key: cord-0006315-s21jrbvn authors: Katsaliaki, K; Mustafee, N title: Applications of simulation within the healthcare context date: 2010-10-13 journal: J Oper Res Soc DOI: 10.1057/jors.2010.20 sha: 1f3b83accf5603ba19f99d05039d1a1d03dcfa98 doc_id: 6315 cord_uid: s21jrbvn A large number of studies have applied simulation to a multitude of issues relating to healthcare. These studies have been published in a number of unrelated publishing outlets, which may hamper the widespread reference and use of such resources. In this paper, we analyse existing research in healthcare simulation in order to categorise and synthesise it in a meaningful manner. Hence, the aim of this paper is to conduct a review of the literature pertaining to simulation research within healthcare in order to ascertain its current development. A review of approximately 250 high-quality journal papers published between 1970 and 2007 on healthcare-related simulation research was conducted. The results present a classification of the healthcare publications according to the simulation techniques they employ; the impact of published literature in healthcare simulation; a report on demonstration and implementation of the studies’ results; the sources of funding; and the software used. Healthcare planners and researchers will benefit from this study by having ready access to an indicative article collection of simulation techniques applied to healthcare problems that are clustered under meaningful headings. This study facilitates the understanding of the potential of different simulation techniques in solving diverse healthcare problems. Healthcare needs grow and healthcare services become larger, more complex and costly (Eveborn et al, 2006; Wand, 2009 ). Moreover, the intrinsic uncertainty of healthcare demands and outcomes dictates that healthcare policy and management should be based on the evidence of its potential to tackle these stochastic problems. It seems apparent that computer modelling should be valuable in providing evidence and insights in coping with these systems. They can be used to forecast the outcome of a change in strategy or predict and evaluate the implications of the implementation of an alternative policy (Wierzbicki, 2007) . The use of modelling in healthcare is not limited to the management of activities necessary to deliver care alone. It is also used for the study of several topics related to healthcare, for example, air pollution, pharmacokinetics and food poisoning. In this paper, we aim at profiling studies that have designed, applied, described, analysed or evaluated healthcare problems with the use of simulation modelling. Computer simulation is a decision support technique that allows stakeholders to conduct experiments with models that represent real-world systems of interest (Pidd, 2004) . It can be used as an alternative to 'learning by doing' or empirical * Correspondence: K research (Royston, 1999) . Furthermore, simulation modelling gives stakeholders the opportunity to participate in model development and, hopefully, gain a deeper understanding of the problems they face. As a result, decision makers and stakeholders can gain a new perspective on the relationships between the given parameters, the level of systems' performance, the cost-effectiveness and its quality, or risk association. In the field of Operations Management, simulation is recognised as the second most widely used technique after 'Modelling' (Amoako-Gympah and Meredith, 1989; Pannirselvam et al, 1999) . Thus far, there have been a number of reviews in the literature on the applications of simulation to health. Fone et al (2003) have conducted a systematic review of the use and value of computer simulation methods in population health and healthcare. reviewed the application of a diverse range of simulation techniques in healthcare settings. Brennan and Akehurst (2000) and Barrios et al (2008) considered the application of simulation in the economic evaluation of health technologies and health products as well as a proposed method for the evaluation of pharmacoecomonic models (Hay, 2004) . Dexter (1999) includes a review of computer simulation and patient appointment systems. A number of reviews have focused on the applications of Discrete-Event Simulation (DES) in healthcare in general (England and Roberts, 1978) , and more specifically in health clinics (Jun et al, 1999) and healthcare capacity management (Smith-Daniels et al, 1988) . gives a personal review of the use of Discrete Event Simulation in health among other fields. However, most reviews limit themselves to either a single application area or/and a single simulation technique. Most of the current reviews lack the breadth of simulation techniques, the width of applications coverage and are published in outlets of different fields (eg medical, OR, health informatics journals, etc), thus potentially hampering the widespread reference and use of such studies. Hence, the purpose of this review is to fill these gaps and categorise and synthesise academic literature pertaining to the use of computer simulation in health problems (a) over a number of unrelated publishing outlets, (b) with a broader scope of simulation techniques and (c) in a variety of health applications. This would, in turn, help in ascertaining the current development in the field of healthcare simulation. In light of the above, by sampling publications pertaining to the application of simulation in the healthcare domain, we hope to realise the following objectives: (1) to classify publications according to the simulation methods they employ; (2) to determine the healthcare problems often investigated by these methods and to analyse their trends; (3) to identify the impact of published simulation research in the healthcare context; (4) to monitor results' demonstration and implementation; (5) to identify funding sources for healthcare simulation studies; (6) to identify software associated with the studies and show their frequency of use. In order to achieve these objectives, we have conducted a review of 251 articles published during the period 1970-2007. The main objective of this review is to offer a broad and extensive picture of the role of simulation techniques in healthcare. To the best of our knowledge, objectives (1) and (2) have not been previously investigated in a single study for all four selected simulation techniques in the health sector, and objectives (3) to (6) have not been presented in a published source-with the exception of England and Roberts (1978) who presented similar results for Discrete Event Simulation and System Dynamics over 30 years ago. It is hoped that the findings of our analysis will be beneficial to the community of simulation and healthrelated academics and practitioners. The remainder of this paper is structured as follows. The next section ('Simulation modelling') provides a discussion of the different simulation methods selected for this study. The methodology employed for the research is explained under the 'Research methodology' section. The section on 'Research paradigm' categorises the applications of simulation under various simulation techniques and healthcare problems-this fulfils objectives (1) and (2) . This is followed by the 'Research impact' section (fulfils objective 3) that identifies some important papers that have been reviewed in our study and measures their impact through a citation-based analysis. The section on 'Results implementation, funding sources and analysis of simulation software' presents statistics pertaining to these variables, and thereby fulfils objectives (4), (5) and (6) . The penultimate section presents a 'Discussion' of the findings of this study, and the paper concludes with 'Conclusions and further reflections' that outline the limitations of our approach and reflect on the contribution of this work. The simulation modelling techniques that were found appropriate for the purposes of this study are Monte Carlo Simulation (MCS), Discrete-Event Simulation (DES), System Dynamics (SD) and Agent-Based Simulation (ABS). Journal papers included in this study have been selected based on the criteria that the papers report on the use of one or more of these simulation techniques in the healthcare settings. The choice of simulation techniques was made through interaction with experts in this area but was also backed by the review of Jahangirian et al (2009) of simulation in business and manufacturing. The latter identifies the following simulation techniques: DES, SD, ABS, MCS, Intelligent Simulation, Traffic Simulation, Distributed Simulation, Simulation Gaming, Petri-Nets and Virtual Simulation, excluding simulation for physical design. According to this study, the first five techniques were the most commonly presented/used in the selected papers for that review. Initially in our study, we also considered papers that reported on the use of Intelligent Simulation and Parallel & Distributed Simulation. However, these categories were later dropped owing to the fact that only a few relevant papers pertaining to the aforementioned categories were found in our sample study (one or two for each category). Moreover, our choice of simulation techniques is further supported by the study conducted by Fone et al (2003) , wherein DES, SD and MCS are discussed as popular simulation techniques in healthcare. Those who wish to have an introduction to the aforementioned techniques can refer to Rubinstein (1981) for MCS, Robinson (1994) for DES, and Sterman (2001) for SD. ABS is the most recent of the four simulation methods used since the mid-1990s. A brief description of ABS is provided below. ABS is a computational technique for modelling the actions and interactions of autonomous individuals (agents) in a network. The objective here is to assess the effects of these agents on the system as a whole (and 'not to' assess the effect of individual agents on the system). ABS is particularly appealing for modelling scenarios in which the consequences on the collective level are not obvious even when the assumptions on the individual level are very simple. This is so because ABS has the capability of generating complex properties emerging from the network of interactions among the agents, although the in-built rules of the individual agents' behaviour are quite simple. In this paper, we have conducted a review of literature in healthcare simulation. Our review method has been influenced by the systematic literature review approach adopted by Eddama and Coast (2008) , wherein (a) databases such as ISI Web of Science ® and MedLine ® were searched using a combination of search terms, (b) papers were screened by reading article titles and abstracts and in accordance to some inclusion criteria, and (c) the contents of the papers selected in the earlier stage were reviewed. Our literature profiling methodology consists of two stages and is illustrated in Figure 1 . Stage 1 is the 'Paper Selection' stage and it describes the methodology used for the purpose of selecting papers for inclusion in this study. Stage 2 is the 'Information Capturing' stage and it identifies the information that is captured from papers that have been included in the study; the latter is analysed in the subsequent sections of this paper. Both the stages of our methodology are further described below. The papers selected for this study were identified from the Web of Science ® database The Web of Science ® is one of the largest databases of quality academic journals and provides access to bibliographic information pertaining to research articles published from 1970 onwards. It indexes approximately 8500 high impact research journals from all around the world spread across approximately 200 different disciplines. Our aim was to identify publications with the highest credibility and thus we looked only at journal articles having an impact factor (note: only journals with an impact factor are included in the ISI Web of Science ® database).We do recognise, however, that other bibliographic databases could have also been looked at. But for the purpose of this research, we decided to include only the Web of Science ® database since this study is not a systematic review but is a sample review of publications in healthcare simulation. The Web of Science ® has a user-friendly search engine that assists in the refinement of a search by allowing the user to incorporate specific search conditions. Our search strategy was driven by the simulation methodology used in the sought after papers. To identify articles that would be incorporated in our study's data set, the following criteria were used: inclusion of the words, 'simulat*' OR 'health*' in the article's title and both of the words/phrases ('Monte SAME Carlo' AND 'health*') OR ('Discrete SAME Event*' AND 'health*') OR ('System* SAME Dynamics' AND 'health*') OR ('Agent SAME Based' AND 'health*') in the abstract or keywords of the published paper. The SAME operator returns records in which the terms separated by the operator appear in the same sentence. The use of the asterisk, '*' in the Boolean keywords combination, allowed for the inclusion of keyword derivatives in the search options. The search identified only articles and review papers written in the English language The second step involved the screening of these papers. The two authors independently and critically reviewed all the abstracts of 251 papers' and read the full text when necessary. The appraisal was carried out based on certain inclusion criteria as follows: The selected papers should evidently demonstrate strong relation with the healthcare sector or have an impact on healthcare and use the chosen simulation method to describe, analyse or assess the situation. The paper should include at least one paragraph describing the applied simulation method that was used in the study. Thus, pure physics simulations and human systems simulations did not fulfil the inclusion criteria. The boundaries between healthrelated papers and non-health-related papers, were not always straightforward. In many papers the impact on human healthcare is provided by a less direct relationship. The reviewers took a flexible approach by including papers in which one could clearly relate the problem described with some kind of health impact. Each of the reviewers assessed all abstracts independently and compared the results were compared. In cases of discrepancies, the full text of the paper was examined and, after discussion between the reviewers, a decision was reached for the paper's inclusion or exclusion. This filtering resulted in a set of 201 relevant papers. The full text papers were collected via online or inter-library loan services. The second stage concentrated on the content of the 201 papers in order to answer the six objectives of our study as identified in the introductory section. Of the selected papers, MCS seems by far to be (69%) the most applied method dealing with health issues. It is followed by DES and SD. Finally, the method with the least number of papers is ABS-this is not a surprise since it is the most recently developed simulation technique. Table 1 (last two columns) lists the results of our screening. The last row of the table ('multiple simulation methods') identifies five papers that use or mention two or more simulation techniques. These ('multiple simulation methods') papers, for simplicity purposes, are described under the research paradigms of the four identified categories as explained in the next section. As this is a sample review, no inferences can be drawn from Table 1 as to the impact of each simulation method in healthcare. Nonetheless, we believe that the statistics below provide the readers with some understanding of the research trends in this area. The papers that have been included in our review are listed in separate tables [Tables 3 -6] . These tables are presented in the relevant sub-sections associated with each simulation technique in question. Every paper has a unique identifier beginning with the initials of the simulation method under which it is categorised (MC, DES, SD, ABS) and is suffixed with a numerical value, for example MC1, MC20, etc. When many papers are listed in a row under the same category, the prefix is entered only at the beginning and is omitted from the rest of the papers for brevity (eg MC11, 27, 81). In the tables, these papers are presented in a descending date of publication order, and this, in turn, shows the research effort over these 37 years. Thus, small numbers correspond to the most recent publications and large numbers to the older ones. The Vancouver reference style is followed. Rather than including the references alphabetically at the end of the paper, we consider this scheme of collecting and tabulating all references pertaining to a particular simulation technique together at the end of each section as important because we feel that it improves the readability of the paper. These tables will also serve as a future reference/study list for the reader. The papers pertaining to the different simulation techniques have been categorised under several general headings/categories. An overview of these categories is presented in Table 2 (objective 1). This is followed by a discussion of the categories under each of the four identified simulation techniques (objective 2). Some papers can be categorised under multiple headings and the decision to favour one classification category over the other was based on the relative importance attributed to specific simulation techniques in the discussion part of the paper. MCS is the most predominantly used simulation technique of the four identified techniques. Of the 163 reviewed papers in MCS, we found 142 to be suitable for inclusion in our dataset (Table 3 ). In the context of healthcare, MCS has generally been used for the following purposes: (a) To assess health risks from exposure to certain elements and determine drug doseresponse portions; this is the most popular sub-category with 60 papers in our sample; (b) as the main approach to modelling used in economic evaluations in healthcare interventions when there is a need to increase the number of states in the model to overcome the homogeneity assumptions inherent in Markov models and decision trees (Barton et al, 2004 ) (18 papers); (c) to evaluate the cost-effectiveness of competing technologies or healthcare strategies that require the description of patient pathways over extended time horizons with 41 papers in this sub-category; and (d) for Miscellaneous taxonomies, literature review and feasibility studies with 23 papers altogether. Each of these four issues will now be looked at in greater depth. Health risk assessment Numerous environmental and occupational studies have shown a link between the measures of public health and intake of contaminants, via different environmental media and exposure routes such as inhalation, skin and ingestion. Twenty-two studies focused on air pollution [MC3, 10, 20, 26, 29, 40, In such health risk assessments or epidemiological studies, the exact amount of a chemical or contaminant that an individual comes into contact with over a lifetime should ideally be estimated. However, for many obvious reasons this estimation is difficult. Simulation studies can fill in data gaps regarding historical exposures by generating these data using parametric functions, which are critical to improving the power of such studies. MCS is the method most commonly used for classical probabilistic risk assessments that uses mathematical or statistical models to estimate the frequency in which an event will occur. This technique is particularly useful when a large number of algorithms are required to address various multipathways of exposure to humans. The use of Monte Carlo analysis has reformed the practice of exposure assessment and has greatly enhanced the quality of the risk characterisation. Moreover, 15 risk assessment studies focus on drug development and dose-response portion [MC4, 14, 17, 19, 31, 42, 47, 52, 53, 54, 67, 71, 80, 106, 137] . MCS can be used to determine the Probability of Target Attainment of pharmacodynamic indices by taking the inherent variation of different populations into account. In MCS, the model parameters are treated as stochastic or random variables, by using a probability density function for example, rather than fixed values. The aim of these studies is to establish a population pharmacokinetic model to study the parameters for the drug being administered through an intravenous escalating dosing regimen in healthy subjects, which could, in turn, be used for design of patient protocols with direct therapeutic benefit and maximal safety. These simulations are dependent on the assumptions in the model, including the types and number of subjects in the pharmacokinetic studies and the data used. Differences in pharmacokinetic parameters (for different patient populations) and/or data can lead to differences in the target attainment rates obtained with these simulations. Studies of these kinds usually derive their data from clinical trials. Prognostic and transmission models of health interventions MCS is extensively used to measure the number and impact of medical interventions for the prevention of disease deterioration or disease transmission. Many intervention procedures with medical treatment show substantial reductions in disease morbidity or mortality. However, their use is expensive and to some extent determined by local practice, with great variation in the rates of these procedures. The optimum level of such procedures may therefore be uncertain, and this uncertainty is a major problem for both clinicians and health service administrators. It is therefore important to have methods that model the requirements for these interventions at the population level by capturing the movement of individuals between different states based on disease and/or procedure history. Such interventions that usually involve patients or disease transmission stages use Markov processes to measure the probabilities of transmission. MCS analysis of the Markov process is the most useful model for this situation, which also allows the enumeration of events as The above research can easily be adapted or expanded to fit economic data, which evaluate the cost-effectiveness of specific interventions, treatments, tests and health programmes. Certain medical conditions have a profound and growing impact on healthcare resource utilisation. In many circumstances the direct expenditures for screening or treatment (with drugs or other therapy) of these conditions have substantially increased due to the overall ageing of the population. Therefore, research in this field tries to assess the economic value of a population-based screen-and-treat strategy for diseases or medical conditions compared to alternative strategies or no intervention [MC6, 7, 9, 12 This is the second most popular category in our study with 40 papers overall after screening (Table 4 ). It is said that DES can create significantly more insight than MCS in areas such as health economics . and work flow, allocation of resources when sizing and planning beds, rooms, and staff personnel) also bear resemblance to our sub-categories in 'Planning healthcare services', as the latter study is focused on a specific area of DES and is more analytic. We now discuss each of our DES categories according to the number of publications identified in each cluster in a descending order. Planning of healthcare services and health interventions DES allows decision makers to effectively assess the efficiency of existing healthcare delivery systems such as hospitals [DES29], to improve system performance or design and to plan new ones in a risk-free and costless environment by investigating the complex relationships among the different model variables (ie rate of arrivals, time spent in the system, etc) and overcoming bottlenecks. The scope of evaluation can be micro in scale, for example by examining resource needs in terms of scheduling staff and measuring bed and equipment capacity at individual clinics, or macro in proportion (healthcare policy for the entire population). DES allows the decision makers to gather insights and obtain approximate results of the differing but competing policies that may be implemented in the future. Moreover, since DES allows the creation of dynamic population-based models, wherein each entity in the simulation represents an individual, the results could indicate the number of people who may be affected by the adoption of a particular strategy. Some As large majorities of the population depend on edible products or by-products from livestock, the health of livestock has a significant effect on public health. Health economic models Health economic models evaluate the health implications and the economic costs of providing healthcare to the population at large. They usually do so by comparing alternative healthcare interventions aiming to maximise welfare through optimal utilisation of the allocated public health funds. With respect to health economic models, the use of DES has been reported for evaluating, among others, the cost of providing dental care to children System Dynamics SD can assist the design of healthcare policies by examining how the fundamental structure might influence the progressive behaviour of a system. It takes into consideration factors such as the time variation of both the tangible elements, such as waiting times and healthcare costs, as well as intangible elements, such as patient anxiety and the effects of various pressures on purchasing decisions (Taylor and Lane, 1998) . Seventeen studies are counted under this technique. The papers pertaining to SD have been categorised under the following headings: (a) public health policy evaluation and economic models, represented in nine papers in our search; (b) modelling healthcare systems and infrastructure disruption (four papers); (c) use of SD as a training tool (three papers); and (d) one review paper of SD for modelling public health matters of disease epidemiology and healthcare capacity [SD6]. The first three categories are described below in the same order as above. The papers are listed in Table 5 . Public health policy evaluation and economic models SD has been applied for the evaluation of several public health policies. With regard to communicable diseases, SD models were developed to estimate the effect of harm reduction policies for HIV/AIDS and tuberculosis (such as 'needle-sharing and injection-frequency among drug users and multi-drug resistant tuberculosis control [SD2]) and to assess economic consequences of testing and treating pregnant women for HIV virus with different regimens to avoid prenatal transmission [SD16]. Moreover, SD was used in several studies to evaluate the long-term health impact of smoking by comparing policies such as increasing cigarette excise taxes, raising the legal smoking age to 21 [SD4] and introducing tobacco harm reduction policies [SD8, 9, 11] . They suggested that a large tax increase would have the largest and most immediate effect on smoking prevalence. Control over the cigarette content would bring a net gain in population health, although 'healthier' cigarettes make smoking more attractive and increase tobacco consumption. SD has also been used by health planners to gain a better understanding of diabetes population dynamics [SD7]; to model the feedback effects of reconfiguring health services [SD10] by shifting towards the primary level and bringing services 'closer to home'; to investigate the impact of privacy legislation in the individual health insurance market and the social costs that are borne when applicants do not divulge private information about their medical conditions [SD14]. Modelling healthcare systems and infrastructure disruptions A healthcare system consists of many individual sub-parts that interact with each other, for example the national health system (NHS) consists of vast numbers of GP clinics, walkin centres, hospitals, tertiary care centres, A&E, IT infrastructure, NHS supply chains, etc. SD allows modelling of several sub-parts of these complex healthcare systems, such as a city's delivery of emergency and on-demand, unscheduled care [SD12], an A&E dynamics of demand pattern, resource deployment and parallel hospital processes [SD15]. In this regard, SD also has the potential to simulate multiple, independent key elements of an infrastructure. Innovative modelling and analysis framework based on SD could study the entire system of physical and economic infrastructures, and specifically of healthcare facilities, and propose public responses to infrastructure disruptions [SD5] and disasters [SD1], as well as to reduce the devastating health effects of such phenomena by modelling into a unified whole the relief effort of evacuations, provision of temporary shelters, restoration of electricity and communication lines, etc. Training SD has also been used as a tool for training health policymakers. It can facilitate the understanding of the dynamics of an epidemic such as SARS [SD3] and explore the applicable combinations of prevention or suppression strategies. Moreover, SD provides an opportunity in some educational environments such as in health sciences by allowing students to experiment in the classroom with the use of professional tools. SD software together with calculator-simulators has been used for teaching pharmacokinetics [SD13], and pharmacological system dynamics models have also been developed for the same purpose [SD17]. Applications of ABS in the healthcare sector are not yet widespread but it has been used to study problems such as the spread of epidemics (Bagni et al, 2002) . The research methodology that we have followed in our review has identified only two papers that have used ABS. The papers are listed in Table 6 . One study reported an ABS model called CancerSIM, which allows researchers to study the dynamics and interactions of cancer hallmarks and possible therapies [ABS1] . The other study [ABS2] used software agents to preserve individual health data confidentiality in micro-scale geographical analyses and showed that by limiting the accuracy of geocodes for the purposes of privacy protection, the ability to identify areas of high disease risk is degraded. The five papers that report on several simulation techniques (refer to Table 1 ) have been included in the MCS and the DES category for the sake of simplicity. Three papers report both on MCS and DES and were described under the 'Prognostic and transmission models of health interventions' [MC48, 58] and the 'Cost-benefit analysis and policy evaluation of medical treatment and disease management programs' [MC65] headings of MCS. Moreover, there are two papers that were described under the 'Review papers' heading of DES. A review paper [DES9] that refers simultaneously to DES, SD and MCS and a taxonomy paper [DES10] that refers to DES and SD among other operational research techniques. In this section, we present the citation statistics of a few highly cited papers in the field of healthcare simulation (objective 3) ( Table 7 ). The table shows the total citations and the average article citations as a means of identifying the impact of these publications. The list is sorted (and therefore publications for inclusion in Table 7 are selected) based on the total citation count. However, the authors recognise that the average citation is also a very useful measure as it eliminates the discrepancies caused by the number of years passed since publication. It is generally expected that review papers have more citations than research papers. It is therefore surprising that none of the papers included in the list are review papers. Even more surprising is the fact that all papers use the MCS technique as their main method Table 7 Publications with high number of citations Average citations Publication of analysis. Many of the papers in Table 7 present costeffectiveness analyses of specific healthcare applications or disease prevention methods, including the first paper that was published in the journal Bone in 1994. It should be noted here that a good number of journals in Table 7 are either medical or health-related journals. It is widely accepted that medical journals generally have citations that are much higher compared to the OR journals, from which it might be concluded that impact is not incomparable between them. A more stratified representation would shed more light. However, this was out of the main scope of this study. In this section, we examine the evidence of results presentation, implementation (objective 4), funding (objective 5) and software usage (objective 6) from among those papers that were selected for inclusion in this study after screening. Of the 201 papers, 184 (91%) present results and have a separate, typically large section supported with tables and graphs to give a full analysis and explanation to the readers. There are seven MCS papers, eight DES, three SDS and one ABS paper, which do not present results. Of these, the majority are review and methodology papers. There are only five papers that fall in other categories (health risk assessment; health economic model; planning of healthcare services) and do not demonstrate results in a numerical format in the way described above. Yet, implementation of research results is hardly mentioned in these publications, with only a few papers (11 out of 201, 5.4%) reporting on the implementation of results to the stakeholder organisations, in which the case studies were based. Six are reported in the MCS category, four in DES and one in SD. However, this is not to say that the case-oriented simulation studies that have not implemented their results have gone astray. Neither should it be implied that their impact is only academic and does not reflect the real world. Looking further at the issue, one may realise that healthcare simulation studies generally have a long gestation period before they reach the ultimate decision makers in a comprehensive format. These decision makers need to decide among a plethora of similar studies, taking into consideration various other factors, and come to a conclusion of turning a specific recommendation from a study into a policy applicable in health organisations and settings. Subsequently, it is unlikely that implementation will be part of the paper. Moreover, researchers are eager to publish once they have the first results in hand and only very occasionally will they wait until the impact of their method is shown in the real world in order to incorporate it into their paper. Perhaps a better measure of the interest in the research being conducted in the healthcare simulation studies is the funding process. Of the 201 studies, 87 (43%) have received full or partial funding. Of the 163 identified MCS studies, around 39% mention their project's funding source, 48% of the DES papers, 65% of the SD papers and 100% of the ABS papers (two papers) report a funding source. Many of these papers refer to various sources of funding. Table 8 illustrates some of these sources. As can be seen from the table, health departments and national foundations are the major sources of funding, closely followed by pharmaceutical companies. Other governmental departments and national institutions also fund healthcare studies. Funds for research are also derived from internal University funding and research council grants. From our sampled list of papers, we find that funding seems to be consistent throughout the years. This suggests that there is no identified trend that more funding is provided for healthcare research over the last years or vice versa. Finally, we conclude by presenting some statistics on simulation software/programming languages that were used to support model development in the selected studies. It is important to mention that, from our sample of 201 selected papers, only 83 papers acknowledge the software or programming language that was used to develop the model. This data is presented in Table 9 (MCS software), Table 10 (DES software) and Table 11 (SD software), respectively. With regard to MCS (Table 9 ), @Risk and Crystal Ball were among the most popular software, followed by Excel. Numerous other software and programs have also been used, some of them specific to health or other applications. The process of building DES models involves some form of software. The software can either be a high-level programming language or a Commercial, Off-The-Shelf (COTS) simulation package. DES software Arena is the most popular in this sample review, followed by the programming language Borland Delphi and COTS package Simul8 (Table 10) . As for SD, the use of only few types of software is reported. Vensim is first in the list, closely followed by STELLA. DYNAMO comes last (Table 11) . Finally, one of the two ABS papers reported the use of the programming language C++ to create CancerSIM. In general, the rapid growth in simulation software technology has created numerous new application opportunities, including more sophisticated implementations, as well as combining simulation and other methods for complex models and processes. Trends from our data analysis suggest that, in the most recent years, COTS packages have taken the lead over one-off models that are coded using programming languages. This is explained by the fact that COTS simulation packages are rapidly evolving through inclusion of more advanced features (eg 3-D graphics, parallel processor support, etc). The field of healthcare simulation has evolved significantly over the past 30 years. A great number of health problems have been approached with simulation techniques, which have offered greater precision with regard to resource allocations, evaluations between health strategies and risk assessments. In this review paper reflecting on 37 years of healthcare simulation, we see some trends that apply to the discipline as a whole. Looking first at the statistics of our sampled papers, we could derive the conclusion that the proportion of papers published in the field has drastically increased, with more than three-quarters published after 2000. Annual paper contributions amounted from one paper in 1988 to 36 in 2007. It is, however, surprising that the oldest paper in our data set is from 1988 as our search strategy concentrated on identifying healthcare simulation papers published from 1970 onwards. One reason for this is possibly that the number of journals indexed by ISI WoS has swelled with the rising popularity of the Internet and the availability of electronic bibliographical information (this may not have been the case during 1970s-1980s). Furthermore, it is arguable that although simulation has been applied to manufacturing, defence, supply chains etc., for a long time, its application in the healthcare context is comparatively new. Figure 2 illustrates the historical trends of the healthcare modelling papers for each simulation technique (the only exception is ABS which has only two papers). The ascending lines show the increasing number of published papers in the field especially after the mid-1990s for all three simulation methods. This is in line with the clear increase in simulation usage in the general service sector from the 1990s onwards (Robinson, 2005) . Year-to-date figures suggest that this gradual upward trend will continue. It is apparent that during the last 4 years the published papers in this field have drastically increased. A reason might be the possible increase of funding in recent years (Murphy and Topel, 2003) . Simulation as a technique in health problems is used both as the main methodology of the research and as a supportive method to evaluate the robustness of other methods in different papers. MCS seems to be the most popular simulation technique in health studies, and the majority of papers fall within the health risk assessment category. In this category studies pertaining to air and water pollution, food poisoning and soil contamination are leading in terms of published papers, and drug development and dose-response portion studies follow. Cost-benefit analyses health studies with the use of MCS are also popular. They assess the economic value of population-based screen-and-treat alternative strategies for diseases and medical conditions. Some of these studies hold the first positions in terms of research impact and are found to have the maximum average number of citations in our dataset. Moreover, it is particularly noticeable that of the 142 MCS papers, none were published in an OR journal (as defined by the Association of Business School-ABS list). One reason for this may be that MCS is extensively used by health professionals/academics who wish to publish in health-related outlets, or that OR academics have lost interest in the use of MCS and have focused in the use of other simulation techniques to tackle health problems. Nevertheless, several of the MCS papers identified in our study would fit the aim and scope of OR journals. For example MC7, 8, 9, 23, 25, 26, 2, 30 , 32, 38 and many more. In the analysis of the research paradigms categories, it is obvious that some overlap exists among the health applications examined by simulation technique. A very apparent example is that all simulation techniques deal with screening strategies and cost-benefit analysis of medical interventions. Assuming that the categorisation of papers was made according to the health problem tackled and regardless of the simulation technique employed, the papers of cost-benefit analysis would be at about the same level of the health risk assessment category. However, many researchers will agree that, although the application area is the same, the extent, the level and the detail at which this is examined differs according to the technique employed. SD takes a holistic approach and thus the health problem or situation is looked at from a more global level to a greater extent. Consequently, this technique is appropriate for facilitating health policy making at the macro-level. DES and ABS examine the health problems in more detail (micro-level), taking into account the properties of individual entities, yet this restricts the extent of the system that can be modelled. Therefore, decisions can usually be reached with the use of DES and ABS only at the operational level. Monte Carlo simulation incorporates the random sampling element at aggregated level, which makes modelling of population-based diseases easy to handle. When the individual aspect is important then DES is more appropriate. Moreover, DES and SD are more suitable for modelling problems in which the time element plays a significant role, such as utilisation of health services' resources and bed/equipment capacity management. Nonetheless, looking at the categories presented in this study, one can see that health risk assessment is pertinent to MCS modelling; planning of health services is most of the times handled with the DES models (and less with SD); and training of health students and managers is prevalent in the SD approach. Unfortunately, we could not make a distinct category for ABS since the sample was so small. Moreover, a year-by-year analysis of the number of papers in each research paradigm showed that there are no chroni gaps in the identified categories, and for that reason published research in these general fields are continuous. Relatively few of the published healthcare simulation articles reported significant effects that simulation had on the healthcare system being studied. This may imply that, although authors document the model, the issues they model and the model results, there are few real implementation results to report. England and Roberts (1978) implied that the reasons behind this are either inadequate models that cannot quantify the impact of the human factor, or the diversity of authority in healthcare facilities, which thwarts the simplicity of a single administrative decision to change the system. The latter problem lies mostly in the political sphere. However, governmental bodies and other national or local council/agency fund a considerable number of studies (43% in our review). In terms of the modelling approach, it seems that the use of COTS packages is quite widespread, although many models are still being developed in high-level programming languages that usually have larger capabilities in accommodating complex behaviours of the system modelled. Yet, the ease of use that is offered by COTS simulation packages allows those who are not computer programmers to develop valid simulation models. This gives the opportunity to a number of people, including some stakeholders of the systems under question to engage in modelling and quantify their problems and the impact of alternative actions. However, in this way, limitations to the models are posed not only by the data availability and the computer operating cost but also by the imagination and capabilities of the modeller and the software. Simulation software costs can be high, yet since the mid-1990s, a number of low cost COTS packages have come to the market. The latter have certainly widened access to simulation (Robinson, 2005) . It is widely accepted that one of the most important results of computer simulation in healthcare, as well as in other sectors, is the increased understanding of the systems being modelled, which results from constructing the models. We hope that in the future it will become more imperative that healthcare modellers seek close ties and cooperation with healthcare administrators to ensure utilisation and implementation of the worthwhile models that are developed. However, the exact same anticipation was expressed some 30 years ago (England and Roberts, 1978) . As stated by Robinson (2005) , simulation techniques have all followed separate paths in both research and practice until now. A closer integration among simulation techniques conjoined with advances in computing and inclusion of the World Wide Web could lead to the development of better designed models with faster execution times, high level of graphics and, most importantly, enhanced user interaction. Such an advance will be in line with the requirements of the new computer literate generation of users. This is a sample review of healthcare simulation studies, which aims at identifying healthcare problems that are modelled using four popular simulation techniques, namely MCS, DES, SD and ABS. The specific selection criteria of articles that were reviewed here may have left out a number of noble publications in the field (eg articles that do not mention health in their title topic but refer to health problems with more specific terms such as hospitals, patients, etc; articles that did not appear in journals indexed by ISI Web of Knowledge ® ). The implications of this are that there may be an unintentional bias introduced by the specific keywords search and by ISI WoS membership, which leaves out newer journals that have not yet met the 'duration of service' required by the ISI WoS and journals where editorial boards do not wish their journal to have an impact factor. These factors may therefore not be taken into account when basing quality on impact factors. However, the debate as to whether this is right or wrong is outside the scope of this article. We merely wish to provide an analysis of literature within the scope of journals with impact factors and therefore provide some reflection as to the 'health' of healthcare simulation within a potentially metric-driven world. We hope that this study gives an indication of the pulse of research being conducted in the healthcare simulation field, although generalisation of the results may not hold. Future research could involve a systematic review of the field including all relevant journals from various academic databases and investigate the relationships between impact factor and non-impact factor journals. This approach could more accurately map the discipline and provide us with statistics of interesting variables similar to the ones presented here and with additional ones, such as popular journals, productive institutions and frequently published authors. Future research could also broaden the scope of our literature review by profiling health-related research with the use of other OR/MS techniques. For the benefit of healthcare and simulation audience, this paper provides an overview of research published in various journals from across different subject areas in health. This research is likely to help authors, reviewers and editors to better understand the potential of different simulation techniques for solving diverse healthcare problems and can also assist upcoming researchers in developing an appreciation of this research area and the various issues considered worthy of research and publication. Furthermore, we hope that healthcare planners, management engineers, as well as researchers will benefit from this study, by having ready access to an up-to-date, indicative collection of articles describing these applications. Finally, our study is likely to stimulate researchers to explore other research areas by undertaking comparative/cross-journal studies. Monte Carlo simulation of animal-product violations incurred by air passengers at an international airport in Taiwan A SAS/IML program for simulating pharmacokinetic data A clinically based discrete-event simulation of end-stage liver disease and the organ allocation process An analysis of data from two general health surveys found that increased incidence and duration contributed to elevated prevalence of major depression in persons with chronic medical conditions Using simulationto assess the sensitivity and specificity of a signal detection tool for multidimensional public health surveillance data Pharmacoeconomic assessment of oseltamivir in treating influenza-the case of otherwise healthy Danish adolescents and adults Human health risk assessment of naturally occurring radioactive materials in produced water-A case study Treatment costs in Canada of health conditions resulting from chronic hepatitis B infection Advances in risk-benefit evaluation using probabilistic simulation methods: an application to the prophylaxis of deep vein thrombosis Health economics studies assessing irbesartan use in patients with hypertension, type 2 diabetes, and microalbuminuria Estimating the demand for health care with panel data: a semiparametric Bayesian approach. Health Econ Improved radial dose function estimation using current version MCNP Monte-Carlo simulation: Model 6711 and ISC3500(125)I brachytherapy sources Health system costs of out-of-hospital cardiac arrest in relation to time to shock Incorporation of statistical uncertainty in health economic modelling studies using second-order Monte Carlo simulations Assessing cost-effectiveness-Mental health: introduction to the study and methods Population pharmacokinetics of APOMINE (TM): A meta-analysis in cancer patients and healthy males Cardiovascular health and economic effects of smoke-free workplaces Costs and net health effects of contraceptive methods Predictive assessment of fish health and fish kills in the Neuse River Estuary using elicited expert judgment Parametric analysis of intercellular ice propagation during cryosurgery, simulated using Monte Carlo techniques A quantitative risk assessment of waterborne cryptosporidiosis in France using second-order Monte Carlo simulation Quantifying human health risks from virginiamycin used inchickens How robust are health plan quality indicators to data loss? A Monte Carlo simulation study of pediatric asthma treatment An environmental decision-making tool for evaluating ground-level ozone-related health effects Use of Monte Carlo simulation to design an optimized pharmacodynamic dosing strategy for meropenem Public health, GIS, and spatial analytic tools Cost-effectiveness of recombinant versus urinary folliclestimulating hormone in assisted reproduction techniques in the Spanish public health care system Cost benefit of influenza vaccination in healthy, working adults: an economic analysis based on the results of a clinical trial of trivalent live attenuated influenza virus vaccine Bayesian analysis of a self-selection model with multiple outcomes using simulation-based estimation: an application to the demand for healthcare A simulation model for estimating direct costs of type 1 diabetes prevention Design techniques for stated preference methods in health economics Oseltarnivir for treatment of influenza in healthy adults: Pooled trial evidence and cost-effectiveness model for Canada Clinical trial simulation using therapeutic effect Modelling: Application to ivabradine efficacy in patients with angina pectoris Breaking out of the silo: One health system's experience. American Journal of Health-System Pharmacy In silico toxicology: Simulating interaction thresholds for human exposure to mixtures of trichloroethylene, tetrachloroethylene, and 1,1,1-trichloroethane. Environ. Health Perspect Using Monte Carlo simulation in life cycle assessment for electric and internal combustion vehicles Cost effectiveness of influenza vaccination for healthy persons between ages 65 and 74 years Chloroform associated health risk assessment using bootstrapping: A case study for limited drinking water samples A Monte Carlo simulation for modelling outcomes of AIDS treatment regimens Health risk assessment on residents exposed to chlorinated hydrocarbons contaminated in groundwater of a hazardous waste site Simulated effect of tobacco tax variation on Latino health in California Monte Carlo simulation of nitrogen oxides dispersion from a vehicular exhaust plume and its sensitivity studies Simulation and assessment of subsurface contamination caused by spill and leakage of petroleum products-A multiphase, multicomponent modelling approach Health and economic consequences of HCV lookback An event-by-event probabilistic methodology for assessing the health risks of persistent chemicals in fish: A case study at the Palos Verdes Shelf Cost-benefit analysis of a strategy to vaccinate healthy working adults against influenza Penetration of particles into buildings and associated physical factors. Part I: Model development and computer simulations A probabilistic model for silver bioaccumulation in aquatic systems and assessment of human health risks Life expectancy as a summary of mortality in a population: statistical considerations and suitability for use by health authorities The use of the Tobit model for analyzing measures of health status The use of Monte Carlo simulation to examine pharmacodynamic variance of drugs: fluoroquinolone pharmacodynamics against Streptococcus pneumoniae A case study of stochastic optimization in health policy: Problem formulation and preliminary results A model of the health and economic impact of posttransfusion hepatitis C: application to cost-effectiveness analysis of further expansion of HCV screening protocols Cost effectiveness of human immunodeficiency virus postexposure prophylaxis for healthcare workers A Monte Carlo simulation study to investigate the potential of diffraction enhanced breast imaging A stochastic model simulating the feeding-health-production complex in a dairy herd Analytic approaches based on life expectancy and suitable for small area comparisons Quasi-REML' correlation estimates between production and health traits in the presence of selection and confounding: A simulation study Health risk assessment of a modern municipal waste incinerator Public health sealant delivery programs: Optimal delivery and the cost of practice acts Modelling and improving emergency department systems using discrete event simulation. Simulation-Transactions of the Society for Modelling and Simulation International Modelling the economic and health consequences of managing chronic osteoarthritis pain with opioids in Germany: comparison of extended-release oxycodone and OROS hydromorphone Emerging methods in economic Modelling of imaging costs and outcomes: A short report on discrete event simulation Choice of modelling technique for evaluating health care interventions A simulation study of scheduling clinic appointments in surgical care: individual surgeon versus pooled lists Using simulation to improve the blood supply chain Combining Data Mining and Discrete Event Simulation for a value-added view of a hospital emergency department Simulation modelling in healthcare: reviewing legacies and investigating futures A taxonomy of model structures for economic evaluation of health technologies. Health Econ Montgomery Countys Public Health Service uses operations research to plan emergency mass dispensing and vaccination clinics. Interfaces Forty years of discrete-event simulation-a personal reflection Modelling the economic and health consequences of cardiac resynchronization therapy in the UK Modelling the health benefits and economic implications of implanting dual-chamber vs. singlechamber ventricular pacemakers in the UK Modelling the treated course of schizophrenia: Development of a discrete event simulation model Cervical screening programmes: canautomation help? Evidence from systematic reviews, an economic analysis and a simulation modelling exercise applied to the UK A clinically based discrete-event simulation of end-stage liver disease and the organ allocation process Planning health services with explicit geographical considerations: a stochastic location-allocation approach Use of discrete-event simulation to evaluate strategies for the prevention of mother-to-child transmission of HIV in developing countries Discrete event simulation of emergency department activity: A platform for system-level operations research Reorganizing the system of care surrounding laparoscopic surgery: A cost-effectiveness analysis using discrete-event simulation Alternative decision modelling techniques for the evaluation of health care technologies: Markov processes versus discrete event simulation. Health Econ Mother-to-child transmission of HIV: a simulation-based approach for the evaluation of intervention strategies Simulation of single start station for Edmonton EMS Towards incorporating human behaviour in models of health care systems: An approach using discrete event simulation Modelling the public health response to bioterrorism: Using discrete event simulation to design antibiotic distribution centers The evaluation of screening policies for diabetic retinopathy using simulation Modelling and analyzing a physician clinic environment using discrete-event simulation Patient-centered simulation to aid decision-making in hospital management Discrete event simulation in the health policy and management program Simulating economic factors in adjuvant breast cancer treatment Planning resources for renal services throughout UK using simulation A simulation model of the cost of the incidence of IDDM in Spain Medical resident work schedules: Design and evaluation by simulation Modelling Discrete event simulation to evaluate screening for diabetic eye disease A veterinary practice simulator based on the integration of expert system and process Modelling Simulating Health Systems-Modelling Problems and Software Solutions Modelling Patient Flows and Resource Provision in Health Systems A Discrete-Event Simulation of the Mcdonnell Douglas Health Information-Systems Online Executive Simulation as a tool to assess the vulnerability of the operation of a health care facility Impact of joined-up HIV harm reduction and multidrug resistant tuberculosis control programmes in Estonia: System dynamics simulation model Teaching through simulation: Epidemic dynamics and public health policies. Simulation-Transactions of the Society for Modelling and Simulation International Limiting youth access to tobacco: Comparing the long-term health impacts of increasing cigarette excise taxes and raising the legal smoking age to 21 in the United States. Health Policy Toward Modelling and simulation of critical national infrastructure interdependencies System dynamics Modelling for public health: Background and opportunities Understanding diabetes population dynamics through simulation Modelling and experimentation Closing the youth access gap: The projected health benefits and cost savings of a national policy to raise the legal smoking age to 21 in the United States. Health Policy Estimating the health impacts of tobacco harm reduction policies: A simulation Modelling approach Modelling the feedback effects of reconfiguring health services Federal policy mandating safer cigarettes: A hypothetical simulation of the anticipated population health gains or losses Emergency and on-demand health care: modelling a large complex system Pharmacokinetic software for the health sciences-Choosing the right package for teaching purposes Managing the costs of informational privacy: Pure bundling as a strategy on the individual health insurance market Looking in the wrong place for healthcare improvements: A system dynamics study of an accident and emergency department HIV screening and treatment of pregnant women and their newborns: Asimulation-based analysis A Dynamo Application of Microcomputer-Based Simulation in Health-Sciences Teaching Preventive Eye Care in People with Diabetes is Cost-Saving to the Federal-Government-Implications for Health-Care Reform Monte-Carlo Techniques for Quantitative Uncertainty Analysis in Public-Health Risk Assessments Cost-benefit analysis of a strategy to vaccinate healthy working adults against influenza The use of Monte Carlo simulation to examine pharmacodynamic variance of drugs: fluoroquinolone pharmacodynamics against Streptococcus pneumoniae New developments in exposure assessment: The impact on the practice of health risk assessment and epidemiological studies Human Interindividual Variability-a Major Source of Uncertainty in Assessing Risks for Noncancer Health-Effects Constructing confidence intervals for cost-effectiveness ratios: An evaluation of parametric and non-parametric techniques using Monte Carlo simulation A comparison of simulation models applied to epidemics Discreteevent simulation models in the economic evaluation of health technologies and health products Modelling in the economic evaluation of health care: selecting the appropriate approach Modeling in health economic evaluation: What is its place? What is its value Design of appointment systems for preanesthesia evaluation clinics to minimize patient waiting times: A review of computer simulation and patient survey studies A systematic review of the use of economic evaluation in local decision-making Simulating economic factors in adjuvant breast cancer treatment Simulation modelling in healthcare: Reviewing legacies and investigating futures Applications of computer simulation in health care LAPS-CARE-an operational system for staff planning of home care Systematic review of the use and value of computer simulation modelling in population health and health care delivery Evaluation and review of pharmacoeconomic models Forty years of discrete-event simulation-A personal reflection Simulation in manufacturing and business: A review Application of discrete-event simulation in health care clinics: A survey The economic value of medical research Operations management research: An update for the 1990s Systems Modelling: Theory and Practice Successful Simulation: A Practical Approach to Simulation Projects Discrete-event simulation: From the pioneers to the present, what next? The use of fractional polynomials to model continuous risk variables in epidemiology Capacity management in health care services: Review and future research directions System dynamics modelling Simulation applied to health services: Opportunities for applying the system dynamics approach The convergence of health care expenditure in the US states Modelling as a way of organising knowledge Acknowledgements-One of the authors was employed as a research fellow in Warwick Business School while working on this paper, and wishes to thank the School for supporting this research. We also thank Dr Simon J. E. Taylor for his comments, which have improved the paper.