Richter A, Mauskopf JA, Research Triangle Institute, Research Triangle Park, NC, USA

In this workshop we will focus on Monte Carlo disease simulations and how they can be used to perform economic evaluations of health care interventions. Monte Carlo disease simulation is a modeling technique that operates on a patient level basis, explicitly estimating the effect of variability among patients in both underlying disease progression patterns and in individual responsiveness to treatments. Typical outputs from these simulations are patient functional status, life years, quality-adjusted life years, and associated costs, all of which can be appropriately discounted. The output information is presented in the form of distributions which can be used to estimate mean or median values and confidence intervals for the outcomes of interest. These results can be used to compute cost-effectiveness ratios and other drug value measures. Monte Carlo disease simulation also allows decision makers to address the question of risk associated with smaller populations who may not tend to the "average" results generated by Markov models or simulations of large populations. In this workshop, we describe how to create a Monte Carlo simulation model and how different types of uncertainty can be incorporated into the model. We will develop a basic Monte Carlo simulation of HIV/AIDS in @Risk showing the step-by-step creation of the model. We will run the simulation and show the typical outcomes that can easily be tracked by this type of simulation. This workshop should be beneficial to outcomes researchers and health care decision makers who need to incorporate uncertainty about the natural history of a disease and the impact of alternative disease management strategies for individual patients into their drug value analyses.

Arcelus A 1, Bhattacharjya A 2, Longo AE 1, Kosik KS 3, Finkelstein SN 4, 1Analysis Group/Economics, Cambridge, MA, USA; 2Janssen Research Foundation, Titusville, NJ, USA; 3Harvard Medical School, Boston, MA, USA; 4Massachusetts Institute of Technology, Cambridge, MA, USA

The model was developed to describe and simulate the natural history of Alzheimer's Disease over a two-year period using an innovative decision database methodology. The decision database models the path-dependent nature of Alzheimer's Disease, whereby a patient's disease outcomes are dependent on the specific secondary conditions that she experiences. The decision model tracks several aspects of Alzheimer's Disease including disease stage, participation in an intervention program, side effects, secondary conditions, clinical complications, institutionalization, and disease progression. This model reports the number and percent of patients experiencing disease outcomes and the total direct and indirect costs associated with treating those patients. The decision database methodology overcomes the "memory-less" limitation of Markov-based models, the extensive time requirement of Monte Carlo models, and the inability of decision tree software to deal with large, complex models. This workshop is designed to provide an overview of the decision database methodology and will focus on the methodology's ability to analyze the path-dependent nature of Alzheimer's Disease. The workshop will also describe the model development process including identification of relevant parameters, model structure, data sources, and application software. In addition, the workshop will demonstrate the decision database software's ability to create customized user interfaces and reports. The decision database methodology enables researchers to examine clinical and economic aspects of Alzheimer's Disease using a rigorous, probabilistic tool that is able to analyze the interaction and path-dependent nature of disease variables. This methodology can be applied to the study of other diseases.

Hasford J, Sangha O, Department of Medical Informatics, Biometry and Epidemiology, University of Munich, School of Medicine, Munich, Germany

The validity of pharmacoeconomic analyses relies on the quality and appropriateness of the medical data sources used. Randomized trials, meta-analyses, epidemiological and observational studies will be presented and the pro's and con's discussed with special emphasis on topics like internal and external validity, comparator treatments, surrogate measures of outcome, assessment of adverse reactions, the role of diagnostics, and censored observations. Analyses of our own and examples from the literature will be presented to describe the pitfalls when using common medical data sources. We aim firstly to sensitize the economist about shortcomings when medical data are used for purposes they were not originally collected for, and secondly to enable physicians to assess critically pharmacoeconomic studies, that are based on empirical data.

Reibnitz C, Silva Saavedra M, School of Public Health, University of Bielefeld, Bielefeld, Germany

The analytic hierarchy process (AHP) is a decision analysis technique to evaluate complex multiattributed alternatives with conflicting objectives among one or more scores. The AHP uses judgements from the decision maker and hierarchical decomposition to derive a set of ratio-scaled utility measures for decision alternatives. With the AHP there is no need to estimate an utility function explicitly since the AHP deals with stated preferences at each step. The model may be used to give a stage 1 breast cancer patient the chance to discover the full extent and weight of the decision in selection of breast cancer treatment, though similar models could be built for any type of treatment e.g. cost benefit analysis. The model allows for the opinions of the physician and the legitimate concerns of the patient. Another problem inherent in the medical field is the lag between new treatments and their long-term statistical confirmation. For this reason, it's imperative that physician and patient alter the relevant weights given the advancements in medical knowledge, particular situational needs and patient concerns. The issues of concern in the model were identified to be the total of psychological effects from the treatment, expected length of time that the patient can reasonably expect to live a normal life after a surgery decision is acted upon, the possibility for, the quality of preservation of the breast tissue and surrounding structure, side effects which may develop as a result of the surgery, patient's background, medical status and cost. Though this model evaluates general treatment options of surgical, non-surgical, a combination, or no treatment, specific alternatives of therapies could also be analyzed.

McGhan WF, University of the Sciences at Philadelphia. Philadelphia, PA, USA

The objective of this workshop is to demonstrate how interactive software programs are useful tools in health care decision-making as well as educational tools on specific diseases and the decision-making process. In this workshop, a specific disease, GERD will be used as the example. Decision support systems are tools that allow systematic approaches to decision making under conditions of uncertainty. There are 4 steps in decision system development: 1) identify and bound the decision problem, 2) structure the decision problem over time, 3) characterize the information needed to structure the decision, and 4) perform the analysis to reveal the preferred course of action. In this workshop a decision tree analysis will be demonstrated. The branches of the tree describing the clinical pathway treatment options will be discussed. The decision nodes (treatment options), probability nodes (healing rates), and resulting outcomes will be demonstrated. Sensitivity analysis (changing one variable such as one of the costs, or changing two or three variables at one time) will be demonstrated. Interactive software programs have the additional benefit as educational tools to self-instruct on the clinical issues of a specific disease. Attendees will see how the use of photos, graphics, and dictionaries within the decision support software program can enhance the value of the program as a user-friendly educational tool.

Langley P, Program in Pharmaceutical Economics and Health Systems Research, University of Colorado Health Sciences Center, Denver, CO, USA

Managed care groups, physician management organizations and independent practice associations in the United States are increasingly seeing formulary submission guidelines as a key management tool for cost and quality control within health systems. The purpose of this workshop is to describe the process of guideline development, the analytical and evaluative standards required from those making submissions and the implementation of guidelines within managed care health systems. The focus is on the guidelines recently put in place by Blue Cross Blue Shield of Colorado, Nevada and New Mexico (with over 0.5 million members) which were developed at the University of Colorado, and similar guidelines in the process of development for the Academy of Managed Care Pharmacy in Washington, DC. Of particular importance in this presentation are the analytical and evaluative standards required in the guidelines. While the guidelines are similar in a number of respects to the revised November 1995 Australian guidelines, the central requirement is for projections to be made of the net cost and outcomes impact on a treating population of the introduction of a new drug or therapy intervention in disease or treatment areas. This is described as a systems impact assessment, which is quite different from the analytical focus of traditional pharmaceutical economics. Rather than base formulary decisions on implications drawn from synthetic decision models, the guidelines ask for cost and outcome impact assessments which are, in principle, verifiable. This means they can serve as a basis for monitoring health impact assessments against targets set and in the context of disease or care management interventions.

Stratchounski L, Rozenson O, Smolensk State Medical Academy, Smolensk, Russia

Pharmacoeconomic data are extremely useful for the development of the optimal treatment strategies in terms of clinical efficacy, safety and cost of drug therapy. Although several pharmacoeconomic trials have been already conducted in Russia, these data are quite scare in our country. In the majority of cases foreign data are the only available source of information. The main advantages of these data are their methodology, design and modern approaches to data analyses and interpretation. However, Russian doctors and Health authorities should be very careful in using data obtained abroad. As far as we are concerned, there are at least four principal limitations for these data in Russia. Firstly, there is a lot of variations in epidemiology of the diseases among different countries. Second reason is a difference in drugs prices, for instance, cefotaxime 1 g costs 2.44 USD in Russia vs 11.7 USD in USA. Thirdly, the costs of medical service and hospitalization are much more cheaper in Russia than in western countries, e.g. cost per day in ICU in Russia is about 100 USD (Smolensk Regional Hospital) whereas in USA it is about 1000 USD (Millard Fillmore Suburban Hospital, Williamsville, NY). It may explain to some extent the lack of sufficient attention paid by physicians to the duration of hospitalization. The total cost of treatment nowadays in Russian predominantly depends on the drug therapy cost. Finally, it is difference in sources of financing of health service between countries. Russian health care system is predominantly funded by the Government and/or insurance companies. Moreover, there are often different source of financing within one region.

Vorobyev PA 1, Aksyuk AV 1, Yakimov OS 1, Novolodsky VM 1, Kuzin VF 1, Kobina SA 2, 1Moscow Medical Academy, Laboratory of Outcomes and Standardization Research, Moscow, Russia; 2Rhone-Poulenc Moscow Representative office, Pharmacoeconomics Department, Moscow, Russia

Since 1998, in Russia the standardization system has been functioning in the public health system, which corresponds with the state system of standardization and with the international requirements in this field. This system is to solve the following major problems: - to assure that citizens shall exercise their rights to generally available free of charge medical care of high quality; - to provide the compulsory health insurance system in Russia with a standardizing base; - to provide the certification and quality assessment system to be set up in the health public system with standards and specifications. At the first stage of development, 16 groups of standardization objects were identified and the basic principles were formulated. The most important principles are a consensuality principle (i.e. a mutual agreement between all the parties engaged in the development and use of standards), an actuality principle (i.e. compliance of the elaborated documents with legislative and other international and state requirements), a complexity principle (i.e. solution of problems of standardization in all spheres of citizens' health care by the uniform methods). At the second stage, a working program on the establishment and development of a public health standardization system in Russia was worked out and approved, which highlights the priorities of work, defines a list of priority regulatory documents and executives of this work. Particular emphasis is placed on a group of system-forming documents, a group of documents that regulate the terms, conditions, and quality of the medical services rendered, a group of documents pertaining to drug provision, medical equipment, informational technologies, etc. Assimilation of the results of new studies in pharmacological economy and evidence-based medicine will be a very important element of development of the documents that regulate the rendering of medical services. The specialists of the pharmaceutical company "Rhone-Poulenc" whose collaboration helps to introduce mathematical programs and informational technologies into the development of a program of standardization program were among the first to initiate new pharmacoeconomic and pharmacoepidemiological developments. At the third stage it is envisaged to establish a standardization service in the branch, which, in addition to regulatory documentation development proper, will be engaged in putting these standards and specifications into practice, in analyzing the results on the introduction of specific regulatory documents. The I.M.Sechenov Moscow Medical Academy (its Laboratory of Outcomes and Standardization Research Problems) has been appointed to be the central research institution to deal with standardization problems. It is entrusted with the duties of a coordinator in research developments, an expert in the existing documents and a developer of a lacking standardizing base. It is scheduled that in 1999, a conference is to be held and a scientific journal in standardization problems will begin to be issued.

Rovira J 1,2, Brosa M 2, Abbas I 2,1 1University of Barcelona, Barcelona, Spain, 2Soikos, Barcelona, Spain

This workshop is intended for individuals who have a basic understanding of the concepts and terminology of economic evaluation of health care who wants to discuss about applied standarization methodologies. Even several software tools are available to assist in the development of pharmacoeconomic models, the lack of a standarised set of rules to present economic evaluation results may limit the comparability of studies carryed out by different analysts. The objective of this session is to define a set of standarized procedures to improve the comparability of results of economic evaluation studies by means of user friendly computer software for Reporting Economic Evaluation Results (the REER software). Two different prototypes will be presented: An application aimed to present economic evaluation results, and a health costs database (HCD). The main program (REER) program is devided in 3 basic modules: 1) Definition of options (treatment/intervention comparators, subjects), 2) Data inputs (resources, unit costs, health outcomes, discounting and sensitivity analysis), and 3) Ouputs (differences in costs and effects, dominance analysis, incremental analysis and sensitivity analysis). A prototype version of the REER software will be presented by means of two published pharmacoeconomic analysis. The Health Costs Database Software (HCDS) enables to collect, categorize, and store unit costs data to be used in different types of studies (e.g. costs analysis, economic evaluation, etc). The data base, which simplifies the retrieval of average and actualized cost figures, makes rapidly available updated unit cost data which can be used directly (e.g., within the REER software), as a first calculation, or be contrasted with later cost estimations. Problems to obtain standarised results of economic evaluation studies and the usefullness of programmes like the REER software will be examined, and future development of complete and fully operational software tools will be discussed.

Evans JMM, Ruta DA, Davey PG, Morris AD, The DARTS / MEMO Collaboration, Dundee University, Scotland

The DARTS database is a validated, population-based register of 9,000 patients with diabetes in the population of Tayside, Scotland, compiled by the record-linkage of independent datasets. These include diabetes clinic records, hospital activity data, biochemistry results and a database of dispensed prescriptions. The database can be used for epidemiological, outcomes and health services research. However, diabetes is a disease with high morbidity and mortality, for which treatments are onerous and time-consuming. Thus the psychological or non-clinical dimension should not be ignored, and the incorporation of quality of life measures into DARTS has been proposed. A wide range of disease-specific and generic measures that tap into many different facets of quality of life have already been used in diabetes. The challenge is to identify a valid and reliable package of measures that could practicably be used for the diabetic patients of Tayside. In this workshop, the approach adopted to identify such a package for routine use in DARTS will be described and discussed. It will include a critical review of quality of life measures that have already been used in diabetic patients, the results of a survey of diabetic patients and diabetic health professionals to identify those aspects of quality of life deemed to be particularly relevant to diabetes, and a discussion of the validation of the measures chosen in a pilot group of patients. The workshop would thus be of interest to researchers working with quality of life outcome measures, or those involved in the development and application of disease-specific databases.

Crawford B 1, Evans C 2, Abetz L 3, 1MAPI Values, Boston, MA, USA; 2Astra Pharmaceuticals, Westborough, MA, USA; 3MAPI Values, Cheshire, UK

The measurement of a product's intrinsic value has become an integral part of a new technologies reimbursement and marketing strategy. The key tactics in the evaluation of a product's value are the economic and humanistic consequences of the new therapy relative to the gold standard. However, in many circumstances the patient is unable to provide an accurate assessment of the health care resources consumed or their quality of life and functional status. Researchers have opted for the use of proxies in these instances, under the presumption that a larger sample size increases the power to detect meaningful differences. This presumption, however, neglects the inaccuracies inherent in proxy reporting and thus may inhibit the reliability of the estimates obtained. This workshop will review and evaluate the use of proxy respondents in the collection of resource utilization and quality of life. A critical examination will be given to their potential influence on study results. Suggestions for overcoming these issues will also be presented and discussed. This session is directed at individuals in pharmaceutical firms, CROs, and consultancy companies who are responsible for the design and conduct of pharmacoeconomic evaluations.

Marquis P 1 , Abetz L 1, Conway K 2, Mear I 2, 1Mapi Values, Lyon, France; 2Mapi Research Institute, Lyon, France

Assessing the impact of disease and the effect of treatment on a patient's Quality of Life (QoL) has become of major importance both to the pharmaceutical industry and the medical profession. The last 20 years have thus seen the development of a large number of QoL questionnaires, which are increasingly being used in clinical trials with a growing emphasis on multinational applications. As a result, there is a continually expanding need for cross-nationally and cross-culturally valid, reliable and responsive QoL instruments. This course will provide key information and operational solutions to implement health status and QoL measures in clinical trials. The content of the course will cover definitions of health status and QoL, type of instruments used, including their strength and weaknesses, identification of concepts to be measured, selection of relevant instruments. The second part of the course will focus on strategy for instrument development, linguistic validation and psychometric validation, as well as interpretation guidelines. Participants will acquire a common baseline understanding of health and QoL evaluations allowing them to have an active involvement in the development of an optimal programme for clinical development and marketing purpose.

Caro JJ 1, Migliaccio-Walle K 1, Thizon de Gaulle I 2, Coniglio A 2, 1 Caro Research, Concord, MA, USA; 2 Bristol-Myers Squibb, Princeton, NJ, USA; 3Sanofi, Paris, France

An important element in judging the worthiness of a new drug's effects is the translation of randomized trial results to actual clinical practice. A key input is the risk experienced by patients managed routinely. This rate is important because the relative risk reduction typically estimated in trials only gains meaning when it is applied to such a reference risk. While the relative risk reduction is widely believed to be generalizable, the reference risk is not. In this workshop, an ongoing study, CAPRA (CAPRIE Actual Practice Rates Analysis) will be used as a case study, along with other published reports, to examine why considering information beyond that obtained in a clinical trial can be critical in assessing the value of a new therapy. The major reasons practice may diverge from trials, along with evidence of their existence, will be introduced. The methods for extending trial results to clinical practice will be demonstrated and discussed along with alternative approaches to estimating actual practice results from clinical trial experience. The focus will be on the use of epidemiologic studies and databases such as that from Saskatchewan Health. Details of the analysis of records for 12,931 patients used to estimate the risk of first subsequent ischaemic events (MI, ischaemic stroke, vascular death) in actual practice will be shown. The magnitude of the distortion, with special reference to number needed to treat, and the implications for cost-effectiveness analysis will also be presented. All available information and methods ought to be employed when considering a new therapy as it is insufficient to evaluate a new therapy based on trial results alone.

Mauskopf J, Research Triangle Institute, Research Triangle Park, NC, USA

Most pharmacoeconomic studies present estimates of the impact of new treatments on expected lifetime costs and health outcomes for typical individuals with that disease. However, national or local health care decision-makers also need to know what impact the new treatment will have on their annual budgets and on the annual health outcomes for their patient populations. In this workshop I will present a method for estimating the population impacts of new treatments. I will show how information about the impact of a new treatment on individual patients can be combined with information about the number and type of patients in the patient population to estimate the impact of the new treatment on annual budgets and health outcomes. For chronic diseases, the impact of a new treatment on annual budget and health outcomes may change over the first few years until a new steady state is reached. I will show how these estimates vary with different assumptions about the extent of use of the new treatment. I will illustrate a method for generating these estimates using a model developed to estimate the budget and health outcome impacts of new HIV treatments for state or federally funded programs. Population estimates allow the health care decision-maker to evaluate the impact on their budgets and patient health of providing the new treatment to their patients and to ensure that they have sufficient funds available. This workshop would be of value both to industry pharmacoeconomists and national or local health care decision-makers. It will show them how to generate analyses of economic and health outcomes in a population format that is likely to be of value for new treatment decisions.

Trippoli S, Messori A, Laboratorio SIFO di Farmacoeconomia, c/o Centro Informazione Farmaci, Azienda Ospedaliera Careggi, Florence, Italy

Handling survival data in cost-effectiveness (C/E) studies generally implies an assessment of the survival data presented in clinical trials (measured survival) together with a long-term prediction of life expectancy for the same patients (predicted survival). Since different methods are needed to manage the data of measured survival and of predicted survival, this workshop is proposed to present an overview of these methods and to discuss their relative advantages and disadvantages. Measured survival. When a single clinical trial is the source of the survival information, traditional methods for constructing survival curves can be utilised in C/E studies. When two or more clinical trials are available, combining these data requires a survival meta-analysis. Although the methodology of survival meta-analysis is still under development, some techniques in this area have adequately been tested and can therefore be proposed for general use. Predicted survival. Traditional life-expectancy calculations remain the mainstay for predicting survival in healthy subjects or in "cured" patients. Specific methods, however, are needed for correcting normal life expectancy predictions on the basis of the presence of a disease condition. The Gompertz extrapolation technique can be used for conditions where the chance of cure is minimal. Furthermore, other methods have recently been proposed to combine the clinical evidence of a specific survival pattern with the assignment of a normal life expectancy to "long term survivors". This workshop is designed to provide an overview of the foregoing methods; to report on their use in clinical and economic evaluation; and to present detailed examples of their application. It is expected that Workshop attendees will be primarily researchers and analysts concerned with clinical and economic evaluation, particularly where multi-centre collaboration is involved.

Seitz R 1, Wasem J 2, Krauth C 3, 1Department for Health Economics, University of Ulm, Germany; 2Institute for Medical Information Management, Biometrics and Epidemiology, University of Munich, Germany; 3Department for Social Medicine and Epidemiology, University of Hannover, Hannover, Germany

Society's perspective on costs and benefits of health care programs has been well established in theory and methodology of outcomes research. Far less examined are issues concerning the payers' perspective. For rehabilitation programs in Germany, the perspective of Social Old Age Insurance Funds (SOAIF) is of particular importance: For the workforce they finance most of rehabilitative interventions, and they finance invalidity pensions in case that their insured can no longer participate in the workforce; rehabilitation programs are designed to avoid invalidity. In the workshop theoretical and methodological issues of a payer's perspective in general and of SOAIF in particular are discussed. Four perspectives of SOAIF can be distinguished: (a) minimizing of SOAIF's expenditure - this requires to finance rehabilitative interventions in case that it can be assumed that the expenditure for invalidity pensions saved through an intervention exceed the expenditure for the intervention itself; (b) according to the (German) social code on rehabilitation: minimizing costs and cases of invalidity - here the costs of the interventions themselves are not considered; (c) maximizing a social welfare function: as part of social insurance and the welfare state arrangements, SOAIF could be required to maximize a social welfare function - which could be identical to society's perspective; (d) maximizing the utility function of SOAIF's bureaucrats - according to this perspective, as many rehabilitation interventions will be produced as is necessary to maximize power, prestige and income of the functionaries of the pension funds. Consequences for study designs are discussed.

Crawford B 1, Evans C 2, 1MAPI Values, Boston, MA, USA; 2Astra Pharmaceuticals, Westborough, MA, USA

Pharmacoeconomic and quality of life analyses are becoming more familiar and hence more important to marketing and reimbursement. The use of piggyback studies as an analytic technique has increased rapidly. A piggyback study is performed alongside an existing clinical trial, leaving the clinical components intact and undisturbed. Since clinical safety and efficacy are the drivers behind the trial, efforts are made to ensure the clinical endpoints are not compromised by the economic sub-study. This growing emphasis on cost-effectiveness and improvements in humanistic attributes (QOL, functioning, ADLs) has caused many researchers to add poor economic and quality of life sub-studies in an attempt to tell a compelling story about their product. PE/QOL studies should be developed with a strong foundation. This workshop will provide a detailed review of the fundamental steps necessary to properly design a piggyback study in the international setting. These steps are: the development of a priori objectives and target audience, hypotheses, endpoints to be measured, definition of the target population, selection of appropriate comparators, the time-frame to be analyzed, determination of how the data shall be collected, creation of an analysis plan, costing the data, analysis of the data, and lastly, the reporting of the results in a concise and meaningful way. Defining the objectives and hypotheses will be the focus of the workshop with a review of the other steps. This session is directed at individuals in pharmaceutical firms, CROs, and consultancy companies who are responsible for the design and conduct of pharmacoeconomic evaluations.

Crawford B 1, Evans C 2, 1MAPI Values, Boston, MA, USA; 2Astra Pharmaceuticals, Westborough, MA, USA

This workshop offers a critical examination of the use of patient self reports in pharmacoeconomic evaluations. This workshop is directed at individuals in pharmaceutical firms, CROs, and consultancy companies who are responsible for the design and conduct of pharmacoeconomic evaluations. This workshop advances current research by concentrating on how the method of data collection in prospective clinical economic evaluations influences study results. Scant attention has been paid to the fact that the high level of internal validity found in prospective studies may be compromised by the application of inappropriate methodologies to data collection. Particular attention in this workshop is placed on how validity may be affected by the elapsed time between admission and reporting, the salience of the treatment event and the perceived social desirability of the condition. As part of this workshop we will demonstrate specific areas where estimates based on patient self report leads to reliable or suspect values. In particular, we examine areas of hospitalization, outpatient consultations, medication use and indirect costs. The impact that patient self-reports have on cost effectiveness ratios is also discussed. Attendees of this workshop will gain an understanding of current methodological shortcomings in this area and researchers and readers of pharmacoeconomic studies will gain the skills necessary to better design and evaluate the validity and potential bias in cost effectiveness analyses.

Dittrich I 1, Jirsa J 2, 1Prague, Czech Republic; 2 Via Nova s.r.o.hopital/clinic, Zabrch, Czech Republic

Development of the disease management program accepted by physicians, payors and patients in the general practice has to respond to challenges and overcome number of obstacles. In countries of eastern Europe, many of those challenges and obstacles are similar as in the western Europe, many of those challenges and obstacles are similar as in the western Europe and in the USA, but there are some other issues specific to the region, which need to be addressed. This workshop summarises experience gained the attempt to launch a disease management program focused on complications associated with climacteric conditions of women in 46 to 58 age category. The emphasis was on osteoporosis and cardiovascular complications. It is expected that this workshop would be attended by representatives of institutions contemplating starting disease management programs in general and in the eastern European region in particular.

LeLorier JDesgagné A 1,2, Castilloux A-M 1, Angers J-F 2, 1, 1 Centre de recherche, Centre hospitalier de l'Université de Montréal, Campus Hôtel-Dieu, Canada; 2 Département de mathématiques et de statistique, Université de Montréal, Canada

In pharmacoeconomics, the comparison of costs generated by the use of two different drugs for the same treatment is of great interest. The problem is especially challenging when the drugs are likely to produce costly adverse effects in a small number of patients. The distribution of cost data is likely to be skewed and therefore traditional statistical methods may be inappropriate to analyse the difference in the mean costs. The objective of this workshop is to discuss the limitations of classical statistical methods used for cost analysis, and to present an alternative method appropriate to deal with skewed data. In this workshop, a pharmacoeconomic cost analysis example with skewed data will be presented. The bootstrap method will be presented as an alternative approach. Cost analysis will be demonstrated on an example with classical and alternative methods. In this workshop, the following statistical methods and their limitations will be discussed: 1) the Student t-test, which assumes normality of data and is sensitive to the skewness of data for small to moderate sample sizes; 2) the t-test on log-transformed data, which assumes log-normal distributions and equality of variances of log-transformed data for both samples; and 3) the nonparametric rank tests, which assumes that the distributions of data have the same shape and variance. In this workshop, the bootstrap method will be shown to produce the most reliable results for the example given. The use of the nonparametric bootstrap test for most pharmacoeconomic cost analyses with skewed data and with small to moderate sample sizes will be discussed. No restricting assumptions are needed for its application and its ability to deal with skewness makes it an appropriate alternative.

Value in Health Index