4 11-2 4-4 13-6 4-7 Mid-Atlantic 1.2 0.4 1.6 0.5 2.7 0.7 Tennessee 1'7 1'8 9‘3 4'1 103 4'2 New York 0 5 0.4 1.2 0.6 1 6 07 Alabama 1.7 2.0 6.6 3.9 8.3 4.3 New Jersey 0'1 0 2 0 6 0 5 0'7 0 6 Mississippi 5.7 3.3 11.4 4.5 16.5 5.3 Pennsylvania 30 1.2 2.9 1.2 5.6 1.6 West South Central 4.0 1.0 5.5 1.1 9.1 . 1.4 North Central 2.1 0.5 3.4 0.6 5.3 0.7 33:52:: 3(5) . :‘2‘ 2: ‘3‘; 12-; g; 0h1_o 2.2 1-0 3-2 1.3 5.0 16 Texas 4.0 1.4 4.6 1.5 8.2 2.0 $139 if (I): 3: f? :2 f: West 1.4 0.4 3.3 0.6 4.5? 0.8 {nois ' ' ' ‘ ‘ ' Mountain 2.3 1.1 5.4 1.6 7.5 1.9 Michlgan 0.8 0.6 2.7 1.2 3.4 1.3 M 5 5 3 1 8 3 3 7 13 7 4 7 Wisconsin 2.9 2.3 2.9 2.3 5.8 3.1 Idgg‘ana 2‘3 2'1 6'7 3‘5 8'7 3'9 0 . . . . . . Minnesota 3.5 2.5 2.8 2.2 6.1 3.2 Colorado 1_2 1.6 6.4 3.6 7.5 39 Iowa 1-8 1.9 4.6 2.9 6.4 3.4 New Mexico 5.3 3.0 5.2 3.0 10.2 4.1 Missouri 3.1 2.1 3.6 2.2 6.7 3.0 Arizona 240 2_1 3_g 2.8 5_4 3.4 North Dakota 6.1 3.1 5.1 2.9 10.7 4.0 Utah ()9 1.3 3.0 2.4 3.7 2.7 South Dakota ‘ 1.9 1.7 6.1 3.0 7.9 34 Nevada 15 2.0 2.3 2.7 4.3 33 Nebraska 1.4 1.6 6.8 3.4 8.0 3.6 Pacific 1.0 0.4 2.6 0.7 3.4 0.8 Kansas 3-3 25 8-6 3-9 1 1-7 4-4 Washington 18 1.9 6.1 3.5 7.1 3.7 South 2-7 0.5 6-0 0.7 8.3 0-8 Oregon 2.7 2.5 5.4 3.4 7.6 4.0 South Atlantic 1.8 0.5 5.2 0.9 6.7 1.0 California 0.7 0.5 1.7 0.7 2.3 0.8 Delaware 0.6 1.2 2.4 2.3 3.0 26 Alaska 2.5 2.2 6.3 3.4 8.8 4.0 Maryland 0.4 0.8 2.1 1.8 2.4 1.9 Hawaii 0.2 0.7 0.4 0.9 0.7 1.2 '1 Estimates were obtained from the September 1985 CPS. NCI MONOGRAPHS, NUMBER 8, 1989 SLT: RECENT ESTIMATES FROM THE CPS 19 example, West Virginia reported the highest use of snuff (11.5%) and chewing tobacco (13.5%) in the nation. West Virginia is the only state in the South Atlantic division that exceeded the regional average for snuff. In the East South Central and West South Central divisions, Mississippi (5.7%), Arkansas (6.0%), Oklahoma (4.8%), and Texas (4.0%) also reported relatively high use of snuff (when compared with other Southern states), whereas Mississippi (11.4%), Kentucky (11.2%), Arkansas (9.5%), and Ten- nessee (9.3%) also reported relatively high use of chewing tobacco. For males, the West had among the lowest rates in the country with respect to snuff use (1.4%) and chewing tobacco (3.3%). Also reported in table 1 are estimates of “any” SLT use (snuff or chewing tobacco). Of special note in this re- gard is the finding that rates of any use are largely ad- ditive across snuff and chewing tobacco, which indicates little overlap in the use of these tobacco products among males. States reporting exceptionally high rates of any use included West Virginia (23.1%), Mississippi (16.5%), Wyoming (15.8%), Arkansas (14.7%), Montana (13.7%), and Kentucky (13.6%). Table 2 provides rankings of states with the highest and lowest rates of snuff and chewing tobacco use for males. As shown, states from the South dominate the rankings for high use of both substances; states from the Northeast dominate the rankings for low use of both substances. In- terestingly, states ranking high in use of SLT also tend to have a high percentage of nonmetropolitan residents. Thus when we correlated state rates of SLT use (among males) with the percentage of the state population residing out- side designated Standard Metropolitan Statistical Areas, we found significant bivariate correlations for snuff (r=.49; P<.001), chewing tobacco (r=.60; P<.001), and any use of SLT (r=.66; P<.001). . Use of SLT is much less prevalent among women in the United States than among men. Overall, an estimated 0.5% of women 16 years of age and over reported using snuff, and they were concentrated almost entirely in the South (1.4%), with all other regions having prevalence rates of 0.1% or less. Southern states with the highest reported use of snuff among women included North Carolina (4.0%), South Carolina (2.2%), Georgia (2.7%), Alabama (2.6%), Mississippi (2.0%), and Tennessee (1.9%). Only one other state in the nation reported a prevalence rate for snuff that exceeded 2.0% among women (Alaska = 2.1%). Use of chewing tobacco is even less prevalent among women than use of snuff, with an estimated 0.2% preva- lence rate nationally. The Northeast, North Central, and West had rates of chewing tobacco use that were 0.1% or less for women. It was highest in South Carolina (1.1%), Mississippi ( 1.4%), and Arkansas (1.7%). No other state in the nation exceeded a prevalence rate of 1.0%. Differences by Race or Ethnicity Tables 3 and 4 depict racial and ethnic differences among males in reported use of SLT. As shown, estimates for the United States show white males reporting noticeably higher rates of using snuff (2.2%) and chewing tobacco (4.3%). Black and Hispanic males are similar in their reported use of snuff, with both groups reporting a much lower rate of 0.7%. With respect to chewing tobacco, black males (2.6%) ranked second to whites and Hispanics (1.1%) ranked third. White males maintain their higher reported use of snuff and chewing tobacco in all regions and divisions within the United States. The use of both among white males was greatest in the South, with the highest rate of snuff use reported in the West South Central division (4.9%) and the highest rate of chewing tobacco use reported in the East South Central division (10.4%). TABLE 2.—States ranking highest and lowest in use of snuff and chewing tobacco by 216-yr-old males Highest rates Lowest rates Tobacco product State Region Rate State Region Rate Snuff West Virginia South 11.5 Washington, DC South 0.0 North Dakota North Central 6.1 New Jersey Northeast 0.1 Arkansas South 6.0 Hawaii West 0.2 Mississippi South 5.7 Massachusetts Northeast 0.2 Montana West 5.5 Connecticut Northeast 0.3 New Mexico West 5.3 Maryland South 0.4 Oklahoma South 4.8 Rhode Island Northeast 0.5 Texas South 4.0 New York Northeast 0.5 Minnesota North Central 3.5 Delaware South 0.6 Wyoming West 3.4 South Carolina South 0.7 California West 0.7 Chewing West Virginia South 13.5 Massachusetts Northeast 0.4 tobacco Wyoming West 130 Hawaii West , 0.4 Mississippi South 1 1.4 Washington, DC South 0.4 Kentucky South 1 1.2 Connecticut Northeast 0.5 Arkansas South 9.5 New Jersey Northeast 0.6 Tennessee South 9.3 Rhode Island Northeast 0.6 North Carolina South 8.6 New York Northeast 1.2 Kansas North Central 8.6 Maine Northeast 15 Montana West 8.3 New Hampshire Northeast 1.5 Georgia South 7.3 California West 1.7 SMOKELESS TOBACCO USE IN THE UNITED STATES 20 MARCUS ET AL TABLE 3.—Percent current users of snuff by race, region, and divisiorL‘ Zl6—yr—old males“ White Black Hispanic Area Percent CI Percent C1 Percent C1 United States 22 0.3 0.7 0.4 0.7 05 Northeast 1.2 0.4 0.1 0.4 0.2 0.6 New England 0.5 0.5 — — — — Mid-Atlantic 1.5 0.5 0.2 0.6 0.2 0.7 North Central 2.3 0.5 0.1 0.4 0.0 0.0 East North Central 2.0 0.6 0.2 0.6 0.0 0.0 West North Central 3.1 1.1 0.0 0.0 — — South 3.1 0.6 1.0 0.7 1.6 1.4 South Atlantic 2.2 0.7 0.7 0.8 0.7 1.7 . East South Central 2.9 1.2 1.8 2.2 —— -— West South Central 4.9 1.3 1.0 1.4 2.0 1.9 West 1.6 0.5 0.9 1.6 0.3 0.5 Mountain 2.6 1.2 — — 1.1 2.0 Pacific 1.2 0.5 1.0 1.8 0.0 0.0 ‘1 Estimates were obtained from the September 1985 CPS. Estimates based on fewer than 100 cases (as indicatedby dashes) were deleted from the table. Use of SLT by women was higher among blacks than whites or Hispanics. Overall, 2.1% and 1.2% of black women are estimated to use snuff and chewing tobacco, re— spectively; the corresponding figures for white women are 0.3% and 0.1%. Among Hispanic women, use of SLT was virtually nonexistent (less than 0.1% for snuff and chewing tobacco). Among black women, use of snuff was estimated at 0.3% in the Northeast, 0.5% in the North Central, 3.5% in the South, and 0.4% in the West. Among white women, use of snuff was estimated at less than 0.1% in the North— east, North Central, and West, and 0.9% in the South. With respect to the use of .chewing tobacco, prevalence rates for black women were estimated at 0.3% in the North- TABLE 4.—Percent current users of chewing tobacco by race, region, and division: Zl6—yr-old males“ White Black Hispanic Area Percent Cl Percent CI Percent CI United States 4.3 0.4 2.6 0.8 1.1 0.7 Northeast 1.6 0.5 0.7 1.0 0.4 0.9 New England 0.8 0.6 — — — — Mid—Atlantic 1.9 0.6 0.7 1.1 0.5 1.1 North Central 3.7 0.6 0.6 0.9 2.7 3.7 East North Central 3.2 0.7 0.6 1.0 0.7 2.1 West North Central 4.8 1.3 0.4 1.8 —— — South 7.0 0.8 3.9 1.3 1.3 1.3 South Atlantic 5.8 1.1 3.6 1.7 0.5 1.4 East South Central 10.4 2.2 4.6 3.4 — — West South Central 6.6 1.5 4.1 2.8 1.6 1.7 West 3.9 0.8 2.2 2.5 1.1 1.0 Mountain 5.8 1.8 — - 2.3 2.9 Pacific 31 0.9 2.6 2.9 0.6 0.9 ‘1 See footnote, table 3. east and..West, 0.1% in the North Central, and 2.0% in the South. Among white women, rates of chewing tobacco use did not exceed 0.1% in any region of the country except the South where it was 2%. Mei-em by Occupation Occupational differences among males in reported use of snuff follows the same gradient typically found for cigarette smoking (2). Thus male white-collar workers reported lower rates, followed by service and blue—collar workers (table 5). Reported use of chewing tobacco among males shows a similar pattern nationally, with rates of white-collar workers typically being the lowest, followed by service and blue—collar workers (table 6). Reported use of snuff among females was virtually nonexistent for white-collar workers (0.0%), followed by service (0.4%) and blue-collar workers (0.5%). Use of chewing tobacco was also virtually nonexistent for female white—collar workers (0.0%), compared with 0.4% for fe- male service and blue-collar workers. Use of snuff and chewing tobacco among female service and blue—collar workers was confined largely to the South. One exception to this pattern occurs in the Mountain division of the West, where 0.9% of female blue-collar workers reported using chewing tobacco. Differences by Age As reported in table 7, young adults reported higher rates of using snuff, which is especially evident in the South. In contrast, no clear age pattern is observed in use of chewing tobacco among males (table 8), although the oldest age group reported slightly higher usage than did other age groups. In the West, use of chewing tobacco among males was higher in the younger age groups (16—29 yr). With respect to females, a modest age gradient occurred in use of SLT, with the 70 and over group reporting the highest rates of snuff (2.0%) and chewing tobacco (0.5%) TABLE 5.—Percent current users of snuff by occupation, region, and division: Zl6-yr-old males Workers White collar Service Blue collar Area Percent CI Percent CI Percent C1 United States 1.0 0.3 1.6 0.7 2.8 , 0.5 Northeast 0.4 0.4 0.5 0.8 1.7 0.8 New England 0.1 0.4 0.8 2.1 0.5 0.9 Mid—Atlantic 0.5 0.5 0.4 0.9 2.0 1.0 North Central 1.1 0.6 1.5 1.4 2.9 0.9 East North Central 0.8 0.6 1.5 1.7 2.4 1.0 West North Central 1.6 1.4 1.7 2.9 4.1 2.0 South 14 0.6 2.8 1.8 3.9 0.9 South Atlantic 0.9 0.7 2.3 2.2 2.4 1.1 East South Central 1.2 1.4 3.0 4.9 3.9 2.0 West South Central 2.3 1.3 3.5 3.5 6.3 2.0 West 1.0 0.6 1.2 1.4 1.6 08 Mountain 1.7 1.5 1.6 2.8 3.2 2.3 Pacific 0.8 0.6 _1.1 1.6 0.9 0.7 NCI MONOGRAPHS, NUMBER 8, 1989 SLT: RECENT ESTIMATES FROM THE CPS TABLE 6.—Percent current users of chewing tobacco by occupation, region, and division: Zl6-yr-old males Workers White collar Service Blue collar Area Percent CI Percent CI Percent CI United States 2.1 0.4 3.1 1.0 5.2 0.6 Northeast 0.7 0.5 0.8 1.1 2.0 0.9 New England 0.3 0.6 0.7 2.0 1.0 1.2 Mid-Atlantic 0.9 0.6 0.9 1.3 2.3 1.1 North Central 1.7 0.8 2.5 1.8 4.5 1.1 East North Central 1.1 0.7 2.4 2.1 4.2 1.3 West North Central 3.0 1.9 2.9 3.7 5.3 2.3 South 3.4 0.9 4.3 2.2 7.6 1.3 South Atlantic 2.3 1.0 4.0 2.9 7.2 1.8 East South Central 5.5 2.9 8.7 8.1 10.8 3.2 West South Central 4.2 1.7 2.7 3.1 6.0 2.0 West 1.9 0.8 439 2.7 4.9 1.4 Mountain 3.5 2.1 7.2 5.8 7.8 3.5 Pacific 14 0.8 3.8 2.9 3.9 1.5 use. As noted previously, use of snuff and chewing tobacco among females is confined almost entirely to the South, and, in keeping with this, the age gradient was most pronounced in the South Atlantic and East South Central divisions. Among males, the younger age groups typically reported higher rates of using snuff than did other age groups. This raises the question of whether this pattern reflects a re- cent development, or whether it has been in existence for many years. In particular, if the higher rates of snuff use in the younger adult age groups is a recent phenomenon, then it would document the widespread adoption of snuff among the demographic subgroup targeted by SLT adver— tising practices in the 19708 and early 1980s (i.e., young adult males). In the 1986 Surgeon General’s Report, rates of SLT use obtained from the 1985 CPS were compared with the corresponding rates obtained from the 1970 Na- 21 tional Health Interview Survey (3). Both surveys 1) adopted reasonably similar survey methodologies, 2) contained large sample sizes (114,000 and 77,000, respectively), and 3) in— cluded approximately the same proportion of proxy re- sponses (45%). As shown in table 9, a substantial increase in use of snuff was evident among males 16—29 years of age, with nearly a tenfold increase in use of snuff among males 16—19 years of age. A similar, though less pronounced, pat- tern was found for males who chewed tobacco. SIGNIFICANCE OF STATE LEVEL ESTIMATES The CPS exists as the only survey of the United States population designed to provide national and individual state level estimates of SLT use for all 50 states and the District of Columbia. The results reported herein afford a unique opportunity for state health administrators, researchers, and public health officials to compare SLT use patterns both across and between individual states and regions. Such information will allow state health agencies to plan more effective control efforts in those states where rates are high. The CPS also affords epidemiologists a tool to examine morbidity and mortality patterns among those states with varying levels of tobacco use rates. Economists can examine these data and determine how patterns in use are affected by differences in tax rates among those states or regions where such taxes are applied. If future data collections by researchers using the CPS are conducted, 1985 results could be compared over time with new estimates as a means of assessing policy issues for control of tobacco use for individual states. DISCUSSION At the turn of the century, the major form of tobacco consumed in the United States was SLT (4). Of the ap- proximately 7 pounds of tobacco consumed per adult in 1900, 4.1 were in the form of chewing tobacco. The re- maining categories of per capita tobacco consumption were 1.63 pounds of pipe tobacco and 0.32 pound of snuff. To- tal cigar and cigarette consumption accounted for less than TABLE 7.—Percent current users of snuff by age, region, and division: 216—yr-old males Age ranges (yr) and CI Area 16—19 CI 20—29 CI 30—39 CI 40—49 CI 50—59 CI 60-69 CI 70+ CI United States 2.9 0.6 2.7 0.3 1.8 0.3 1.5 0.3 1.2 0.3 1.1 0.3 1.9 0.5 Northeast 1.9 1.0 1.2 0.5 1.2 0.5 0.8 0.5 0.9 0.6 0.2 0.3 0.8 0.6 New England 0.5 1.0 0.8 0.8 0.5 0.6 0.1 0.3 0.3 0.7 0.2 0.5 0.3 0.8 Mid-Atlantic 2.3 1.2 1.4 0.6 1.4 0.6 1.0 0.6 1.1 0.7 0.2 0.3 1.0 0.8 North Central 2.6 1.0 2.3 0.6 1.7 0.5 2.2 0.7 1.9 0.7 1.8 0.8 2.7 1.1 East North Central 2.5 1.2 2.2 0.7 1.2 0.5 1.6 0.8 1.3 0.7 1.7 0.9 2.4 1.3 West North Central 2.8 2.0 2.5 1.2 3.0 1.3 3.5 1.7 3.6 1.9 2.1 1.5 3.3 2.1 South 4.7 1.2 4.5\ 0.7 2.5 0.6 2.0 0.6 0.8 0.4 0.9 0.5 2.2 0.9 South Atlantic 3.9 1.6 2.8 0.8 1.7 0.7 1.3 0.7 0.6 0.5 0.8 0.6 1.6 1.0 East South Central 4.3 2.6 5.5 1.8 1.8 1.1 2.1 1.5 0.8 1.0 0.5 0.8 2.4 2.0 West South Central 6.2 2.5 6.4 1.4 4.0 1.2 3.0 1.3 1.0 0.9 1.3 1.1 3.3 2.0 West 0.9 0.7 . bg 0.6 1.2 0.5 0.8 0.5 1.2 0.7 1.8 0.9 1.6 1.0 Mountain 2.2 2.1 V 3. 1.5 1.9 1.2 1.6 1.3 1.8 1.6 2.8 2.2 2.2 2.3 Pacific 0.4 0.6 1.2 0.6 1.0 0.5 0.5 0.5 1.0 0.7 1.5 1.0 1.5 1.1 SMOKELESS TOBACCO USE IN THE UNITED STATES 22 MARCUS ET AL. TABLE 8.—Percent current users of chewing tobacco by age, region, and division: 216-yr-old males Age ranges (yr) and Cl Area 16—19 CI 20—29 CI 30—39 CI 40—49 CI 50—59 CI 60—69 CI 70+ CI United States 3.0 0.6 4.2 0.4 3.7 0.4 3.3 0.4 3.9 0.5 4.2 0.6 4.9 0.7 Northeast 0.7 0.6 1.6 0.6 1.7 0.6 1.3 0.6 0.6 0.4 1.2 0.7 2.7 1.1 New England 1.5 1.7 1.0 0.9 0.7 0.7 0.6 0.8 0.3 0.7 0.8 1.1 0.9 1.4 Mid—Atlantic 0.4 0.5 1.8 0.7 2.0 0.8 1.5 0.8 0.7 0.6 1.3 0.8 3.2 1.4 North Central 3.1 1.1 4.0 0.8 3.6 0.8 3.0 0.9 3.0 0.9 2.8 0.9 3.6 1.3 East North Central 2.6 1.2 3.0 0.8 3.1 0.9 2.5 0.9 3.1 1.1 2.7 1.1 2.8 1.4 West North Central 4.4 2.5 6.7 1.9 4.6 1.6 4.1 1.9 2.7 1.7 3.1 1.8 5.2 2.6 South 4.1 1.1 4.8 0.7 5.6 0.8 5.5 1.0 7.6 1.3 8.0 1.4 8.9 1.7 South Atlantic 3.3 1.4 4.6 1.0 4.8 1.1 4.6 1.3 6.9 1.7 6.6 1.7 6.1 1.9 East South Central 7.3 3.4 7.0 2.1 7.9 2.3 8.7 2.9 11.7 3.6 12.3 3.9 15.5 4.8 West South Central 3.3 1.8 3.8 1.1 5.5 1.4 5.3 1.7 6.2 2.1 8.0 2.5 9.6 3.4 West 3.4 1.4 5.9 1.0 2.7 0.7 2.2 0.8 2.3 0.9 2.6 1.1 1.8 1.0 Mountain 2.8 2.4 11.1 2.7 4.2 1.8 3.1 1.9 3.4 2.2 2.3 2.0 4.2 3.1 Pacific 3.6 1.7 3.9 1.0 2.2 0.8 1.8 0.9 1.8 1.0 2.7 1.3 1.1 1.0 a pound Of tobacco per capita annually. The first blended cigarette was introduced in the United States in 1913. En- couraged by heavy advertising and promotion by manu- facturers, cigarette smoking had become the predominate form of tobacco consumption by the end of the First World War, a pattern which is still evident today. Per capita consumption Of cigarettes peaked in 1963, the year immediately prior to the release of the original Sur- geon General’s Report on smoking in January 1964 (I ). AS cigarette smoking became increasingly popular in the decades prior to the 1960s, use Of other forms Of tobacco, including SLT products, declined, both in the proportion of the population as well as the total pounds of tobacco con- sumed annually in the United States. In the 1970s, how- ever, SLT manufacturers began to advertise and promote these products aggressively, including new product lines of moist snuff. These products were increasingly promoted on television and other broadcast media (use of these media was banned by Congress for cigarettes and small cigars in the early 1970s). The packages did not contain health warnings, and much of the advertising was targeted toward adolescents and young adult males by prominent sports and entertainment figures. Unlike cigarette smoking, which still enjoys a much broader social appeal and acceptance nationally, use of SLT TABLE 9.—Prevalence of male SLT use by age, 1970 and 1985“ Snuff Chewing tobacco Age, yr 1970 1985 1970 1985 16—19” 0.3 2.9 1.2 3.0 20-29 0.6 2.7 1.9 4.2 30—39 0.7 1.8 2.8 3.7 40—49 1.2 1.5 3.0 3.3 250 2.7 1.4 6.5 4.2 “ Sources of data were the 1970 National Health Interview Survey and the 1985 CPS. b For 1970, this age group was composed Of 17- to l9-yr olds. is largely confined to much narrower segments Of the popu— lation. According to estimates derived from the 1985 CPS, users Of these products are primarily males, especially white males, who reside in the southern region of the country. For example, Of all users Of chewing tobacco in the United States, 90% are male, and of these, 90% are white and 50% reside in the South. A similar pattern is seen among snuff users; nearly 80% are male, Of whom 92% are white and more than 50% live in the South. In addition, it should be noted that SLT use is two to three times more prevalent among blue— than white-collar workers. Also noteworthy is that among males, reported use of snuff in 1985 was highest in the young adults. As reported earlier in table 9, this was not always true. In 1970, the youngest adult age group actually reported the lowest rate Of snuff use, with a perfect linear age gradient leading to higher rates Of use in the older age groups. However by 1985, this positive age gradient had been replaced by a striking inverse age gradient, highlighted by nearly a tenfold increase in reported use of snuff in teenage boys (16—19 yr). A similar (although less pronounced) 15-year trend was also evident for chewing tobacco, with teenage boys reporting over a twofold increase in use. Extrapolation of these trends to the total population in the United States indicates an aggregate increase Of more than a million new SLT users in 1985 compared with 1970. In contrast, the total number of cigarette smokers between 1970 and 1985 has remained virtually unchanged, despite large increases in the total population base. What impact these additional million or more tobacco users will have on national and regional cancer incidence and death rates will not become fully evident until after the turn of the century. REFERENCES (I) CENTERS FOR DISEASE CONTROL, PUBLIC HEALTH SERVICE: Smoking and Health. Report of the Advisory Committee to the Surgeon General of the Public Health Service. DHEW (PHS) Publ No. 1 103. Washington, DC: Centers for Disease Control, 1964 NCl MONOGRAPHS, NUMBER 8, 1989 SLT: RECENT ESTIMATES FROM THE CPS (2) OFFICE ON SMOKING AND HEALTH, PUBLIC HEALTH SERVICE: The Health Consequences of Smoking: Cancer and Chronic Lung Disease in the Workplace. A Report of the Surgeon General. DHHS (PHS) Publ No. 85-50207. Rockville, MD: Office Smoking and Health, 1985 (3) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health SMOKELESS TOBACCO USE IN THE UNITED STATES 23 Consequences of Using Smokeless Tobacco: A Report of the Advisory Committee to the Surgeon General. DHHS (NH-I) Publ No, 86-2874. Bethesda, MD: Natl Cancer Inst, 1986. (4) SHOPLAND DR: Smoking. In 1987 Medical and Health An- nual. Chicago: Encyclopedia Britannica, 1986, pp 420—424 Smokeless Tobacco Use in the United States: The Adult Use of Tobacco Surveys Thomas E. Novotny, John P. Pierce, Michael C. Fiore, and Ronald M. Davis1 ABSTRACT—Prevalence of smokeless tobacco use is reported for adults aged 21 years and older in the Adult Use of Tobacco Surveys, 1964-1986. Data from the 1986 survey on prevalence, beliefs, ages of initiation, and demographic correlates of use by males aged 17 years and older are also reported. The prevalence of smokeless tobacco use declined slightly among persons aged 21 and older between 1966 and 1986. However, 5.2% of the males aged 17 and older used smokeless tobacco in 1986, and prevalence was highest among those 17 to 19 years old (8.2%). The median age of initiation for both products was 19 years. Smokeless tobacco use was most common among white men who were 1) living in the southeastern United States, 2) unemployed, and 3) in blue-collar or service/labor employment. Most users (77.4%) and nonusers (83.4%) believe that smokeless tobacco is a health hazard. Many current users (39.1%) had attempted to quit. Some current (6.4%) and former smokers (7.0%) have used smokeless tobacco as an aid to smoking cessation. National survey data such as these permit the identification of high-risk groups, so that interventions against smokeless tobacco use may be specifically targeted.—NCI Monogr 8:25-28, 1989. In the United States, SLT has reemerged as a public health problem (I, 2). National surveys have identified young white males as the group at highest risk of using these products and have shown a higher prevalence among persons living in the southern and north central regions of the United States (3—5). In the longitudinal Bogalusa Heart Study, 8— to 17-year-old white males reported a large increase in SLT use between 1976 and 1981, including an eightfold increase in the use of snuff (6). American Indians and Alaskan Natives also have a high prevalence of SLT use (7, 8). Most disturbing is the national trend toward higher use among men who are less than 21 years old. Between 1970 and 1985, snuff and chewing tobacco use increased tenfold and more than twofold, respectively, among males aged 17 to 19 years. Smaller increases were observed among the middle-aged groups, and a decrease in the use of both products was noted for older men aged 50 and above (I ). The National Institutes of Health Consensus Develop— ment Conference Statement on the Health Implications of Smokeless Tobacco Use called for additional studies of SLT use based on national probability sample data (9). In 1986, ABBREVIATIONS: SLT = smokeless tobacco; AUTS = Adult Use of Tobacco Survey(s). 1Office on Smoking and Health, Center for Chronic Disease Preven- tion and Health Promotion, Centers for Disease Control, Department of Health and Human Services, Park Building, Room 1-10, 5600 Fishers Lane, Rockville, Maryland 20857. Address reprint requests to Thomas E. Novotny, MD. the Office on Smoking and Health conducted the AUTS on a nationally representative sample of the resident adult population in the United States (10). Questions regarding SLT use were included in this survey, as well as in earlier AUTS conducted in 1964, 1966, 1970, and 1975 (11—13). In this article, we compare data on its use among persons aged 21 or older from the AUTS. We also report a more detailed analysis of the correlates of SLT use, including attitudes and beliefs, from the 1986 AUTS for males aged 17 and older. METHODS Prevalence data for the use of chewing tobacco and snuff separately were collected by the National Clearinghouse for Smoking and Health (now the Office on Smoking and Health) through the AUTS for 1964, 1966, 1970, 1975, and 1986 (10—13). In each survey year, smokers were over- sampled, and the final data were weighted to adjust for this oversampling. For the 1964, 1966, 1970, and 1975 sur- veys, data were collected from the civilian, noninstitution- alized United States population aged 21 years and older. In 1986, data were collected from persons aged 17 years and older. The respondents were selected as a national prob- ability sample in each year. The 1970 and 1975 surveys were telephone based, but the 1964 and 1966 surveys were conducted by personal interview. The prevalence of current SLT use among persons aged 21 and older was stratified by product and sex and was reported for each survey year. In the 1986 AUTS, data were collected from a sam- ple [drawn according to the Waksberg-Mitofsky random digit dialing technique (14)] of 13,031 persons (6,377 men and 6,654 women) aged 17 years and older. These data were weighted to the United States population, based on the 1986 Current Population Survey of the Bureau of the Census (15). This procedure controlled for the race, sex, ed- ucation, and regional distribution of the population in the United States. The 1986 survey asked more detailed ques- tions regarding the use of SLT. Because so few women surveyed in 1986 had ever used such products (0.5%), we restricted the remainder of our analysis to data from male respondents. We report the prevalences of use of any, ei— ther, or both SLT products for men aged 17 years and older. We also calculated the prevalence of ever and current use of any SLT product stratified by demographic categories, including age, race, region, education, poverty level [as de- fined by the Census Bureau criteria (16)], employment, and household income. These variables were included in a mul- tivariate model using the CATMOD procedure in the Statis- tical Analysis System or SAS (1 7); the final model includes 25 26 NOVOTNY ET AL. only those variables found to be significant in predicting the outcome of current SLT use. Ninety-five percent confidence limits about the odds ratios for each parameter included in the final model are reported. Respondents were also asked if they had attempted to quit using it, whether they experienced difficulty in quitting, and when had they begun to use SLT products regularly (initiation). We reported the age group at initiation and median ages at initiation by the product used as well as by product and birth cohort (before 1950 and 1950—1969). In addition, both users and nonusers were asked about their beliefs regarding the harmfulness of SLT. To determine if SLT is used as an alternative to smoking, interviewers asked current and former cigarette smokers whether they had used it as an aid to quitting smoking; this percentage and the percentage of current and former smokers using formal cessation programs are reported. RESULTS The prevalence of snuff and chewing tobacco use among men and women aged 21 and older from 1964 to 1986 has declined slightly (table 1). Among adults aged 21 or older in 1986, the weighted prevalence of snuff use was 2.2% for men and 0.5% for women, and for chewing tobacco it was 3.1% for men and 0.1% for women. For 1986, overall prevalence of “ever use” and “current use” of SLT among males aged 17 and older is shown in table 2. More than 12% of the male respondents had ever used SLT products; chewing tobacco appears slightly more commonly used than snuff. Few (0.5%) men use both products simultaneously. For men, the prevalence of ever use and current use of any SLT product, stratified by selected sociodemographic variables, was highest among younger white men who 1) were living in the Southeast, 2) had lower levels of ed— ucation, 3) had lower income, 4) were below the poverty level, and 5) were unemployed (table 3). All sociodemographic parameters in table 3 were entered in a multivariate model for which current use was the outcome variable and were then eliminated if they did not contribute significantly to predicting current use of SLT. The final model included race, occupation, and region. In this model, white men were more than twice as likely as black men to use SLT; men employed in blue-collar or service/laborer jobs or who were unemployed were three times more likely than white-collar workers to use it; and men in the Southeast and West were more likely than men in other regions to use SLT (table 4). ' TABLE 1.——Prevalence (%) of SLT use: 1964—1986, AUTS“ Men Women Year Snuff Chewing tobacco Snuff Chewing tobacco 1964 2.0 5.1 2.0 0.5 1966 3.1 7.1 2.1 0.4 1970 2.9 ~ 5.6 1.4 0.6 1975 2.5 4.9 1.3 0.6 1986 2.2 3.1 0.5 0.1 ‘1 Adults surveyed were aged 21 yr and older. TABLE 2.—Preva1ence (%) of ever use and current use of SLT: males aged 17 and older, 1986 Product used Ever use Current use Any SLT 12.6 5.2 Snuff 5.8 2.4 Chewing tobacco 9.9 3.3 Both" 3.1 0.5 " Percentages were obtained by subtraction of any SLT from the sum of snuff and chewing tobacco. Two—thirds of the men who ever used SLT began use before they were 21, and more than one-third began before they were 16 (table 5). Because the ranges for age of initiation of SLT use are wide (8 to 59 yr for snuff and 6 to 75 yr for chewing tobacco) and the distribution is skewed toward the younger years, the median age of initiation was ‘ used as the measure of central tendency. The median age of SLT initiation for both products is 19. Categorized by birth cohort, men born from 1950 through 1969 had a lower median age of initiation than those born before 1950 for both snuff (18 vs. 25) and chewing tobacco (18 vs. 20) as shown in table 6. TABLE 3.—Prevalence (%) of ever use and current use of any SLT by sociodemographic categories, males aged 17 and older, 1986 Ever Current Sample Category use use Population size Age group, yr 17—19 12.3 8.2 5,365,000 299 20—29 1 1.4 5.9 20,345,000 1,208 30-39 7.3 4.1 18,593,000 1,495 40—49 9.7 5.0 12,778,343 1,128 >50 11.5 4.8 26,844,167 2,247 Race White 11.1 5.6 72,073,331 5,703 Black 6.6 3.0 8,588,942 465 Other 7.7 2.9 3,263,237 209 Geographic area Southeast 14.5 7.5 29,231,720 2,110 West 9.6 4.5 17,147,702 1,290 Midwest 9.5 4.3 19,915,891 1,570 Northeast 5 ,5 3 .0 17,630,198 1,407 Completed years of school 311 14.6 7.3 22,222,266 1,258 12 11.1 5.6 28,887,445 2,240 13—15 9.1 3.8 15,877,148 1,448 216 4.8 2.9 16,938,651 1,431 Poverty level Below 16.1 8.5 6,582,762 390 Above 9.9 4.9 65,039,406 5,115 Employment Unemployed 13.0 8.3 3,270,903 173 Service/ laborer 12.3 6.4 12,100,216 794 Blue collar 70 3.6 31,645,988 2,367 White collar 2.3 1.0 33,128,857 2,754 Annual household income <$10,000 16.1 8.6 6,124,778 416 $10,000—29,999 4.7 2.2 32,744,319 2,439 2$30,000 3.0 1.6 32,753,069 2,650 NCI MONOGRAPHS, NUMBER 8, 1989 ADULT USE OF TOBACCO SURVEYS 27 TABLE 4.—Significant sociodemographic correlates of any current SLT use, males aged 17 and older, 1986 TABLE 6.—Median age of initiation of male SLT users aged 17 and older, by birth cohort, 1986 Parameter Odds ratio 95% confidence limits Region Southeast 3.0 1.8, 4.8 West ' 1.9 1.1, 3.3 Midwest 1.4 0.8, 2.5 Northeast Referent Race White 2.4 1.3, 4.3 Black Referent Employment Unemployed 3.8 1.9, 7.6 Blue collar 3.0 2.1, 4.3 Service/laborer 2.9 1.8, 4.6 White collar Referent Among current users, 77.4% believe that SLT use is a health hazard; among nonusers, 83.4% believe its use is hazardous. Among current users, 39.1% had attempted to quit, and of these, 46.7% reported experiencing difficulty in doing so. Among current and former smokers, 6.4% and 7.0%, re- spectively, used SLT to help them quit smoking. In contrast, 2.4% of current and 1.7% of former smokers used organized programs to help them quit smoking. DISCUSSION The AUTS indicate overall that the use of SLT products among persons aged 21 and older declined slightly between 1964 and 1986. However, the production of these products in the United States increased 40% between 1970 and 1986, from 95.2 million to 132.8 million pounds (18). The discrepancy between prevalence and production data is only partly explained by the increase in population. Between 1970 and 1986, the population over the age of 14 years (the potential users) grew from 151 to 193 million, an increase of almost 28% (19, 20). The increase in production above what might be expected from the growth in population has three possible explanations: Either the number of persons under the age of 21 who use SLT has increased, or users have increased their use, or both. Age-stratified prevalence of snuff and chewing tobacco use from the 1985 Current Population Survey has been compared with similar data from the 1970 National Interview Survey (21). For males aged 17 to 19 years, the prevalence of snuff and chewing tobacco use increased from 0.3% to 2.9% and from 1.2% to 3.0%, respectively; prevalence among men 20 to 49 years old increased less dramatically and declined for men aged 50 and older. This behavior change occurred despite the TABLE 5.~Reported age and median age of initiation of male SLT users, aged 17 yr and older, 1986 Age group at initiation, yr“ Product <16 16-18 19-20 21+ Median Any SLT 37.1 7.8 21.4 33.8 19 Snuff 35.5 8.6 23.0 32.8 19 Chewing tobacco 36.6 6.7 20.3 36.3 19 ‘1 Values represent percentages of males reporting. SMOKELESS TOBACCO USE IN THE UNITED STATES Birth cohort Product Before 1950 1950—1969 Any SLT 21 17 Snuff 25 18 Chewing tobacco 20 18 fact that most men were well aware of the hazards of using SLT in 1986. Because of the relatively small proportion of users among respondents in the nationally based 1986 AUTS, prevalence estimates among subgroups may be unstable. Nevertheless, the multivariate model shows that SLT use among men, like cigarette smoking, is a greater problem for those who are unemployed, employed in blue-collar or service/ laborer jobs (22), or live in the South (23). In contrast to cigarette smoking (22), SLT use is more common among whites than among blacks. Most importantly, nearly two-thirds of the current users of snuff and chewing tobacco began using them before they were 21 years old. Our data also suggest that younger men (born in 1950 or later) began to use SLT earlier in life than did men in the older birth cohorts. However, because 50% of the men in our sample began using it after they reached age 19, this cohort effect may be biased, insofar as some nonusers in the most recent birth cohort will begin to use SLT after their teenage years and therefore would not be detected in this survey. Among high school seniors, the prevalence of regular cigarette smoking has consistently been higher for women than for men (21); some of this difference in smoking prevalence between genders has undoubtedly been offset by the dramatically higher prevalence in recent years of SLT use among young men (21). As cigarette smoking becomes less socially acceptable in the United States, adult smokers may try to quit smoking by using alternative nicotine delivery systems such as nicotine polacrilex gum or snuff and chewing tobacco. They may also use SLT products in schools, workplaces, or other en- vironments that have restrictions against smoking. Several brands of SLT have been advertised as acceptable substi- tutes when smoking is not permitted. Policies regulating exposure to environmental tobacco smoke at worksites and public places are proliferating (24), and a comparison of this situation to the regulatory climate at the turn of the century proves interesting. At that time, tobacco spitting was identified as an environmental health risk. As a re— sult, laws regulating public spitting were established, and the tobacco industry developed a “cleaner” product: the mass-produced cigarette (25). The spittoon disappeared as the ashtray became commonplace. Now that environmen- tal tobacco smoke has been identified as a health hazard, SLT use has reemerged among young adults, perhaps partly because of restrictions against smoking. The reemergence of use among children was also highly correlated with the increase in advertising for these products (26). Nicotine is an addictive drug, and all tobacco products contain substantial amounts of nicotine. Although the deliv- 28 NOVOTNY ET AL. ered biologic dose cannot be accurately predicted from the nicotine content of the product, the resulting blood levels of nicotine produced by SLT use are similar in magnitude to those produced by smoking cigarettes (21 ). Therefore, it is not surprising that almost 50% of the men in our sam— ple who had attempted to quit using SLT reported difficulty in doing so. We did not ascertain the percentage of former users who went on to smoke cigarettes. Additional longitu— dinal research is needed before the natural history of SLT use relative to both smoking and switching products can be described. Young users of snuff and chewing tobacco may switch to cigarette smoking to satisfy their nicotine addiction when they reach adulthood. Although among males use of SLT may be perceived as more socially acceptable by teenagers, it also may be perceived as less socially acceptable by adults. This perception may prompt young adults to change to cigarette smoking and thus increase their health risk. We have shown that both current and former smokers use SLT as an aid to smoking cessation more frequently than they use organized cessation programs. However, we were unable to ascertain the comparative success of these techniques with our survey. We have learned a great deal about cigarette smoking as a national public health problem. Although the prevalence of SLT use is far less than that of smoking in the United States, i.e., approximately 30% (10), we must apply the same public health strategies for prevention and cessation to this health risk behavior that we do to smoking. Sev- eral governmental efforts to decrease the use of SLT have recently been enacted, including health warnings on prod- uct packages and advertisements, banning SLT advertising in the broadcast media (27), and imposing a small fed- eral excise tax. Although the adverse health effects of SLT have not become widespread in the United States, they have been carefully documented (1 ). The public health commu- nity and policy makers now have a unique opportunity to prevent the potential impact of these health problems; to accomplish this end, their targeted efforts to prevent SLT use, especially among young men, are needed. REFERENCES (1) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco. A Report of the Advisory Committee to the Surgeon General. DHHS (NIH) Publ No. 86-2874. Bethesda, MD: Natl Cancer Inst, 1986 (2) CONNOLLY GN, WINN DM, HECHT SS, ET AL: The reemer- gence of smokeless tobacco. N Engl J Med 314:1020— 1027, 1986 (3) ROUSE B: Epidemiology of smokeless tobacco use: A na- tional study. NCI Monogr 8:29—33, 1989 (4) CENTERS FOR DISEASE CONTROL: Smokeless tobacco use in the United States—Behavioral Risk Factor Surveillance System, 1986. MMWR 362337—340, 1987 (5) BOYD GM, ARY DV, WIRT R, ET AL: Use of smokeless tobacco among children and adolescents in the United States. Prev Med 16:402—421, 1987 (6) HUNTER SM, CROFT JB, BURKE GL, ET AL: Longitudinal patterns of cigarette smoking and smokeless tobacco use in youth: The Bogalusa Heart Study. Am J Public Health 76:193—195, 1986 (7) CENTERS FOR DISEASE CONTROL: Smokeless tobacco use in rural Alaska. MMWR 37:140—143, 1987 (8) SCHINKE SP, GILCHRIST LD, SCHILLING RF, ET AL: Smokeless tobacco use among Native American adolescents. N Engl J Med 314:1051—1052,1986 (9) NATIONAL INSTITUTES OF HEALTH: Conference Statement: Health implications of smokeless tobacco use, January 13—15, 1986. CA 36:310—316, 1986 (10) CENTERS FOR DISEASE CONTROL: Cigarette smoking in the United States, 1986. MMWR 36:581—585, 1987 (I 1) NATIONAL CLEARINGHOUSE FOR SMOKING AND HEALTH: Use of Tobacco: Practices, Attitudes, Knowledge, and Beliefs, United States—Fall 1964 and Spring 1966. Washington, DC: DHEW, PHS, July 1969 (12) NATIONAL CLEARINGHOUSE FOR SMOKING AND HEALTH: Adult use of tobacco—1970. Washington, DC: DHEW, Center for Disease Control, June 1973 (13) NATIONAL CLEARINGHOUSE FOR SMOKING AND HEALTH: Adult use of tobacco—1975. Washington, DC: DHHS, Center for Disease Control, 1977 (14) WAKSBERG J: Sampling methods for random digit dialing. J Am Stat Assoc 73:40—46, 1978 (I5) BUREAU OF THE CENSUS, DEPARTMENT OF COMMERCE: United States population estimates, by age, sex, and race: 1980 to 1987. Current Population Rep Ser P—25, No. 1022. Washington, DC: Dept Commerce, 1988 (16) BUREAU OF THE CENSUS, DEPARTMENT OF COMMERCE: Char- acteristics of the population below the poverty level: 1983. Current Population Rep Ser P-60, No. 147. Washington, DC: Dept Commerce, 1985 (17) SAS INSTITUTE INC. SAS Version 5. Cary, NC: SAS Inst, 1985 (I8) ECONOMIC RESEARCH SERVICE, DEPARTMENT OF AGRICUL- TURE: Tobacco situation and outlook report. ERS Rep No. TS—199. Washington, DC: Economic Res Serv, June 1987 (19) BUREAU OF THE CENSUS, DEPARTMENT OF COMMERCE: Pre- liminary Estimates of the Population of the United States, by Age, Sex, and Race: 1970 to 1981. Current Popula- tion Rep Ser P-25, No. 917. Washington, DC: US Govt Print Off, 1982 (20) BUREAU OF THE CENSUS, DEPARTMENT OF COMMERCE: Pre- liminary Estimates of the Population of the United States, by Age, Sex, and Race: 1980 to 1987. Current Popula- tion Rep, Ser P-25, No. 1022, Washington, DC: US Govt Print Off, 1988 (2]) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Smoking: Nicotine Addiction. A Report of the Surgeon General. DHHS (PHS) Publ No. 88-8406. Rockville, MD: Office Smoking and Health, May 1988 (22) NOVOTNY TE, WARNER KE, KENDRICK JS, ET AL: Smoking by blacks and whites: Socioeconomic and demographic differences. Am J Public Health 78:1187—1189, 1988 (23) CENTERS FOR DISEASE CONTROL: Regional variation in smoking prevalence and cessation: Behavioral risk fac— tor surveillance, 1986. MMWR 36:751—754, 1987 (24) MARWICK C, MERZ B, GUNBY P: New rules extinguish “smoking lamp” in growing number of public places. JAMA 259:2809-2810, 1988 (25) CHRISTEN AG, SWANSON BZ, GLOVER ED, ET AL: Smoke- less tobacco: The folklore and social history of snuff- ing, sneezing, dipping, and chewing. J Am Dent Assoc 1051821—829, 1982 (26) ERNSTER V: Advertising and promotion of smokeless to- bacco products. NCI Monogr 8:87—94, 1989 (27) US. CONGRESS: Comprehensive Smokeless Tobacco Health Education Act of 1986, 99th Congress, Public Law 99- 252, February 27, 1986 Epidemiology of Smokeless Tobacco Use: A National Study1 Beatrice A. Rouse 2» 3 ABSTRACT—The prevalence and patterns of smokeless to- bacco use and its correlates were assessed in the National In- stitute on Drug Abuse National Household Survey of residents 12 years of age and older. Overall, 11% of the general popula- tion have “ever tried” chewing tobacco, snuff, or other smokeless tobacco. Of these, 5% were former users and 3% used smokeless tobacco almost daily in the past year. Rates of its use differed significantly by sex, age group, race, region, and metropolitan area size. Although females were far less likely to try it, those who did were as likely as males to be daily users. Smokeless tobacco users were also more likely to use alcohol, cigarettes, and marijuana. In general, those who used smokeless tobacco almost daily were more likely to report poor health and hospi- talization for illness or injury in the past year, even when other substance use was controlled. Smokeless tobacco users also were more likely to report symptoms of depression. Finally, some sub- stituted smokeless tobacco for cigarettes, but youths (12-17 yr old) were more likely than older tobacco users to use both forms of tobacco regularly.—NCI Monogr 8:29—33, 1989. Tobacco chewing first became popular in the early nine- teenth century when it became known as “The American Habit” while other parts of the world mainly used dry snuff (I ). Although the first American tobacco advertisement was published in 1789 by Lorillard, most tobacco users sniffed or chewed tobacco until the first quarter of the twentieth century. Then smoking tobacco increased due to the mass production, advertising, and reduced cost of cigarettes. As a result of this technology, as well as the increased health concerns regarding the spread of airborne diseases from to— bacco spitting, SLT use decreased. Smoking cigarettes has remained the most popular form of tobacco use. In the lat— ter part of the twentieth century, however, the public has become increasingly aware of health risks associated with cigarette smoking and its associated passive effects and with tobacco’s addictive properties (2). Now SLT is being considered by some as an alternative to cigarettes. Com- parable nicotine levels are achieved with either form of ABBREVIATION: SLT = smokeless tobacco. 1The views expressed in this paper are those of the author and are not to represent the official position of the National Institute on Drug Abuse or the Department of Health and Human Services. 2 Epidemiology Research Branch, Division of Epidemiology and Statisti- cal Analysis, National Institute on Drug Abuse, Department of Health and Human Services, 5600 Fishers Lane, Room 11A-55, Rockville, Maryland 20857. 31 am grateful to Gayle Saunders and Sidford Sand of the National Institute on Drug Abuse for their word processing assistance, Dr. Robert Santos of Temple University, Philadelphia, for his population estimation techniques, and to Dr. Leonard LoSciuto and his staff at Temple University for their data collection efforts. tobacco (3). Therefore, medical concerns are being raised regarding carcinogenesis, nicotine dependence or addiction, dental problems, and other health effects associated with the use of SLT (3, 4). In the 19708, regular SLT use in the United States was low and. remained fairly stable. Use was generally higher among males, nonsmokers, and residents in the southern states; moreover, chewing tobacco was used more than snuff (5— 7). The national surveys of adults aged 21 years and older conducted by the Office on Smoking and Health found that rates of lifetime SLT use increased since 1964, peaked in 1970, and remained relatively stable in 1975. In the peak year of 1970, among males, 25% had “ever tried” chewing tobacco and 7% had ever tried snuff; among females, 2% had ever tried chewing tobacco and 3% had ever tried snuff. Six percent of the males and 1% of the females chewed tobacco, and 5% of the males and 1% of the females used snuff currently. National rates of SLT use among youth, however, were unknown for the 19708. In the 1980s, small-scale studies in various regions have produced estimates of SLT use in adolescents that have ranged from 5% to 36% (8—18). Moreover, such stud- ies suggest that the lifetime and current prevalences of its use by youth have increased in the last decade. For example, longitudinal data from the Bogalusa (Louisiana) Heart Study indicated a doubling in tobacco chewing and a tripling in snuff dipping between 1976 and 1981 among 8- to 17-year-old white males. In 1981, the highest rate of chewing tobacco was 43% in 14- to 15-year-olds and the highest rate of snuff dipping was 32% in 12- to 13-year-olds (12). This report presents data from a recent national study of the general population. The prevalence, patterns, and cor- relates of SLT use were examined, so that how generalized and widespread this form of tobacco use has become could be determined. The self-perceived health status of the re- spondents was also examined. METHODS Data on SLT use were collected as part of the National Household Survey on Drug Abuse. This survey is conducted periodically by the National Institute on Drug Abuse to gather data on the prevalence, correlates, and trends of drug abuse in a representative sample of the household popula- tion 12 years of age and older in the coterminous United States. In 1985, a national stratified multistage probability sample of household residents was drawn that included an oversampling of blacks and Hispanics and of the younger age groups. Only one eligible resident was randomly drawn from each household. Sampling weights and the appropri- ate statistical adjustments were applied to obtain the preva- 29 30 ROUSE lence rates and population estimates of drug use. Group dif- ferences were tested by chi-square analyses appropriately adjusted for the design effect of the complex sampling. Sta- tistical levels of .05 or less were considered significant. Re- gional classifications were based on the Bureau of Census geographical divisions. Data were collected by trained interviewers who com- bined personal interview procedures with a self-admin- istered questionnaire. The interviewers recorded the in— formation about the respondents’ tobacco use, health, and sociodemographic characteristics. For the questions regard- ing illicit drug use, the respondents recorded their answers on self-administered questionnaires during the interview to maintain confidentiality and anonymity. Data were col- lected between June and December 1985. The overall re- sponse rate was 84% and the sample size was 8,038. The 1985 National Household Survey on Drug Abuse was the eighth in the series but the first to gather data on SLT. Respondents were asked: “On the average, in the past 12 months, how often, if ever, have you used chewing tobacco or snuff or other smokeless tobacco?” RESULTS Prevalence and Patterns of Population Use An estimated 11.1% of the household population aged 12 years and older used chewing tobacco, snuff, or other SLT at least once in their lifetime. When applied to current population census figures (19), this lifetime prevalence rate produced a point prevalence estimate of 21.2 million house— hold residents who have ever tried SLT. Of the estimated 12.3 million who used it at least once in the past year, 5.5 million did so almost daily. The lifetime prevalence rate for any SLT use for adults aged 21 years and older was 19% for males and 3% for females; for 12— to 20—year-olds, it was 28% for males and 3% for females. Overall, 2.9% in the household population 12 years old and older used it al— most daily, 3.6% used it less often in the past year, 4.7% were former users, and 88.8% never chewed or dipped. Significant differences were found in rates of SLT use by sex, race and/or ethnicity, and age group (table 1). Any lifetime use was relatively rare among females (3%) compared with males (20%). In the total sample, lifetime TABLE 1.-—Percentage of household residents 12 yr of age using SLT ever (E) and past year (P) by race and/ or ethnicity, sex, and age group Whites Hispanics Blacks Total Sex Age group, yr E P E P E P E P Males 12—17 30 25 8 6 8 4 25 20 18-25 38 25 8 4 9 5 32 21 26—34 20 1 1 4 3 9 7 18 10 35+ 17 9 9 2 18 5 14 8 Total 22 14 8 3 13 5 20 12 Females 12-17 3 2 l 1 3 l 3 2 18—25 3 1 3 1 2 2 3 1 26—34 3 2 l a 3 1 3 2 35+ 2 ‘1 2 1 1 1 6 3 1 Total 2 l 2 1 7 3 3 1 prevalence was higher among non-Hispanic whites (12%) than blacks (9%) or Hispanics (5%). Among males, the younger age groups and non-Hispanic whites had higher lifetime and past—year use rates of SLT. Among blacks, the highest rates for both males and females were among the older adults. Furthermore, older black females who had ever tried SLT were more likely to be current daily users. Among black males aged 55 years and older, 32% had used SLT in their lifetime and 9% were currently using it daily. Among black females in the same age range, 19% had used it in their lifetime and 12% were currently daily users. Among males who had ever tried SLT, 60% used it in the past year; among females who had tried it, 42% used it in the past year. The females who were users in the past year, however, were older, black, and more likely to use SLT almost daily. Although few females had ever tried it, those who did use it in their lifetime were as likely as the males to have used it almost daily in the past year (26%). Overall, among males who used SLT in the past year, 43% used it almost daily, 8% on 1 to 2 days per week, 29% on about 3 to 51 days in the past year, and 20% on 1 to 2 days. The frequency of use in the past year among male users is shown by age group and race or ethnicity in table 2. Most (93%) of the older blacks who used it in the past year chewed or dipped almost daily. Regional differences in lifetime and past—year prevalence rates are shown by sex and age in table 3. Prevalence rates were generally lower in the Northeast but comparable in the other regions. Furthermore, over 50% of those who had ever tried SLT were using it in the last year in all areas. Given the low rate of use among females, the large proportion of lifetime and past-year users among the 12- to 17-year-olds in the North Central region is especially noteworthy. Further examination of these users indicated that none lived in large metropolitan areas, and all the daily users were in rural areas or small cities. Indeed, size of metropolitan area was significantly re- lated to SLT use for males and females and all age groups. Lifetime, past—year, and daily use were highest in non- metropolitan areas. For example, among the total sample of males, 15% of those in large (population more than 250,000) and 17% of those in small metropolitan areas (50,000—250,000), and 27% of those in rural areas have used SLT at least once in their lifetime. Furthermore, the rate of current daily use among males in the nonmetropoli— tan areas was over four times that in the large metropolitan areas (9% vs. 2%). TABLE 2.—Percentage distribution of frequency of SLT use among past—year male users for whites (W), Hispanics (H), and blacks (B) by age group Age group, yr 18—25 26—34 35+ H B 12—17 FrequencyWHBWHBWHBW Most days/wk 23 a a 48 40 31 49 39 32 49 a 93 1-2 days/wk 7 8 8 6 15 5 12 a 19 7 ” a 3—51 days/yr 41 13 24 31 9 18 19 41 26 27 40 7 1—2 days/yr 29 79 67 15 35 46 20 19 24 18 60 ’1 “ Use was <0.5%. ‘1 See footnote, table 1. NCI MONOGRAPHS, NUMBER 8, 1989 EPIDEMIOLOGY OF SLT USE: A NATIONAL STUDY 31 TABLE 3,—Percentage of household population using SLT ever (E) and past year (P) by region, sex, and age group North- east South West North Central Sex Age group, yr E P E P E P E P Males 12-17 13 11 30 24 28 25 29 22 18—25 24 18 33 24 35 19 36 21 26—34 8 5 l8 13 20 12 21 8 35+ 8 3 22 12 15 6 18 9 Total 11 7 24 16 21 12 23 13 Females 12—17 1 a 2 1 3 l 8 4 18—25 1 l l a 5 3 4 1 26-34 3 l 4 3 3 2 1 1 35+ 3 1 2 1 2 a 3 1 Total 2 l 3 2 3 1 3 l ‘1 See footnote, table 1. The relationship between educational status and SLT use was examined in the age group of highest use, i.e., 12—25 years. Use for youth who were still in school is shown by sex and grade in table 4. Daily use was highest among older male students and ranged from less than 0.5% for 6th grade to 12% for the 12th-grade males. Among females, rates of daily use were consistently low across all grade levels. Young adults were classified by highest level of education attained: college students, high school graduates, or high school dropouts. For both sexes, lifetime prevalence and past-year use were lowest for college students. Among males, daily use was highest among school dropouts (13%) and lowest among college students (6%). Cigarette and Other Substance Use Cigarette smokers at all ages were more likely to have tried SLT, and current smokers were also more likely to use it. Among the 12- to 17—year-old males, 14% of those who never smoked cigarettes, 32% of the former smokers, and 41% of those who smoked in the last year have tried SLT in their lifetime. Among females of the same age group, 1% TABLE 4.—Percentage of SLT use among students by grade and sex Almost Less than Not in Sex Grade Ever daily“ daily past year Males 6 30 b 21 10 7 20 b 14 6 8 18 4 7 7 9 29 4 23 2 10 22 4 12 5 11 27 5 l9 3 12 36 12 17 7 Females 6 4 ” 4 7 l a 1 8 1 [I a 9 4 l 3 10 4 2 2 1 1 5 2 l 2 12 5 3 l ‘1 Use for female students in grades other than 11 was <0.5%. b See footnote, table 1. SMOKELESS TOBACCO USE IN THE UNITED STATES of those who never smoked, 5% of the former smokers, and 7% of those who smoked in the last year have tried SLT in their lifetime. Younger smokers were more likely than older smokers to use cigarettes and SLT regularly. Among the 12- to 17-year-old males, 4% of those who smoked no cigarettes in the past month, 8% of those who smoked about one-half pack, and 7% of those who smoked at least a pack daily either chewed tobacco or dipped snuff almost daily. In contrast, among males aged 35 years and older, 4% of the nonsmokers, 18% of the half—pack smokers, and less than 1% of the smokers of a pack daily used SLT daily. Those who smoked cigarettes and drank alcoholic bever- ages in the past year were the most likely to have ever tried SLT; however, those who only drank were almost as likely to use it. Among males aged 12—17, 30% of those who only drank also used SLT at some time in their life, compared with 11% of those who neither smoked nor drank and 41% of those who did both in the past year. The relationship was the same among older males, but the rates of substance use were lower. Tobacco chewing and snuff dipping were also associated with recent use of marijuana even in the older age groups. Among males, a comparison by age group of the percentage reporting simultaneous use of SLT and marijuana in the past year is shown by the following data: Marijuana status Age group, yr User Nonuser 12—17 37 15 18—25 23 17 26—34 11 8 35+ 16 8 Health in Past Year Respondents were asked, “Would you describe your health for the past 12 months as excellent, very good, good, fair, or poor?” Those who had ever used SLT were more likely than nonusers to describe their health status for the past 12 months as less than good. In the total sample, 18% of the daily users, 13% of the former users, and 11% of the nonusers reported poor health in the past year. Furthermore, SLT users in general and daily users in particular reported higher rates of hospitalization overnight because of injury or illness in the past 12 months. Among males, 9% of the nonusers and 16% of the daily users and, among females, 9% of the nonusers and 30% of the daily users reported such hospitalization. Rates of hospitalization among youth and young adult daily users compared with nonusers, however, were not significantly different. Rates of hospitalization are shown by SLT use, age group, and sex in table 5. In gen- eral, past-year daily users reported poorer health and more hospitalizations even when the use of other substances was controlled. For example, of the males who neither smoked cigarettes nor drank in the past year, 8% of those who also did not use SLT were hospitalized, compared with 13% of the daily SLT users. As a measure of emotional health, the respondents were asked to indicate how often in the past week they felt var- 32 ROUSE TABLE 5.—Percentage of household residents hospitalized in past year by SLT use, sex, and age group SLT use Almost Less than Not in Sex Age group, yr Never daily daily” past year Total Males 12—17 8 7 5 9 8 18—25 8 5 17 3 9 26—34 6 22 14 5 7 35—54 9 15 9 15 9 55+ 1 l 36 26 20 13 Females 12— l 7 5 b 9 b 5 18—25 8 16 b 8 26-34 1 1 30 14 1 1 35—54 7 70 10 7 55+ 14 34 8 14 ‘1 Use by female residents in age groups other than 12—17 was <0.5%. [7 Use was <0.5%. ious symptoms of depression. Among males and females, more former SLT users reported feeling depressed often in the past week than either daily users or nonusers. Among the males, those who chewed tobacco or dipped snuff were more likely than nonusers to report that most days every- thing that they did “was an effort” (24% vs. 12%) and that most days they could not “get going” (10% vs. 3%). DISCUSSION Rates of SLT use in the general population were highest among white male 12— to 25-year—olds who were residents of nonmetropolitan areas. This finding is consistent with that of regional and other small—scale studies, which sug- gests that such use is not a localized phenomenon. Although males are more likely to try SLT, those females who did try it were as likely as males to become daily users. Although the unsightliness of the practice is likely to prevent most persons from using SLT, several factors are currently oper- ating to increase its use: 1) increased public emphasis on a smokefree environment, 2) increased availability Of various products, and 3) advertisements featuring prominent sports figures that teach how the product is to be used. The results from this national study suggest that there are two distinct types of users with differing implications for the health care delivery system. First are the young users. Youths are more likely to experiment with SLT products, but almost 50% of those 18 and Older who use it are daily users. Some substitute it for cigarettes, but they tend to use both forms of tobacco. Often their range of substance use is related to rebellious attitudes toward adults and author- ity and to risk-taking. Public health and civic leaders who conduct SLT prevention- and treatment-related activities targeted to youth and young adults need to take into ac- count their cigarette smoking and illicit substance use and the motivations for such behavior. The second type is the older cohort who has chewed tobacco, dipped snuff, or used other smokeless products daily for many years. Although there are long-term health implications for the young SLT users, the older ones have more immediate and severe medical needs. These users include especially those blacks over 55 and a cohort of older white southern women who continue to use SLT and who are more likely to prefer it to cigarettes. Therefore, prevention and intervention programs must focus on the health effects of tobacco, regardless of the form used. Finally, SLT users in general are at risk for numerous health problems especially given their multiple drug use. In— deed, they are more likely to perceive their health as poor, to report symptoms of depression, and to be hospitalized for illness or injury. As a result, where medical and men- tal health care is accessible, even the young SLT user is likely to come to the attention of health care profession- als. Although such individuals may not have been referred because of their use of it, such contact provides the health professional the opportunity to identify for these individuals and to educate them about the fact that SLT is but a form Of tobacco and as such carries many of the same undesir— able dependence-producing behaviors and adverse health consequences as cigarettes. REFERENCES (I) CHRISTEN AG, SWANSON BZ, GLOVER ED, ET AL: Smokeless tobacco: The folklore and social history of snuffing, sneez- ing, dipping and chewing. J Am Dent Assoc 105:821—827, 1982 (2) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Smoking: Nicotine Addiction. A Report of the Surgeon General. DHHS (PHS) Publ No. 88-8406. Rockville, MD: Off Smoking and Health, May 1988 (3) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco: A Report of the Advisory Committee to the Surgeon General. DHHS (NIH) Publ No. 86-2874. Bethesda, MD: Natl Cancer Inst, 1986 (4) CONNALLY GN, WINN DM, HECHT SS, ET AL: The reemer— gence of smokeless tobacco. N Engl J Med 314:1020— 1027, 1986 (5) NATIONAL CLEARINGHOUSE FOR SMOKING AND HEALTH: Use of Tobacco: Practices, Attitudes, Knowledge, and Beliefs, United States Fall 1964 and Spring 1966. Washington, DC: DHEW, PHS, July 1969 (6) DEPARTMENT OF HEALTH, EDUCATION AND WELFARE: Adult Use Of Tobacco—1970. Washington, DC: DHEW, PHS, 1973 (7) DEPARTMENT OF HEALTH, EDUCATION AND WELFARE: Adult Use of Tobacco—1975. Washington, DC: DHEW, PHS, 1976 (8) EDMUNDSON EW, GLOVER ED, HOLBERT D, ET AL: Person- ality profiles associated with smokeless tobacco use pat— terns. Addict Behav 13:219—223, 1988 (9) GLOVER ED, O’BRIEN K, HOLBERT D: Prevalence of smoke- less tobacco use in Pitt County. Int J Addict 22:557—561, 1987 (10) GRITZ ER, KSIR C, MCCARTHY WJ: Smokeless tobacco use in United States: Past and future trends. Ann Behav Med 7:24—27, 1985 (I I) GUGGENHEIMER J, ZULLO TG, KRUPER DC, ET AL: Changing trends of tobacco use in a teenage population in western Pennsylvania. Am J Public Health 76:196—197, 1986 (12) HUNTER SM, CROFT JB, BURKE GL, ET AL: Longitudinal patterns of cigarette smoking and smokeless tobacco use in youth: The Bogalusa Heart Study. Am J Public Health 76:193—195,1986 (I3) MCCARTHY WJ, NEWCOMB MD, MADDADIAN E, ET AL: NCI MONOGRAPHS, NUMBER 8, 1989 EPIDEMIOLOGY OF SLT USE: A NATIONAL STUDY 33 Smokeless tobacco use among adolescents: Demographic differences, other substance use, and psychological corre- lates. J Drug Ed 16:383—402, 1986 (I4) MARTY P], MCDERMOTT RJ, WILLIAMS T: Patterns of smokeless tobacco use in a population of high school stu- dents. Am J Public Health 762190—192, 1986 (15) NEWMAN IM, DURYEA EJ: Adolescent cigarette smoking and tobacco chewing in Nebraska. Nebr Med J 66:243—244, 1981 (I6) POULSON TC, LINDENMUTH J E, GREER R0: A comparison of the use of smokeless tobacco in rural and urban teenagers. CA 34:248—261, 1984 SMOKELESS TOBACCO USE IN THE UNITED STATES (I 7) YOUNG M, WILLIAMSON D: Correlates of use and expected use of smokeless tobacco among kindergarten children. Psychol Rep 56:63—66, 1985 (18) SCHAEFER SD, HENDERSON AH, GLOVER ED, ET AL: Patterns of use and incidence of smokeless tobacco consumption in school age children. Arch Otolaryngol Head Neck Surg 111:639-642, 1985 (I9) BUREAU OF THE CENSUS: Current population survey data tape, estimate of civilian, noninstitutionalized population of the US. on July 1, 1985. Washington, DC: Dept Com- merce, 1986 Use of Smokeless Tobacco by Age, Race, and Gender in Ten Standard Metropolitan Statistical Areas of the Southeast United States1 Karl E. Bauman, Gary G. Koch, Lynn A. Fisher, and Elizabeth S. Bryanza 3 ABSTRACT—Most surveys of smokeless tobacco use have been limited to young people, and in the few studies of adults, re- searchers have not considered age, race, and gender simultane- ously, although broad age groups have been used. Data on smoke- less tobacco use by race and gender for 5-year age groups up to age 70 and older were compiled from 21,203 households in 10 Standard Metropolitan Statistical Areas of the southeastern United States.—NCI Monogr 8:35-37, 1989. The harmful effects of SLT (I) and public policy regard— ing its use (2, 3) have attracted substantial attention. These concerns can be viewed in the context of many studies of SLT use by young people. Boyd and associates (4) pre- sented original findings on the prevalence of use by race, gender, and grade in school for 14 surveys of children and adolescents conducted since 1983, and they cited the stud- ies of adolescents that had been reported earlier. In sharp contrast to this wealth of information for young people is the paucity of data that describes adults according to ba— sic demographic characteristics. The few earlier studies of SLT use by adults, which are described in the report to the Surgeon General (I ), did not present results by age, race, and gender simultaneously. Moreover, the studies showing use by age were of broad age groups. Our purpose is to de- scribe the prevalence of SLT use by race and gender for a wide range of ages in 10 SMSA in the southeastern United States. METHODS The data reported here are from a field experiment that was conducted for an assessment of the influence of a mass media campaign on adolescent cigarette smoking. The study required area probability samples of households in 10 SMSA in the southeastern United States. As part of the screening procedure to identify eligible adolescent subjects, we asked adult members of 21,267 households to provide information on the race, gender, age, and SLT use of household members. ABBREVIATIONS: SLT = smokeless tobacco; SMSA = Standard Metropolitan Statistical Area(s). ‘Supported in part by Public Health Service grant CA-38392 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services. 2School of Public Health, CB #7400, Rosenau Hall, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599. Address reprint requests to Karl E. Bauman, PhD. 3 We thank Ms. Jill Bissette for performing some of the computer analy- ses and Ms. Connie A. Padgett for providing valuable comments. The procedures for selecting the 10 SMSA were as fol- lows: The 81 SMSA in the Southeast were systematically studied with the use of available 1980 Census data. To en- hance internal validity for the field experiment, we identi— fied the SMSA that had similar social and economic charac— teristics. There were 16 such SMSA, and 10 were needed for the research. We chose the 10 by using a table of random numbers and stratification to eliminate proximate SMSA as necessary to prevent overlap in mass media markets. The 10 SMSA are Chattanooga, TN; Columbia, SC; Jack- son, MS; Lakeland, FL; Lexington, KY; Macon, GA; Mo- bile, AL; Montgomery, AL; Roanoke, VA; and Savannah, GA. Averages of selected characteristics for the 10 SMSA, with ranges across SMSA in parentheses, are 322,142 in— habitants (224,341 to 443,536), 29.2 years of age (27.2 to 32.7), 73.9% white (60.1 to 88.2), 15.7% with 4 or more years of education (11.4 to 21.3), and $18,078 median fam- ily income ($16,512 to $19,388). Within each of the 10 SMSA, 1980 Census enumera- tion districts and block groups were used to form area seg— ments, which contained one or more Clusters of household units with an estimated 2.8 persons 12 to 14 years of age per cluster. Approximately 80 clusters were randomly se- lected in each SMSA, and the 30,561 housing units within the chosen clusters were then visited by field personnel to obtain the information on household members that was nec- essary to identify adolescent subjects. Field personnel made at least two attempts to obtain the information from a mem- ber of the household unit who was 16 years of age or older. If these attempts were unproductive, the staff tried to con- tact another knowledgeable person, such as a neighbor, to learn whether a person approximately 12 to 14 years of age lived in the house. When information from a nonmember of the household suggested that the household might con— tain an eligible 12- to 14-year-old subject, our efforts to contact an adult household member continued, whereas if no eligible subject was reported to live in the household, no further attempts were made to screen the household. Of the 30,561 household units, 2,788 were eliminated from further study because they were not occupied perma- nent residences. A member of each household provided in- formation for 21,267 of the 27,773 remaining households. Information was unavailable from 6,506 households be- cause members of 43 refused to participate, 647 were not contacted, and, for 5,816, a knowledgeable person (who did not live in the household) indicated that no persons aged 12 to 14 lived in the household. Sixty-four of the 21,267 house- holds with information from a household member were lost to study for unknown reasons, yielding 21,203 for study. Interviewers contacted the available adult member in 35 36 BAUMAN ET AL. each of the 21,203 households to identify the race, gender, and age of each household member; a total of 55,709 mem- bers were identified. The interviewer also asked whether anyone living in the house now used chewing tobacco or snuff. If the answer was yes, the interviewer asked who was the user of chewing tobacco or snuff to identify each user in the household. The 349 household members who were not white or black were omitted from these analyses because there were too few for reliable estimates across age and gender categories, and 852 household members were elim- inated because information on gender, race, age, or SLT use was incomplete. Thus a total of 54,508 household members were used in these analyses. RESULTS Reported use of SLT by age, race, and gender is shown in table 1. Few in the two youngest age groups (0 to 4 and 5 to 9 yr) were identified as users of SLT, with no more than 0.3% use in any race or gender group for those ages. The remainder of our report deals with persons 10 years of age and older. The evaluation of trends in use across age was done for each race and gender group by correlation chi-square tests, which were specifically directed at this type of potential association (5). The results for white females, black males, and black females were statistically significant (P < .01) and thereby revealed increases in use with age. Among white males, those 15 to 24 years of age were found by regular chi—square tests to be significantly more likely than younger or older males to use SLT (P < .01). Also for white males, the age groups of 25 to 29 years and older were not significantly different in use. Comparisons of use by males and females within each race and age group were based on chi—square tests for 2 X 2 tables or on Fisher’s exact test when expected frequencies for users were small. For whites in each age group, males were more likely than females to use SLT (P < .01). There was no clear gender difference pattern for blacks. Among blacks 15 to 19, 25 to 29, and 65 to 69, males were significantly more likely than females to use SLT; the only other age group with a significant gender difference was for those 70 and older, with females more likely to use it than males (P < .01). Use by whites and blacks within gender and age groups was also compared. For most of these analyses, chi—square tests were applied; for females in the 15- to 39-year age groups, Fisher’s exact test was used because of the small expected frequencies of users. Among males for each 5-year age group less than 45 years, whites were more likely to use SLT than were blacks (P < .01). White and black males aged 50 to 64 and 70 and older did not differ in use. Of the males 65 to 69 years of age, blacks were more likely than whites to use snuff or chewing tobacco (P < .01). Use among white and black females less than 30 years of age was similar. For females 30 years of age and older, blacks were significantly more likely than whites to be users (P < .01). The race, gender, and age groups with the highest preva- lence of SLT use were 1) black females aged 70 and older, 2) black males 65 and older, and 3) white males 15 years of age and older. In the above analyses, separate attention was given to age comparisons for each race and sex, sex comparisons for each age and race, and race comparisons for each age and sex. This perspective for interpretation was justified by the significant (P < .01) results of tests for age—race—sex interactions in two analyses of the data for all persons 10 years old or more and of either race or sex. One of these analyses was based on the maximum likelihood fit of a logistic regression model (6), and the other was based on the weighted least squares fit of a strictly linear model for use (7). Both were undertaken with the CATMOD procedure in the Statistical Analysis System or SAS (8). The presence of the age—race—sex interactions implies that the TABLE 1.—Use of SLT in 10 SMSA of the southeastern United States by race, gender, and age White Black Male Female Male Female Age, yr Percent use No. Percent use No. Percent use No. Percent use No. 0—4 0.2 1,372 0.1 1,345 0.0 617 0.0 649 5—9 0.3 1,471 0.0 1,309 0.0 740 0.0 695 10—14 3.5 1,588 0.1 1,533 0.3 670 0.2 655 15— 19 11.4 1,664 0.2 1,574 1.4 713 0.0 764 20—24 10.6 1,526 0.5 1,568 1.6 558 0.7 708 25—29 8.8 1,601 0.5 1,736 2.7 520 i 1.0 724 30-34 7.8 1,721 0.1 1,752 2.6 506 1.9 636 35—39 9.4 1,565 0.4 1,640 4.0 372 2.8 471 40-44 9.7 1,384 0.6 1,438 3.7 298 3.9 438 45-49 10.0 1,029 0.3 1,055 6.2 225 5.0 322 50—54 8.4 1,064 0.8 1,155 9.2 251 5.6 357 55-59 7.8 961 1.2 969 7.0 215 7.8 296 60-64 7.4 914 1.0 1,061 8.4 227 9.3 324 65—69 8.7 733 2.3 912 15.9 189 8.9 248 270 9.4 1,087 3.9 1,626 11.1 316 18.6 451 Total 19,680 20,673 6,417 7,738 a, NCI MONOGRAPHS, NUMBER 8, 1989 SLT USE BY AGE, RACE, AND GENDER 37 pattern of variation for each factor varies across the levels of the other two and thereby merits separate consideration. DISCUSSION Generalization of these findings to other populations and comparisons to future studies of SLT use should be made only with the appropriate considerations. First, these data do not necessarily reflect use in SMSA other than those stud- ied. Second, as described above, information was not avail- able for all households originally identified for study. Third, similar to other large-scale surveys such as the Census, in- formation about all household members was obtained from one adult household member rather than directly from each person for whom we show data. Fourth, in comparing our findings with those from other studies, researchers should also carefully consider the questions that were used. We would have identified more users if we had asked whether SLT had been used within a specified time, such as within 30 days or ever, but we considered such questions to be too specific for respondents who provided information about other household members. Patterns of use across race, gen- der, and age should be similar for many different measures of its use, and our purpose was to identify those patterns. Finally, the design is cross-sectional, and therefore age and cohort effects cannot be separated. For example, the rela- tively high use of SLT by black females 70 years of age and older could reflect an increase in use when black women get older or it could indicate that black women who also had high use when they were younger are now in this age group. CONCLUSIONS In the households studied, SLT use varied substantially by race, gender, and age. Use generally increased with age for white females, black males, and black females. Among white males, use increased to the 15 to 19 age group and SMOKELESS TOBACCO USE IN THE UNITED STATES then remained similar through the oldest age group. Among whites in each age group, males were more likely than females to use SLT. The gender difference pattern was not as consistent for blacks, but when there was a difference, use was greater for males except among those 70 years old and older, an age group in which females were more likely to be users than males. Among males less than 45 years of age, whites were more likely than blacks to use SLT; the races were the same for males aged 50 to 64 and 70 and older, and black males 65 to 69 were more likely than white males of the same age to be users. Among females under 30 years of age, whites and blacks had similar use rates, whereas among older females, blacks were more likely to use it than were whites. REFERENCES (I ) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco: A Report of the Advisory Committee to the Surgeon General. DHHS (NIH) Publ No. 86—2874. Bethesda, MD: Natl Cancer Inst, 1986 (2) IGLEHART JK: The campaign against smoking gains momen- tum. N Engl J Med 314:1059—1064, 1986 (3) KOOP CE: The campaign against smokeless tobacco use. N Engl J Med 314:1042—1044, 1986 (4) BOYD G, ARY DV, WIRT R, ET AL: Use of smokeless tobacco among children and adolescents in the United States. Prev Med 16:402—421, 1987 (5) LANDIS JR, HEYMAN ER, KOCH GG: Average partial associa- tion in three-way contingency tables: A review and discus- sion of alternative tests. Int Stat Rev 25:237—254, 1978 (6) KLEINBAUM D, KUPPER L, MORGENSTERN H: Epidemiological Research: Principles and Quantitative Methods. Belmont, CA: Lifetime Learning, 1982 (7) GRIZZLE JE, STARMER CF, KOCH GG: Analysis of categorical data by linear models. Biometrics 25:489—504, 1969 (8) SAS INSTITUTE: SAS User’s Guide: Statistics. Cary, NC: SAS Inst, 1985 Native Youth and Smokeless Tobacco: Prevalence Rates, Gender Differences, and Descriptive Characteristics1 Steven P. Schinke,2 Robert F. Schilling II,2 Lewayne D. Gilchrist,3 Marianne Rolland Ashby,3 and Eiji Kitajima3 ABSTRACT—This is a report on smokeless tobacco use among Native American youth from Indian reservations in Washington State. Study findings indicate that snuff and chewing tobacco are used frequently, heavily, and at an early age by Native Americans. Nearly one-half of our subjects had used smokeless tobacco on 11 to 20 or more occasions; close to one-third of all the females had used smokeless tobacco on more than 20 occasions. Weekly users in this study were young. Of those Native subjects who used snuff or chewing tobacco weekly, 72% were under 12 years of age. Among youth who reported weekly smokeless tobacco use, about 74% of all females and 90% of all males had first used snuff or chewing tobacco before they were 10 years old. Study results have implications for efforts toward detection, treatment, and prevention of snuff and chewing tobacco use among Native American adolescents.—NCI Monogr 8:39-42, 1989. The use of SLT poses serious health risks for Native American youth, for there is evidence of early, frequent, and heavy snuff and chewing tobacco use among them and Alaskan Native adolescents (1 ). Early use of snuff and chewing tobacco may precede later, habitual use of tobacco products (2). On the addictive potential of snuff and chew— ing tobacco use, the Office of Medical Applications of Re- search of the National Institutes of Health reported: “Blood levels achieved by smokeless tobacco use are similar to those of cigarette smoking . . . . Addicted users of nico- tine become tolerant to the drug . . . ; such users increase their dosage until it levels off at one that fulfills their need. Such users need nicotine continually” (3). Emerging data indicate that young SLT users may later switch to cigarette smoking as a source of nicotine (4, 5). Traditionally, Native American youth have enjoyed low rates of smoking relative to others in minority and majority cultures. However, their early introduction to SLT increases the likelihood of progressing to smoking and longer term addiction. Pollin (6), recalling tobacco use mortality rates, said, “We can conservatively estimate that more than 60% of these yearly deaths do not result from ignorance or from a freely chosen risk . . . but instead represent persons who became addicted to nicotine as adolescents.” Nicotine ABBREVIATION: SLT = smokeless tobacco. 1Supported in part by Public Health Service grants CA-29640 and CA-44903 from the National Cancer Institute, National Institutes of Health, and grant DA-03277 from the National Institute on Drug Abuse, Department of Health and Human Services. 2 School of Social Work, Columbia University, 622 West 1 13th St., New York, NY. 10025. Address reprint requests to Steven P. Schinke, PhD. 3 School of Social Work, University of Washington, Seattle, Washington 98195. addiction may be partly responsible for the recalcitrance of SLT habits. Some tobacco users succeed, but most cessation attempts fail, initially and repeatedly (7). Due to these failures, tobacco use relapse rates are higher than for other substances (8). Added risks of SLT use by Indian youth are implied by patterns of progressive substance abuse when adolescents move into adulthood. One study found “clear temporal de- velopmental stages in the use of licit and illicit drugs from adolescence through young adulthood” (9). Indeed, ear- lier and more consistently than non-Indian youths, Native American youths progress to and sustain alcohol, drug, and other substance abuse (IO—I3). To learn more about SLT use among the young Indians, we studied snuff and chewing tobacco use patterns and correlates among early adolescent females and males from Washington Indian reservations. METHODS Subjects—We drew from a representative sample of American Indian reservations in Washington State. Re- cruited at reservation and tribal schools, recreation centers, and social and health agencies by Indian research assistants and indigenous staff, subjects were eligible for study partic- ipation if they were American Indians or Alaskan Natives and if they gave their informed consent. Of the sample of 119 informed and consenting subjects, who represented 98% of all eligible youths asked to par- ticipate in the study, 51 were female and 68 were male. Their ages ranged from 8 to 16 years old; the mean age was 11.3. Most (89%) lived within the boundaries of feder- ally recognized Indian reservations; the remaining subjects lived within a few miles of a reservation. All relied on their respective reservations for educational, health, social, recre- ational, and commercial services. Procedure—In small groups at each site, subjects re- ceived a description of the study from the research assis- tants and staff teachers, counselors, health care profession— als, and nonprofessional volunteers. Native staff described the questionnaires that subjects would complete, then out- lined provisions for response anonymity and data confiden- tiality. The youths were assured that their questionnaires and responses would be anonymously masked to preclude knowledge of their identity. Staff further explained how the information obtained would be kept confidential. Their questions answered, subjects were administered brief, anonymous questionnaires. They were told that their names were not required on the questionnaires and they were free to omit answers on any items. If they had read- ing difficulties, the staff read the questions aloud and sub- 39 40 SCHINKE ET AL. jects privately recorded their answers. Questionnaires in- cluded demographic items about them and their families. Other items asked about their ever and recent use of snuff, chewing, plug, and smoking tobacco and about their to— bacco use consumption patterns. The questionnaire had a Cronbach alpha of .79. For analytic purposes, snuff use was scored positively if subjects reported oral use of moist snuff, dry snuff, or pre- measured snuff packets. Chewing tobacco use was scored positively if they reported use of loose cut, plug, or leaf to- bacco products. Smoking tobacco use was scored positively if regular cigarette, cigar, clove cigarette, or tobacco pipe smoking was reported. RESULTS By descriptive statistics, study results showed that among Native Americans more than 4 in 5 females and more than 7 in 8 males had ever used snuff or chewing tobacco (table 1). More males than females had used SLT once and between two and five occasions. Smokeless use rates of 11 to 20 occasions were similar for females and males. Heavier SLT use (> 20 occasions) was more prevalent among males. Over one-third of the sample reported SLT use on more than 20 occasions. More than 83% of all weekly SLT users began their current habit before they were 10 years old. For subjects who reported weekly snuff or chewing tobacco use, the data were categorized by user characteristics. By these findings, female and male weekly SLT users were close in present age (table 2). Compared with females, males began using SLT products at an earlier age. Although they reported weekly SLT use, over 43% and 47% said that they had used no snuff or chewing tobacco, respectively, in the last 7 days. These discrepancies suggest that subjects who considered themselves weekly users actu- ally used snuff and chewing tobacco less often than weekly. Lending credence to this suggestion are findings of 44 sub- jects reporting SLT use on over 20 occasions as noted in table 1 relative to the 54 who reported weekly use, as noted in the present age category of table 2. Of those adolescents who had used snuff in the last 7 days, most (43.7%) had used it between one and five oc- casions. Gender differences revealed slightly more females than males reporting snuff use on two or more occasions in the past week. Over 50% of the Native American youth who reported chewing any tobacco in the last 7 days had used the product on one to five occasions. Males were more apt than females to report chewing tobacco more than twice TABLE 1.—SLT use reported by subjects“ . Females Males Total Snuff or chewmg tobacco use No. Percent No. Percent No. Percent Never 10 19.6 9 13.2 19 16.0 Once 7 13.7 17 25.0 24 20.2 2 to 5 occasions 4 7.8 8 11.8 12 10.1 6 to 10 occasions 7 13.7 0 0.0 7 5.9 11 to 20 occasions 7 13.7 6 8.8 13 10.9 > 20 occasions 16 31.4 28 41.2 44 37.0 ”Percentages may not add to 100 because of rounding. in the last 7 days. Amounts of SLT use reported by sub- jects were uniformly low for both genders. Finally, 60% of those who said they used SLT in the last 7 days reported having never smoked tobacco. Relative to females, males were inclined to have smoked once. Rates of weekly and daily tobacco smoking did not differ between the genders. DISCUSSION These data indicate that among Native American youths from Washington State reservations, snuff and chewing tobacco are used more often by both genders than among comparably aged youths in other minority cultures (14—16). Close to 50% of the Native American adolescent subjects had used SLT between 11 and 20 or more times. Females in this study reported relatively high, although less than males, “ever use” rates of snuff and chewing tobacco. Nearly one- third of all our female participants had used SLT on more than 20 occasions. To date, the heavy ever use frequencies reported for the present female adolescents are the highest rates of SLT use published for young women of any ethnic and/or racial group. Also noteworthy are study findings on the characteristics of our weekly SLT users. Overall, the sample of weekly to- bacco users was young; only 28% of them were more than 12 years old. Understandably, the age at which they re— ported initial SLT use was similarly young. Among those who reported weekly use, about 74% of the females and 90% of the males had first used snuff or chewing tobacco before the age of 10. Results on participants reporting oc- casions of snuff and chewing tobacco use in the last 7 days are puzzling. Despite reports that 54 described themselves as weekly users, only 27 and 23 reported any snuff and chewing tobacco use, respectively, in the last 7 days. Possi- bly, they were exclusively snuff or chewing tobacco users, although this possibility still leaves 4 unaccounted for when tobacco use rates in the last 7 days are subtracted from re- ports of weekly SLT use. More likely because of the young age represented in our sample, some regarded themselves as weekly users, yet they did not literally use snuff and chew- ing tobacco every week. Amounts of SLT use reported by the youth indicate fairly light snuff and chewing tobacco habits, relative to adult users. Given their age, however, even these light SLT use patterns are disquieting. Most in- teresting perhaps were the low rates of tobacco smoking reported by the participants. Only 20% and 17% of the In— dian females and males, respectively, reported weekly or daily smoking. In light of our findings, explanations for snuff and chew- ing tobacco use among Native American youth warrant consideration. Early adolescents, not unlike their major- ity culture counterparts, may View SLT use as a rite of passage into adulthood (17—19). Furthermore, smokeless product use patterns observed in this study might reflect the success of advertising by the SLT industry, new and eas- ily used snuff and rolled leaf products, the absence of laws governing the sale and labeling of SLT, and antismoking campaigns (20—24). Deceptively, SLT use may appeal to American Indian youths as a benign substitute for smoking. Whatever the explanation, SLT may introduce these young people to nicotine. The dangers of this introduction were al- NCI MONOGRAPHS, NUMBER 8, 1989 NATIVE YOUTH AND SLT USE 41 TABLE 2.—Characteristics of weekly SLT users” Females Males Total Characteristic No. Percent No. Percent No. Percent Present age, yr 9—10 5 22.7 10 31.3 15 27.8 11—12 12 54.5 12 37.5 24 44.4 13-14 4 18.2 6 18.8 10 18.5 >14 1 4.5 4 12.5 5 9.3 Age at first use, yr < 6 2 10.5 7 24.1 9 18.8 6—7 4 21.1 12 41.4 16 33.3 8—9 8 42.1 7 24.1 15 31.3 10—11 3 15.8 3 10.3 6 12.5 >11 2 10.5 0 0.0 2 4.2 Snuff use in last 7 days None 9 47.4 12 41.4 21 43.8 Once 2 10.5 6 20.7 8 16.7 2 to 5 occasions 5 26.3 7 24.1 12 25.0 >5 occasions 3 15.8 4 13.7 7 14.6 Chewing tobacco use in last 7 days None 8 50.0 13 46.6 21 47.7 Once 4 25.0 2 7.1 6 13.6 2 to 5 occasions 2 12.5 9 32.1 11 25.0 >5 occasions 2 12.5 4 14.3 6 13.6 Cans, pouches, plugs used weekly <1 7 46.7 9 31.0 16 36.4 1 7 46.7 16 55.2 23 52.3 2 to 4 1 6.7 3 10.3 4 9.1 >4 0 0.0 1 3.4 1 2.3 Current tobacco smoking Never smoked 12 60.0 18 60.0 30 60.0 4.0, P < .05) for: number of offers of alcohol, offers of cigarettes, of- fers of marijuana, number of alcohol uses, and number of cigarette offers accepted. When number of offers of alco- hol was entered first in the function, none of the other vari- ables resulted in significant additional change in Rao’s V, and they were not added to the equation. With only the of- fers of alcohol in the function, the significant Wilks’ A was .96, P < .003. The relationship of offers to use of these substances and subsequent use of SLT was consistent, although the explained variance was small, 4.3%. Predictors of Change in Daily Use of Smokeless Tobacco Central to any SLT cessation effort is a greater un- derstanding of the factors related to change in use pat- terns. The predictors of change in the pattern of daily use were examined with discriminant analysis so those factors could be identified that differentiated males who remained daily users at follow-up from those who did not. Of the 52 ARY TABLE 2.—Concurrent discriminant analyses of SLT use among adolescent males Tried chewing tobacco or snuff” b Daily users vs. triers Daily users vs. never used‘ Standardized Structure Predictor variables coefficients coefficients Structure coefficients Standardized coefficients Structure coefficients Standardized coefficients No. of friends using SLT .52 .66 Tried smoking .34 .67 Level of smoking experience .46 .72 Intention to smoke — .35 No. of cigarettes last wk — — No. of cigarettes yesterday — — No. of cigarette offers last wk —.18 — No. of cigarette offers accepted last wk — — Best friend smokes — — No. of friends smokers — .37 Father smokes — — Mother smokes — — No. of brothers smoke -— — No. of sisters smoke - — No. of marijuana uses last wk — — Use marijuana daily — — No. of marijuana offers last wk — .32 No. of alcohol uses last wk —— .43 Use alcohol daily — — No. of alcohol offers last wk .29 .48 Summary statistics Total Wilks’ A Significance of A Canonical correlation Explained variance Group 1 centroid Group 2 centroid Percent correct classification Overall 74 Group 1 85 Group 2 51 .78 P < .0001 .47 .22 .36 —.77 .86 .93 .76 .84 — — — .31 .26 ' .34 .44 .49 .39 P < .0001 .78 .61 1.86 —.82 .75 P < .0001 .50 .25 1.13 —.29 83 90 53 82 91 94 “In group 1, 560 students tried, 260 did not (group 2). bln group 1, l 15 Students were daily users, 455 were triers (group 2). ‘In group 1, l 15 students were daily users, 260 had never used (group 2). 115 males who reported daily use at initial assessment, 86 were reassessed at 6-month follow—up and 77 were re- assessed at 12—month follow-up. For maintenance of a bet- ter subject-to-variable ratio in the discriminant analysis, the number of variables was reduced to seven. Several compos- ite variables were formed by standardization of item values and summation of component items. Marijuana and alco- hol use composites combined number of uses in the last week with daily use. A family smoking composite com- bined items on mother, father, sister, and brother smoking. A peer smoking composite included items on best friend smokes and the number of friends who smoke. A cigarette use in the last week composite combined cigarettes smoked yesterday and those in the last week. The remaining ques- tionnaire items used in this analysis were the number of friends who use SLT, tried smoking, and had the intention to smoke. The discriminant analysis of the 6-month follow-up data yielded a significant Wilks’ A of .85 (P < .002) that ex— plained about 15% of the variance and classified 77% cor- rectly. The structure coefficients indicated that the “friends use of smokeless tobacco” was strongly related to contin- ued daily use (.66) and peer smoking was negatively re- lated (—.36). For the 12-month data, the Wilks’ A was .91, P < .01. Nine percent of the variance was explained, and 82% of the cases were correctly classified. Only the num— ber of friends using SLT and the marijuana use composite were related to the function. Peer influence seems to be important in maintaining a pattern of daily SLT use. Multiple Drug Use Among Male Adolescents In an earlier study, we (5) found evidence of significant concurrent use of drugs among male adolescents and par- ticularly among daily users of snuff and chewing tobacco. Among the male adolescents in this study, 10.5% reported smoking cigarettes in the last week, 14.4% indicated mari— juana use in the last week, and 37.0% responded that they had consumed alcohol in the last week. Forty-five percent of the respondents reported using at least one of these four substances in the last week. Forty-seven percent of those reporting some use of these substances indicated that they used more than one. Among those who reported daily use NCI MONOGRAPHS, NUMBER 8, 1989 SLT USE AMONG MALE ADOLESCENTS 53 of SLT, 80.0% reported use of at least one additional sub- stance. Specifically, use in the last week was reported by 20% for cigarettes, 35% for marijuana, and 73.9% for alco- hol. The relationships between daily SLT use and any use of cigarettes, marijuana, and alcohol in the last week were examined by formation of 2 X 2 contingency tables and cal- culation of X2 and¢> statistics. All three X2 were significant at the P < .001 level. The ¢coefficients were: .13, .24, and .31 for cigarette, marijuana, and alcohol use, respectively. In addition, subject use of cigarettes, marijuana, and alcohol were significantly interrelated. The number of cigarettes in the last week was significantly correlated (P < .001) with the number of marijuana (r=.48) and alcohol (r=.27) uses in the last week. Similarly, use of marijuana in the last week was significantly related to alcohol consumption (r=.49). Smokeless Tobacco Use As A Risk Factor For Other Drug Use Male adolescents who report that they have used or tried SLT appear to be at increased risk to begin use of cigarettes, alcohol, and marijuana. Table 3 shows that SLT users who did not smoke at baseline were significantly more likely to report smoking at 1-year follow-up (5.8%) than those who did not use either substance at baseline (0.5%). Similarly, subjects who reported no use of alcohol in the last week at baseline were more likely to report use of alcohol at 1-year follow-up if they had reported use of SLT at initial assessment (29.0% vs. 11.6%). For marijuana, a significantly greater proportion of the baseline SLT users who had not reported use of marijuana at initial assessment reported it in the last week at 1—year follow-up (12.4% vs. 2.0%). The strength of the relationship between new use of cigarettes, alcohol, and marijuana at follow—up and having used SLT at baseline was indicated by ¢coefficients of .13, .21, and .18, respectively. Having used SLT was also related to increased use of cigarettes, alcohol, and marijuana at follow-up for all males, regardless of their previous status for these substances. Baseline SLT users were significantly more likely to have increased their use of cigarettes, alcohol, and marijuana at l-year follow-up than were males who had not used snuff or chewing tobacco (table 3). For increased use of cigarettes, alcohol, and marijuana, the ¢coefficients were .12, .15, and .20, respectively. DISCUSSION Although it may be true that the rate of cigarette smok- ing among adolescent females is higher than that of males, the findings of this Study reveal that the combined use of cigarettes and SLT may be substantially higher for male adolescents. The extent to which these SLT users later take up cigarette smoking is a critical issue, and additional lon- gitudinal studies are needed to examine it. This “gateway” transition from SLT use to cigarette smoking may be occur- ring with surprising regularity as adolescent males become older and face the possible social consequences of chew- ing tobacco and snuff use (e.g., dating-age females who do not find this behavior attractive). Given the increased re- strictions on smoking in public places, the prevalence of the opposite transition (i.e., from cigarette smoking to SLT use) may be an issue of public concern as well. The analysis of the concurrent data replicates findings of the earlier study (5). Both analyses indicate that peer use of SLT and experience with cigarette smoking were the primary discriminators between male adolescents who had tried snuff and chewing tobacco and those who had not. Similarly, results of both studies indicated that peer use of SLT discriminated between male daily users and those who had never tried it. This replication of the importance of peer use supports the adoption of social Skill training to resist peer pressure as an effective intervention strategy. Although some differences did occur in secondary factors across grade level, the major discriminating factors appear to be consistent. The current study provides evidence that peer use of SLT is the primary factor distinguishing between male adolescents who have become daily users and those who have tried it but have not gone on to become daily users. It seems that peer influence is important not just at onset but in the development of a daily use pattern. The prospective relationships to the onset of SLT use were relatively small. However, one interesting pattern that is consistent with a social pressure model did emerge in the 12-month follow—up analysis. Offers of alcohol, cigarettes, and marijuana were all Significant predictors of the onset of SLT use. Additional research including the assessment of offers is needed if investigators are to examine more fully the role of social pressure in the onset process. TABLE 3.—Percentage of male users and nonusers of SLT and other substances Use (last wk) at 1-yr follow-up, Increased use (last wk) at l—yr follow—up, SLT use nonusers at baseline all students at baseline Cigarettes Alcohol Marijuana Cigarettes Alcohol Marijuana User, % 5.8 29.0 12.4 6.8 25.4 14.7 Nonuser, % 0.5 1 1.6 2.0 1.5 12.5 2.0 x2a 8.3” 16.2” 16.0d 7.07” 13.10 21.74 ¢ coefficient .13 .21 .18 .12 .15 .20 ”Degree of freedom = 1. 1’10 < .01. C P < .001. dP < .0001. SMOKELESS TOBACCO USE IN THE UNITED STATES ' 54 It appears that once male adolescents establish a daily pattern of SLT use, most of them continue to use it daily. In this study, 76.6% of the daily users still reported daily use at 12—month follow-up. My associates and I (5) found 71% of the youths were still using it daily at a 9-month follow-up. How this pattern relates to the finding that only 30% of the daily cigarette smokers in this sample reported con- tinued daily smoking at 12-month follow-up is not clear. One explanation is that the daily SLT users in this study may be more addicted than the daily smokers. Another is that a more consistent use pattern is possible for those who use snuff or chew tobacco because they can be used less obtrusively within school and home environments. In ad- dition, social pressure from peers and adults to stop SLT use may be lessened. Clearly, replication by more direct measurement methods (e.g., self-monitoring and direct ob- servation) are needed for careful and rigorous examination of the self—quitting process. Also examined in the current study were changes in habits among daily SLT users, and again, peer use was the best predictor of continued daily use at 6- and 12—month follow-up. This finding is consis- tent with a social influence model. Parent and sibling use of SLT was not related to either the beginning or continuation of its use by the participants in this study or the previous one (5). This is an interest- ing finding because investigators (9—11) who conducted a number of prospective studies of adolescent smoking have also reported no relationships between these family mod— eling factors and adolescent smoking. Additional research specifically directed at the role of family members in the onset and maintenance of adolescent SLT use would help to clarify these relationships. In both the current study and Ary et a1. (5), tobacco, alco- hol, and marijuana use appear common in adolescent males, with 42%—45% reporting use of at least one of these sub— stances in the last week. Among those reporting some use, 43%-45% stated they had used more than one substance in the last week. Multiple substance use seems particularly high among daily SLT users, with 80%—83% reporting one or more other substances. In both studies, an unusually high percentage of daily SLT users indicated drinking alcohol in the last week (73%—74%). A closer examination of ado— lescent drug use patterns, with particular attention to the situations, antecedent events, and the consequences of use for each substance would be helpful in the development of efficacious interventions. Data from this study support an earlier finding (5) that SLT use was identified as a risk factor for the use of other drugs. Those who use SLT were more likely to begin use of cigarettes, marijuana, and alcohol than were other subjects; having used SLT was related to increased use of cigarettes, marijuana, and alcohol among adolescent males. Although the consistency of these findings does not establish a causal relationship between SLT use and that of these other sub- stances, it does point out the need for a more rigorous effort by researchers to examine these relationships. Longitudinal multimethod research might include in-depth interviews, di- rect observation, a series of telephone interviews over time to monitor change, and self—monitoring of use including corroboration by significant others. The limitations of the current study should be noted. The “generalizability” of the findings may be restricted to this geographic region of the country. In addition, the conclu- sions are based on self-report questionnaire measures, the validity of which may be a concern, despite the inclusion of physiologic pipeline procedures. The “true” use rates for these substances may be a bit higher than reported here, due to the voluntary nature of student participation in the study. Although the decline rate was only 4%, those who did not participate may be more likely to use drugs (12). The implications of this type of bias on the generalizabil- ity of the correlational relationships explored here are not clear. Future research efforts in this area might emphasize multimethod longitudinal research with special attention to the gateway transition process from use of SLT to that of cigarettes, marijuana, and alcohol. The development of effective prevention measures might include the integra- tion of SLT components into existing school—based smoking prevention curricula (13), the involvement of dentists and physicians in office-based interventions, and the creation of cessation clinics modeled after those used with smoking cessation. REFERENCES (1) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco: A Report of the Advisory Committee to the Surgeon General. DHHS (NIH) Publ No. 86-2874. Bethesda, MD: Natl Cancer Inst, 1986 (2) DEPARTMENT OF HEALTH AND HUMAN SERVICES: Health Im- plications of Smokeless Tobacco Use: National Institutes of Health Consensus Development Conference Statement, vol 6, No. 1. Bethesda, MD: Natl Inst Health Off Med Application Res, 1986 (3) BOYD G, ARY DV, WIRT R, ET AL: A preliminary report on the prevalence of the use of smokeless tobacco: Regional survey results. Prev Med 16:402—421, 1987 (4) DEPARTMENT OF HEALTH AND HUMAN SERVICES: Youth Use of Smokeless Tobacco: More Than a Pinch of Trouble (Control No. P-O6—86-005 8). Washington, DC: DHHS Off Inspector General, Off Analysis Inspections, 1986 (5 ) ARY DV, LICHTENSTEIN E, SEVERSON HH: Smokeless to- bacco use among male adolescents: Patterns, corre- lates, predictors, and the use of other drugs. Prev Med 16:385—401, 1987 (6) BAUMAN KE, DENT CW: Influence of an objective measure on self-reports of behavior. J Appl Psychol 67:623—628, 1982 (7) EVANS RI, HANSEN WB, MITTELMARK MB: Increasing the validity of self-reports of smoking behavior in children. J Appl Psychol 62:521—523, 1977 (8) DEPARTMENT OF HEALTH AND HUMAN SERVICES: Highlights from Drugs and American High School Students, 1975— 1983. DHHS (ADM) Publ No. 84-1317. Rockville, MD: Natl Inst Drug Abuse, 1984 (9) ARY DV, BIGLAN A: Longitudinal changes in adolescent cigarette smoking behavior: Onset and cessation. J Behav Med 11:361-382, 1988 (10) MCCAUL KD, GLASGOW RE, O’NEILL HK, ET AL: Predicting adolescent smoking. J School Health 52:342—346, 1982 NCI MONOGRAPHS, NUMBER 8, 1989 SLT USE AMONG MALE ADOLESCENTS 55 (11) MITTELMARK MB, MURRAY DM, LUEPKER RV, ET AL: Pre- cedures with adolescents. Addict Behav 8:433—438, 1983 dieting experimentation with cigarettes: The Childhood (13) SEVERSON HH, LICHTENSTEIN E: Smoking prevention pro- Antecedents of Smoking Study (CASS). Am J Public grams for adolescents: Rationale and review. In Child Health 76:1—3, 1986 Health Behavior (Krasnegor N, Cataldo M, eds). New (12) SEVERSON HH, ARY DV: Sampling bias due to consent pro- York: Wiley, 1985 SMOKELESS TOBACCO USE IN THE UNITED STATES Activity Involvement, Risk-taking, Demographic Variables, and Other Drug Use: Prediction of Trying Smokeless Tobacco1 Steve Sussman,2’ 3 Liana Holt,2 Clyde W. Dent,2» 3 Brian R. Flay,4 John W. Graham}, 3 William B. Hansen,2v 3 and C. Anderson Johnsonz’ 3 ABSTRACT—Four activity participation variables (clubs, sports, church, and parties); two indices of “risk-taking” (prefer- ence for risk-taking, getting into trouble at school); three demo- graphic variables (sex, ethnic group, socioeconomic status); and two drug use variables (trial of cigarettes and alcohol) were exam- ined as correlates and prospective predictors of trial of smokeless tobacco in two cohorts of seventh graders in urban Los Ange- les. The data were analyzed separately for males and females. Cross-sectional logistic regression analyses indicated that corre- lates of trying smokeless tobacco among the seventh-grade co- horts or among these same cohorts in the eighth grade (con- sidering those persons who had not tried smokeless tobacco in seventh grade) generally included being white, trying cigarettes, risk-taking, and attending parties. Prospective logistic regression analyses with data from subjects who had not tried smokeless to- bacco in the seventh grade indicated that predictors of subsequent trial of it generally included only being white and having tried cigarettes. Sports participation predicted onset only in one cohort of female subjects but not in males. Some activities that have been proposed as being predictive of smokeless tobacco use (e.g., sports participation) are generally irrelevant for a large sample of young adolescents in urban Los Angeles. White male cigarette smokers, regardless of the activities they have engaged in, are most likely to try smokeless tobacco.—NCI Monogr 8:57-62, 1989. Involvement in relatively conventional activities (e.g., church, clubs, and sports) has been reported as less likely to be associated with drug use than involvement in rel- atively unstructured and less conventional activities, such as attending parties with peers (I, 2). Of course, activity participation may be differentially associated with various drugs. For example, Huba and associates (3) found a high association of party attendance and marijuana use, but only a slight association between party attendance and cigarette smoking and alcohol use, and no association between party attendance and heroin or hallucinogen use. Chassin et al. (4, 5) suggested that use of SLT, because it receives relatively little social disapproval compared with ABBREVIATION: SLT = smokeless tobacco. [Supported in part by grant DA-03046 from the National Institute on Drug Abuse; Public Health Service grants CA-34622 and CA-44907 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services. 2 Institute for Health Promotion and Disease Prevention Research, 35 N. Lake Ave., Pasadena, California 91101. Address reprint requests to Steve Sussman, Ph.D. 3 Division of Health Behavior, Department of Preventive Medicine, Uni- versity of Southern California School of Medicine, Los Angeles, California 90024. 4Prevention Research Center, School of Public Health, University of Illinois, Chicago. use of other drugs, is adopted by conventional adolescents (e.g., athletic, otherwise nondrug using) as one avenue of shared peer group experience. This suggestion is particu— larly interesting because other drugs, including cigarettes, tend to be used by risk-taking and rebellious youths (1). However, Dent and co-workers (6) found that being white, preferring to take risks, being male, and using cigarettes and alcohol were predictors for trying SLT 1 year later in a cohort of urban or suburban Los Angeles—based, young ado- lescents. They suggested that, aside from gender and ethnic variables, youths who use SLT are those who are willing or enticed to engage in a variety of other risk—taking activ— ities consistent with Jessor and Jessor’s (I ) perspective on deviant-prone behavior. Unfortunately, Dent et al. were unable to assess the re- lationship between conventional activities and trial of SLT. Anecdotal observations and a few rural studies suggest that its use is widespread among young male athletes (7, 8). Un— til recently, tobacco industry campaigns designed to change the visibility and social acceptance of SLT products among teenagers and young adults had been employing famous sports figures as actors, such as Bobby Murcer and Walt Garrison (9). Although television advertising of SLT has been curtailed (10), magazine advertisements are prevalent (e.g., in TV Guide), use on television at baseball games is no- ticeable (I I ), and manufacturers of beef jerky and bubble gum continue to offer their products packaged like snuff and chewing tobacco. Thus SLT may be used by youths who are actively engaged in school athletics. Of course, apparent relationships between participation in sports and use of snuff and chewing tobacco could have been due to a high rate of participation in sports by young males, rather than to a unique contribution of athletic participation to the prediction of trial of SLT. We (12) found that 80% of the seventh-grade males from a large sample of Los An- geles area schools report participation in sports activities, and we suggested that relationships between sports partici- pation and SLT use are statistically redundant with gender. Jones and Moberg (13) reported data that would seem to reconcile the results obtained by Dent et al. (6) with those of other researchers who found positive relationships between use of SLT and athletic team participation. In a large sample of seventh- through twelfth-grade students, they found that SLT users were more likely to report taking risks and were slightly more likely to be involved in team sports than nonusers. They examined use in males only, so that the relationships they obtained were not confounded by gender. Thus conventional activity participation by youths may not always be indicative of their being conservative in behavior, at least regarding participation in team sports. 57 58 SUSSMAN ET AL. Unfortunately, their data are cross-sectional; in such studies, it is unclear whether psychosocial variables, e.g., risk—taking or sports participation, precede or are preceded by trial of SLT. Only longitudinal data can disclose a priori predictors of its use (14). A knowledge of a priori pre— dictors of use is essential in the development of tobacco use prevention programming, which should be acquisition oriented (15). In addition, Jones and Moberg (I3) exam- ined only one activity, sports participation. An exploration of how use of SLT is related to several different activities would be informative. Our present study served as a replication and extension of the study by Dent et al. (6). We examined concurrent and l-year prospective relationships between trial of SLT and participation in club, party, sports, and church activities, as well as between trial use and risk-taking, demographic variables, and trial of cigarettes and alcohol in two samples of heterogeneous, young adolescents in urban Los Angeles. METHODS Subjects—Subjects were from two cohorts of seventh- grade students. Cohort A consisted of 422 students from eight junior high schools in the Los Angeles metropolitan area. Cohort B consisted of 771 students from eight Los An— geles area schools, who were measured 1 year after cohort A (five of eight schools overlapped). Those schools were a random sample of 11 of 76 schools in the Los Angeles Unified School District, the school district of the City of Los Angeles and the largest in Los Angeles County. School participation was determined by district— and school-level agreement to participate in a longitudinal school-based study of drug use behavior, and all seventh graders were assessed in cooperating schools. In cohort A, 45% of the students were male, 46% were white, and 54% were non- white (i.e., 26% Hispanic, 19% black, and 9% Asian). In cohort B, 51% of the students were male, 49% were white, and 51% were nonwhite (i.e., 34% Hispanic, 12% black, and 5% Asian). Total attrition from cohorts A and B was 35% each and was due primarily to students transferring to nonparticipating schools. Trial of SLT, cigarettes, and al- cohol by seventh- grade students did not differ between the full and the reduced longitudinal samples, which suggested that the sample was unbiased due to attrition. Measures—Students completed in—class, self-report ques- tionnaires while in the seventh grade (May 1982 for cohort A, and May 1983 for cohort B) and 1 year later. On each testing occasion, students reported whether they had ever used cigarettes, SLT, or alcohol. To increase the validity of self-reports regarding tobacco products, we (16) collected saliva thiocyanate test samples. We scored drug use item re— sponses as either dichotomous (SLT: never or one or more times) or trichotomous (cigarette smoking: never, once, or more than once; drinking alcohol: never, once, or more than once) to equalize cell size of response categories. Be- cause of the greater use of SLT by whites than nonwhites in these samples and the desire for equal cell size, this pre- dictor was dichotomized (whites vs. nonwhites). Other questionnaire items included participation in four activities (clubs, parties, sports, and church); risk-taking and delinquent behavior (two indices, each composed of four items); and three demographic variables (sex, ethnic group, and socioeconomic status). The club item requested the student to indicate the number of groups, such as Scouts, Young Men’s Christian Association, or Boys or Girls Clubs, that he or she had been part of in the past 12 months. The party item requested that the number of parties be indicated that the student had attended with friends in the past 4 weeks. The sports item requested that the student indicate the number of times he or she played a competitive sport, e.g., baseball, swimming, basketball, or track in the last 7 days other than in a gymnastics class. The church item requested the student to indicate how often he or she attends services at a church, synagogue, mosque, or temple. These items were on 4- to 7-point rating scales. The delinquent behavior items were on 4—point rating scales and included a student’s being sent out of class for bad behavior, cheating on a test, being suspended from school, and destroying things on purpose that were not per- sonal possessions. The risk-taking items included liking to take chances, thinking school was a waste of time, believ— ing it was worth getting into trouble to have fun, and doing things the student is told not to do. The test—retest reliability of all the above items ranged from 0.5 to 0.8, and the risk-taking and juvenile delin— quency indices showed adequate discriminant and con- vergent validity (I 7, I8). Socioeconomic status was an 11-point rating-scale item that was derived from an open—ended response to a pretest item requesting the stu- dents to indicate their father’s occupation and was scored on the basis of occupational status levels adapted from Hollingshead and Redlich (19). Analysis: Prediction of trying smokeless t0bacc0.—-The analysis strategy used involved the logistic multiple regres- sion model (20) to predict a binary (0, l) outcome vari- able that indicated initial trial of SLT. The model assumes that the probability of a student trying it is described by: 1/ (1 + exp(—alpha—XB) ), where exp indicates natural base exponentiation, alpha is an intercept (level) parameter, X is a vector of (given) predictor variables, and B is a vector of regression parameters to be estimated. We tested two basic forms of this model: a cross-sectional form in which variables were used to predict previous trial of SLT and a prospective form in which variables were used to predict onset of its use 1 year later. The cross- sectional form allows identification of correlates of SLT trial, whereas the prospective/ longitudinal form allows us to establish an order of precedence among activities or other possible predictors of SLT use (6, 14). Although causal inference should not be made, longitudinal analyses enable us to identify prospective factors that can identify youths at risk for beginning use of SLT. We tested models in several subsamples. Gender differ- ences exist in frequency of trying snuff and chewing tobacco (6); we postulated that a different model of onset may ex- ist as well. Therefore, all analyses were done separately within male and female subsamples. Cohort subsamples were formed‘ as a means of cross—validation. Similar results in both cohorts were taken as stronger indications of model effects compared with a single sample analysis. Because the cohorts were measured 1 year apart and shared only 50% of the same schools, replication of effects was of primary NCI MONOGRAPHS, NUMBER 8, 1989 PREDICTION OF TRYING SLT 59 interest (as opposed to some calculation of shrinkage). We also used grade level to define student subsamples when studying the cross-sectional form of the models. Prospective analyses include one sample of subjects at two grade lev— els. The cross—sectional analyses of seventh graders, which predicted their trial, included all students regardless of con- current use (within each of the previously defined gender and cohort subsamples). All other analyses were performed in a subsample of students who reported that they had not used SLT at the time of their seventh-grade survey. In all, we tested the cross-sectional form of the model in eight sub- samples (2 gender X 2 cohorts X 2 grades), and we tested the prospective form of the model in four subsamples (2 gender X 2 cohorts). Predictor variables were 10 of the measures described above: the four activities items, the two risk-taking indices, the two “other drug” use items, and two of the three demo- graphic items (socioeconomic status and ethnicity). Model parameter estimation was achieved by the LO- GIST procedure in the SAS or Statistical Analysis System (21). In that program, the model likelihood ratio X2 re- ported is twice the difference in the log-likelihood between the tested model and a model with only intercept terms included (i.e., an omnibus test that no variables are sig- nificantly predictive). The interpretation of this X2 test is identical to likelihood ratio X2 tests performed in two-way contingency tables; i.e., large X2 values indicate rejection of the notion that the outcome and predictor variables are only randomly related. The program also computes a statis— tic labeled “R,” the square of which equals the percentage of the intercept-term-only (i.e., null) model log-likelihood accounted for by the predictors, analogous to R squared in ordinary least squares regressions. The R associated with individual variables (effect R) are related to the percent of log-likelihood reduction when that variable is included (given all other variables are included in the model). An ef- fect R is analogous to a standardized regression coefficient in ordinary least squares regression. RESULTS Prevalence of Trying Smokeless Tobacco Table 1 shows the prevalence of the two cohorts trying SLT at two times (seventh and eighth grade) by race and sex. The results are similar to those found by Dent et al. (6). Males are approximately three times more likely than fe- males to have tried SLT (23% vs. 6% and 27% vs. 10% in seventh-grade cohorts A and B, respectively; 38% vs. 12% and 38% vs. 11% in eighth- grade cohorts A and B, respec- tively). Whites generally report the highest percentage of trying SLT for males and females. Hispanics and blacks report the next highest percentage, followed by Asians. Correlates of Trying Smokeless Tobacco: Cross-sectional Analyses All model X2 were significant (tables 2 and 3).5 Results for seventh-grade females were similar across cohorts. At- tendance at parties and, less consistently, getting into trou- ble at school and trying cigarettes were positively associated with trying SLT. For seventh— grade males, being white and trying cigarettes were associated with trying SLT across co— horts. Alcohol use by cohort A males was also significantly associated with their trying SLT. Results for eighth- grade females were similar across co- horts. Being white and of lower socioeconomic status were associated with trying SLT across cohorts, and risk—taking and cigarette smoking were significant correlates as well in cohort B females. Risk-taking, cigarette smoking, and alco- hol use for eighth- grade males were consistently associated with trying SLT; also, being white and attending parties were associated with its use in cohort A males, whereas participation in sports for cohort B males was marginally related (P<.08) to trying SLT. Predictors of Trying Smokeless Tobacco: Prospective Analyses All model X2 tests were significant (tables 2 and 3). Being white and having tried cigarettes in the seventh grade were the only consistent predictors of trying SLT by the eighth grade across gender and cohort (cigarette smoking was not a significant predictor, but alcohol use was a significant predictor among cohort A females). Sports participation was a significant predictor of beginning SLT use, and risk- 5All two-way interactions between ethnicity and the other nine predictor variables were examined. We used gender, cohort, and grade indicators to construct analysis subsamples, and therefore they were not included as statistical variables. No significant two-way interactions with ethnic group were found, and in the interest of parsimony, they are not included in tables 2 and 3. TABLE 1.—Prevalence of trying SLT by cohort, sex, ethnic group, and grade No. of students Grade 7 Grade 8 Sex and ethnic group Cohort A Cohort B Cohort A, % Cohort B, % Cohort A, % Cohort B, % Females Asian 20 17 l 6 15 6 Black 49 49 4 12 6 12 Hispanic 50 130 2 8 2 8 White 108 179 8 11 19 15 All 227 375 6 10 12 11 Males ' Asian 21 21 5 14 10 19 Black 30 42 23 26 27 38 Hispanic 53 134 13 21 36 27 White 76 199 34 33 51 47 A11 180 396 23 27 38 38 SMOKELESS TOBACCO USE IN THE UNITED STATES 8: H :8qu“ mo 08038 E0 0000. 8.00 8.0 :000. 0:00 80 :000. :00 80 800. 8.3 000 800. 0000 000 :000.v 8.3 000 300: 00.0 E. 000 00.0 8. 00.0 2.0 800. 00.2 00.0 00. :1 00.0 00. S: 00.0 00. 000 80 6082 0:0 :000. 00.2 8.0 :000. 8.: S0 :000. 00.: 0:0 000. :0 000 0000. 800 000 800. 00.: a: 002006 00.0 00. 8.0 000 00. 8.0 8.0 2. 00.0 :0 8. 03 :0 0000. 00.: 00.0 00. :00 02:» 255 00.0 E. :0 00.0 00. 00.: :00I 2. 00.0 80| 0:. 00.0 00.0 8. 0:0 00.0 00. 80 20% 252.8868 00.0 :0. S0 000 0:. 8.0 00.0 00. 8.0 00.0 00. :00 8.0 0:. $0 000 _ _. $0 20000 .8000 00.0 S. 000 8.0 8. :0 8.0 00. $10 000 v0. 0; 8.0 : 0 00.0 00.0 00. a0 8:3ng 00.0 00. 3.0 00.0 00. :2 00.0 00. 00.0 8.0 E. 00.0 00.0 E. 80 :00I 8. 8.0 0826 00.0 00. 80 000 00. 00.0 8.0 00. _ :0 00.0 00. .00.: 00.0 3. 000 00.0 00. $0 8000 00.0 8. 8.0 00.0 8. 000 00.0 00. $0 00.0 8. 0:0 000 00. 3.0 000 00. 5.0 305 00.0 00. 80 8.0I 0:. 00.0 00.0 00. E0 00.0 00. :0 00.0 00. 00.: 00.0 00. 000 820 x m 00 x .0 000 x .0 0x 2 m 000 x ,0 0x 0 .0 0x 252 033:3 080m 800m 080m 080m 800m 080m 0202 moEEom 0202 385$ 0222 020:5”. oéoommoa ouEmfiEEo 9 .£:o>om 302800-820 oufiw zEmE 302008.820 008w £55m m 20:00 3 (rm—m main: .8 :ozoEEn—l.m mfim 080m 080m 080m— :éfi 080m 38m 0202 020.com 0202 020.com 0202 02080”— u>uoommoa oufimfifims 9 .£:u>um 302800-820 ouEw :Ewi 302080-820 093w fico>om 60 < :28 3 (film ME?“ 00 :o:o%o.&|.~ Mimi. NCI MONOGRAPHS, NUMBER 8, 1989 PREDICTION OF TRYING SLT 61 taking was a marginally significant predictor for females in cohort B. Seventeen percent (19 of 110) of the girls who reported participating in four or more competitive sports activities over the last 7 days reported having tried SLT, compared with only 8.5% (22 of 255) of them who reported participating in three or less competitive sports activities over the same time. DISCUSSION Our results suggest that the male youth who enjoys tak— ing risks, smokes cigarettes, and drinks alcoholic beverages is the one who is most likely to have previously tried SLT by the eighth grade. Compared with females, males were three times more likely to try SLT. Most females may sim- ply prefer not to use a substance that makes them spit. Pos- sibly, prevention programs should include females in novel ways. The negative responses of females to the practices of chewing and dipping might be integrated into programs designed to build resistance to adoption of these behaviors (22). For example, female students in the classroom might give low popularity ratings for those types of males who use SLT and spit, indicate their unwillingness to be friends with users, and role-play their physical reactions to some- one spitting as ways to deter use among males. On the other hand, at least 8% of the females in each cohort reported having tried SLT by the seventh grade, which suggests that its use could become a problem drug for some females. As with the males, females who previ- ously tried SLT tend to attend parties, smoke cigarettes, be white, and take risks. If these predictors were to oper— ate as acquisition variables for males and females, other treatment implications might include teaching of alterna— tive prosocial risk—taking activities (e.g., rock-climbing), self-management and coping skills, and self-confrontation skills, as well as needed education on the health conse- quences of SLT use (23). In general, the prospective analyses indicated that being white and having tried cigarettes were predictive of onset of SLT use across gender. Participation in sports was a significant predictor of beginning use of it for females (but not for males) that gave some support to the suggestion that athletes are at higher risk for its use. This result must be interpreted with caution because it did not replicate across cohorts. In general, activity participation was not a significant predictor of SLT use. Also, we did not provide a strong replication of the findings by Dent et al. (6) that risk-taking predicted first use. In the present study, risk— taking and party attendance were significantly correlated (P < .05) with trial of SLT in the cross-sectional analyses only. Perhaps participation in activities and preference to take risks are not relevant to the initiation of SLT use, at least in student samples from primarily urban areas, although they are associated with previous trial of the substance. At a minimum, these variables are not directly predictive of trial of SLT. Although not measured herein, the perception of relative safety in the use of snuff and chewing tobacco (4, 5) could lead to the initiation of the substance by a wide variety of youths in several contexts. SMOKELESS TOBACCO USE IN THE UNITED STATES REFERENCES (I) JESSOR R, JESSOR SL: Problem Behavior and Psychosocial Development: A Longitudinal Study of Youth. New York: Academic Press, 1977 (2) SKINNER WF, MASSEY JL, KROHN MD, ET AL: Social influ- ences and constraints in the initiation and cessation of adolescent tobacco use. J Behav Med 8:353—376, 1985 (3) HUBA GJ, WINGARD JA, BENTLER PM: Beginning adolescent drug use and peer and adult interaction patterns. J Consult Clin Psychol 47:265—276, 1979 (4) CHASSIN LA, PRESSON CC, SHERMAN SJ, ET AL: The social image of smokeless tobacco use in the three different types of teenagers. Addict Behav 13:107—112, 1988 (5) CHASSIN L, PRESSON CC, SHERMAN SJ, ET AL: Psychosocial correlates of adolescent smokeless tobacco use. Addict Behav 10:431—435, 1985 (6) DENT CW, SUSSMAN S, JOHNSON CA, ET AL: Adolescent smokeless tobacco incidence: Relations with other drugs and psychosocial variables. Prev Med 16:422—431, 1987 (7) CHRISTEN AG, MCDAVID RK, DORAN TE: Snuff dipping and tobacco chewing in a group of Texas college athletes. Tex Dent J 97:6—10, 1979 (8) WILLIAMS TK: Smokeless Tobacco Behaviors Among Ru- ral High School Students in Arkansas. Ph.D. dissertation (unpublished), Univ Arkansas, 1985 (9) CHRISTEN AG, GLOVER ED: Smokeless tobacco: Seduction of youth. World Smoking Health 6:20-24, 1981 (10) US. CONGRESS: Comprehensive Smokeless Tobacco Health Education Act of 1986. Public Law 99-252, Sect 3f, February 27, 1986 (II) JONES RB: Use of smokeless tobacco in the 1986 World Series. N Engl J Med 316:952, 1987 (12) SUSSMAN S, HOLT L, DENT CW, ET AL: Activity involvement, risk taking, and demographic variables aS predictors of trying smokeless tobacco. Presented at the 20th Annual Convention of the Association for the Advancement of Behavior Therapy, Chicago, 1986 (I3) JONES RB, MOBERG DP: Correlates of smokeless tobacco use in a male adolescent population. Am J Public Health 78:61—63, 1988 (I4) COLLINS LM, SUSSMAN S, RAUCH JM, ET AL: Psychoso- cial predictors of young adolescent cigarette smoking: A sixteen-month, three-way longitudinal study. J Appl Soc Psychol 17:554-573, 1987 (I5) CHASSIN LA, PRESSON CC, SHERMAN SJ: Stepping backward in order to step forward: An acquisition-oriented approach to primary prevention. J Consult Clin Psychol 53:612— 622, 1985 (I6) EVANS RI, HANSEN WB, MITTELMARK MB: Increasing the validity of self-reports of smoking behavior in children. J Appl Psychol 62:521—523, 1977 (I7) GRAHAM JW, FLAY BR, JOHNSON CA, ET AL: Reliability of self-report measures of drug use in prevention research: Evaluation of the Project SMART questionnaire via the test—retest reliability matrix. J Drug Educ 14:175—193, 1984 (I8) GRAHAM JW, SOBEL JL, GROSSMAN LM, ET AL: Reliabil- ity of self-report measures in drug abuse prevention re- search: Evaluation of psychosocial measures in the Project SMART Questionnaire. Tech Rep Ser, Health Behavior Res Inst. Los Angeles: USC Press, January 19, 1985 (I9) HOLLINGSHEAD AB, REDLICH FC: Social Class and Mental Illness. New York: Wiley, 1958 62 SUSSMAN ET AL. (20) WALKER SH, DUNCAN DB: Estimation of the probability of tobacco use. In National Institutes of Health Consen- an event as a function of several independent variables. SUS Development Conference on Health Implications of Biometrika 54:167—179, 1967 Smokeless Tobacco Use. Bethesda, MD: Natl Inst Health (21) HARRELL FE: The LOGIST procedure. In SUGI Supplemen- Off Med Applications Res, 1986 tal Library User’s Guide. Cary, NC: SAS Inst,'1983, pp (23) SUSSMAN S: Two social influence perspectives of tobacco 181—202 use development and prevention. Health Educ Res Theory (22) GRITZ ER: Prevention and treatment aspects of smokeless Practice 411989 Initiation and Use of Smokeless Tobacco in Relation to Smoking1 Arthur V. Peterson}, 3 Patrick M. Marek,2 and Sue L. Mannz» 4 ABSTRACT—Questionnaire data obtained from 1,631 tenth grade students in 14 school districts in the State of Washington are used in this investigation of the relationship between the onset processes for smokeless tobacco use and smoking. Emphasized is the use of time-to-event data on the ages of occurrence of six events in these onset processes. Concepts and methods for the statistical analysis of time-to-event data are demonstrated. The occurrence of events in the smoking onset process are strongly related to increases in the subsequent onset rate for smokeless tobacco use. Compared with before initial smoking has occurred, the onset rates for weekly smokeless tobacco use after initial smoking has occurred are 2.03 (P<.001) and 6.72 (P<.001) times as large for males and females, respectively. Furthermore, both initial and weekly use of cigarettes contributes to the risk of subsequent weekly smokeless tobacco use. Conversely, the steps in the onset process of smokeless tobacco use are strongly related to increases in the subsequent smoking onset rate. Possible implications for intervention in prevention of smokeless tobacco use and for further research are discussed.—NCI Monogr 8:63— 69, 1989. The use of SLT among adolescents, especially snuff among adolescent males, has skyrocketed in recent years (I). This development has ominous health implications, because SLT contains known carcinogens and because a growing body of epidemiologic evidence indicates that its use carries the risk of various adverse health effects includ- ing oral cancer (2). Scientists are expending considerable effort to establish the circumstances and factors related to SLT use and its onset process among youth and to incor- porate these findings in the designs of effective prevention programs. Determining the relationship between smoking and the use of SLT is important for their investigation of 1) smok- ing as a possible risk factor for the onset of SLT use and 2) the extent to which such use is associated with subse- ABBREVIATION: SLT = smokeless tobacco. 'Supported in part by Public Health Service grants CA—38269 and CA-34847 from the National Cancer Institute, and GM-24472 from the National Institute of General Medical Sciences, National Institutes of Health, Department of Health and Human Services. 2 Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, 1124 Columbia St., Seattle, Washington 98104. Address reprint re- quests to Arthur V. Peterson, Ph.D., Cancer Prevention Research Program MP603. 3Department of Biostatistics, University of Washington, Seattle. 4We thank the entire data collection and management staff of the Hutchinson Smoking Prevention Project for their contributions to the study, Bobbie Nielsen for preparation of the manuscript, our colleagues who provided comments on the manuscript, and the personnel of the participating school districts for their collaboration and cooperation with the Hutchinson Smoking Prevention Project. quent smoking. That both cigarettes and SLT are tobacco products and contain absorbable nicotine indicates strongly that an associated nicotine dependence may result in indi- viduals taking up one when quitting the other to maintain a habituated nicotine level. It is clear that the concurrent use of cigarettes and SLT are associated (3—6). The purpose of our investigation is to use data on the ages at which young people begin to smoke and use SLT to examine the relationship between the smoking and SLT onset processes. Some basic concepts and methods for the statistical analysis of time—to-event data are demonstrated. The concept of a smoking onset process, i.e., a series of events that describe an increasing level and/or frequency of cigarette use, has been advanced by Flay et a1. (7), Lev- enthal and Cleary (8), Hirschman and co—workers (9), and others. Our description of such a process consists of 1) spec— ifying meaningful events of tobacco use and 2) determin- ing transition rates between the events with data on the times (ages) at which the events occur. The investigations reported here are restricted to such events; other important aspects of the smoking onset process, such as social influ— ences, the environment, and motivation, will be added in subsequent investigations. We used data on the ages of oc- currence of six tobacco use events: initial, tenth, and first weekly use of SLT, and initial, tenth, and first weekly use of cigarettes. METHODS Survey procedures—Tobacco use, including cigarette smoking and SLT, was assessed through a questionnaire ad— ministered in the classroom to entire grades of tenth-grade students in 14 rural and suburban school districts in the State of Washington in January 1986. Through an infor— mational letter to parents and by in-class procedures, par— ents and students were fully informed in advance and were given an opportunity to ask questions and to decline to par- ticipate. The tobacco survey was part of a baseline assessment of tobacco use among students in school districts participat- ing with the Fred Hutchinson Cancer Research Center in the Hutchinson Smoking Prevention Project, a long—term, randomized controlled trial in school-based smoking pre- vention. Of the total enrollment of 2,214 tenth graders, 1,918 (87%) took part in the survey. Twelve percent were absent from class; 0.2% (parents) and 0.8% (students) declined participation. Data for all questionnaire items pertinent in this investigation are available on 1,631 students. All results reported below are based on analyses of data from 1,631 students (840 males, 791 females). 63 64 PETERSON ET AL. Measures we used to enhance the accuracy of the re- sponses to the questionnaire items included 1) administra- tion of the questionnaire on an unannounced date; 2) pro- cedures to maintain confidentiality and assurances about these to the parents and students; 3) classroom procedures designed to maintain and demonstrate confidentiality, in- cluding the use of study identification numbers and the handling of questionnaires by project data collectors only; 4) explanation and collection of saliva samples from all participating students concurrent with administration of the questionnaire; and 5) explanation of the data collection ob- jectives and the important role of the students in achieving them. The questionnaire included items that assessed various aspects of current, past, and future intended use of cigarettes and SLT products. The wording for questions and multiple- choice responses for cigarettes was similar to that for SLT, so that differences in patterns for smoking and SLT use could be ascertained without confounding from differences in the nature or wording of the items. Analysis—For binary data items (e.g., whether a cer- tain level of past or current use of tobacco was achieved), simple proportions (prevalences) are reported. Data on time-to-smoking and time-to-SLT-use are analyzed by standard time-to-event statistical methods (survival analy- sis methods) that accommodate data on individuals for which the event (smoking, SLT use) has not occurred (cen- soring). Two time-to—event statistical methods are used: 1) Kaplan-Meier survival curves provide a descriptive display of time—to-event data (e.g., age at initial smoking) obtained on a set of individuals. When data on time-to—event are completely available for all participants, then this curve (at any age t) is simply the fraction of individuals whose observed times—to—event are greater than t (e.g., a fraction of individuals who have not smoked at age t). The Kaplan-Meier curve can also accommodate the sit- uation characteristic of time-to-event data, when, for some individuals, the age at initial smoking is not known, but only that no smoking has occurred by a certain age (e.g., the age at which data collec- tion occurs). Mathews and Farewell (10), Lawless (I I ), Miller (12), and Kalbfleisch and Prentice (13) provide further descriptions of the Kaplan-Meier survival curve, including formulas for its compu- tation and assumptions for its use. 2) We used the Cox regression method (1 0—14) to analyze the impact of the occurrence of one to- bacco use event (e.g., initial smoking) on the sub— sequent rate of onset of another event (e.g., weekly use of SLT). By such analyses, one can investigate directly the interrelationship between the smoking and SLT use onset processes. The Cox regression method models the onset rate Mt) for some speci- fied event (e.g., weekly use of SLT) as a function of the follow-up time t(e.g., age). The model specifies that the onset rate A(t,'z) for any individual with ex- planatory (regression) variables 21, 22, . . . zp is just the product of a “baseline” onset rate A00) and a function g(zB) of the covariates, often taken to be the exponential function g(zB) = exp(zB): mm) = x00) - eXP(2B), where A00?) > 0 is a completely unspecified base- line onset rate, 2 = (21, . . . ,2!) is a regression vector consisting of the p explanatory variables, and B’ = (Bl, . . . ,3!) is a vector of regression coefficients to be estimated from the data. The Cox regression model offers a number of desirable features and improvements over more traditional methods that make it particularly helpful in investigations of onset processes, such as those of smoking and SLT use: 1) The age-specific onset rate Mt), a meaningful mea- sure of smoking onset as a function of age, is modeled directly. 2) No assumption is made about the shape or magni- tude of the onset rate as a function of age. It is data determined. 3) The quantities exp(Bl), . . . . , for which estimates are readily obtained by the usual partial likeli— hood analysis, have the useful interpretation of rel- ative onset rates, e.g., the estimated smoking onset rate for prior SLT users relative to that for prior nonusers. 4) As in other regression models, the effect of other variables can be conveniently controlled by their inclusion as covariates in the regression model. 5) Unlike binary data methods, the model and analysis can accommodate censored data. 6) The model can be generalized in numerous ways for adaptation to a wide range of applications. Used in this paper is a generalization that allows an explanatory variable to depend on follow-up time t. In our application below, we let 21 = 21(t) depend on the follow-up time (age) and define it to be the indicator function for the occurrence of a specified prior event (e.g., the occurrence of initial smoking), taking the value 0 before the event occurs and 1 afterward. The quantity exp(Bl) is interpreted as the relative onset rate (e.g., of weekly SLT use after the prior event of initial smoking compared with before). See the references above for a complete description of this model, its generalizations and assumptions, and method of analysis. RESULTS Prevalence of Smoking and Smokeless Tobacco Use Table 1 presents, first for cigarettes and then for SLT, the fraction of boys and girls who have ever used, currently use, and have attained certain events of the onset process. The percentage of males who have ever smoked cig- arettes is about the same as have ever used SLT. More boys have dipped more than five pieces of SLT than have smoked more than five cigarettes (40.6% vs. 32.5%). Almost 70% of the females have ever smoked cigarettes and about 31% have ever used SLT; almost 6% of the girls have used more than five pieces of SLT. More boys are currently using SLT than cigarettes: 17.7% versus 14.4% (weekly use). Although 19.4% of the girls NCl MONOGRAPHS, NUMBER 8, 1989 INITIATION AND USE OF SLT AND SMOKING 65 TABLE l.—Current use, lifetime use, and attainment of onset events: cigarettes and SLT Boys, % Girls, % Use/onset/ transition Cigarettes SLT Cigarettes SLT Lifetime use Ever 69.8 71.1 68.6 31.2 >Five times 32.5 40.6 39.1 5.9 Current use ZOnce/mo 19.7 25.3 26.1 2.8 ZOnce/wk 14.4 17.7 19.4 1.4 Onset of use First 69.8 71.1 68.6 31.2 Tenth 30.8 40.5 37.7 5.8 First weekly 22.4 31.1 31.2 4.4 Transition probabilities Between no use and 69.8 71.1 68.6 31.2 first Between first and 44.1 57.0 55.0 18.6 tenth use Between tenth and 70.3 73.6 79.9 69.8 weekly use smoke cigarettes at least weekly, only 1.4% of them use SLT at least weekly. Consistent with the results just presented, more boys attained the tenth use for SLT than for cigarettes (40.5% vs. 30.8%), and more boys attained weekly use of SLT than cigarettes (31.1% vs. 22.4%). Also, far fewer girls attained both tenth and weekly use of SLT than for cigarettes. It is noteworthy that 4.4% of the girls did use SLT weekly at one time in their lives. Because tobacco use onset is a process of increasing use, presentation of the results in transition probabilities (lower portion of table 1) from one event to another is helpful. These results reinforce the ideas that 1) the higher prevalence for male use of SLT compared with smoking is attributed mostly to the higher first-to-tenth use transition probability (57.0% vs. 44.1%), and 2) the drastically lower prevalence for female use of SLT compared with smoking is attributed to both a lower prevalence of initial use and a lower first-to-tenth use transition probability. The extent to which cigarettes and SLT are used sepa- rately and concurrently is shown in table 2. Consistent with results from other studies, a strong relationship is evident between smoking and SLT use, for both lifetime and cur- rent use. First, a majority of males (60.2%) have used both cigarettes and SLT. Only about 1 of 3 males who have never tried SLT have tried cigarettes; 6 of 7 males who have tried SLT have also tried cigarettes. Fewer than 1 of 10 males who do not use SLT weekly smoke weekly, but more than 1 of 3 males who use SLT weekly also smoke weekly. Of the 25.5% of the boys who use tobacco at least once a week, 7.6% use cigarettes only, 11.7% use SLT only, and 6.2% use both. Among females, 29.5% have used both cigarettes and SLT. Over 50% of the females who have never tried SLT use cigarettes, but more than 9 of 10 who have tried SLT have also tried cigarettes. On average, fewer than 1 in 5 girls who do not use SLT weekly smoke weekly, but more than 1 of 2 who are weekly SLT users smoke weekly. Onset of Smoking and Smokeless Tobacco Use The results presented to this point have described preva- lence of current use, lifetime use, and the frequency of oc- currence of certain onset events for smoking and SLT sepa- rately and together. These results have described the extent to which various smoking events occurred but not at what ages they occurred. Attention is now focused on the ages at which the onset events occurred, with emphasis on a de— scription of the age-specific onset rates for first, tenth, and first weekly use for smoking and SLT. First, smoking and TABLE 2.——Relationship between smoking and SLT use, % Cigarettes Cigarettes Sex Use/onset only SLT only and SLT Neither Males Lifetime use Ever 9.5 10.8 60.2 19.4 >Five times 10.2 18.4 22.3 49.3 Current use ZOnce/mo 7.8 14.6 11.1 66.5 ZOnce/wk 7.6 11.7 6.2 74.5 Onset of use First 9.6 10.7 60.5 19.2 Tenth 9.4 19.0 21.4 50.1 First weekly 9.0 17.7 13.3 59.9 Females Lifetime use Ever 39.1 1.6 29.6 29.7 >Five times 33.7 0.6 5.3 60.4 Current use ZOnce/mo 24.0 0.8 2.0 73.3 ZOnce/wk 18.5 0.8 0.7 80.1 Onset events of use First 38.9 1.6 29.8 29.6 Tenth 32.4 0.5 5.3 61.8 First weekly 27.7 0.9 3.5 67.9 SMOKELESS TOBACCO USE IN THE UNITED STATES 66 PETERSON ET AL. 1 _ 0.9 _ '— CIgarette Use 0.8 - """"" 5LT USO first 5LT Use 0.7 _‘ ............. - F ........ First Cigarette r 0.6 - O c 0.5 - ......... t Tenth SLT Use FIGURE l.—Onset curves for smoking l 0.4 fl .. _ _ _ . _ .. """""" and SLT for 840 adolescent males. 1: .......... _. ......... Week/y SLT Use 0.3 - E 0.2 - __l—J— Weekly Cigarette 0.1 " Jse 0 I I l 4 15 16 17 SLT are considered separately and then together. We used basic methods of presenting and analyzing time-to-event data. The onset curves for first, tenth, and weekly use for cigarettes and SLT are shown in figures 1 and 2 for males and females, respectively. The onset curves indicate that the rates are greatest (onset curves increase the fastest) during certain age ranges shown in table 3. It is clear that substantial differences exist between the age ranges when smoking and SLT onset occur. For males and females, initial SLT use occurred later than initial smoking. For males, tenth and weekly SLT use did not occur later than tenth and weekly use of cigarettes. Rather, the onset rates were similar until the boys were 11 years old, after which more boys achieved tenth and weekly SLT use than tenth and weekly smoking. For females, tenth and weekly SLT use occurred later than tenth and weekly use of cigarettes. Relationship Between Smoking Onset and Onset of Smokeless Tobacco Use The relationship between the onset of smoking and that of SLT use in adolescents was investigated by a number of methods. FIGURE 2.—Onset curves for smok- ing and SLT for 791 adolescent fe- males. 17 1 .. 0.9 fl I Cigarette Use 0'8 " --------- SLT Use ......... first Cigarette 0.7 - __'_’_ F r 0.6 - a c _ t 0.5 | Tenth Cigarette o 0.4 d _I— n o 3 Em ' Weekly Cigarette ........ Use 0.2 -‘ 0.1 - ......... Tenth SLT Use 0 l i """" I I I I I i ...... i ............. II weeid] 5” LI!“ 4 5 6 7 8 9 1O 1 1 12 13 1 4 15 1 6 Age NCI MONOGRAPHS, NUMBER 8, 1989 INITIATION AND USE OF SLT AND SMOKING 67 TABLE 3.—Age ranges for high onset rates Use First Tenth First regular Age range, Percent Age range, Percent Age range, Percent Sex yr onset” onset” yr onset“ Males Smoking 6—14 55 9—15 26 11—15 14 SLT 9—14 55 11—15 30 11—15 25 Females Smoking 8—14 51 9—15 33 10-15 27 SLT 12—16 26 11—15 5 12-16 5 ” Percent onset is for age range indicated. First, joint distributions were computed for the age of occurrence of an event in the smoking onset process and that of the corresponding event in the SLT onset process. Time-to-event censoring was handled by inclusion of a “never started” category. From the joint distributions (not shown) of: 1) age at first use of cigarettes and at first use of SLT, 2) age at tenth use of cigarettes and age at tenth use of SLT, and 3) age at regular use of cigarettes and at regular use of SLT, a number of conclusions can be obtained. One summary point of the joint distribution of ages of initiation of smoking and SLT use is the extent to which the smoking event occurs before the SLT use event, and vice versa, among those experiencing both smoking and SLT use events. As shown in table 4, smoking and SLT each occurred first in a substantial number of males. Among boys who used both, 25% tried SLT first and 60% tried cigarettes first. Fifteen percent first tried both at the same age. In contrast, the vast majority of females tried cigarettes first (76%) rather than SLT first (12.5%). Finally, the following question is addressed: How are the age—specific onset rates for regular smoking related to the prior occurrence of steps in the SLT onset process for the 840 boys and 791 girls? Conversely, how are the onset rates for regular SLT use related to the prior occurrence of steps in the smoking onset process for both sexes? We investigated these questions using time—to—event regression methods developed by Cox (14). Data for all participants are included in these analyses; those not experiencing the end point of interest (e.g., weekly smoking, in table 5) are treated as censored. The occurrences of steps in the SLT onset process are strongly related to increases in subsequent onset rate for TABLE 4,—Precedence of smoking vs. SLT among individuals using both, % Smoking event SLT use Simultaneous Use occurred first occurred first occurrence Males First 60.0 24.9 15.1 Tenth 44.2 28.2 27.6 Weekly 43.4 31.0 25.6 Females First 76.3 12.5 11.2 Tenth 73.8 11.9 14.3 Weekly 60.7 14.3 25.0 SMOKELESS TOBACCO USE IN THE UNITED STATES weekly smoking. From runs (i.e., individual analyses) 1 and 4 of table 5, the weekly smoking onset rate is 1.65 (P = .002) and 2.13 (P < .001) times as large for males and females, respectively, after initial SLT use has occurred compared with before initial SLT use. From runs 2 and 5, the weekly smoking onset rate is 1.83 (P = .002) and 3.25 (P = .021) times as large for males and females, respectively, after weekly SLT use began compared with before weekly SLT use. Furthermore, evidence suggests that each of the SLT steps, initial and weekly use, contributes to the risk of sub- sequent weekly smoking. When both steps are included in the analyses for males (run 3), the occurrence of initial SLT multiplies the risk of smoking onset by 1.45 (P = .047), and the subsequent occurrence of weekly SLT use multiplies the risk of smoking onset by an additional 1.47 (P = .08) for a net multiple (1.47 X 1.45) of 2.13 times the smoking onset rate when no SLT event has occurred. For females (run 6), the relative risks are 2.04 (P < .001) and 1.80 (P = .27); not enough females use SLT regularly to provide the data needed in this data set for us to determine whether weekly SLT use provides an added risk of weekly smoking beyond the risk provided by initial SLT use. The results for the converse relationship are similar (table 6). The occurrences of steps in the smoking onset process are strongly related to increases in the subsequent onset rate for SLT use. From runs 1 and 4 of table 6, the onset rate TABLE 5.—Results of relative risk regression analyses of relationship of steps in onset process of SLT with onset of weekly smoking“ Relative rate and 95% confidence interval of weekly smoking onsetb Run After initial SLT use P After weekly SLT use P Males 1 1.65 (1.20, 2.29) .002 — 2 — 1.83 (1.24, 2.69) .002 3 1.45 (1.00, 2.09) .047 1.47 (0.95, 2.27) .08 Females 4 2.13 (1.50, 3.02) <.001 — 5 — 3.25 (1.20, 8.81) .021 6 2.04 (1.42, 2.93) <.001 1.80 (0.64, 5.07) .27 ‘1 Analyses used data on 840 males and 791 females, of whom 188 and 247, respectively, attained weekly smoking. ’7 Values in parentheses represent 95% confidence intervals. 68 PETERSON ET AL. TABLE 6,—Relative risk regression analyses of relationship of steps in onset process of smoking with onset of weekly SLT use” Relative rate and 95% confidence interval of weekly SLT useb Run After initial smoking P After weekly smoking P Males l 2.03 (1.56, 2.62) <.001 — 2 —- 3.50 (2.55, 4.82) <.001 3 1.65 (1,25, 2.18) <.001 2.74 (1.95, 3.85) <.001 Females 4 6.72 (2,34, 11,92) <.001 — 5 — 4.57 (2.31, 9.06) <.001 6 4.56 (1.49, 14.0) <.001 2.63 (1.29, 5.40) <.001 ‘1 Analyses used data on 840 boys and 261 events and 791 girls with 35 events. b See table 5, footnote b. for weekly SLT use is 2.03 (P < .001) and 6.72 (P < .001) times as large for males and females, respectively, after initial smoking has occurred than that before it occurred. This result for females is particularly striking: Females who have tried cigarettes are at almost seven times the risk for using SLT as those who have not. From runs 2 and 5, the onset rate for weekly SLT use is 3.50 (P < .001) and 4.57 (P < .001) times as large for males and females, respectively, after weekly smoking than before it occurred. Furthermore, evidence is clear that each of the smoking onset steps, initial and weekly use, contributes to the risk of subsequent weekly SLT use. When both steps are included in the analysis for males (run 3), the occurrence Of initial smoking multiplies the risk of onset of weekly SLT use by 1.65 (P < .001), and the (subsequent) occurrence of weekly smoking multiplies the risk of onset of weekly SLT use by an additional 2.74 (P < .001), for a net multiple (1.65 X 2.74) of 4.5 times the SLT onset rate when no smoking event has occurred. The corresponding multiples for females (run 6) are 4.56 (P < .001), 2.631 (P < .001), and 12.0 (4.56 X 2.63). DISCUSSION AND CONCLUSION A consistently strong relationship is observed between the onset processes of SLT and smoking among adolescents. In particular, the occurrence of events in the smoking onset process is strongly related to increases in the subsequent onset rate for SLT use. Conversely, the occurrence of steps in the onset process of SLT use is strongly related to in- creases in the onset rate of subsequent smoking. The finding that prior use of SLT is a risk factor for smoking indicates that prevention of its use may also help prevent smoking. Conversely, the finding that prior smok- ing is a risk factor for SLT use indicates that prevention of smoking may also help prevent SLT use. These results in- dicate the possible desirability of combining the prevention components of both within an overall intervention program. Such integration also makes practical sense in light of the tobacco common denominator between smoking and SLT use and the needs of schools for integrated interventions. Several limitations of this investigation should be noted. The data used in these analyses on the onset processes for smoking and SLT use are recall data collected retrospec- tively from a cross—sectional survey. Resulting limitations include: 1) Recall bias may be present because the data are limited to those individuals who can remember, and the re- call may be biased among those who remember. 2) The sample doeS not correspond to a defined cohort but is a modification (by in- and out-migration) of some identifi— able original cohort. However, to the extent that in- and out-migrating students are similar in their SLT and smok- ing onset patterns, no bias would result. Also, these data on occurrence of events in the smoking and SLT onset processes span a period (1975—1985) during which the prevalence of SLT use was increasing rapidly. As a result, the relationship between the onset of both during such a period necessarily includes the effects of temporal changes in prevalence. These investigations illustrate how survival analysis methods, and in particular survival analysis regression methods, can help to provide insight into the onset of in— dividual steps of the smoking onset process, the relation— ship between age and the onset rate of various tobacco use events, and the degree to which onset Of different events are related. Results of such investigations can contribute to the design of health—promoting interventions by guiding the choice of component, delivery method, and age and grade at which they are provided. Further research is indicated in several directions: how the effect of SLT use on subsequent smoking onset depends on age and inclusion of other aspects of the tobacco use onset processes including social, environmental, and moti- vation variables. Finally, cohort studies are needed. REFERENCES (I) BOYD G, ARY DV, WIRT R, ET AL: Use of smokeless tobacco among children and adolescents in the United States. Prev Med 16:402—421, 1987 (2) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco: A Report of the Advisory Committee to the Surgeon General. DHHS (NIH) Publ No. 86-2874. Bethesda, MD: Natl Cancer Inst, 1986 (3) MURRAY DM, ROCHE LM, GOLDMAN AI, ET AL: Smoke- less tobacco use among ninth graders in a north-central metropolitan population: Cross-sectional and prospective associations with age, gender, race, family structure and other drug use. Prev Med 17:449—460, 1988 (4) LICHTENSTEIN E, SEVERSON H, FRIEDMAN L, ET AL: Chewing tobacco use by adolescents: Prevalence and relation to cigarette smoking. Addict Behav 9:351—355, 1984 (5) SEVERSON H, LICHTENSTEIN E, GALLISON C: A pinch or a pouch instead of a puff? Implications of chewing tobacco for addictive processes. Bull Psych Addict Behav 4:85—92, 1985 (6) ARY DV, LICHTENSTEIN E, SEVERSON HH: Smokeless to- bacco use among male adolescents: Patterns, correla- tions, predictors, and the use of other drugs. Prev Med 16:385-401, 1987 (7) FLAY BR, D’AVERNAS JR, BEST JA, ET AL: Cigarette smoking: Why young people do it and ways of preventing it. In Pediatric and Adolescent Behavioral Medicine (McGrath P, Firestone P, eds). New York: Springer-Verlag, 1983, pp 132—183 (8) LEVENTHAL H, CLEARY PD: The smoking problem: A review NCI MONOGRAPHS, NUMBER 8, 1989 INITIATION AND USE OF SLT AND SMOKING 69 of the research and theory in behavioral risk modification. Psychol Bull 88:370—405, 1981 (9) HIRSCHMAN RS, LEVENTHAL H, GLYNN K: The development of smoking behavior: Conceptualization and supportive cross-sectional survey data. J Appl Soc Psychol 14:184— 206, 1984 (10) MATHEWS DE, FAREWELL VT: Using and Understanding Medical Statistics, 2nd ed. New York: Karger, 1988 SMOKELESS TOBACCO USE IN THE UNITED STATES (II) LAWLESS JF: Statistical Models and Methods for Lifetime Data. New York: Wiley, 1982 (12) MILLER RG Jr: Survival Analysis. New York: Wiley, 1981 (I3) KALBFLEISCH JD, PRENTICE RL: The Statistical Analysis of Failure Time Data. New York: Wiley, 1980 (I 4) COX DR: Regression models and life tables (with discussion). J Stat Soc [B] 34:187—220, 1972 Prevalence and Predictors of Smokeless Tobacco Use: Iowa’s Program Against Smoking1 Julie A. Burke,2 Rebecca Arbogast,2 Samuel L. Becker,2 Michelle Naughton,3 and Ronald M. Lauer4a 5 ABSTRACT—Data from surveys of adolescents were analyzed so that we could determine the prevalence of smokeless tobacco use and identify and compare the concurrent correlates of its use and cigarette smoking. Panel data from seventh through eleventh and eighth through twelfth graders between 1980 and 1984 and cross-sectional data from seventh graders in 1980 and 1985 suggest that boys are more likely to use SLT than are girls and that the boys’ use has increased with age and over time. Concurrent correlates of seventh-grade boys’ and girls’ SLT use and cigarette smoking were identified with discriminant analyses. Predictors of smokeless tobacco use were compared with those of weekly smoking for boys and girls separately. Predictors of use by boys were also compared with those by girls. Trying alcohol and the other form of tobacco were the only predictors that related to the use of either form by boys and girls. Differences among other predictors were noted and their implications for prevention are discussed.—NCI Monogr 8:71—77, 1989. Following decades of decline, the sale of SLT has in- creased steadily since 1970 (I ). This increase and the mounting evidence of deleterious health consequences from its use (I, 2) have prompted interest in the demographic and psychosocial characteristics of those responsible for recent increases in consumption. Although SLT use by adults has remained relatively con— stant at 3%—4% for males and 1% for females (1 ), that for adolescent males, however, has emerged (in a recent na— tional survey) as representative of a group whose SLT use is increasing. According to the 1985 National Institute on Drug Abuse Household Survey, 16% of the males under 21 years of age had used SLT within the preceding year and 10% were using it at least once a week (I ). Increased use by adolescent males is confirmed by cross-sectional data from several recent regional surveys (1—19) as well as longitu- dinal data from the Bogalusa Heart Study (20, 21). Taken collectively, these data show that SLT use among adoles- cent males 1) is more prevalent than among females, 2) is increasing with age, and 3) is increasing over time. In this paper, we provide additional information on the prevalence of SLT use among adolescents and report our ABBREVIATION: SLT = smokeless tobacco. 1Supported in part by Public Health Service grants HL-32847 and HL-20124 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Department of Health and Human Services. 2Department of Communication Studies, University of Iowa, lOSCSB, Iowa City, Iowa 52242. Address reprint requests to Julie A. Burke, Ph.D. 3Sociology Department, University of Iowa. 4Department of Pediatrics, University of Iowa Hospitals and Clinics. 5 We thank Virginia Daugherty for reading and commenting on an earlier draft. initial efforts to identify demographic and psychosocial cor- relates of such use. Because a recent study (17) reported that some of the social factors influencing adolescent smok— ing were also related to the use of SLT, our examination of its correlates concentrated on variables theoretically and empirically linked to adolescent smoking. The role of social institutions and relationships in con- straining deviant behavior, such as smoking, is emphasized by the social bonding theory (22). In studies of adoles— cent smoking based on this theory, researchers found that adolescents were more likely to smoke if they experienced problems in relationships with parents or peers; were less committed to, or successful in, conventional arenas such as school or church; or were less committed to conventional systems of beliefs (23—28). On the basis of this work, we chose to examine the influence on SLT use of variables as- sessing attachment to parents and peers, commitment to and success in school, adherence to a religion or faith, and com- mitment to conventional values. Because advocates of sev- eral theories (e.g., those on social learning and differential association) argue that adolescents’ smoking is influenced by the smoking of significant others (25—28), we assessed the smoking behavior of parents, siblings, and friends. We also examined the relationship between SLT use and demo- graphic variables such as sex, age, and head of household’s occupation. METHODS Population and procedures—We drew on survey data from two research projects. In the Muscatine Smoking Study, researchers designed a questionnaire to assess to- bacco use that was administered annually between 1980 and 1984 in two middle schools and one high school in Muscatine, Iowa. Public school students in grades 7 through 12 were surveyed in the first 4 years of the study and grades 9 through 12 in the final year. Less than 9% of the sam- ple was nonwhite. Longitudinal data were available from 443 students who were surveyed all 5 years of this project. Data from 427 seventh graders surveyed in 1980 are also reported. Additional data are presented from the first year of the Three Community Smoking Prevention Project. All seventh-grade students in three demographically compa- rable Iowa communities were surveyed twice during the 1984—1985 school year. In the fall, 1,207 students and in the spring 1,170 were surveyed. Analyses were performed on the data from 1,064 students who were surveyed in both waves. Reflecting their home communities of Burlington, Clinton, and Muscatine, these students were predominantly white and from middle- or working-class families. 71 72 BURKE ET AL. Measures—Several questions elicited information on to- bacco use; one was about the frequency of SLT use, and three were on smoking. Respondents were asked about the frequency and quantity of their smoking. They were also asked to place themselves in one of five categories along a continuum from nonsmoker to heavy smoker. No biochemical measure validated self-reported SLT use. Saliva samples collected and tested for thiocyanate, a com- pound found in higher concentrations in the saliva of reg— ular smokers, validated self-reported smoking. Mean sali- vary thiocyanate levels, compared across smoking groups, indicated that confidence in the self-report measures was warranted. For a discussion of the use of such group com- parisons for validating self-reports, see Akers et a1. (29). To determine whether adolescents experience problems in their social relationships, we assessed attachment to family, peers, and community with items used in pre- vious research. An attachment-to-mother index measures the frequency with which respondents’ mothers understand them, praise them, talk with them about their plans and problems, make them feel close, behave in a way they admire, and seem willing to listen (alpha = .89). An attachment-to-father index is made up of the same six items but with the father as the referent (alpha = .92). An attachment—to—friends index measures the frequency with which respondents’ friends understand them, praise them, talk about their thoughts and feelings, make them feel close, behave in a way they admire, and do things that make them happy (alpha = .86). A single item measured attachment to hometown. We also assessed several variables related to commit— ment to and success in school with questions asked in pre- vious research. Single items measured educational aspira- tions, amount of time spent on homework, self-reported grade-point average, and concern about teachers’ appraisal of them. A school trouble index assessed the extent to which students have problems adjusting to the school environment, as indicated by their agreement with three items: People at school blame my group if there is trouble, people at school think of me as a troublemaker, and I would like to quit school (alpha = .69). We constructed an index to measure social integration and success in school that asked respon- dents whether they perceive themselves as being leaders, having friends who are active in school activities, being popular, and being in the “top crowd” (alpha = .73). Because religion represents another arena in which ado- lescents may learn conventional values, we also assessed religiousness. A single item asked respondents how closely they adhered to religious precepts. To determine adolescents’ commitment to conventional values, we assessed their beliefs about obedience to author- ity. Respondents were asked to indicate the extent of their agreement with the following statements: We all have a moral duty to obey the law, the fact that it is against the law makes the use of drugs wrong, the rules or moral be- liefs that my parents have are good enough for me. These items have also been used in previous research based on the social bonding theory. We measured the smoking behavior of several significant others: mother, father, sister, brother, best female friend, and best male friend with five-point scales ranging from never to every day. We assessed demographic variables including age and sex. Using the 1970 detailed occupation codes of the Cen- sus Bureau, we also classified head of household’s occupa- tion to obtain a rough approximation of occupational pres- tige. Because tryng other illicit substances has been related to the likelihood of smoking, we also assessed the frequency of alcohol consumption. Finally, we explored whether sev- eral beliefs about the consequences of smoking might make SLT a more attractive alternative. We asked respondents whether they associated smoking with the following con- sequences: hurting sports ability, endangering health, and leaving a bad smell on breath and clothing. RESULTS Prevalence of Tobacco Use The panel data allowed us to assess trends among ado- lescents as they matured. Table 1 reports the SLT use of two cohorts of boys and girls over a 5—year period as they progressed from seventh through eleventh or eighth through twelfth grade. In this panel, for all grades and years, boys were sub— stantially more likely than girls to use SLT. Most of the SLT use among girls was occasional rather than weekly or daily. For boys, the pattern was one of steadily increasing use. Only 16% of the younger and 21% of the older cohort had tried SLT at least once in 1980, but its use increased annually for both cohorts; by 1984, 46% of the younger and 57% of the older cohort had tried it at least once. The pro- portion of boys who had tried SLT was larger for the older than for the younger cohort for each year between 1980 and 1984. A comparison of cross-sectional data from Muscatine seventh graders also revealed dramatic increases across time in boys’ use of SLT (table 2). In 1980, 18% of the Muscatine seventh-grade boys and 1.5% of the girls had tried SLT; in 1985, 59% of these boys and 18% of the girls reported having tried it. For boys, both occasional and regular use of SLT increased substantially between 1980 and 1985; occasional use was responsible for girls’ increased use. Parallel data for smoking indicated that experimentation with cigarettes also increased for boys and girls between 1980 and 1985. However, during this period, we found a decline in weekly use of cigarettes by boys. Concurrent Predictors of Tobacco Use Although the data from seventh graders in the 1984- 1985 wave of the Three Community Smoking Prevention Project revealed that some individuals used both forms of tobacco, the overlap between groups was not substantial. In the overall sample, 39.5% had neither smoked cigarettes nor tried SLT, and only 3.9% of the individuals had tried SLT at least once and smoked an average of one cigarette a week. We ran several stepwise discriminant analyses to identify the predictors of SLT use and cigarette smoking by seventh- NCI MONOGRAPHS, NUMBER 8, 1989 IOWA’S PROGRAM AGAINST SMOKING 73 TABLE 1.—Self—reported frequency (percent) of SLT use among two cohorts of 211 boys and 232 girls: Muscatine, Iowa, 1980—1984“ Year of data collection Boys Level of use Girls by grade 1980 1981 1982 1983 1984 1980 1981 1982 1983 1984 Never 7th 84.4 8th 78.8 71.6 9th 64.6 10th 1 lth 12th Occasional 7th 14.6 8th 21.2 25.6 9th 30.2 10th 1 1th 12th Weekly 7th 0.0 8th 0.0 28 9th 2.1 0.9 10th 3.1 0.9 1 lth 0.0 12th Daily 7th 0.9 8th 0.0 0.0 9th 3.1 0.9 10th [.0 0.9 1 1th 10.0 12th 64.2 51.5 56.6 43.3 34.0 44.3 41.5 46.7 99.1 99.1 98.2 96.4 92.8 93.0 95.5 92.6 90.9 90.9 53.7 43.4 0.9 3.6 7.2 7.0 4.5 39.8 7.4 8.2 48.5 6.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 4.6 0.0 0.0 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.9 0.0 0.9 6.1 2.7 ‘1 Respondents were asked to indicate how often they used chewing tobacco or snuff on a six-point scale with response options ranging from “never” to “every day.” Those who had tried SLT but used it less than once a week were classified as occasional users. TABLE 2,—Prevalence (percent) of SLT use and cigarette smoking among seventh graders: Muscatine, Iowa, 1980 and 1985 Boys Girls Form and use 1980 1985 1980 1985 SLT" Never 82.4 40.6 98.5 82.1 Occasional 16.1 47.8 1.5 17.3 Frequent 1.5 l 1.6 0.0 0.5 Cigarettes"7 Never 71.8 46.2 75.5 52.6 Occasional (experimenter) 22.2 49.2 21.1 44.3 Weekly 7.1 4.5 3.4 3.1 ‘1 In 1980, 205 boys and 197 girls were asked to indicate how often they used chewing tobacco or snuff. Those who used SLT less than once a week were classified as occasional users. When they reported weekly or daily use, they were classified as frequent users. In 1985, 207 boys and 196 girls were asked about lifetime SLT use. When they reported lifetime use of one to ten times, they were classified as occasional users. When they reported having used SLT more than ten times, they were classified as frequent users. ’7 In 1980, 212 boys and 204 girls were classified according to use of cigarettes; in 1985, there were 199 and 192, respectively. SMOKELESS TOBACCO USE IN THE UNITED STATES grade boys and girls from three Iowa communities. Because the stepwise procedure capitalizes on chance variation, we randomly selected and set aside 35% of the cases for cross- validation. We report the results from the analyses of the total sample as well as the classification results from the cross-validation analyses. For the classification analyses, prior probabilities were set to equal the proportion of cases in each group. Predictors of Smokeless Tobacco Use Because of the striking difference in prevalence of SLT use among boys and girls, we ran separate discriminant analyses to see whether different factors predispose their use of it. Respondents were divided into 2 groups according to their self-reported SLT use; respondents who reported never having tried SLT were classified as never users; those who had tried it even once were classified as SLT users. Table 3 reports the variables that best differentiatedrboys and girls who had never used from those who had tried SLT. For boys, the discriminant function yielded a canonical correlation of .60 and a Wilks’ lambda of .64. On this func- tion, the centroid for the 152 never users was —.83, com- 74 BURKE ET AL. TABLE 3.—Discriminant analysis results of boys’ and girls’ SLT use in 3 Iowa communities, 1984—1985 Standardized discriminant Step function Sex entered Variable coefficient“ Boys 1 Cigarette smokingb .71 2 Alcohol drinking” .44 3 Mother smokingb .27 4 Female friend(s) smokingb —.25 5 Attachment to motherc .25 6 Smoking harmful to sports abilityb .20 7 Educational aspirationc —.l7 8 Head of household occupationf .14 9 Homework hoursc .11 10 Brother(s) smoking” —.1 1 1 1 Social success index‘ —.12 12 Grade point average” .12 Eigenvalue .57 Girls 1 Cigarette smokingb .52 2 Drug use legally wrong" .31 3 Alcohol drinkingb .24 4 Mother smokingb —.27 5 Head of household occupation“ .24 6 Sister(s) smokingb —.28 7 Parents’ rules good“ .23 8 Observance of laws a moral duty‘ .27 9 Attachment to hometown“ —.25 10 Ageb ~27 11 Female friend(s) smokingb .24 12 Social success indexc —.27 13 Attachment to friends“ .18 Eigenvalue .22 " P < .0001. b On these variables, higher scores reflect more of whatever was being measured (e.g., more smoking). C On these variables, higher scores reflect less of whatever was being measured (e.g., less attachment). pared with a centroid of .67 for 188 SLT users. Twelve variables were selected to differentiate these 2 groups. Boys who had tried SLT were more likely to have smoked cigarettes and drunk alcohol. They spent less time on home- work and had lower grades but claimed greater educational aspirations and more social success in school. They also had a parent with a less prestigious occupation, were more likely to believe that smoking would hurt their sports abil- ity, and were less attached to their mothers. Their mothers were more likely to smoke, but their brothers and best fe- male friends were less likely to smoke. For girls, the discriminant analysis produced a canoni- cal correlation of .42 and a Wilks’ lambda of .82. The 232 never users had a centroid of —.18, whereas the 135 SLT users had a centroid of 1.19 on the discriminant function. Thirteen variables were selected to differentiate girls who had never used from those who had tried SLT. Like boys, girls who had tried SLT had a parent with a less presti— gious occupation and a higher probability of having tried cigarettes or alcohol. On the other hand, the girls whose mothers or sisters smoked cigarettes were less likely to try SLT; those whose best female friends smoked were more likely to try it. The greatest difference from boys, though, was the girls’ commitment to conventional values; those who agreed with parental or legal rules were less likely to try SLT. Those who were less attached to their friends but more attached to their hometowns were more likely to try SLT. Although the age range in the sample was small, girls who had tried SLT tended to be younger. In the cross—validation, we correctly classified 79% of the original and 71% of the set-aside subsample of boys. For girls, we correctly classified 87% of the original and 83% of the set-aside subsamples. Closer inspection revealed errors primarily in classifying girls who had used SLT. In the original subsample, we were able to classify correctly 80% of the boys who used SLT but only 17% of the girls. In the set-aside sample, we classified correctly 73% of the boys but only 21% of the girls who had used SLT. Predictors of Cigarette Smoking To parallel the analyses for SLT, we ran separate step— wise discriminant analyses on smoking for seventh-grade boys and girls. In these analyses, we sought to differentiate boys and girls who were never smokers, experimenters, and weekly smokers. Responses to three questions were used for determining membership in these groups. Those who re- ported never having smoked on the frequency and quantity question and who described themselves as nonsmokers were classified as never smokers. Those who reported smoking at least once a week were defined as regular smokers. Those who smoked less often than once a week were classified as experimenters. Nine respondents who provided inconsistent responses were excluded from subsequent analyses. Results are reported in tables 4 and 5. For boys, two discriminant functions were significant (ta- ble 4), as were the differences among the three centroids. The first discriminant function provided the greatest dis- crimination between the 15 weekly and 172 never smokers. Boys who were weekly smokers were distinguished from the never smokers by having tried alcohol or SLT, having male or female friends or a brother who smoked, and hav- ing less success in or commitment to school (as reflected by TABLE 4.—Summary statistics from discriminant analyses of seventh-grade boys’ and girls’ smoking groups, 1984—1985 Boys Girls Summary statistics Function 1 Function 2 Function 1 Function 2 Centroids Never smoker —.84 —.23 —.86 .19 Experimenter .62 .39 .73 —.38 Weekly smoker 3.31 —1.40 3.61 1.24 Eigenvalue 1.02 .18 1.22 .15 Percent of variance 84.68 15.32 89.09 10.91 in discriminant space Canonical R .71 .39 .74 .36 Wilks’ lambda .42 .84 .39 .87 Chi-square 289.01 56.05 239.97 35.67 Degrees of freedom 20 9 34 16 P .0001 .0001 .0001 .003 NCI MONOGRAPHS, NUMBER 8, 1989 IOWA’S PROGRAM AGAINST SMOKING 75 TABLE 5.—Predictors of seventh-grade boys’ and girls’ cigarette smoking status, 1984—1985 Standardized discriminant coefficient Step Sex entered Variable Function 1 Function 2 Boys 1 Alcohol drinking“ .38 —.15 2 SLT“ .44 .76 3 Female friend(s) smoking“ .25 -.22 4 Smoking harmful to sports ability“ —.15 .37 5 Smoking dangerous to health“ .28 —.29 6 Brother(s) smoking“ .14 .37 7 Grade point average“ .19 —.08 8 Homework hours“ .17 —.09 9 Male friend(s) smoking“ .22 —.05 10 Observance of laws a moral duty“ .09 -.19 Girls 1 Female friend(s) smoking“ .30 .ll 2 Alcohol drinking“ .42 .07 3 Grade point average“ .31 .00 4 Brother(s) smoking“ .33 .01 5 Attachment to hometown“ .26 .22 6 SLT“ .14 —.57 7 Age“ .27 —.02 8 School trouble index“ —. 10 —.47 9 Religiousness“ .02 .39 10 Concern for teachers’ appraisal“ —.13 .44 11 Parents’ rules good“ .09 —.35 12 Male friend(s) smoking“ .16 —.19 13 Attachment to father“ .18 —.05 14 Mother smoking“ .17 .05 15 Smoking harmful to sports ability“ .05 .27 16 Social success index“ .11 .25 17 Attachment to friends“ —.16 .01 “ On these variables, higher scores reflect more of whatever was being measured (e.g., more drinking). “ On these variables, higher scores reflect less of whatever was being measured (e.g., student was less likely to believe smoking hurts sports ability). grades and amount of time devoted to homework). Weekly smokers were less likely than never smokers to believe that smoking endangered their health or hurt their sports ability. They were also less likely to believe that one had a moral duty to obey the law. The second discriminant function differentiated most be- tween boys who were weekly smokers and the 153 who were experimenters. The variables that increased the prob- ability that a boy was an experimenter, rather than weekly smoker, were: having used SLT, having a brother who smokes, and believing that smoking hurts sports ability or endangers one’s health. On the other hand, boys were more likely to be weekly smokers than experimenters if they had female friends who smoked. For girls, two discriminant functions were significant, and again, the differences among the three centroids were sig- nificant. The first discriminant function provided the great— est discrimination between the 144 who never smoked and the 12 who smoked weekly. Girls were more likely to be weekly than never smokers if substance use was relatively common in their environment: using SLT or alcohol them- selves, or having female or male friends, mothers, or broth- ers who smoke. Being older and feeling a close attach- ment to their friends also increased the likelihood of being a weekly smoker, but having good grades and feeling more attached to their fathers or hometowns decreased that like- lihood for girls. SMOKELESS TOBACCO USE IN THE UNITED STATES The second discriminant function accounted for little variance in the discriminant space (table 4). It distinguished most between girls who were weekly smokers and those 111 who were experimenting. Weekly smokers were less likely than experimenters to have tried SLT or to be reli- gious. They cared little about what teachers thought and had more trouble and less social success in school. They were also more likely to believe that smoking hurts sports ability and that their parents’ rules are good enough. In the cross-validation analyses, the discriminant func- tions correctly classified 76% of the boys and 77% of the girls overall in the original subsample. In the set-aside sub- sample, we were able to classify 67% of the boys and 64% of the girls overall. In both the original and set-aside sam- ples, a large number of girls who were experimenters were misclassified as never smokers (31% in the original and 53% in the set-aside sample). In the latter sample, 87.5% of girls who were weekly smokers were misclassified as experimenters. CONCLUSIONS Our data, consistent with other recent reports on the prevalence of SLT use among adolescents, suggest that boys are more likely than girls to use SLT and more likely to use it as they get older. Use by male adolescents has increased over the last several years. 76 BURKE ET AL. Many of the predictors of boys’ smoking were the same as those for girls’, and none of the predictors of smoking had opposite effects for boys and girls. This is in contrast to the results from the discriminant analyses for boys’ and girls’ use of SLT; few of the predictors for boys’ were the same as those for girls’ use. The exceptions were trying alcohol or smoking, having a parent with a less prestigious occupation, or having more social success in school; for boys and girls these increased their likelihood of using SLT. If their same- sex sibling smoked, boys and girls were both less likely to try SLT. Two predictors had opposite effects on boys’ and girls’ SLT use: if their mothers smoked, boys were more and girls less likely to have tried SLT; if their best female friend smoked, girls were more and boys less likely to try it. Less academic commitment and achievement increased the probability that boys would use SLT but were unrelated to girls’ use. Questioning conventional values increased the likelihood that girls would try SLT but was unrelated to boys’ use. We did much better in predicting boys’ use than that of the girls. We only explained 18% of the variance in girls’ use, compared with 36% of the variance for boys. When we compared the predictors of boys’ smoking and SLT use, we found that boys who had tried one form of tobacco or alcohol were more likely to have tried the other form of tobacco. Boys who were weekly smokers or SLT users spent less time on homework and had lower grades. On the other hand, several variables had opposite effects on boys’ smoking and SLT use. Those who smoked were more likely to have a female friend or brother who smoked, than were those who used SLT. Those who smoked were less likely to believe that smoking hurt sports ability; those who used SLT were more likely to believe this. Educational aspirations and popularity with classmates were related to SLT use but unrelated to smoking. There were few similarities in the predictors of smoking and SLT use among girls. Besides trying alcohol and the other form of tobacco, only the association with a female friend who smoked was related to both the smoking and SLT. A large number of variables were related in opposite ways to their smoking and SLT use. Girls who smoked were more likely to have a mother who smoked than those who had tried SLT. Girls who smoked were older; girls who had tried SLT were younger. Girls who smoked had less social success in school; girls who used SLT were more popular with classmates. Girls who smoked were more attached to friends and less attached to their hometown; girls who used SLT were less attached to friends and more attached to their hometown. Smoking by male friends and brothers was related to girls’ smoking but not to their SLT use. Lack of commitment to conventional values was more highly related to girls’ use of SLT. Given these results, we would expect the typical adoles— cent SLT user to be a male who has experimented with alcohol and cigarettes. His mother is more likely to smoke, but his brother and best female friend are less likely to smoke. He has less academic success, although he has higher educational aspirations and perceives himself to be more popular with classmates. He is also more likely to believe that smoking hurts sports ability. The typical ado- lescent smoker is not necessarily male. Nevertheless, if we focus only on boys for purposes of comparison, we see that the typical adolescent male smoker is similar to the typi- cal SLT user because he too is likely to have experimented with alcohol and cigarettes. He, like the SLT user, is less likely to spend time on homework and more likely to earn low grades. Boys who are weekly smokers are more likely than SLT users to have friends and brothers who smoke. They are also less likely to believe that smoking hurts sports ability. These findings have a number of implications for the pre- vention of SLT use. First, taken together, the small amount of variance we accounted for in girls’ SLT use and the dis- similarities between predictors of their smoking and SLT use indicate that interventions for preventing smoking may not reduce girls’ SLT use. Clearly, we need more research on psychosocial correlates of girls’ SLT use if we are to de- sign effective programs for preventing it. However, those conducting prevention programs may want to consider the tendency of female SLT users to question parental and legal rules. This finding suggests that girls’ SLT use is perceived as more rebellious than is that of boys. Put another way, unlike boys’ tobacco use, girls’ SLT use, and to a lesser extent their smoking, may be a form of rebellion against societal strictures. On the other hand, the greater similar- ity among predictors of boys’ SLT use and their smoking suggests that similar interventions are effective for both. However, we should note that prevention programs that try to convince boys that smoking will hurt their sports abil- ity may reduce their smoking but increase their use of SLT as a substitute, unless they are convinced that SLT use will hurt their athletic performance as well. REFERENCES (1) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco: A Report of the Advisory Committee to the Surgeon General. DHHS (NIH) Publ No. 86-2874. Bethesda, MD: Natl Cancer Inst, 1986 (2) DEPARTMENT OF HEALTH AND HUMAN SERVICES: Health Im- plications of Smokeless Tobacco Use. National Institutes of Health Consensus Development Conference, vol 6, No. 1. Bethesda, MD: Natl Inst Health Off Med Applications Res, 1986 (3) BONAGURO JA, PUGY M, BONAGURO EW: Multivariate anal- ysis of smokeless tobacco use by adolescents in grades four through twelve. Health Educ 1714—7, 1986 (4) BOYD GM, ARY DV, WIRT R, ET AL: Use of smokeless tobacco among children and adolescents in the United States. Prev Med 16:402—421, 1987 (5) CHASSIN L, PRESSON CC, SHERMAN SJ: Stepping backward in order to step forward: An acquisition-oriented approach to primary prevention. J Consult Clin Psychol 53:612—622, 1985 (6) CHASSIN L, PREss0N CC, SHERMAN SJ, ET AL: Psychosocial correlates of adolescent smokeless tobacco use. Addict Behav 10:431—435, 1985 (7) GLOVER CD: Regional prevalence and patterns of smokeless tobacco use. In Health Implications of Smokeless Tobacco Use. National Institutes of Health Consensus Development Conference, vol 6, No. 1. Bethesda, MD: Natl Inst Health Off Med Applications Res, 1986, pp 27—31 (8) GUGGENHEIMER J, ZULLO TG, KRUPEE DC, ET AL: Changing trends of tobacco use in a teenage population in western Pennsylvania. Am J Public Health 76:196—197, 1986 NCI MONOGRAPHS, NUMBER 8, 1989 IOWA’S PROGRAM AGAINST SMOKING 77 (9) JONES RB: Smokeless tobacco: A challenge for the 80’s. J Wis Dent Assoc 10:717—721, 1985 (10) LICHTENSTEIN E, SEVERSON HH, FRIEDMAN LS, ET AL: Chew- ing tobacco use by adolescents: Prevalence and relation to cigarette smoking. Addict Behav 9:351-355, 1984 (11) MARTY PJ, MCDERMOTT RJ, WILLIAMS T: Patterns of smokeless tobacco use in a population of high school stu- dents. Am J Public Health 76:190—192, 1986 (12) MARTY PJ, MCDERMOTT RJ, YOUNG M, ET AL: Prevalence and psychosocial correlates of dipping and chewing in a group of rural high school students. Health Educ 17:28— 31, 1986 (I3) NEWMAN IM, DURYEA EJ: Adolescent cigarette smoking and tobacco chewing in Nebraska. Nebr Med J 66:243—244, 1981 (14) POULSON TC, LINDENMUTH JE, GREER R0: A comparison of the use of smokeless tobacco in rural and urban teenagers. CA 34:248—261, 1984 (I5 ) ROUSE BA: National prevalence of smokeless tobacco use. In Health Implications of Smokeless Tobacco Use. National Institutes of Health Consensus Development Conference, vol 6, NO. 1. Bethesda, MD: Natl Inst Health Off Med Applications Res, 1986, pp 25—26 (16) SCHAEFER SD, HENDERSON AH, GLOVER ED, ET AL: Patterns of use and incidence Of smokeless tobacco consumption in school-age children. Arch Otolaryngol 111:639—642, 1985 (17) SEVERSON HH: Psychosocial factors in the use of smoke- less tobacco by adolescent males. In Health Implica- tions of Smokeless Tobacco Use. National Institutes of Health Consensus Development Conference, vol 6, NO. 1. Bethesda, MD: Natl Inst Health Off Med Applications Res, 1986, pp 38—43 (18) IOWA DEPARTMENT OF PUBLIC INSTRUCTION SUBSTANCE EDUCATION PROGRAM: 1984—85 Iowa Study of Alcohol and Drug Attitudes and Behaviors Among Youth. Report 1. Normative and Trend Data. Des Moines, IA: Iowa Dept Public Instruction, June 1985 SMOKELESS TOBACCO USE IN THE UNITED STATES (I9) YOUNG M, WILLIAMSON D: Correlates of use and expected use of smokeless tobacco among kindergarten children. Psych Rep 56:63—66, 1985 (20) HUNTER SM, CROFT JB, BURKE GL, ET AL: Social influences and longitudinal patterns of smokeless tobacco use. In Health Implications Of Smokeless Tobacco Use. National Institutes of Health Consensus Development Conference, vol 6, No. 1. Bethesda, MD: Natl Inst Health Off Med Applications Res, 1986, pp 32—37 (21) HUNTER SM, CROFT JB, BURKE GL, ET AL: Longitudinal patterns of cigarette smoking and smokeless tobacco use in youth. Am J Public Health 76:193—195, 1986 (22) HIRSHI T: Causes Of Delinquency. Berkeley, CA: Univ Calif Press, 1969 (23) AKERS RL, KROI—IN MD, LANZA-KADUCE L, ET AL: Social learning and deviant behavior: A Specific test of a general theory. Am Sociol Rev 45:636—655, 1979 (24) KROHN MD, MASSEY JL, SKINNER WF, ET AL: Social bonding theory and minor deviant behavior: A longitudinal analy- sis of adolescent cigarette smoking. J Health Social Behav 24:337—349, 1983 (25) KROHN MD, NAUGHTON MJ, SKINNER WF, ET AL: Social disaffection, friendship patterns and adolescent cigarette use: The Muscatine Study. J School Health 56:146—150, 1986 (26) LAUER RM, AKERS RL, MASSEY J, ET AL: Evaluation of cigarette smoking among adolescents: The Muscatine Study. Prev Med 11:417—428, 1982 (27) MASSEY JL, KROHN MD: A longitudinal examination of an integrated social process model of deviant behavior. Social Forces 65:106—134, 1986 (28) SKINNER WF, MASSEY JL, KROHN MD, ET AL: Social influ- ences and constraints on the initiation and cessation of adolescent tobacco use. J Behav Med 8:353—376, 1985 (29) AKERS RL, MASSEY JL, CLARKE W, ET AL: Are self-reports of adolescent deviance valid? Biochemical measures, ran— domized response, and the bogus pipeline in smoking be- havior. Social Forces 62:234—251, 1983 Marketing Smokeless Tobacco in California Communities: Implications for Health Educationl’ 2 Marc T. Braverman,3 Carol N. D’Onofrio,4 and Joel M. Moskowitz“, 5 ABSTRACT—In the first phase of a 5-year research project aimed at curtailing smokeless tobacco use among youth, we ex- amined the marketing of smokeless tobacco within California. Observational data were collected from almost 200 retail stores located near high schools and colleges in 14 California coun- ties. Interviews were conducted with over 100 school personnel and youth leaders in 13 counties and with seven retailers from 4 counties. Overall, 81% of the stores in the sample carried smoke- less tobacco, and even in urban areas most stores (71%) had these products for sale. Urban stores were more likely to have materials promoting them (e.g., posters, displays, coupon offers). Almost all (98%) stores that carried smokeless tobacco sold moist snuff. Most (78%) school personnel and youth leaders, especially in ur- ban areas, were aware of advertisements for it, particularly on television and in magazines. Rogers’ theory regarding the diffu- sion of innovations was the basis of the discussion of our results. The implications of marketing for the development of health ed- ucation programs are examined.—NCI Monogr 8:79-85, 1989. The tobacco industry attributes much of the credit for recent jumps in SLT consumption to “well planned and executed marketing and advertising strategies” (1). Con- cerned health professionals also recognize its visibility and advertising as major factors in increased prevalence and use (2—5). Therefore, the promotion of such tobacco products is a force that must be considered in the design of prevention programs. In this report, we provide an initial exploration of marketing issues, investigating the availability and salience of SLT across communities, as well as the perceptions of community informants regarding product advertising and promotion. These findings are discussed with regard to the ABBREVIATIONS: SLT = smokeless tobacco; R = coefficient of multiple correlation. 1 Supported by Public Health Service grant CA-41733 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services. 2Address reprint requests to: Project 4-Health, University of California, 1919 Addison St., Suite 204, Berkeley, California 94704. 3 Department of Applied Behavioral Sciences, University of California, Davis, California 95616. 4School of Public Health, University of California, Berkeley, California 94720. 5The authors gratefully acknowledge the assistance in data collection provided by Linda Araujo-Wilson, Evelyn Conklin, Harry Dasher, Richard Enficld, Peggy Frazee, Jeannette George, Roger Ingram, Ray Lyon, Susan Osaki, Harriet Polansky, David Snell, Catherine Speer, and Deryl Waldren (California 4-H advisors); Kathy Rivas (4—H volunteer leader); and by Judy Antell, Dorothy Snow, Howard Tokunaga, and Lois West (University of California, Berkeley, students). development of community-based interventions to curtail SLT use among youth. An understanding of product availability and production is particularly important for an educational program in- tended for delivery throughout a state as geographically, demographically, and socially diverse as California. A crit- ical question is whether the marketing of SLT varies by region within the state, for this might well affect exposure to products, inducements to use them, and, ultimately, use patterns. These issues were investigated within a larger program of exploratory research undertaken by the staff of Project 4—Hea1th as the first phase of a 5—year project aimed at curtailing the use of SLT among California youth. Data were gathered to address the following questions: 1) How widely available are various forms of SLT in California communities? Operationally, availabil- ity was assessed in the percentage of retail stores carrying snuff and chewing tobacco, the types and brands stocked, and the pricing of these products. The latter variable was included because adoles- cents, having limited funds, are likely to respond to variations in product prices, as has been found in the case of cigarettes (6). 2) Is the availability of SLT related to differences in community size and “urbanicity”? 3) Are there regional differences in the promotion of SLT? Here both point-of—sale inducements and community perceptions of advertising and other promotional activities were considered. Variability in these dimensions of marketing would have clear implications for targeting prevention efforts to youth within California, as well as for matching program content to the degree of young people’s exposure to SLT and its commercial promotion. METHODS Project 4-Hea1th is a multiphase research and interven- tion study being collaboratively conducted by the School of Public Health at the University of California, Berkeley, and the California 4-H program, a unit of University of California Cooperative Extension. In each of Califomia’s 58 counties, the 4-H program is administered by univer- sity academic personnel known as 4-H advisors, who are knowledgeable about their communities and generally fa- miliar with techniques of objective data collection. Advi- sors from 13 counties were invited to participate in the exploratory research that initiated Project 4—Hea1th. These counties were purposefully selected to represent a range 79 8O BRAVERMAN ET AL. of urbanicity, size, demographic population characteristics, and geographic location. After receiving training on the specific methods and in— struments to be used, participating advisors assisted by graduate students interviewed key informants in their coun- ties about SLT and completed structured observations con- cerning its availability in local communities. Project staff completed additional observations in a 14th county; the data were collected during June and July 1986. Formal Observations Sample selection—Within each county, the advisor se- lected two municipalities; the largest city (or town) and a small-to-moderate-sized town. In each municipality, one public high school was randomly selected. Exceptions were made for large cities, and thus four high schools were se- lected in Los Angeles, whereas two were selected in both San Francisco and San Jose. Local college campuses also were chosen as observation areas in smaller communities. Around each selected school, an “observation area” ex- tending three to five blocks in each direction was mapped out and used to define the sampling frame for retail estab- lishments. Within this area all supermarkets, convenience, drug, and liquor stores were listed, and two stores from each category were randomly selected for observation. Most areas had fewer than two stores in one or more of the cat- egories, and, in the final sample, the number of total stores selected per area ranged from one in residential areas to eight in commercial areas. In those communities in which only one or two stores were in the surrounding area of the selected high school, the advisor selected for observation a third community within the county as well. Altogether, ob- servations were completed in 199 retail stores around high schools and college campuses in 43 observation areas lo- cated in 14 California counties. Instrumentation and procedure.—The instrument for for— mal observations of retail establishments was a checklist that included, in addition to the location and type of store, the following variables: 1) The number of varieties of the following kinds of products: moist snuff, moist snuff pouches, “loose- leaf chew,” plug, and dry snuff. For each product, one of the following categories was checked: No varieties present, one or two varieties present, three or more varieties present. 2) The presence of posters, displays, or other induce- ments to buy SLT. 3) The location of SLT products within the store. 4) The products adjacent (“within two feet”) to the SLT products. 5) The presence of “look-alike” products, e.g., bubble gum or beef jerky processed and packaged to re— semble moist snuff or loose-leaf chew. 6) The price of selected brands of various types of SLT. If the store owner or manager refused to allow information to be recorded, the advisor selected an alternate store. Data collection was interrupted in fewer than 2% of the stores visited. Data analysis—In analyzing data from the formal ob- servations of SLT in stores, we used the local community (operationally defined here as the observation area) as the conceptual unit. Therefore, for most of the variables, scores were averaged across all stores within each observation area to produce community means (n = 43). Preliminary data analysis revealed that the category “drug store” in- cluded two distinct clusters: large chain drug stores, which sell a wide variety of nondrug products, and local pharma- cies. Thus this category was subdivided for analysis. Using multiple regression, we analyzed the data to deter— mine whether product availability and price could be pre— dicted from selected community descriptors. The following were designated as dependent variables: (1—5) the means for each community on the availability index for, respectively, moist snuff, snuff pouches, loose—leaf chew, plug, and dry snuff; (6) the percentage of stores in a community that car- ried any SLT products; (7—10) the average price for a rep; resentative brand of moist snuff, snuff pouches, loose-leaf chew, and plug; and (11) the number of store-based induce- ments to purchase SLT. (The numbers in parentheses refer to the numbers given to the variables.) The independent variables were categorized into two blocks. Block 1 described the stores within the observation area that contributed the data. Four variables were created reflecting the percentage of the kinds of stores within the observation areas sampled (i.e., percentage of supermarkets, convenience stores, etc.). A fifth variable was the total number of stores sampled in the observation area. Block 2 included three variables that pertained to the community itself. Population size was included by the rank— ing of numbered municipalities associated with each obser- vation area, with “1” being assigned to Los Angeles. Obser- vation areas within the same city were coded with the same rank. The population size of municipalities represented in the sample ranged from 2,900 to over 3 million. To represent an urbanicity dimension, we rank-ordered counties in the percentage of the county’s population that lives in Census—defined “urban areas,” with “1” being the highest and “14” the lowest. We created a third variable, representing the interaction between community size and urbanicity for each community by calculating the product of its urbanicity and size rankings. Thus high values rep- resented small towns in rural counties, whereas low values represented large cities in urban counties. The level of edu- cational institution that formed the core of the observation area, high school or college, was not an even marginally significant predictor for any of the dependent variables and was omitted from the analyses reported here. A hierarchical procedure was used in which block 1 vari— ables were entered first, followed by block 2. For statisti— cal reasons, we deemed it more meaningful to present the change in R2 for the block as a whole rather than break it down into the contribution from each component variable. Interviews Within the selected counties, advisors drew upon their knowledge of the community, resource lists, and recom- mendations from others to arrange interviews with key informants in several categories: school personnel, youth NCI MONOGRAPHS, NUMBER 8, 1989 SLT MARKETING AND IMPLICATIONS FOR HEALTH EDUCATION 81 leaders, health professionals, youth, and young adult users of SLT. As time and access permitted, interviews also were conducted with retail personnel in stores selling the prod- ucts. Six types of interview forms were prepared that cor— responded to the six categories of informants and included both open-ended and fixed response items. All those inter- viewed provided informed consent. If the informant agreed, interviews were tape-recorded. The interviews averaged ap- proximately 30 minutes in length. Responses to selected items from interviews with school personnel and youth leaders were copied verbatim from in- terview forms and audiotapes and were subjected to content analysis for identification of variables and related coding categories. All responses were then coded according to a common system. Descriptive and nonparametric statistics were used for analysis of the results. RESULTS The report of results is organized to highlight the follow- ing components of marketing: product placement, product diversification, price, point—of—sale inducements, and adver- tising and promotion. The tables summarize findings from the observations. Table 1 presents descriptive results at the store level for the 199 stores; table 2 shows the correla- tions among observed variables, when the 43 communities (i.e., observation areas) were used as the units of analysis; and table 3 gives the results of the hierarchical regression analyses, also with the community as the unit of analysis. Placement One component of product placement is the type of store in which it is sold. As table 1 shows, in areas around schools, over 90% of the supermarkets and liquor stores, as well as over 80% of convenience and chain drug stores, carry some form of SLT product; however, almost no pharmacies did so. Another aspect of placement concerns the location of these products within the 161 stores that carry them. The SLT was adjacent to 1) cigarettes in 62%, 2) candy or snacks in 42%, and 3) alcohol in 23%. (This latter fig- ure is heavily weighted by observations in liquor stores.) These proportions exceed 100%; typically, several prod- uct types are adjacent in each case. In many stores, SLT was kept behind the counter or near the cash register to prevent shoplifting, which was repeatedly cited by retailers as a significant problem. In some stores, moist snuff was kept in refrigerated cases to prevent dehydration. This was found most often in stores in large towns and near college campuses that were associated with a more sophisticated clientele, according to several retailers. The characteristics of the communities in which SLT is found were explored in the regression analyses (table 3). Within a community, the percentage of stores that carry some form of it was related to both the type of stores in the community (block 1, R2 = .276, P < .05) and the urban- icity and size of the community itself (block 2, R2 = .174, P < .05). The coefficients of the three variables in block 2 were all significant, which indicated a complex relationship between town size, urbanicity, and the size—urbanicity inter- action. Follow-up analysis of the interaction revealed that large cities in urban counties had the lowest percentage of stores selling SLT (an average of 71% across the communi- ties that fit this classification), whereas large cities in rural counties (91%) and small towns in urban counties (87%) TABLE 1.—Descriptive observation information, store-level data Convenience Chain Liquor Variable Supermarket store Pharmacy drug store store Total No. in sample 60 51 19 28 41 199 Percent of stores in which 92 84 5 88 95 81 any SLT is sold Percent of stores carrying Moist snuff 90 82 5 88 82 79 Snuff pouches 78 67 5 88 82 71 Loose leaf chew 83 57 5 85 86 70 Plug 68 31 0 51 82 51 Dry snuff 37 0 0 10 61 22 Percent of stores carrying 25 30 100“ 26 46 32 SLT displaying inducements to purchase Percent of stores with a 15 24 0 14 20 17 look-alike product Percent of stores carrying SLT in which it is adjacent to Cigarettes 65 67 100” 74 44 62 Alcohol products 1 1 12 0“ 13 59 23 Candy or snacks 35 51 100“ 35 44 42 Mean price :t SD Can moist snuff 1.24 i .14 1.35 :1: .11 1.36” 1.32 i .11 1.32 i .09 1.30 i .12 Can snuff pouches 1.21 :t .12 1.36 i .10 1.36” 1.30 :1: .14 1.31 i .10 1.29 :1: .13 Package ofchew 1.04:l:.10 1.22:1:.15 — 1.10i.08 1.16i.12 1.12i.13 Package ofplug 991.12 1.14:1:.17 — 1.09:t.12 1.08:.19 1.06:t.16 " Only one case was in the subcategory of pharmacies carrying SLT. SMOKELESS TOBACCO USE IN THE UNITED STATES 82 BRAVERMAN ET AL. TABLE 2,—Zero-order correlation matrix for community-level observational variables Variable 2 3 4 5 6 7 8 9 10 11 12 13 1. No. of varieties moist snuff .72a .64“ .3017 .13 —.87“ .42“ .19 24C .26“ .44a .250 .42b 2. No. of varieties snuff 1.00 .64“ .50“ .08 —.68“ .52“ —.03 .01 .37“ .48“ .33“ .21 pouches 3. No. of varieties loose-leaf 1.00 .62“ .35“ -.66“ .37“ .11 .03 .18 .25 " .16 .09 chew 4. No. of varieties plug 1.00 .36“ —.42“ .10 .13 .03 .08 .07 —.12 .22 5. No. of varieties dry snuff 1.00 —.08 —.20 .00 —.05 —.27“ —.23“ —.19 —.30“ 6. Percent of stores with no 1.00 —.35“ —.27“ —.31 b -.40“ -.40“ —.31“ —.20 SLT 7. No. of store-based 1.00 —.21 f —.13 .44a —.51a .17 .31 b inducements 8. Community urbanicity 1.00 .60“ —.07 —.12 —.11 .00 9. Community size 1.00 .09 .01 .09 .11 10. Average price, can of moist 1.00 .86“ .57“ .76“ snuff 11. Average price, can of snuff 1.00 .39“ .58“ pouches 12. Average price, package of 1.00 .66“ loose-leaf chew 13. Average price, package of 1.00 plug “ P < .01. b P < .05. C P < .10. had the highest percentages. Both slightly exceeded small towns in rural counties (85%). Thus, although as a group these factors significantly predict availability, availability does not increase linearly as community size or urbanicity decrease. Product Diversification In View of the marketing emphasis on moist snuff, we examined the number of brands available for each type of SLT product, hypothesizing that the variety of moist snuff brands would be greater than the variety of other SLT types. This expectation was confirmed by the community averages on the 3-point index of availability completed for each store (0 = no brands present; 1 = one or two brands; 2 = three or more brands). Across all communities, moist snuff stood highest (1.42), followed by loose-leaf chew (1.15), snuff pouches (.84), plug (.63), and dry snuff (.27). Because 63% of the observed stores were coded as stock- ing “3 or more” brands of moist snuff, it is likely that a moderate ceiling effect existed in the data. In some stores, the actual number of brands ranged as high as six or seven. For the other product types such as chew and plug, a large variety of brands was much less common. If a wider range had been incorporated at the upper level of the scale, moist snuff would probably show even greater product diversifi- cation than is indicated here, but other smokeless products would probably not show substantial change. The number of varieties of moist snuff, snuff pouches, and loose—leaf chew were all strongly intercorrelated (table 2). A wide va— riety of one form in a given community was associated with a wide variety of the others. As shown in table 3, regressions were conducted for each type of SLT so we could determine whether the presence of different varieties could be predicted from either store TABLE 3.—Increments in R2 for hierarchical regression equations“ R2, block 1 + Dependent measure R2, block 1 P, block 1 block 2 R2 change P, block 2 Average availability index Moist snuff .155 NS“ .365 .209 <02 Snuff pouches .314 <.O2 .385 .071 NS Loose-leaf chew .281 <.05 .383 .102 NS Plug .467 <.001 .486 .019 NS Dry snuff .343 <.01 .378 .035 NS Percent of stores with no SLT .276 <.05 .450 .174 <.05 Store-based inducements .319 <.02 .359 .040 NS “ Note: Independent variables constituting block 1 (all except 5 expressed as percentage) include: 1) supermarkets; 2) convenience stores; 3) pharmacies; 4) chain drug stores; 5) No. of stores in observation area. Independent variables constituting block 2 include: 1) size of city or town (by rank); 2) urbanicity of county (by rank); 3) interaction of size—urbanicity. “ NS = not significant. NCI MONOGRAPHS, NUMBER 8, 1989 SLT MARKETING AND IMPLICATIONS FOR HEALTH EDUCATION 83 type or community characteristics. For all products except moist snuff, store type accounted for a significant propor- tion of variance, whereas community characteristics did not. For moist snuff, community characteristics predicted vari- ety, but store type did not. Furthermore, the main effects for town size, county urbanicity, and the size—urbanicity in- teraction all produced significant coefficients. In 17% of all stores, a look-alike product, usually bubble gum packaged to resemble chewing tobacco, was sold. The other major type of look-alike product was beef jerky out like moist snuff and sold in small round tins. Nearly one of every four convenience stores stocked some form of look-alike product. Price Table 1 shows the average prices and standard devia— tions for each form of SLT product, categorized by type of store. Prices for chewing tobacco appear more likely to vary across sites than prices of snuff. Store type accounted for a significant amount of variance (P < .01) for snuff pouches and for a marginally significant (P < .10) propor- tion of variance for moist snuff with prices being lower in supermarkets than in other stores. Store type did not ac- count for a significant amount of variance for loose-leaf chew or plug tobacco. More important, community charac- teristics did not account for a significant amount of variance for any of the products. This indicates that product prices do not vary appreciably across the dimensions of town size or urbanicity. Point-of-sale Inducements In general, some form of inducement (posters, special displays, or tear—off coupon offers) was present in 32% of the stores that sold SLT (table 1). Furthermore, as table 2 shows, inducements were more likely to be present in urban than in rural stores (r = —.21), and they were moderately correlated with higher prices for moist snuff, moist snuff pouches, and plug tobacco. The results of the regression analysis for point-of-purchase inducements are given in table 3. Store type accounted for a significant proportion of variance. Inducements were most prevalent in liquor stores, followed by convenience stores. Advertising and Other Promotion in the Community Over 78% of the school personnel and youth leaders in- terviewed had seen advertisements for SLT. The propor- tion of respondents having seen advertisements was greater in urban than in rural counties. Fifty—seven percent of the respondents named specific brands that they had seen or heard advertised. The proportion of youth leaders naming brands was significantly greater in urban counties (76.9%) than in rural ones [38.5%, x2(1) = 7.88; P < .01], but no urban—rural differences were found among school person— nel on this variable. Seven specific brands were identified, with Skoal (including Bandits) accounting for 44.6% of all that were mentioned and Copenhagen accounting for an- other 26.5%. Slightly over 50% of the interviewees who had seen SLT advertisements reported seeing them in magazines. A sig- nificantly greater proportion of respondents from urban SMOKELESS TOBACCO USE IN THE UNITED STATES (60%) than from rural counties (38%) mentioned this source [#0) = 3.89; P < .05]. Particular magazines mentioned were most often related to sports, agriculture, and automo- biles. Also mentioned were magazines on camping, outdoor life, and fishing, as well as men’s magazines, trade journals, and general magazines. Only 6 respondents said they had seen SLT advertised in newspapers, and this was typically in the Sunday magazine section. Nearly 1 of every 4 respondents mentioned seeing SLT advertised in stores, particularly commenting on eye-level displays at cash registers and signs at check-out stands in liquor and convenience stores. In addition, about 20% of those interviewed claimed they had seen it advertised on billboards. Interviewees from 10 counties described adver- tisements for it at special events such as rodeos, car races, car shows, sporting competitions, and county fairs. Promo— tional devices in these instances included signs, banners, posters, bandannas, and decals on beer glasses offered for sale, as well as overall sponsorship of the event. Respon— dents who mentioned seeing SLT advertised on television frequently described similar promotional methods on tele- vision broadcasts of sporting events. Awareness of the distribution of free samples was re- ported by 30% of the respondents. The proportion observing free sample distribution was greater in urban than in rural counties. Two-thirds of those who knew of this promotional technique named specific brands. Seven brands were iden- tified, with Skoal Bandits (40%), Skoal (24%), and Copen- hagen (12%) being the most widely reported. Public events were most frequently named as places where free sam- ples were given away. These included county fairs, baseball games, regional events, auto shows, and farm equipment and livestock shows. An important issue related to marketing is the sale or free distribution of tobacco to minors, which is forbidden by California State law. A number of respondents expressed concern about this topic, with regard to SLT. For example, one reported that a distributor at an air show in a rural county was “passing it out to anyone.” Another respondent said that her son came home with free samples given away in oil fields. DISCUSSION Clearly, SLT has penetrated markets throughout Califor- nia, and it is present to a very high degree in the community environments of youth. Some form was offered for sale by 81% of the stores near high schools and college campuses in 43 geographically diverse areas. If pharmacies are ex- cluded, this percentage rises to 90. Illustrating the striking degree to which moist snuff leads the SLT market, all but 4 of the 161 stores in our sample that had some form for sale also carried moist snuff. In addition, the advertising and promotion of it appear nearly as pervasive as the product. The finding that the proportion of stores carrying SLT was lower in large cities within urban counties (71%) than in other types of communities may reflect urban-rural dif- ferences in use rates found in national and regional surveys (7). Still these data show that it is readily available even in urban areas and suggest that efforts are under way to in- crease demand in metropolitan communities. Point-of-sale 84 BRAVERMAN ET AL. inducements were more prevalent in urban areas, and urban respondents were 1) more aware of SLT advertising than were those in rural areas, 2) more likely to identify adver- tisements from multiple sources, and 3) knew of more free sample distributions. Variations in the convenience store category may explain why SLT was found in a slightly higher proportion of stores in large cities within rural counties (91%) and small towns within urban counties (87%) than in small towns in rural counties (85%). In larger towns and urbanized counties, convenience stores almost invariably belong to large chains which, like the tobacco industry, use modern marketing techniques, including incentives for their sales force (8). However, convenience stores in rural towns also include a percentage of small, independent, general stores with less systematic product stocks and less aggressive marketing approaches. A critical issue is whether marketing practices influence youth to use SLT. The results of this study, interpreted in light of research on the diffusion of innovations (9), may be useful in describing the growing popularity of these products, particularly moist snuff, among youth. The ready availability of moist snuff would appear to be a necessary, if not sufficient, condition for use by youth. Thus moist snuff is stocked by a high proportion of several types of stores near high schools and college campuses, whereas other forms are less widely available. Accessibility is additionally assured by distributors locating moist snuff in convenience stores, chain drug stores, and supermarkets that youth visit frequently. Users thus are able to obtain the product with little effort, and nonusers are continually exposed to SLT. According to Rogers (9), such awareness or knowledge is the first stage in the process of adopting an innovation. The finding that SLT was frequently found next to cigarettes, candy, snacks, or alcohol implies the association of moist snuff with other products purchased by consumers for pleasure. This association is likely to increase interest in the product, corresponding to the second or persuasion stage of the adoption process. Awareness and interest are enhanced by advertising, which portrays the association of SLT with attractive role models, and by product sponsor- ship of exciting events. Creative packaging and point-of— purchase inducements arouse additional interest in a setting where the opportunity to try moist snuff is readily available. Decision and implementation, the third and fourth stages of the adoption process, may well be influenced by these fac— tors. Trial of SLT is facilitated in numerous ways. Product di- versification enables the experimental user to exert his in- dividuality by exercising choice. Moreover, if an initial trial is not satisfactory, the experimenter can try another form. Moist snuff pouches initiate the new user to SLT in pre— measured amounts and minimize problems of controlling loose tobacco in the mouth. Instructions distributed at the point of purchase with free samples and television advertis- ing guide the experimenter through early use experiences. The provision of free samples and modest product pricing permit trial to occur at little or no cost. Very young or ten- tative experimenters may begin by using familiar products processed and packaged to look like SLT. Advertising again is important in establishing confirma- tion, the final stage of adoption. In addition to the identifi— cation of self with attractive role models seen in advertise- ments, the new user finds himself identified with other users in the community and with a culture promoted by banners, hats, and T-shirts sporting brand names and slogans. The evaluation of moist snuff after experimental use thus involves much more than simple like or dislike of the product. Newfound status, self-identification, and social acceptance all contribute to the user’s decision to use the product again. Because these benefits meet major develop- mental needs of adolescents, youth who experiment with moist snuff may be especially prone to continue use. With the gradual development of addiction, physiologic processes augment social—psychological reasons for use and take over as primary factors in habit maintenance. Finally, increases in the number of local adopters are likely to stimulate adop— tion by others. In summary, designers of programs to prevent and re- duce the use of SLT among youth must recognize the high availability and visibility of moist snuff within young peo— ple’s immediate environments. The results of this study in- dicate that moist snuff is no longer a new or unfamiliar product in California communities. It is widely available in urban and rural areas, and in a variety of flavors, forms, and packages. More traditional forms of SLT are also available throughout the State, but in fewer stores. The promotion of SLT, and particularly moist snuff, is also pervasive. A range of marketing practices promotes adoption of the SLT habit; therefore, programs aimed at prevention must be sensitive to the local context in which information about it appears and must be flexible enough to permit consideration of local customs and culture related to its use. Prevention programs must recognize, as the tobacco industry has (4), that America is not so much a mass market as a collection of micromarkets. An analysis of the strategies through which SLT is promoted in local communities can help us identify high-risk target groups, as well as various potential intervention points. Additional research is needed on the relationship between product availability and use. Further research is also needed on other dimensions of the community context for SLT use, requiring conceptualization that expands upon models of peer pressure and social modeling. The results of this study support what we consider to be an important new direction for research and intervention relating to the SLT problem. REFERENCES (I ) SHELTON A: Smokeless sales continue to climb. Tobacco Re- ports 109:42—44, 1982 (2) CHRISTEN AG, SWANSON BZ, GLOVER ED, ET AL: Smokeless tobacco: The folklore and social history of snuffing, sneez- ing, dipping, and chewing. J Am Dent Assoc 105:821—829, 1982 (3) CONNOLLY GN, WINN DM, HECHT SS, ET AL: The reemer- gence of smokeless tobacco. N Engl J Med 31:1020—1027, 1986 (4) ERNSTER VL: Advertising and promotion of smokeless to- bacco products. NCI Monogr 8:87—94, 1989 (5) GLOVER ED, CHRISTEN AG, HENDERSON AH: Smokeless tobacco and the adolescent male. J Early Adolescence 221—13, 1982 NCI MONOGRAPHS, NUMBER 8, 1989 SLT MARKETING AND IMPLICATIONS FOR HEALTH EDUCATION 85 (6) LEWIT EM, COATE D: The potential for using excise taxes to (NIH) Publ No. 86-2874. Bethesda, MD: Natl Cancer Inst, reduce smoking. J Health Economics 12121—145, 1982 1986 (8) ENGLANDER TI: The c—store scene. US Tobacco Candy J, June (7) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health 20—July 10, 1985 Consequences of Using Smokeless Tobacco: A Report of (9) ROGERS EM: Diffusion of Innovations, 3rd ed. New York: The the Advisory Committee to the Surgeon General. DHHS Free Press, 1983 SMOKELESS TOBACCO USE IN THE UNITED STATES Advertising and Promotion of Smokeless Tobacco Products1 Virginia L. Ernsterz’ 3 ABSTRACT—This paper is focused on the approaches used to advertise and promote smokeless tobacco products during the early to mid-19805. These included traditional motifs that fea- tured rugged-looking masculine models in sporting and outdoor settings as well as an expanded white-collar appeal. Smokeless tobacco was not affected by the ban on broadcast advertising of cigarettes that went into effect in 1971, and, until 1986, both print and broadcast media were used to advertise it. Promotional activities ranged from sponsorship of sporting events to offers for clothing bearing smokeless tobacco product logos. Despite the claims of manufacturers that advertising and promotional ef- forts were not targeted to youth, smokeless tobacco companies sponsored tobacco-spitting contests with teenage participants, a college marketing program, and college scholarships. In efforts that appeared designed to bolster their public image in the face of growing concern over the consequences of smokeless tobacco use by young people, companies like U.S. Tobacco Company con- tributed to major social programs, including, ironically, alcohol- and drug-abuse prevention programs. Spurred by public health groups, federal legislation was passed in 1986 that banned televi- sion and radio advertising of smokeless tobacco products and re- quired manufacturers to include warning labels on their products on the potential health hazards of smokeless tobacco use.—NCI Monogr 8:87—94, 1989. Given the increased prevalence of SLT use among young males in recent years, growing concern has been expressed about the role of advertising and promotion in fostering such use (1—5). Between 1970 and 1985, the domestic pro- duction of all forms of SLT increased 42%, and the pro- duction of fine-cut tobacco, used in moist snuff, tripled (6). From 1980 on, the increase in sales was largely a func- tion of a continuing increase in the moist snuff category; loose-leaf sales increased slightly until 1984 and then de- clined. All other categories (plug, moist plug, dry snuff, twist/roll, and loose leaf) declined over the same period (7, 8), as shown in table 1. Although firm data on SLT advertising and promotional expenditures are not readily available, various estimates have been published. In 1984, an estimated $8 to $10 million was spent on advertising by US. Tobacco alone (9, 10), with estimates of industry-wide ABBREVIATION: SLT = smokeless tobacco. 1Supported in part by Public Health Service grant CA-13556 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services. 2Department of Epidemiology and International Health, 1699 HSW, School of Medicine, University of California, San Francisco, California 94143-0560. 3I acknowledge the assistance of Bonnie Fox in compiling the back- ground materials for this paper and Maureen Morris for manuscript prepa- ration. advertising expenditures for that year ranging from $20 to $31 million (11, 12). It appears that moist snuff experi- enced the greatest proportional increases in advertising of all the major categories of SLT. The US. Tobacco Company, which commands nearly 90% of the domestic moist snuff market (8), introduced a variety of new SLT products, most notably those in the Skoal Bandits line, and the company’s advertising expendi- tures increased dramatically, from $800,000 to $4.6 million between 1972 and 1984 for television advertising alone (13). These efforts brought marked increases in sales of the company’s moist snuff brands. By 1984, US. Tobacco had become a Fortune 500 company, ranking 485th in to- tal sales but a high 44th in growth rate in earnings per share over the previous 10 years (14). By 1985, the com- pany’s ranking among the Fortune 500 had moved to 476th place (15). In this paper, I describe the types and themes of SLT advertiSements and promotions, address the issue of youth targeting in the advertisements and promotional activities, and review recent legislative efforts to limit its advertising in the United States. Focus is primarily on activities in the first half of the 19805. TYPES AND THEMES OF SMOKELESS TOBACCO ADVERTISING Advertisements for SLT products have traditionally used a rugged, masculine image, reflective of the brand names of such products (e.g., Red Man, Kodiak, Chattanooga Chew, Applejack, Levi Garett, Cannon Ball, and Country Blend). They have appeared in magazines that cater to the sport- ing and outdoor public, such as Field and Stream, Road and Track, and Sports Illustrated, and on sports programs on radio and television. As an official sponsor of the 1980 Winter Olympics, U.S. Tobacco spent $2.5 million to pro- mote snuff (16). For the 1984 Winter Olympics, the com- pany sponsored a Sports Medicine Program and conducted a $1-million ad campaign for Skoal Bandits that was aired during American Broadcasting Company’s coverage of the games (16, 17). For a time, professional athletes served as spokesmen for SLT products, including Pittsburgh Steelers quarterback Terry Bradshaw, Kansas City Royal George Brett, Texas Ranger Sparky Lyle, and Carlton Fisk of the Chicago White Sox (1, 18). New York Yankee Bobbie Murcer recorded the song “I’m a Skoal-dipping Bandit” (I9), and popular country musician Charlie Daniels has appeared in adver- tisements for US. Tobacco’s smokeless products (20). Industry representatives have publicly described product design and promotional strategies developed to attract new types of consumers. The president of Helme Tobacco Com— 87 88 ERNSTER TABLE 1.—SLT sales in thousands of pounds, 1981-1985“ Category 1981 1982 1983 1984 1985 Plug 11,348.0 10,249.0 9,578.0 8,687.0 8,019.0 Moist plug 6,519.0 5,073.0 4,357.0 3,956.0 3,842.0 Dry snuffb 11.8 11.2 10.7 10.2 9.5 Moist snuff and fine cutb 30.2 32.7 35.0 37.5 39.1 Twist/roll 1,861.0 1,767.0 1,709.0 1,632.0 1,516.0 Loose leaf 70,519.0 70,886.0 70,924.0 72,976.0 71,732.0 Total 132,221.0 131,845.0 132,205.0 134,921.0 133,739.0 ”Source for 1981-1982 data is (7); 1983—1985 data (8). b Values for this category are in millions of pounds. pany said in a 1984 interview, “I feel there is a genuine op- portunity to expand the numbers and types of people who are using smokeless tobacco” (21). A 1983 article in Ad- vertising Age described a new urbane marketing strategy for US. Tobacco and quoted the company’s marketing services director as saying, “What we’re doing is directing oral to- bacco at an entirely new demographic group” (19). The text of a 1973 advertisement from US. Tobacco clearly re- veals the company’s attempt to broaden the demographic appeal of its SLT products: “It’s easy to see why cowboys are into ‘smokeless tobacco.’ And scientists and lab technicians are us- ing it, too. . . . Even lawyers and judges are taking it into their courtrooms. Students and teachers enjoy it in their classrooms” (22). The Skoal Bandits line was introduced to the New York market with considerable fanfare in the summer of 1983, with expenditures of $150,000 for a single day’s public- ity events and $2 million or more for related activities in the New York area over that summer (19, 20, 23). Con- sistent with a strategy to gain customers outside the tra- ditional rural, backwoods market, advertisements began to feature white-collar models and to appear in more general circulation magazines, including TV Guide and the National Enquirer, as well as in major newspapers. Televised spots during the 1984 Winter Olympics featured male models with a “big city” image (9). The advertisements suggested that SLT is an acceptable alternative to cigarette smoking. Television and print ads included such slogans as: “Take a pouch instead of a puff” (24). “Just a pinch between your cheek and gum will give you real pleasure—without lighting up” (25). “When you can’t smoke but you want to enjoy tobacco, try Skoal Bandits” (26). Another common theme was the ease of using SLT (fig. 1). The text of a brochure included with a Skoal Bandits product display read: “It’s as easy as 1-2-3 . . . . All you do is put it between your cheek and gum—the refreshing taste comes right through.” Some advertisements provided instructions for use, such as the one in 1980 (20) titled, “Walt Garrison answers your questions about smokeless tobacco.” The text posed such questions as, “Walt, just what is Moist Smokeless Tobacco?”; “How do you use it?”', and “Is it hard to use?” The printed reply to the last question was, “When you first try it, the tobacco may move around in your mouth more than it should, and your mouth may water a bit more than you’re used to, but getting the hang of ‘going smokeless’ is all part of the fun. In a couple of weeks, you’ll be a ‘pro’ ” (fig. 2). Around the time that the Skoal Bandits brand was intro- duced in 1983, the president of the tobacco division of US. Tobacco candidly described the intended “graduation pro- cess” for consumers using his company’s various brands of snuff (2 7). Consumers could begin with Skoal Bandits, in its easy-to-use tea-bag format, then progress to Skoal itself, and finally on to Copenhagen, the strongest of their snuff products. PROMOTIONAL EFFORTS Coupons for free SLT samples have frequently been in- corporated into magazine advertisements (fig. 3), with ac- companying slogans that include “Something for nothing” (28), “Try it for free!” (29), or . . have a dip on us” (30). A 1985 advertisement read, “Shooting the rapids ain’t easy. Using Skoal Long-cut is. Getting a FREE sample is easy too” (3 I ). Free samples have also been distributed in public places and at events such as rodeos, auto races, and other sporting events. Various other forms of promotion have been used. Swiss Army knives were offered to consumers and retailers by US. Tobacco in exchange for a coupon and slide label from Skoal Bandits mint or Skoal Long-cut mint and $2 (32). The Skoal Long-cut Sportsman’s Choice Sweepstakes in 1985 included a grand prize of transportation and lodging expenses plus $500 for a fishing trip for two in Alaska, Canada, or Costa Rica. To enter the competition, one had to submit the answers to three questions (fig. 4) to US. Tobacco: “1) What flavor is Skoal Long-cut? 2) Who makes Skoal Long-cut? 3) How many colors are on the Skoal Long-cut can?” (33). In the spring of 1986, US. Tobacco sponsored the “Once in a Lifetime Liberty Sweepstakes,” which offered consumers the chance to win one of five all-expense—paid trips for two to the Fourth of July Statue of Liberty rededication ceremonies. Entry blanks for the contest, which was intended to promote Copenhagen SLT, appeared in popular magazines (34, 35). Apparel and other items bearing product logos are avail- able free or for purchase, as advertised in popular maga- zines (36, 37). Offers for free merchandise and gift catalogs are commonly included in advertisement coupons (29). A brochure entitled “Smokers Traders” features items bear- NCI MONOGRAPHS, NUMBER 8, 1989 ADVERTISING AND PROMOTION OF SLT PRODUCTS 89 ing U.S. Tobacco product logos, including hats, automotive aprons, car protectors, and golf bags. Underwriting sporting events has been a major form Of promotion. Professional racing car drivers such as Coca Cola 500 winner Benny Parsons, Indianapolis 500 winner A. J. Foyt, and NASCAR driver Harry Gant (38—40) have been sponsored by companies producing SLT. In 1984, US. Tobacco sponsored five of the starting cars at the Indi- anapolis 500. Exposure of cars bearing the Skoal Bandits and Copenhagen logos on the telecast of the race has been estimated at over $500,000 in advertising value, not includ- ing promotional benefits derived from other broadcasts of the race and the appearances of product logos in related stories in Sports Illustrated, People Weekly, and elsewhere. Publicity surrounding Italian driver Teo Fabi’s pole posi— tion in a Skoal Bandits car in the 1983 Indianapolis 500 race was thought to be worth close to $3 million (41). SMOKELESS TOBACCO USE IN THE UNITED STATES FIGURE 1.—Advertisement promoting the ease of using SLT. In professional baseball, U.S. Tobacco has presented the Skoal Pinch Hitter of the Year award. An announcement of the 1985 award in the The Sporting News added, “For a hit pinch, try Skoal” (42). Skoal also sponsored the 1985 season broadcasts of the Atlanta Braves’ baseball games on cable station WTBS, which reaches over 26 million homes in the United States (43). Even after he stopped appear- ing in advertisements for SLT products, former professional football player and rodeo star Walt Garrison continued as spokesman for US. Tobacco, and, in 1985, former Miami Dolphin Nick Buoniconti became president and chief oper- ating officer of US. Tobacco (10). Although currently active athletes no longer appear in the ads, the companies continue to link sports with SLT, and they encourage professional athletes to use their products. For example, visors bearing the labels “Skoal Bandits” and “World Champion 49ers” were produced in 1985. A sur- 90 ERNSTER Walt Garrison answers your questions about smokeless tobacco. Q: Walt, just what is Moist Smokeless Tbbacco? A: It‘s just what it says: ’Ibbacco you enjoy without lighting up. Q: How do you use it." A: First, you don‘t chew it. Just take a small pinch between your thumb and forefinger, put it between your cheek and gum, and leave it there. The tobacco will slowly release its great flavor to give you real tobacco satisfaction Q: Is it hard to use? A: Not really When you first try it, the tobacco may move around in your mouth more than it should, and your mouth may water a bit more than you’re used to, but getting the hang of "going smokeless“ is all part of the fun. In a couple of weeks you’ll be a "pro.” Q: Is there a difference between the three most popular brands? A: Sure. . .HAPPY DAYS is a mild, satis- fying blend of mint-flavored tobacco, while SKOAL is full-bodied with the added good taste of Wintergreen. COPENHAGEN is a stronger, natural blend of choice tobaccos. I r)‘tKi)U\ [o All three are packed in con- venient cans and each is dated for freshness. Q: Is the date on the can the expiration date? A: No, it’s the date of manu- facture. It’s our way of letting you know how fresh and moist our tobacco is. Q: How much does “Going Smokeless” cost? A: An average user ”dips" about 1V2 cans per week, and that’s about 3 dollars worth. Not bad, when you think how much every- thing else costs these days. Q: Do a lot of people use smokeless tobacco? A: A lot more than you think. Last year we sold over 325 million cans. And more and more people from every part of the country are "going smokeless" all the time. (Even loose-leaf chewers are mixing it in with their brands for extra flavor.) Q: Where can I buy it? A: Ask for it at your favorite tobacco coun— ter: or mail the coupon below and you’ll get a free can of HAPPY DAYS to try. Thanks a lot, Walt. . .. A pinch is all it takes!” SEND ME MY FREE CAN OF HAPPY DAYS. Fill out and send to: ”Smokeless Tobacco." PO Box 2900. Dept. FS 003. Greenwich, Conn., 06830 Name and complete address must appear on outside ofenvelope. Nani» W Address C l ly 7 Si illt‘ «V77 7.1;) 77 Age" 7 w” iiiii it ‘OFFEI NOTAVAILAILE 10 IINOIS, FIGURE 2.—Advertisement providing in— structions on the use of SLT. vey of major and minor league baseball teams conducted by the Public Citizen Health Research Group in February 1985 found that almost all the teams responding (30 of 32) answered yes to the question, “Do manufacturers of chew- ing tobacco and tobacco snuff provide these products to your team free of charge?” (Greenberg A: Personal com- munication). Well—known race—car drivers continue to drive cars emblazoned with SLT logos on national television. CORPORATE CONTRIBUTIONS Manufacturers of SLT have cleverly created links be- tween their harmful products and some of our most val— ued social institutions. With its pledge of $10 million (44), US. Tobacco was the largest corporate contributor to the Statue of Liberty restoration program, and snuff tins bore the Statue of Liberty “funding sponsor” logo. The com- pany also contributed $15,000 to a Farm Aid concert, and a blow—up of the company’s check was displayed by Willie Nelson (45). In the mid-1980s, over one—half of US. Tobacco’s charitable contributions went to health and hu— man services agencies, including United Way, Yale Medical Center, Greenwich Hospital, the Alcoholism Council, Save the Children Foundation Inc., the New York Public Library, and Meals on Wheels (46). Similar efforts on the part of the cigarette manufacturers have been viewed as attempts to gain legitimacy, innocence, and institutional dependence on the industry (4 7—49). Such public relations efforts would certainly seem to diffuse or NCl MONOGRAPHS, NUMBER 8, 1989 ADVERTISING AND PROMOTION OF SLT PRODUCTS 91 silence criticism of the SLT industry’s fundamental enter- prise. Indeed, industry executives have been praised for their public service (50). Ironically, U.S. Tobacco has contributed to drug pre— vention programs and has actively supported the National Federation of Parents for Drug-free Youth. Moreover, its second-quarter report for 1985 states that the company contributes to drug prevention and alcoholism programs in the belief that “addiction is one of the most serious prob— lems confronting American families today” (46). Thus the company has become identified with anti-drug abuse and alcoholism programs even as it promotes an addictive and harmful substance. EVIDENCE OF SMOKELESS TOBACCO PROMOTION TO YOUNG PEOPLE The Smokeless Tobacco Council has adopted its Adver- tising and Sampling Code. The Council maintains that the SAW! [-0er cm“ The taste of mint in smokeless tobacco: Skoal‘Long Cut” Easy-to-use pinch. Skoal Bandits= ,. Easy-to-use pouch. . .5 ‘ .L‘m industry has stopped using currently active athletes and em- ploys only models who are 25 years of age or older for its advertisements and commercials. A public service relations promotion from the Smokeless Tobacco Council that ap- peared in The Washington Post in 1985 stated that “chewing tobacco and snuff are intended for adults only.” It defined “adult” as age 18 and older. Official policy notwithstanding, a number of industry activities appear to be youth oriented, e.g., manufactur- ers of SLT have sponsored tobacco-spitting contests, with teenagers as active participants, and company represen- tatives handed out free samples as well as frisbees and T-shirts in southwest Connecticut (51, 52). By defining an adult as someone 18 years of age or older, the industry has given itself permission to promote its products among older teens and college students. In the name of Copen- hagen/Skoal, U.S. Tobacco has contributed $135,000 an- nually in college scholarships in conjunction with the Na- FIGURE 3.—Advertisement incorporating 4 coupons for free samples. Samplethembothforfree. For your free sample of both Skoal Long Cut Mint and Skoal Bandits Mlnt fill out and send to: THE MINT CHOICE. De t. #SH-SOI Box 2900, Greenwich. 06836 Name Address Cilv State Are you a regular user of any smokeless tobacco? Yes No If yes, what brand? I certify that I am years of age.‘ ‘0'!!! no! lvaillhl! lo mlnors Oller good only In U SA. Please allow 46 weeks delivery. Oller vold where prohibited by law 3“; 1985 U 5 Tobacco Co Look for the special “Mint Choice” display Wave: 3ka Bandits and Skoal Long Cut are sold for a valuable premium offer. SMOKELESS TOBACCO USE IN THE UNITED STATES 92 ERNSTER ‘ CANADA-«Plummets was: Choose 5 Crg‘ 326?". $KOAL® was LONG cur» - Easy to Use ~ Great Taste GRAND PRIZE: One week for two at one of the above ”Sportsman's Choice” locations. All FIGURE 4,—Advertisement featuring a con- test with prizes. OFFICIAL RULES: N0 PURCHASE NECESSARY. here is all you have to do. transportation and lodging expenses are paid. $500.00 additional cash (incidental expenses). 100 FIRST PRIZES: Choice of On a 3" x 5" card or piece ol paper hand-print your name. address and zip code. and answer the lollowmg questions (Answers may be Iound on can ol Skoal Long Cut) 1 What llavor is Skoal Long Cul’l - 2 Who makes Skoal Long Cut? - 3 How many colors are on the Skoal Long Cut can’7 (In the event you do not have a can at Skoal Long Cut you can send in tor a label by writing to UST Marketing Communications, ioo West Putnam Avenue Greenwrch. CT 06830 i - Enter as often as you WlSh our mail each entry separately (only one entry per envelope) to SKOAL LONG CUT . ' ' v ‘ ‘ SPORTSMAN‘S CHOICE SWEEPSIAKES. PO BOX 9742. BRIDGEPORI‘ CI (5699 EnIIIeS Lightning ROd f'Sh'ng must be received by July 30, l985 pole and matching 'M X" Line. Winners WI" be selected in random urawrngs conducted by an independent running organization whose ueciSion is linal Sweepstakes open to persons pl 18 years and over Employees and their lamilies at U S 100 SECOND PRIZES: Tobacco Co ,their uealers distributors and advertising agencies are not eligible VOID WHERE PROHIBITED by law or regulation All leaeral and local regulations apply Winners will be notiliea by mail arm odds at winning Wlll depend upon the number at entries received No substitutions Ior prizes allowed unless ollerep by the iudging organizarion whose decision on such matters is linal Taxes are the sole responsibility at the wrnners Winners may be required to sign an allidavn oi eligibility Gian Sportsman Cooler. For a tree sample ol Skoal ' Long Cut." write to Skoal Long Cut. Box 2900. GreenWich, CT 06830 Enclose your name. address. age, and the words "Dept OL-SOI“ on your request Allow 4-6 weeks tor delivery. Otter good only in U S A Oller not available to minors rriuriicatioris tIJlJ West Putnam Avenue Greenwmh, CT 06 3 tional Intercollegirte Rodeo Association (53). An article on rodeo scholarships from the June 1985 issue of Better Homes and Gardens featured a picture of US. Tobacco’s three brands of snuff (54). Moreover, U.S. Tobacco had a College Marketing Program, which included having col- lege students serve as campus representatives for free sam- pling activities as well as sponsoring spring break activities (55). The 1973 US. Tobacco advertisement quoted ear- lier (22) that claimed “Students and teachers enjoy it in their classrooms” is an overt indication of youth targeting. Promotional materials prepared in conjunction with recent attempts to market Skoal Bandits in Scotland included the sentence, “Like your first beer, Skoal Bandits can be a taste that takes time to acquire and get the most out of.” Louis Bantlc, chairman of the board of US. Tobacco Company, once told a reporter, “In Texas today, a kid wouldn’t dare Acceptance oi prizes constitutes permission to use the name. photograph. and likeness ol wrnneis tor purposes at advertising and trace loi no tuither compensation . For names at winners, send a stamoed sell-admessed envelope to Skoal Long Cut ’Spoitsman‘secrinice" Sweepstakes USI Marketing Com 0 go to school, even if he doesn’t use the product, without a can in his Levis” (56). A May 1985 report on the US. Tobacco Company issued by Montgomery Securities specu- lated that “the 18- to 25-year-old age group has been a key element in the company’s growth, with Skoal and Copen- hagen each benefiting as males in their early prime move to demonstrate their masculinity” (57). LEGISLATIVE EFFORTS TOWARD REGULATION OF ADVERTISING AND PROMOTION The SLT products were not affected by the Federal Cigarette Labeling and Advertising Act of 1965 that re— sulted in a mandatory health hazards warning label on cigarette packages. Nor did the consent orders negotiated by the Federal Trade Commission in 1971—1972 to require warning labels on cigarette advertisements apply to SLT. NCI MONOGRAPHS, NUMBER 8, I989 ADVERTISING AND PROMOTION OF SLT PRODUCTS 93 The ban on radio and television advertising that went into effect in January 1971 likewise applied only to cigarettes and not to other forms of tobacco. In June 1984, US. Tobacco yielded to the order of the New York State Attorney General barring the company from using the slogan “Take a pouch instead of a puff” in its advertising of Skoal Bandits. The Attorney General held that “The slogan, used without further qualification, implies that the product is a safe alternative to cigarette smoking when it is not.” The slogan had appeared in commercials aired during coverage of the 1984 Winter Olympics, in print advertising, in subway posters, and on tote bags distributed at Yankee Stadium (10, 58). The Public Citizen Health Research Group filed a peti- tion in February 1984 with the Federal Trade Commission asking it to require health warnings on all packaging and advertising for SLT products (17). A year later, in January of 1985, the Federal Trade Commission asked the Surgeon General for a full-scale investigation of scientific evidence on the dangers of snuff and chewing tobacco. Considerable legislative activity in 1985 at the state and federal levels was designed to require warning labels, pro- hibit SLT advertising via the broadcast media, prohibit free distribution of products, and ban sponsorship of sporting events by SLT companies (11, 12, 59). The State of Mass- achusetts ruled in 1985 to require the following warning label on all packages of snuff: “Warning: Use of snuff can be addictive and can cause mouth cancer and other mouth disorders.” At the federal level, a bill to ban radio and tele- vision advertising of SLT products and to require three rota- tional warning labels on them and in advertisements passed both houses of Congress and was signed into law (PL. 99-252) in early 1986. The warning labels read: “Wam- ing: This product may cause oral cancer.” “Warning: This product may cause gum disease and tooth loss.” “Warning: This product is not a safe alternative to cigarettes.” The American Medical Association, American Public Health Association, American Academy of Pediatrics, and the major voluntary health organizations have all taken the public position that SLT advertising should be banned. A bill introduced into the US. House of Representatives in 1986 called for a total ban on the advertising and pro— motion of all tobacco products (HR. 4972). Congressional oversight hearings on tobacco advertising and promotion have since been under way. CONCLUSIONS Among the many parallels between the history of adver— tising and promoting cigarettes and that of SLT are: Use of sports figures and other celebrities for product promotion, College campus marketing programs and other youth- oriented activities, Adoption by industry of voluntary codes officially con- demning advertising and promotion to children, Implication of product safety in advertisements, Debates over proposed requirements for health hazards warning labels and bans on radio and television adver- tising, and Sponsorship of socially valued events and institutions. SMOKELESS TOBACCO USE IN THE UNITED STATES Unlike cigarette advertising, however, advertisements for SLT have not been directed to women. More information about the target audiences of advertising and promotional efforts is needed, especially if we are to ascertain the effect of such efforts on use patterns by various groups. We need to assess the impact on youth of sponsorship that permits such product logos to be visually associated with sports. Many will watch with great interest the impact of legislation to mandate SLT warning labels and restrict advertising. For example, will implementation of the ban on broadcast advertising of SLT result in dramatic increases in print advertising, as was true for cigarettes in the-19705, and will there be increases in other forms of promotion? If such changes in industry strategy can be documented, they should be taken into account in any future attempts of investigators to examine the effects of legislation on the prevalence of SLT use. Many of the recent changes in SLT advertising and pro- motion (removal of active athletes, withdrawal of the “Take a pouch instead” campaign, and the like) have come about as the result of the airing of public concerns. Continued monitoring of advertisements for themes and validity of claims is necessary for the ongoing formulation of public policy on tobacco advertising. REFERENCES (1) CHRISTEN AG: The case against smokeless tobacco: Five facts for the health professional to consider. I Am Dent Assoc 101:464—469, 1980 (2) GLOVER ED, CHRISTEN AG, HENDERSON AH: Just a pinch between the cheek and gum. J School Health 51:415-418, 1981 (3) CoNNOLLY GN, WINN DM, HECHT SS, ET AL: The reemer- gence of smokeless tobacco. N Engl J Med 31421020— 1027, 1986 (4) KOOP CE: The campaign against smokeless tobacco. N Engl J Med 314:1042—1043,1986 (5) NASH DB: Health implications of smokeless tobacco. Ann Intern Med 104:436—437, 1986 (6) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco: A Report of the Advisory Committee to the Surgeon General. DHHS (NIH) Publ No. 86-2874. Bethesda, MD: Natl Cancer Inst, 1986 (7) MAXWELL JC JR: Smokeless tobacco market evening out. Advertising Age, July 9, 1984 (8) MAXWELL JC JR: Smokeless tobacco volume sales drop. Advertising Age, June 16, 1986, p 62 (9) DOUGHERTY PH: Advertising: Moving smokers to snuff. New York Times, January 13, 1984 (10) GLOEDE WF: Buoniconti tackles U.S. Tobacco post. Adver- tising Age, July 1, 1985, p 38 (11) COLFORD SW: Another tobacco ad ban closer. Advertising Age, October 28, 1986, p 2, 106 (12) COLFORD SW: Smokeless tobacco ad limits creep closer. Advertising Age, November 25, 1985, p 10 (13) ROSENTHAL J: Son of the Marlboro man. The Washington Monthly March 1985, p 51 (14) ANONYMOUS: 500 largest US. corporations ranked by sales. Fortune April 29, 1985, pp 266—285 (15) ANONYMOUS: UST records 13% rise in earnings. US Tobacco Candy J May 15—June 4, 1986, p 8 94 ERNSTER (I6) BLUM A: Using athletes to push tobacco to children. NY State J Med 83:1365—1367, 1983 (I 7) GORDON RL: Health warnings urged for smokeless-product ads. Advertising Age, February 20, 1984, p 14 (I8) WALLIS C: Into the mouths of babes. Time July 15, 1985, p 68 (I9) O’CONNOR JJ: Bandits out to steal bite from cigarets. Ad- vertising Age, June 27, 1983, p 2, 77 (20) FEIGELSON J: Skoal Bandits blitz kicks off N.Y. entry. Ad- vertising Age, August 8, 1983 (21) ANONYMOUS: See long—term gains for smokeless. US To— bacco Candy J June 14—July 4, 1984, p 1 (22) Advertisement. Mechanix Illustrated May 1973, p 133 (23) ANONYMOUS: U.S. Tobacco tries to promote chewing. Boston Herald, July 8, 1984, p 22 (24) GLOEDE W: Ruling signals ad changes at U.S. Tobacco. Advertising Age, June 21, 1984 (25) Advertisement. Sports Afield June 1981, p 124 (26) Advertisement. TV Guide, September 7—13, 1985, p 25 (27) Advertisement. Field and Stream November 1980, p 2 (28) Advertisement. Field and Stream March 1981, p 191 (29) Advertisement. TV Guide, May 24—30, 1986, pp 40—41 (30) Advertisement. Field and Stream June 1982, p 129 (31 ) Advertisement. Field and Stream October 1985, p 35 (32) Advertisement. US Tobacco Candy J October 3—23, 1985, p 15 (33) Advertisement. Outdoor Life June 1985, p 17 (34) Advertisement. US Tobacco Candy J April 3-23, 1986, p 55 (35) Advertisement. US Tobacco Candy J May 15—June 4, 1986, p 57 (36) Advertisement. Outdoor Life October 1983, p 97, 110 (37) Advertisement. Field and Stream October 1984, p 122 (38) ANONYMOUS: Parsons wins race in Copenhagen car. US Tobacco Candy J June 14—July 4, 1984 (39) ANONYMOUS: A.J. Foyt driving for U.S. Tobacco. US To- bacco Candy J June 20—July 10, 1985, p 38 (40) ANONYMOUS: ‘Hurrying Harry’ Gant. US Tobacco Rev 7—9, Spring 1982 (41 ) ANONYMOUS: Fornoro, the grand old Indy. US Tobacco Rev 112—11, 1985 (42) Advertisement. The Sporting News April 8, 1985 (43) ANONYMOUS: Skoal sponsors Braves. US Tobacco Candy J July 11—31, 1985, p 15 ‘ (44) ANONYMOUS: Skoal, Copenhagen lids promote Statue of Lib— erty restoration. US Tobacco Candy J May 9—29, 1985, p 18 (45) ANONYMOUS: Helps farmers. US Tobacco Candy J Novem- ber 14—December 4, 1985, p 11 (46) ANONYMOUS: U.S. Tobacco supports corporate contributions program. US Tobacco Rev 1:14, 1985 (47) WARNER KE, ERNSTER VL, HOLBROOK JH, ET AL: Public pol— icy on smoking and health: Toward a smoke—free genera- tion by the year 2000. Circulation 73:38lA—395A, 1986 (48) PIFER AI (producer): Transcript of “20/20”: “Growing Up in Smoke.” American Broadcasting Company Network, October 20, 1983 (49) TAYLOR P: The Smoke Ring: Tobacco, Money, and Multi- National Politics. New York: Pantheon, 1984 (50) ANONYMOUS: Louis Bantle recognized in the Congressional Record. US Tobacco Rev 1:1985 (51) ANONYMOUS: Tobacco spitting contests: Marketing ploy at- tracts minors to smokeless tobacco products. Smoking Health Reporter 1:5, 1983 (52) ANONYMOUS: U.S. Tobacco supports Junior Achievement Program. US Tobacco Rev 3:1, 1985 (53) ANONYMOUS: 10th year for UST’s rodeo scholarships. US Tobacco Candy J June 14—July 4, 1984, p 62 (54) ANONYMOUS: Cowboy scholarships. Better Homes and Gar- dens June 1985 (55) ANONYMOUS: Bandits on campus. Advertising Age, January 30, 1984 (56) GOODMAN E: Snuff suit. San Francisco Examiner, June 13, 1986 (57) GOLDMAN E: Montgomery Securities. Company Report. New York: U.S. Tobacco Co., May 16, 1985, p 2 (58) ABRAMS R: Attorney General Abrams speaks out against smokeless tobacco. NY State J Med 85:471—472, 1985 (59) BAILEY WJ: Smokeless tobacco update. Smoking Health Re- porter 3:4, 1985 Development and Evaluation of a Smokeless Tobacco Cessation Program: A Pilot Studyl’ 2 Elizabeth Eakin, Herbert Severson, and Russell E. Glasgow3’ 4 ABSTRACT—A multicomponent, cognitive-behavioral inter- vention program, offered to 25 chronic, adolescent male users of smokeless tobacco, was divided into three sessions and in- volved self-monitoring of smokeless tobacco use, awareness of health risks, behavioral coping strategies, frequent phone contact, and relapse prevention training. Biochemical (carbon monoxide and cotinine) verification of self-reports was obtained, informa- tion about subject and environmental characteristics collected, and a 3-month follow-up conducted. Twenty-one of the 25 sub- jects completed treatment; 9 participants were abstinent at the conclusion of the program, and 4 remained successful in quit- ting at the 3-month follow-up. Participants who did not achieve complete abstinence reported substantial reductions in smokeless tobacco use, and no increase in cigarette consumption occurred as a result of reducing or quitting use of it. Subjects successful in achieving cessation had consumed a smaller amount of smoke- less tobacco at baseline and were more likely to be involved in school athletics.—NCI Monogr 8:95—100, 1989. Although SLT was once extolled as being a relatively harmless substance, we now have sufficient evidence, both epidemiologic and experimental, to conclude that its use in- creases the risk of oral cancer and that this risk increases with the degree of exposure (I, 2). In addition to the health hazards directly associated with SLT, its use may increase the probability of subsequent cigarette smoking or use of other nicotine—containing substances (3, 4).5 At the time this study began, no reports of experimental evaluations of programs for the prevention or cessation of adolescent SLT use had been published. One recent report of an SLT cessa- tion program for young adults, which was an adaptation of the American Cancer Society’s “Fresh Start” adult smok- ing cessation program, reported a 6-month success rate of only 2.3% (5). A few investigators (6, 7)5 have found consistency be- tween the correlates of SLT use and those of cigarette ABBREVIATION: SLT = smokeless tobacco. ISupported in part by Public Health Service grant CA-382773 from the National Cancer Institute, National Institutes of Health, and grant DA-03635 from the National Institute on Drug Abuse, Department of Health and Human Services. 2This study was done as an undergraduate honors thesis by Elizabeth Eakin. 3Oregon Research Institute, 1899 Willamette St., Eugene, Oregon 97401. Address reprint requests to Herbert Severson, Ph.D. 4The authors thank Stephen Francis, Edward Lichtenstein, and Henry Alley, whose help was instrumental in the success of this study. 5Dent CW, Sussman S, Flay BR: Manuscript submitted for publication. smoking among male adolescents. These results suggest that the components involved in smoking cessation pro- grams are applicable to adolescent SLT use. Although con- siderable research has been devoted to adult smoking ces— sation, few studies have focused on adolescents in respect to smoking cessation programs (8, 9). The primary purpose of our study was to develop and evaluate a cessation program for adolescent users of SLT. Based on feedback and experience gained from pilot work, a three-session group intervention program was developed. Our secondary goal was to identify variables that might predict treatment outcome. For example, questions such as whether initial number of dips per day at the beginning of the program or years of SLT use were predictive of success in quitting were investigated. Similarly assessed was whether subjects who were not successful in quitting had higher than average levels of alcohol, marijuana, and other drug use. Concern has been voiced that efforts aimed at SLT ces- sation should not be undertaken until more is known about the underlying psychosocial factors involved in its use (10). Although dissemination of large-scale cessation programs would be premature, a pilot study such as the one described below can help illuminate the underlying aspects of adoles— cent SLT use and provide preliminary information about the efficacy of a treatment program for users. For example, sports participation has been hypothesized as being posi— tively related to SLT use (11). In this study, the relation- ship of sports participation to cessation attempts was also explored. METHODS Subjects, recruitment, and therapists—The majority of the 25 subjects (80%) were recruited from local high schools by referrals from counselors, health teachers, and coaches. Two subjects were referred by local dentists, and 3 were re— ferred by a subject in one of our initial groups. The contact persons (e.g., the teachers, dentists, etc.) were given pack- ets containing information about the cessation program and parent/participant consent forms. They gave these packets to male adolescents who they knew to be “chewers,” and kept a list of the names and telephone numbers of potential subjects who were willing to be called by a staff member of the Oregon Research Institute. Inclusion criteria were that the subjects had been daily users of SLT for a minimum of 6 months prior to entering the program and were between 13 and 19 years of age. Subjects ranged in age from 14 to 18 years and had used SLT regularly for an average of almost 3 years. The mean 95 96 EAKIN ET AL. number of dips per day at baseline for.the entire sample was 5.78 (SD = 3.1). Five of the subjects were also daily smokers. Twenty-three subjects (92%) indicated occasional to weekly use of alcohol or marijuana, and 16 (64%) indi- cated use of other substances in the past 6 months. Fourteen (56%) participated in high school athletics. Of the three counselors, one female and one male were advanced undergraduate research assistants who had been trained in cessation counseling; the third counselor was a psychologist and research scientist in charge of the study. Design—Due to the formative stage of this research, a within-subject, replicated AB design was used (12). Each subject was measured at baseline (A) and then during and after treatment (B). A quasi-experimental comparison group of 11 subjects, for whom treatment was delayed, also provided a crude standard for our evaluating quit rates associated with the intervention. As had been expected, not all 25 subjects were recruited at the same time, and 11 were assessed while they were waiting for treatment. The within-subject design allowed for commencement of a treatment group each time the necessary number of sub- jects (3—6) for a group was recruited. The inclusion of a randomized control condition would have significantly in- creased confidence in the results of this study; however, we found it impossible to recruit enough subjects to con- stitute a sufficiently large control group, given the consid- erations of statistical power. Although the replicated AB design is methodologically weak, the percentage of suc— cessful outcomes will provide some degree of confidence that the treatment is effective, especially given low base rates of cessation of adolescent SLT use (10, I 3 ). Procedure.—Adolescents who gave their permission to be contacted were telephoned by the first author. They were given an overview of the program, and verbal intent to volunteer for the program was obtained. Twenty-five of 34 persons contacted by telephone agreed to participate. They were then asked to estimate their frequency of SLT use and cigarettes during the past week. It was explained that this information was confidential and that it would be destroyed should they decide not to participate in the cessation program. The students were encouraged to return an informed consent form, which required both subject and parent signatures before baseline data collection began. Baseline data were collected during the week before the first treatment session; subjects were called three to four times and asked how many dips (and cigarettes) they had had that day and the preceding day(s). They were requested to dip (and smoke) at their normal level and not to attempt to quit during that week. A 15-minute pretreatment interview was also conducted over the phone. During this phone call, subjects were reminded that they were expected to quit on the day of the first treatment session. They were asked if they felt comfortable with that expectation and if they anticipated having any problems in quitting. A treatment group was formed each time at least four consent forms were received. Two groups were seen ini- tially in March 1986. Subjects whose consent forms were received while these 2 groups were in treatment were told that, because of scheduling constraints, they would be put on a waiting list and that, if they did not quit on their own volition while waiting to enter the program, they would be seen when the first 2 groups had completed the program. Subjects on the waiting list were called once a week and asked if they were still “chewing,” and, if so, how many dips they had had the previous week. Treatment. —The intervention consisted of three small group meetings with 2—6 participants, 2—3 counselors, and lasting 1—11/2 hours. There was 1 week between the first and second sessions and 2 weeks between the second and third sessions. The 6 students who missed a session during the program were given the opportunity to come to a make-up session. Session I began with a program orientation in which sub- jects were told that they would receive a $10 payment at the end of the program, regardless of the degree of success in quitting. Confidentiality procedures were reviewed, and subjects were encouraged to be honest so that we could do our best to help them quit. The two primary treatment com- ponents were discussed: subjects’ motivation to quit and the acquisition of coping skills, which would help make quit— ting easier. Subjects were asked why they wanted to quit and were reinforced for their participation in the program. A discussion of previous attempts to quit ensued that em- phasized learning from past quitting attempts. Then a dis- cussion of previous attempts to quit and current chewing situations followed that focused on having the participants identify situations in which it would be most difficult for them to resist taking a dip. Coping skills for SLT cessation (e.g., “the 4 A’s”: Avoid, Alter, Alternatives, Activities) were introduced. Subjects selected the skill(s) they thought would work best for them and were asked to explain specifically how they would use such strategies to cope with high—risk situations without dipping. It was stressed that if one of the coping strategies did not work, another probably would. Subjects were asked to rate their motivation to quit and their confidence that they would not be dipping a year from that date on 10-point scales. They then completed a Smoke- less Tobacco Use Scale while the counselors collected saliva and breath samples. This was a standard procedure at ev— ery session and follow-up. It was explained before the saliva and breath tests that the samples obtained would be ana— lyzed for cotinine and carbon monoxide and would provide a measure of how much subjects had dipped and smoked in the last several days. This “pipeline” procedure has been demonstrated to increase the self-reported use of cigarettes and was presumed to encourage more honest self-reports of chewing (7). The subjects were given a copy of the “Big Dipper” video on SLT (14) and asked to watch it with their parents sometime before session 11. The few subjects without a VCR in their homes arranged to watch it with a friend or neighbor who had one. They were told that ses— sion 11 would take place in 1 week and that they would receive three phone calls during the week to check on their progress, problem- solve any difficult situations, and encour- age them to continue their quit attempts. Each student com- pleted a standardized 65-question drug survey before leav- mg. Much of session 11 was devoted to a discussion of their progress in quitting: their successes and difficulties. This in- cluded a debriefing of slips in which subjects were asked to NCI MONOGRAPHS, NUMBER 8, 1989 AN SLT CESSATION PROGRAM: A PILOT STUDY 97 describe the stress situation and to suggest a strategy that would enable them to handle that situation without dipping (15). Group problem-solving was encouraged. The view that one slip does not represent the end of their quitting attempt (16) was emphasized. Review of subjects’ impres— sions of the video program was followed by discussion of the health risks of SLT use and parental support of the subjects’ attempts to quit. A Quitting Reward Contract was distributed; it was proposed that subjects might want to en— list their parents’ support of their quitting by contracting with them. They were told that the third and final session would take place in 2 weeks and they would receive three phone calls during that time. The students then identified any particularly difficult situations they anticipated during the coming weeks, and the group problem-solved these. The final session began with a discussion of progress in quitting in which slips were debriefed and coping strategies were generated by the group. Subjects who had success- fully quit were encouraged to share the strategies that had worked for them. Options were discussed for those who had cut down but had not quit; these included continuing their efforts to quit completely (the preferred option) or main- taining their current low level of dipping. The counselors suggested that these subjects work on quitting completely because of the danger of occasional dipping gradually in- creasing into regular use of SLT. All subjects completed a “Practice Makes Perfect” exercise (see Dependent and Pre- dictor Variables sections), which was presented as a method of rehearsing the coping strategies that would help them think on their feet and make it through difficult situations. The group then discussed implementation of the quitting contracts distributed in session 11. The schedule of follow-up phone calls was explained, and subjects were told that the counselors would be glad to continue working with them on quitting during those phone calls. They were paid $10 for their participation in data collection activities as they departed. Follow-up contact—A series of eight follow-up phone calls were conducted during the 3 months after treatment (one call/wk for the first mo, and one call every other wk during the second and third mo). During these calls, subjects were asked to report the number of times they dipped, smoked cigarettes, drank alcoholic beverages, and the number of times marijuana was smoked during the previous week(s). For subjects who expressed the need for help in problem-solving and encouragement regarding their continued efforts to quit, these phone calls also included counseling on handling difficult situations and refraining from SLT use. The average length of a phone call was approximately 5 minutes. Subjects were paid $5 to participate in each of the 3— month follow—up meetings, and $10 to attend the 6—month posttreatment meeting. During those meetings, discussion centered on subjects’ progress in abstaining and collection of saliva and breath samples. Dependent variables—Primary dependent variables were the number of 1) self-reported dips and cigarettes smoked, 2) alcoholic beverages consumed, and 3) times marijuana was used. Subjects were asked to monitor these activities, but they were not asked to keep a continuous written record SMOKELESS TOBACCO USE IN THE UNITED STATES of these behaviors.6 Cotinine content in saliva verified re— ports of SLT use. The saliva samples also served to en- courage subjects’ honesty in self-reporting (17). We ana- lyzed breath samples for carbon monoxide on a Mini CO gas analyzer to corroborate self-reported cigarette and/or marijuana smoking. Predictor variables—Information about various subject and environmental characteristics was collected includ- ing: motivation and degree of confidence at session I, the Smokeless Tobacco Use Scale, drug use survey, and the number of close friends who chew. Addiction was assessed with an eight-item adaptation of the Fagerstrom Tolerance Questionnaire (I 8), which has been widely used in smoking cessation research. RESULTS Outcome Of the 25 subjects who began treatment, 21 completed the program. Nine of the 21 (43%) who completed treat- ment or 36% of those who started in the program were successful in quitting by the final session. Success was de- fined as a self—report of one slip or fewer during the week before the end of treatment and a verifying cotinine analy- sis (8 reported no slips, and 1 reported one slip). Saliva samples collected at the final treatment session were analyzed for cotinine, which is a primary metabolite of nicotine. Cotinine in saliva has a half-life of approxi- mately 20 hours (Benowitz NL: Personal communication). A nonuser of tobacco would not be expected to have a saliva cotinine level over 10 ng/mL. A saliva cotinine level of 10 ng/mL is considered a “grey zone,” and levels greater than 25 ng/mL indicate use of tobacco products within the past 20 hours. Because of the disparity between the definition of end-of—treatment success (e.g., one slip or fewer during the wk prior to the end of treatment) and the 20-hour half-life of cotinine, the results of the saliva samples cannot be used as an exact verification of the end—of—treatment success rate. Nevertheless, the data indicate that at least 7 of the 10 sub— jects who reported having one or no slips during the week before the end of treatment were indeed honest in their self-reports. Of the three remaining subjects, one (No. 4) was in the “grey zone” with 16.7 ng/mL of saliva cotinine; subject No. 7 appears to have been lying (given his value of 86.2 ng/mL); and subject No. 18, with 212.4 ng/mL, smoked a half-pack of cigarettes/day, which makes verifi- cation of his self-reported nonuse of SLT impossible. The correlation between self-reported dips during the week prior to the end of treatment and the saliva cotinine analyses was .49. Excluding cigarette smokers from the samples in— creased this to r = .87. 6Due to the almost total noncompliance of subjects in an initial pilot study in completing and returning self-monitoring forms, the staff decided not to require “on the spot” self-monitoring. They hoped that the frequency of phone calls would circumvent the problem of inaccurate self-reporting resulting from the students forgetting how many dips they had during a given time. 98 EAKIN ET AL. Figure 1 displays the quitters’ and nonquitters’ average number of dips/ day at baseline and over the course of treat— ment. At the end of treatment, even the nonquitters had re- duced their use of SLT by 77% from baseline levels. There was little evidence of increased smoking associated with re- ductions in SLT use among smokers at baseline. From base- line, the smokers showed an average decrease of cigarette use of 43% during treatment. However, 1 nonsmoker at baseline began smoking occasionally after the end of treat— ment. One of the 11 students on the 3-week waiting list quit on his own, and another reduced his average number of dips/ day from 7 to 1.8. The remainder (82%) reported no change in dipping behavior while on the waiting list. The subject who quit while waiting for treatment had a num- ber of friends who were participating in the initial cessation program at the same time that he quit on his own. Six-month Follow-up Outcome We were able to obtain follow-up information on 20 sub- jects at the 6-month follow-up. Success in quitting was again defined as one or fewer slips during the week be- fore follow-up and was biochemically verified. Five sub- jects self-reported abstinence at 6 months. Two of these were confirmed with low saliva cotinine levels, 1 was a daily cigarette smoker, and 2 were not confirmed by co- tinine analysis. Thus, conservatively, basing abstinence es- timates on all subjects who began treatment and required biochemical verification, we determined a long-term ces- sation rate of 12%. Those subjects who did not achieve abstinence reduced their daily use of SLT by 45% from baseline levels. 7 6 5 E E ff 4 a it (”5 3 < C! LIJ 3: 2 NONQUITTERS \\ 1 \ \ ’/*\ 1v” \\ QUITTERS o 1 2 3 5. BASELINE TXI TXII TXlll WEEKS FIGURE l.—Average number of dips per day at baseline and over the course of treatment for quitters and nonquitters. Predictors of Cessation Table 1 shows a comparison of those who quit after testing and nonquitters (including the 4 dropouts) on various baseline measures. The quitters had a significantly lower average number of dips/day at baseline than those who did not quit (4.4 vs. 6.4 dips/day, P < .05). Eight of the 9 quitters (88%) were high school athletes, compared with 38% of the nonquitters (P = .022, Fisher’s exact test). Of the 5 subjects who were daily smokers, only 1 was successful in quitting his use of SLT, compared with 8 of 16 nOnsmokers. Other measures were nonsignificant given the small sample size, although some means were in the predicted direction. DISCUSSION This was a small-scale efficacy study that assessed the feasibility of a multicomponent treatment program adapted from recent developments in the smoking cessation litera- ture in aiding male adolescents to quit using SLT. Although the end—of—treatment success rate of 36% (43% counting only subjects who completed treatment) and 6-month ab- stinence rate of 24% (29%) were not as high as expected, these results are far from discouraging. When the study be- gan, we had no precedent to follow in designing the pro- gram. There is a paucity of literature on smoking cessation with adolescents, and, to our knowledge, no SLT cessation studies with adolescents have been published. A relatively high-risk group of adolescents participated in the program. Twenty—three of the 25 (92%) reported being occasional to daily users of alcohol, marijuana, or cigarettes, and 16 (64%) indicated use of other hard drugs, such as amphetamines, hallucinogens, or psychedelic mush- rooms during the 6 months prior to treatment. In addition, 1 subject moved out of his parents’ home because of family conflict, another was expelled from school and began at- tending a “continuation school,” and 2 attended night school in addition to their regular classes because of attendance, academic, or behavioral problems. Seven of the 9 adoles- cents who quit by the end of treatment engaged in a number of problem behaviors (I 9). As described by Jessor and Jes- sor, these behaviors include drug use, sexual intercourse, activism, drinking, problem drinking, and general deviant behavior. This fact increases our confidence in the efficacy of our program, because these adolescents are often diffi— cult to keep in treatment and resistant to attempts to change their problem behaviors. We hypothesized that subjects who were not successful in quitting would be heavier users of SLT, cigarettes, alcohol, marijuana, and other drugs and they would score higher on the SLT addiction scale compared with those who were able to quit. A number of these hypotheses were confirmed: The successful quitters had a significantly lower baseline rate of SLT use than did the nonsuccessful subjects. The nonquit— ters also had a higher, although statistically nonsignificant, use of marijuana and other drugs than did the students who successfully quit. Given the small sample size, it is not sur- prising that many of these differences did not reach con- ventional levels of significance. However, the direction of the data is consistent with what one would expect to find in most substance abuse cessation programs, i.e., that higher levels of use lead to greater difficulty in quitting. NCI MONOGRAPHS, NUMBER 8, 1989 AN SLT CESSATION PROGRAM: A PILOT STUDY 99 TABLE 1.——Characteristics of 9 quitters and 16 nonquitters“ Quitters Nonquitters Baseline Significance data Mean Percent Mean Percent SD testsb Average No. of dips/day 4.4 6.4 3.4 t(23) = 1.6 Athlete 88 38 P < .10 Daily smoker ll 25 P = .01 SLT use 9.3 12.23 2.97 (addiction scale 0-16) Alcohol 6.6 7.9 6.4 8.2 (No. of drinks/wk) Marijuana 2.7 3.6 6.88 (No. of times used/wk) Other drugs 4.4 5.0 (No. of times used 6 mo before treatment) Quit before 44 68 No. of five close friends 3.3 2.8 1.25 who chew ” The 4 subjects who dropped out of treatment were categorized as nonquitters. 1’ We used t—tests to contrast groups on continuous variables and Fisher’s exact test for categorical variables. Significance was noted only for average number of dips/day and participation in athletics. Most successful participants seemed ready to quit, as ex- emplified by the fact that they quit completely (except for an occasional slip) immediately after the first session. Eight of these 9 were members of athletic teams at their schools. In fact, one of the strongest predictors of success was partic- ipation in school sports. Perhaps the self—discipline charac- teristic of athletes was an important factor in their success in quitting. School contingencies related to team member- ship may also have been a factor in subjects’ success in quitting, or at least they were an initial motivator. Twelve of the 25 participants came from the same high school. We first believed that the school’s health teacher was responsi- ble for their recruitment. However, he reported that he had done little in the way of recruiting. In fact, students had come to him for information about the program, often at the insistence and prodding of other students. This “pyramid referral” may have been spurred by a number of new sub— stance use rules that were put into effect at the beginning of the school year; these included a mandatory suspension for anyone caught using a substance (including SLT) on school grounds, and an automatic 3-week suspension from the team for any athlete caught using a substance. The re- cruitment was, by chance, timely, with the school’s “crack— down” on student drug use. However, this factor may limit the generalizability of the results, because we could not de- termine what effect the school’s antidrug policy had on our subjects’ success in quitting. Future research is needed for an investigation of possible interactive effects between ces- sation programs and school antidrug policies. This study was also limited by its small sample size and our inability to rule out placebo effects. Additional research with larger samples that would also allow for evaluations of peer group and counselor effects is indicated. The Big Dipper video and the Quitting Reward Contract were not as successful in motivating the subjects to quit as had been expected. Subjects complained that the video was not graphic enough in its depiction of the health risks SMOKELESS TOBACCO USE IN THE UNITED STATES of SLT use, and many did not watch it with their parents. During the course of treatment, only 3 students contracted with their parents, and 2 contracted during the period of follow-up. Most students either forgot about the contract or did not want to do so with their parents. The video and contract might have been more successful had there been more parental involvement in the treatment, such as having the parents attend a special session in which they would learn how to support their sons’ efforts to quit. The present cessation program could be modified in fu- ture studies in a number of ways. Many of the changes will likely arise from time constraints. The program as it stands was time-consuming due to the large number of phone calls to the subjects. Although the number of calls (20/subject) could certainly be decreased, these calls and the program sessions may have been the most powerful components of the treatment. Nicotine gum is an option that could be explored for more addicted students experi— encing nicotine withdrawal symptoms or other difficulties in quitting. Finally, other avenues of recruitment might be explored. School recruitment was the primary source of our referrals, but some came from local dentists. This profes- sional group can be a valuable source of referrals, because an oral examination can provide a unique opportunity for a dentist to encourage an SLT user to quit the habit (20). This study suggests that a significant number of adoles- cent SLT users are responsive to the opportunity to partic- ipate in a cessation program. Given the rising number of adolescent users, the authors recommend that any com— prehensive drug—use reduction program also address the issue of SLT cessation. We hope this study will provide the impetus for increasing our efforts toward identifying a cost—effective cessation program to help young users to quit. REFERENCES (1) DEPARTMENT OF HEALTH AND HUMAN SERVICES: The Health Consequences of Using Smokeless Tobacco: Summary of 100 EAKIN ET AL. the Advisory Committee’s Report to the Surgeon General. Public Health Rep 101144, 93, 130—132, 1986 (2) DEPARTMENT OF HEALTH AND HUMAN SERVICES: Health Im- plications of Smokeless Tobacco Use. National Institutes Of Health Consensus Development Conference, vol 6, NO. 1. Bethesda, MD: Natl Inst Health Off Med Applications Res, 1986 (3) CHRISTEN AG, GLOVER ED: Smokeless tobacco: Seduction Of youth. World Smoking Health 6:2024, 1981 (4) SEVERSON HH, LICHTENSTEIN E, GALLISON C: A pinch or a pouch instead of a puff? Implications Of chewing tobacco for addictive processes. Bull Soc Psychol Addict Behav 4:85—92, 1985 (5) GLOVER ED: Conducting smokeless tobacco cessation clin— ics. Am J Public Health 76:207, 1986 (6) ARY DV, LICHTENSTEIN E, SEVERSON HH: Smokeless to- bacco use among male adolescents: Patterns, corre- lates, predictors, and the use Of other drugs. Prev Med 16:385—401, 1987 (7) LICHTENSTEIN E, SEVERSON HH, FRIEDMAN LS, ET AL: Chew- ing tobacco use by adolescents: Prevalence and relation to cigarette smoking. Addict Behav 8:351—355, 1984 (8) BIGLAN A, SEVERSON HH, ARY DV, ET AL: Do smoking prevention programs really work? The effects Of attrition on the internal and external validity of an evaluation of a refusal skills training program. Behav Med l02159—171, 1987 (9) WEISSMAN W, GLASGOW R, BIGLAN A, ET AL: Develop- ment and evaluation of a cessation program for adolescent smokers. Psychol Addict Behav 1:84—91, 1987 (10) CHASSIN L, PRESSON CC, SHERMAN S, ET AL: Predicting the onset of adolescent cigarette smoking: A longitudinal study. J Appl Soc Psychol 142224—243, 1984 (11) GLOVER ED, CHRISTEN AG, HENDERSON AH: Just a pinch between the cheek and gum. J School Health 51:415—418, 1981 (12) FREDERIKSEN LW: Single-case designs in the modification Of smoking. Addict Behav 1:311—320, 1976 (I3) ARY DV, BIGLAN A: Longitudinal changes in adolescent cigarette smoking behavior: Onset and cessation. J Behav Med 11:361—382, 1988 (I4) OREGON RESEARCH INSTITUTE: Big Dipper (video). Eugene, OR: Independent Video Services, 1986 (15) SHIFFMAN S, READ L, MALTESE J, ET AL: Preventing relapse in ex-smokers: A self—management approach. In Relapse Prevention (Marlatt GA, Gordon JR, eds). New York: Guilford, 1985 (I6) MARLATT GA, GORDON JR (eds): Relapse Prevention. New York: Guilford, 1985 (I7) EVANS RI, HANSEN WB, MITTELMARK MB: Increasing the validity of self-reports Of smoking behavior in children. J Appl Psychol 62:521—523, 1977 (I8) FAGERSTROM KO: Measuring the degree of dependence in tobacco smoking with reference tO individualization Of treatment. Addict Behav 3:235—241, 1978 (19) JESSOR R, JESSOR S: Problem Behavior and Psychosocial Development: A Longitudinal Study Of Youth. New York: Academic Press, 1977 (20) SEVERSON HH, LECHANCE PA, EAKIN E, ET AL: Smokeless tobacco use by adolescent males in Oregon: Prevalence, patterns Of use and cessation. J Oregon Dent ASSOC 56:28— 35, 1986 Adolescent Smokeless Tobacco Use: Future Research Needs1 Laurie Chassin,2 Clark C. Presson,2 Steven J. Sherman,3 and Lynne Steinberg3 ABSTRACT—Future research needs in the area of adolescent smokeless tobacco use are addressed, based on the studies re- ported in this volume covering methodologic issues and substan- tive directions. In addition, we outline some implications for de— veloping preventive interventions to deter smokeless tobacco use among adolescents.—NCI Monogr 8:101-105, 1989. The papers in this volume provide a descriptive data base that is a useful start toward our understanding of adoles- cent SLT use. The prevalence data obtained from diverse geographic areas and subpopulations clearly document the magnitude of its use among adolescent boys and the recent increases in such behavior patterns. In addition, these pa- pers have identified important correlates of SLT use that will facilitate the development of theoretical models of this behavior. Identifying these correlates helps us to specify the target audience for preventive intervention as well as to suggest potentially modifiable risk factors that will be the focus of these interventions, e.g., attitudes, beliefs, and social influences (1—3). Although these papers represent a promising beginning, many problems and research questions concerning SLT use remain. We will identify some of these important directions for future work. However, before substantive questions can be raised, several issues pertaining to research methodol— ogy must be addressed, but once they have been discussed, directions for future research and implications for interven- tion can be provided. METHODOLOGIC CONSIDERATIONS As the authors of the various papers in this volume note, the present data bases consist largely of cross-sectional correlational surveys of tobacco use behaviors. Although providing good descriptive data, this methodology cannot distinguish the antecedents of a behavior from its conse- quences, nor can it spell out the mediating factors involved in initial SLT use or continuation. From the standpoint of preventive intervention, it is, of course, the antecedents of the behavior that are most important. The difficulties in in- terpreting cross-sectional data are illustrated in our studies, in which cross-sectional and longitudinal analyses of the ABBREVIATION: SLT = smokeless tobacco. ‘Supported in part by Public Health Service grants HD-13449 from the National Institute of Child Health and Human Development and CA- 37001 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services. 2 Department of Psychology, Arizona State University, Tempe, Arizona 85287. Address reprint requests to Laurie Chassin, PhD. 3 Psychology Department, Indiana University, Bloomington, Indiana 47405. same sample yielded different patterns of age effects for adolescent cigarette smoking (4). Cross-sectional and longitudinal analyses can produce different findings for several reasons. For example, cross— sectional analyses can mask potentially important differ- ences between adolescents who are long-standing users and those who have only recently started using tobacco. Thus the factors that lead to the onset of tobacco use are con- founded by other factors that act to maintain regular use. This problem of interpretation is not a new one. It is found in epidemiologic and public health research on dis— ease prevalence (all individuals who have a disease at a given period) and disease incidence (all new cases of a dis— ease). Prevalence and incidence questions often produce conflicting data. A more detailed discussion of these issues as applied to adolescent cigarette smoking has been pre- sented (4). For our purposes here, the conflicting findings of cross—sectional and longitudinal investigations highlight the need for prospective longitudinal (or sequential) stud— ies that can examine more directly the processes involved in the onset of SLT use. Researchers in this area should use these designs to test specific models of behavior with regard to SLT use. Within longitudinal studies, its use should be conceptual— ized as consisting of unique stages of initiation, regular use, and cessation. These stages have been shown to be use- ful in the study of cigarette smoking (5) in which distinct determinants are associated with each stage. The current common practice of making simple distinctions between “nonuse” and “any use” is problematic. The inclusion of one-time, occasional, and regular uses in a single category will limit the extent to which factors leading to regular use can be identified because the group of users is too hetero- geneous. Even within the experimental user category, im- portant differences may be found between a single use and repeated experimentation. Empirically, the significance of a single instance of SLT use has to be established. That sig- nificance may be determined by the particular context of that use. For example, a female who participated in a sin- gle use of snuff or chewing tobacco for a sorority initiation would be classified as a user. However, in this social con— text, a single use is unlikely to signal the beginning of more habitual consumption. Unfortunately, these distinctions between stages and con— texts of use can only be made with a large sample size. Given the relatively low base rate of SLT use (especially among females), researchers in the past have not always been able to investigate separate stages and types of use. Historically, research in SLT use has grown out of stud- ies of cigarette smoking. This history has its advantages be- cause some similarity is likely in the factors and processes involved in both forms of tobacco use. However, it is also 101 102 CHASSIN ET AL. important that one remember that unique determinants of SLT use cannot be captured by simply measuring general smoking-related variables. It may be these uniquenesses are most important in the design of preventive interventions. In an investigation of the unique features of SLT use, methodologic questions arise from the natural co- occurrence of its use and cigarette smoking. As mentioned by several authors, the two forms of tobacco use are cor— related, although substantial numbers of SLT users do not smoke cigarettes. For example, Ary (6) found that 20% of users were also regular cigarette smokers, substantially higher than the base rate of regular cigarette use. How- ever, most (80%) users of snuff and chewing tobacco did not smoke cigarettes. Thus SLT use is a behavior that has significant overlap with cigarette smoking, but it also has substantial independence from smoking as well. This overlap with smoking can create methodologic and interpretative dilemmas when researchers set out to study the unique features of SLT use. Simple comparisons be— tween users and nonusers might prove misleading if re- searchers fail to distinguish between users who do and those who do not smoke cigarettes. Observed differences between user and nonuser groups could be due to the presence of different levels of concurrent cigarette smoking rather than to SLT use itself. For example, our data suggest that adoles- cent males who use SLT and cigarettes differ significantly from their abstaining peers in their perceived family envi- ronments. However, adolescent males who use SLT only do not differ from their abstaining peers on these same variables (7). Thus adolescents who use it in addition to cigarettes (or multiple other substances) may differ in many important ways from those whose use is restricted to snuff and chewing tobacco. Researchers interested in the unique features of SLT use should consider the confounding effects of other substance use when interpreting their data. Trying to recognize both the commonality and the uniqueness among tobacco use behaviors and between to- bacco and other substance use, we will now raise some substantive directions for future research on adolescent SLT use. In each case, we will try to spell out how these direc— tions may also have implications for preventive interven- tions. SUBSTANTIVE AREAS FOR FUTURE RESEARCH Relationship of Smokeless Tobacco Use to Cigarette Smoking One of the most basic and frequently raised issues con— cerning SLT use is the relationship of this behavior to cigarette smoking. Previous research (including the stud— ies reported in this volume) indicates that cross-sectional associations occur between the two forms of tobacco use. However, investigators’ attempts to unravel the temporal sequence underlying this association have produced find— ings that are far from clear. One pattern reported by Ary et al. [cited in (8)] is that cessation of SLT use was asso- ciated with later smoking. Yet another pattern is suggested by Dent et a1. (9) and by Peterson and associates (10). Their data suggest that two pathways occur. Use of SLT increased the probability of smoking onset, just as cigarette smoking increased the probability of initial SLT use. Despite several efforts, it is clear that we do not have a precise specification of the relationship between cigarette smoking and chewing tobacco and snuff use. Yet under- standing of the precise relationship and the temporal se- quence involved has important implications. It may be that adolescent males abandon cigarette smoking to adopt SLT use as a perceived safe alternative. If this is the temporal sequence, it may be that our antismoking campaigns have been too narrowly targeted. These antismoking messages could inadvertently increase SLT use by successfully edu- cating the public only about the health dangers of smoking. By focusing only on smoking, we may also have created the impression that SLT use is a reasonably safe option. An alternative sequence is that its use is a precursor to cigarette smoking. The likely mechanism here would be that SLT would establish a dependency on nicotine and thereby increase the probability of cigarette smoking. This sequence would imply that prevention programs ought to be aimed more directly at SLT use and perhaps that consciousness should be raised about the nicotine-addicting potential of snuff and chewing tobacco products. A third possibility is that cigarette smoking and SLT use are functionally equivalent and are associated because of a common third variable pathway, such as motivation for re- belliousness. This possibility is indicated by the results of Dent et a1. (9). Such a model would have important im- plications for public health policy with regard to tobacco use. If these behaviors serve a common purpose, then the use of either can substitute for the other. Unless the under- lying common purpose was addressed, interventions aimed only at one of the behaviors would likely lead to an in- crease in the use of the other. One example of this is in- terventions that seek to limit the availability and supply of the substance. If marijuana and cocaine use are motivated by identical third variables and if they serve the same func- tions, then programs aimed at decreasing the supply of mar- ijuana may have the unintended consequence of increasing cocaine use. These considerations underline the importance of inves- tigators understanding the functions of adolescent SLT use, so that effective prevention programs can be devised. If in- tervention programs ignore the functions of the target be- havior, then these factors will continue to operate and will push people to engage in the behavior. Unless the interven- tion itself is powerful, it may not override these motivations. For example, fear appeals have been widely but unsuccess- fully used to deter cigarette smoking (11 ). Although these fear appeals may provide temporary motivation, they fail to address the functions that cigarette smoking serves for ado- lescents. These fear appeals were not powerful enough to override the positive functions of cigarette smoking. Else- where, we have discussed in more detail the importance of a functional analysis of substance use for the development of interventions (12). We will now discuss some possible functions of SLT use behavior, many of which have been suggested by other authors in this volume. Self-image and Social Image Functions One possible function of SLT use is that it serves to ex- press or achieve certain kinds of images. Our work with NCI MONOGRAPHS, NUMBER 8, 1989 ADOLESCENT SLT USE: FUTURE RESEARCH NEEDS 103 cigarette smoking and SLT use has pointed to the impor— tance of social image factors (13, 14). Use of snuff ands, chewing tobacco is associated with an image of masculin- ’ ity, $th brave “ grnd'love of the out- doors. Those' «6 " ar efi‘abl’éifo‘ mam numbers of adolése nt boys, who mayv'adeptfisvl‘ use as a way of projecting this image either in their eyes or in those of their friends. As reported in this volume, it is the image functions of its use that are projected in marketing appeals (15, 16). One issue raised by some authors is whether SLT adver- tisements are targeted to adolescent audiences or are aimed only at adults (as claimed by the tobacco industry). Even if they are successful in portraying SLT use as an adult be- havior, this does not mean that they will be ineffective in influencing adolescents to use tobacco. One important ap- peal of tobacco use behaviors may be to project an image of precocity. Thus it may be attractive for these adolescents to engage in a behavior that is defined as strictly adult in nature. Our work indicates that the use of SLT may help ado- lescents attain an image that has Specific social benefits. If so, preventive interventions cannot be done simply by our teaching adolescents to “say no,” because they may lack the motivation to do so. Instead, interventions should provide adolescents with other ways to attain these social image benefits. Leaders of several promising programs in smok- ing prevention have incorporated social skills training to achieve these goals (I 7). Sex-role Image Functions One part of the social image associated with SLT use involves ruggedness or masculinity. In fact, its use in most subcultures is largely limited to males. One function that it may serve for males is to allow them to engage in a behavior that is explicitly defined as unfeminine. Such a behavior would set them apart from girls and prove or express their masculinity. At one time, a similar function may have been served by cigarette smoking, but with rising rates of smoking among women, sex-typed distinctiveness for cigarette smoking has all but disappeared. To achieve an image of masculinity, adolescent boys may be turning to a different kind of behavior that thus far has not been adopted by adolescent girls. The sex-typed nature of SLT use raises the question of the meaning or significance of it for females. Studies of cigarette smoking have generally found few sex differences in the dynamics of initiation [although girls have been re- ported to be more susceptible to social influences (18)]. Burke et al. (2) reported some sex differences in the corre- lates of SLT use that they found difficult to interpret. They also found that its use by males was more readily predict- able from social—psychologic factors than was its use by females. This difficulty in the understanding of female use may result partly from their low base rate. The small num— ber of females who use SLT may be motivated by vari- ous idiosyncratic factors rather than a homogeneous set of causes that would be more readily detectable. At this time, female use may be more easily studied in the few sub- cultures in which it is more common (19). However, we SMOKELESS TOBACCO USE IN THE UNITED STATES should not ignore the possibility that the use of SLT may follow the history of Cigarette smoking and eventually be- come widespread among females. The sex-typed nature of its use may also have impli- cations for interventions. If boys use it partly to project a masculine image, then their image in the eyes of adolescent girls should be extremely important in influencing their be- havior. For example, Gritz (20) suggests that information about females’ negative reactions to snuff and chewing to- bacco could be incorporated into prevention programs. Parental Influence Parental influences, of course, have been demonstrated in many areas of substance use, and SLT is no exception. One interesting finding is reported by Bauman and co-workers (1 ), who observed that at higher levels of father’s education more similarity occurred between father and son in its use. In the past, many reported similarity between parents and adolescents in substance use behaviors. However, surpris- ingly little work has been done on the question of whether such similarity changes with indices of socioeconomic sta- tus, such as parental education. The effect of socioeconomic status found by Bauman et al. (1) raises several intriguing possibilities about the processes of parental influence. First, it is possible that parents in the lower socioeco- nomic levels are generally less powerful models for their children. In other words, it is possible that similarity (across a range of behaviors) between parents and children is gen- erally less at the lower levels of socioeconomic status. This is consistent with a previous report that adolescents who are poor feel more independent from their parents in var- ious behavioral domains than do those in the middle— and upper-income classes (21 ). If parental behaviors are, in fact, less imitated at lower levels of socioeconomic status, then further questions arise about the ways in which such sta- tus weakens parental modeling effects. One possibility is that status differences in modeling are due to differences in parenting styles across socioeconomic status. Alternatively, the greater impact of SLT use by a fa- ther who has a high level of education might be due to the relative distinctiveness of its use at that level of socioeco- nomic status. At high levels, generally fewer models for SLT use are observed among adults and peers, so that the pres- ence of a parental model is salient and powerful. However, at lower levels, the adolescent will be exposed to multiple models, and the power of a single parental model might be diminished. If this interpretation is correct, then the reverse finding would be expected when the behavior in question was relatively rare at low levels of socioeconomic status. For example, playing golf may be relatively uncommon for a person who is at the lower levels of socioeconomic status; but in this case, there should be more similarity in playing golf among parents and adolescents in the lower socioeco— nomic levels than among their higher income counterparts. This interpretation suggests a more general principle that distinctive or unique parental behaviors are imitated more regardless of socioeconomic status. The finding of greater similarity in parent—child behavior at higher levels of parental education thus indicates sev— eral interesting possibilities. First, it may be that parents 104 CHASSIN ET AL. of lower socioeconomic status are generally less power- ful models. Second, it may be that distinctive or unique parental behaviors are imitated more than common behav- iors regardless of status. Although the finding was reported in the context of SLT use, it may have wider implications for the understanding of parental influences on adolescent socialization generally. One side benefit of adolescent SLT use is that its study may evoke more general questions of adolescent development. Smokeless Tobacco Use as a Problem Behavior One influential theory of substance use by adolescents has been proposed and tested by Jessor and Jessor (22). According to their problem behavior theory, substance use behaviors represent premature transitions to adult behaviors in violation of age norms. These behaviors are adopted by adolescents who are relatively unconventional, rebellious, and tolerant of deviant behaviors. Support for this position has been obtained from studies of alcohol and marijuana use and cigarette smoking. When a behavior is adopted for these reasons, implications for prevention and intervention are important. For example, this theory suggests that the at-risk group for substance use is relatively deviance prone and unconventional. Such a high-risk group is unlikely to be reached or influenced by traditional school-based prevention programs. The major question for our concerns is whether adoles- cent SLT use also represents a problem behavior in Jessor and Jessor’s sense of the term. Some evidence indicates that it does not. Our data (14) suggest that SLT use is a behav- ior that is 1) relatively acceptable to adults compared with other substance use and 2) engaged in openly rather than hidden from parents. In some social contexts, the behavior is even displayed publicly by adolescents and is encouraged by adults, e.g., in tobacco—spitting contests at county fairs. We believe it is impossible to conceive of similar publicly rewarded displays of marijuana smoking or alcohol con— sumption for young adolescents. In addition to its relative social acceptability, some data on personality correlates cast doubt on the role of SLT use as a problem behavior. For example, Edmundsen et al. (23) found that college students who used SLT were more reserved, more conforming, and less outgoing than nonusers. These are not the personality characteristics of deviance-prone adolescents who engage in problem behav- iors. Another personality correlate relevant to the problem behavior theory is risk-taking, which has been associated with a variety of adolescent drug-use behaviors (24 ). The evidence with respect to risk-taking and adolescent SLT use is mixed. For example, Sussman and associates (3) found that risk—taking was not a significant prospective predictor of experimentation with its use. However, Dent et al. (9) determined that risk-taking was a predictor of initial use. Thus data on personality correlates are not strongly supportive of its use, per se, as an adolescent problem behavior. If use of SLT does not represent a classic problem behav- ior, then such use may be a function of positive socialization rather than of “reactance” and rebelliousness. In this model, changing the perceived social acceptability of the behavior should decrease its use. If parents 1) were to be better in- formed about the health risks of the behavior, 2) decreased their use, and 3) were clearly unfavorable to SLT use by their children, then fewer adolescents would use it. However, it is possible that SLT use represents a prob- lem behavior in some populations and not in others. For example, among rural and agricultural subcultures, such as those represented in the Edmundsen et al. study (23 ), SLT use may represent more of an accepted, normative behavior for adolescent boys than in other subcultures. Even within a single population, it may be that its use represents a prob- lem behavior function for some users and not for others. For example, its use among athletes may not reflect a re— bellious, problem behavior. An understanding of the extent to which SLT use repre- sents a problem behavior in particular subpopulations can have important implications for interventions. For example, Eakin and co—workers (25) determined that an intervention program aimed at users was more successful at a 3-month follow-up for less deviance-prone individuals (i.e., for those who were involved in school athletics and who were less likely to use marijuana). The authors noted that their pro- gram impact may have been strengthened by changes in school rules to enact more stringent penalties for SLT use. Such a combination of formal intervention with more broad changes in the school environment (initiated by school ad- ministrators) is indeed ideal from the point of view of mag— nifying and maintaining treatment effects. However, rein- forcement from a school setting may be less able to influ— ence deviance-prone adolescents. CONCLUSIONS We have pointed to several important directions for fu- ture research in the area of SLT use. Methodologically, we underlined the importance of longitudinal studies that look upon its use as a unique behavior (examining the impact of concurrent multiple substance use) and as a process that oc- curs overtime in distinct stages. Substantively, we suggested that researchers examine the temporal sequences underly- ing the relationship between cigarette smoking and SLT use and the functions of its use (including self—image, sex—role, and problem-behavior functions); they should also investi- gate different mechanisms of parental influence. To get a better understanding of adolescent SLT use, we focused on developing better theoretical models of these processes as well as refining and testing these process models. Most work in the area of adolescent substance use arises as a response to an already emergent problem. Thus re- search and intervention have tended to be reactive rather than proactive. Perhaps researchers should turn to the his— tory of patterns of substance use and to use that information to attempt to anticipate future trends in SLT use behavior. REFERENCES (1) BAUMAN KE, KOCH GG, LENTZ GM: Parent characteristics, perceived health risk, and smokeless tobacco use among white adolescent males. NCI Monogr 8:43—48, 1989 (2) BURKE JA, ARBOGAST R, BECKER S, ET AL: Prevalence and predictors of smokeless tobacco use: Iowa’s program against smoking. NCI Monogr 8:71—77, 1989 NCI MONOGRAPHS, NUMBER 8, 1989 ADOLESCENT SLT USE: FUTURE RESEARCH NEEDS 105 (3) SUSSMAN S, HOLT L, DENT CW, ET AL: Activity involve- ment, risk-taking, demographic variables, and other drug use: Prediction of trying smokeless tobacco. NCI Monogr 8:57—62, 1989 (4) CHASSIN L, PRESSON CC, SHERMAN SJ, ET AL: Changes in peer and parent influence during adolescence: Longitudi- nal versus cross-sectional perspectives on smoking initia- tion. Dev Psychol 22:327—334, 1986 (5) FLAY B, D’AVERNAS J, BEST JA, ET AL: Cigarette smoking: Why young people do it and ways of preventing it. In Pediatric and Adolescent Behavioral Medicine (McGrath PJ, Firestone P, eds). New York: Springer-Verlag 1983, pp 132—183 (6) ARY D: Use of smokeless tobacco among male adolescents: Concurrent and prospective relationships. NCI Monogr 8:49—55, 1989 (7) CHASSIN L, PRESSON CC, SHERMAN SJ: Family correlates of adolescent smokeless tobacco use in relation to cigarette smoking. Int J Fam Psychiatry 9:49—66, 1988 (8) ORLANDI MA, BOYD G: Smokeless tobacco use among ado- lescents: A theoretical overview. NCI Monogr 825-12, 1989 (9) DENT CW, SUSSMAN S, JOHNSON CA, ET AL: Adolescent smokeless tobacco incidence: Relations with other drugs and psychosocial variables. Prev Med. In press (10) PETERSON AV, MAREK PM, MANN SL: Initiation and use of smokeless tobacco in relation to smoking. NCI Monogr 8:63—69, 1989 (I 1) THOMPSON EL: Smoking education programs 1960—1976. Am J Public Health 68:250—260, 1978 (12) CHASSIN L, PRESSON C, SHERMAN SJ: Stepping backward in order to step forward: An acquisition-oriented approach to primary prevention. J Consult Clin Psychol 53:612—622, 1985 (I3) BARTON J, CHASSIN L, PREsSON CC, ET AL: Social image factors as motivators of smoking initiation in early and middle adolescence. Child Dev 53:1499—1511, 1982 (I4) CHASSIN L, PRESSON CC, SHERMAN SJ, ET AL: Psychosocial correlates of adolescent smokeless tobacco use. Addict Behav 10:431—435, 1985 SMOKELESS TOBACCO USE IN THE UNITED STATES (15) BRAVERMAN MT, D’ONOFRIO CN, MOSKOWITZ J: Market- ing smokeless tobacco in California communities: Impli- cations for health education. NCI Monogr 8:79—85, 1989 (I6) ERNSTER VL: Advertising and promotion of smokeless to- bacco products. NCI Monogr 8:87—94, 1989 (I 7) BOTVIN G: Prevention of adolescent substance use through the development of personal and social competence. In Preventing Adolescent Drug Abuse: Intervention Strate- gies (Glynn T, Leukefeld C, Ludford J, eds). NIDA Res Monogr 47:115—141, 1983 (I8) HUBA GJ, BENTLER PM: The role of peer and adult models for drug taking at different stages in adolescence. J Youth Adol 9:449—465, 1980 (I9) SCHINKE SP, GILCHRIST LD, ASHBY MR, ET AL: Native youth and smokeless tobacco: Prevalence rates, gender differences, and descriptive characteristics. NCI Monogr 8:39—42, 1989 (20) GRITZ ER: Prevention and treatment aspects of smokeless tobacco use. In Health Implications of Smokeless Tobacco Use. National Institutes of Health Consensus Development Conference, vol 6, No. 1. Bethesda, MD: Natl Inst Health Off Med Applications Res, 1986 (21) PSATHAS G: Ethnicity, social class, and adolescent indepen- dence from parental control. Am Sociol Rev 222415—423, 1957 (22) JESSOR R, JESSOR SL1 Problem Behavior and Psychosocial Development: A Longitudinal Study of Youth. New York: Academic Press, 1977 (23) EDMUNDSEN EW, GLOVER ED, ALSTON PP, ET AL: A com- parison of personality traits of smokeless tobacco users and nonusers. IntJ Addict 22:671—683, 1987 (24) CHASSIN L: Adolescent substance use and abuse. In Ad- vances in Adolescent Behavior Analysis and Therapy (Karoly P, Steffen J, eds), vol 3. Lexington, MA: Heath, 1984, pp 99—152 (25) EAKIN E, SEVERSON H, GLASGOW RE: Development and evaluation of a smokeless tobacco cessation program: A pilot study. NCI Monogr 8:95—100, 1989 U.C. BERKELEY LIBRARIES ‘ IIMIR’EWIHIII ' CDE’EDLBLBD slam NCI Monographs _ i . } / 1 Foreword 3 Preface 5 Smokeless Tobacco Use Among Adolescents: A Theoretical Overview 13 Smokeless Tobacco: Association With Increased Cancer Risk 17 Use of Smokeless Tobacco in the United States: Recent Estimates From the Current Population Survey 25 Smokeless Tobacco Use in the United States: The Adult Use of Tobacco Surveys 29 Epidemiology of Smokeless Tobacco Use: A National Study 35 Use of Smokeless Tobacco by Age, Race, and Gender in Ten Standard Metropolitan Statistical Areas of the Southeast United States 39 Native Youth’ and Smokeless Tobacco: Prevalence Rates, Gender Differences, and Descriptive Characteristics 43 Parent Characteristics, Perceived Health Risk, and Smokeless Tobacco Use Among White Adolescent Males . 49 Use of Smokeless Tobacco Among Male Adolescents: Concurrent and Prospective Relationships 57 Activity Involvement, Risk-taking, Demographic Variables, and Other Drug Use: Prediction of Trying Smokeless Tobacco ' 63 Initiation and Use of Smokeless Tobacco in Relation to Smoking 7] Prevalence and Predictors of Smokeless Tobacco Use: Iowa’s Program AgainsteSmoking 79 Marketing Smokeless Tobacco in California Communities: Implications for Health Education 87 Advertising and Promotion of Smokeless Tobacco Products ‘ 95 Development and Evaluation of a Smokeless Tobacco Cessation Program: A Pilot Study 101 Adolescent Smokeless Tobacco Use: Future Research Needs NIH Publication No. 89—3055 Number 8, 1989