Clinical study methodology issues including pharmacoepidemiology (COS)

W1: EVALUATION OF THE COST-EFFECTIVENESS OF INTERVENTIONS TO CHANGE PROFESSIONAL BEHAVIOUR
Davey P
University of Dundee, Dundee, Scotland, UK

Workshop Purpose: Participants will gain an understanding of the evaluation methods that are accepted by the Cochrane Effective Practice and Organisation of Care (EPOC) Group. The aim is to improve awareness of rigorous methodology and to discuss how healthcare organisations can best evaluate interventions.

Who Would Benefit: Health services researchers in academia, government or agencies responsible for quality improvement; decision makers in healthcare organisations or in clinical practice; referees for research grants or journal articles.

Workshop Description: The workshop presenter is lead author for a Cochrane review of interventions to improve antibiotic prescribing in hospitals. Materials for the workshop will be based on this review and a systematic review of guideline implementation covering all clinical specialties. Participants will be asked to consider evaluation of the cost-effectiveness of a care pathway for community acquired pneumonia and will be given results from a completed study to discuss. The workshop will close with a debate about priorities for future research. Participants will leave with a clear understanding of how to (and how not to) plan an evaluation.
 

W8: ESTIMATING LIFE-EXPECTANCY ACCOUNTING FOR SUBJECT-SPECIFIC CHARACTERISTICS
Caro JJ¹, Ishak K², Proskorovsky I^2
1Caro Research Institute, Concord, MA, USA; ²Caro Research Institute, Dorval, QC, Canada

Workshop Purpose: Participants will learn: 1) modeling techniques to extrapolate survival beyond the time window of available data; 2) how to account for patients’ characteristics in predicting life expectancy; and 3) how to apply prediction equations to estimate LE for populations or individual patients.

Who Would Benefit: Health researchers and research sponsors interested in cost-effectiveness analyses and life expectancy estimation in general.

Workshop Description: Cost-effectiveness analyses often incorporate the impact of medications on the life expectancy of patients or life-years lost following medical events. Life-expectancy can be estimated as the area under the full cumulative survival curve. In practice, however, survival is hardly ever observed completely due to losses to follow-up and administrative restrictions on the duration of studies which can lead to incomplete follow-up for most of the population. It is therefore necessary to extrapolate survival beyond the available time-window to predict survival times. Standard parametric survival analysis methods are not always appropriate for this purpose since the variation in the underlying death hazards over time are not adequately described by any single parametric distribution. In this workshop, we will describe two approaches: the first uses a piecewise parametric method to model observed death hazards in suitably defined time windows; the second employs fractional polynomials, a flexible regression technique that fits a single model to the hazards over time. We illustrate how these derived equations can be used to predict the complete survival curve, which is then integrated numerically to estimate the life-expectancy of the population. It is possible to extend this approach to take into account patient characteristics to produce estimates for specific types of patients or populations. We use Cox regression models to measure the effect of the characteristics of interest; to account for the non-proportionality of these effects we demonstrate two alternative methods, corresponding to the two methods for predicting hazards. For the piecewise parametric approach, a separate Cox model is fitted for each time window used to model the hazards; we use a Cox model with time-dependent effects to complement the fractional polynomial model approach. These approaches are illustrated with actual analyses estimating life-expectancy following severe medical events like myocardial infarction and stroke.  
 

W15: PARAMETRIC SURVIVAL MODELS AND DECISION MODELS: RELATING CONTINUOUS HAZARDS TO DISCRETE-TIME TRANSITION PROBABILITIES
Briggs A
HERC - University of Oxford, Oxford, UK

Workshop Purpose: The aim is to increase understanding of the methods for moving between continuous hazard rates and discrete time transition probabilities -- something that is commonly required when building decision models. Participants should come away with the ability to confidently translate the results of survival analyses into transition probabilities through the hands-on exercise.

Who Would Benefit: Those involved in building discrete time (e.g. Markov) decision models for medical decision making or health economic evaluation.

Workshop Description: State transition models of disease often form the basis of health economic evaluations. Such models are most commonly characterised in discrete time periods with transition probabilities of moving between the model states. Where patient level data exist to inform the estimates of transition probabilities, it is often most natural to consider the use of standard survival models to represent these data as outcomes are often subject to censoring. Survival models are based on a hazard rate of the event of interest and the conversion of this rate into a probability represents a potential source of confusion. In this workshop the standard methods of survival analysis are reviewed and methods for conversion of hazard functions from parametric survival models to discrete transition probabilities for a corresponding state-transition model are illustrated. Attention will be given to the appropriate representation of a treatment effect estimated in continuous time in a discrete time model. A number of examples of different survival analyses to estimate the time-dependent transitions in a decision model are used to illustrate the approach. Particular focus will be given to capturing uncertainty such that transitions probabilities defined from parametric survival models can be made probabilistic. A hands-on illustration will be distributed to participants based on a Microsoft Excel spreadsheet. Participants should have some previous knowledge of state-transition models (such as Markov models) and an elementary understanding of survival analysis. A laptop computer running Microsoft Excel will be required for the hands-on exercise.

W20: DISEASE REGISTRIES: TRICKS AND TIPS IN THE ANALYSIS OF OBSERVATIONAL DATA
Silva SJ, Pasta DJ, Miller DP
Ovation Research Group, San Francisco, CA, USA

Workshop Purpose: Observational outcomes studies, or registries, present a novel way of collecting naturalistic outcomes data associated with a particular disease or product. This workshop reviews the ways registries differ from clinical trials and the procedural and methodological challenges faced when analyzing data collected in a registry setting.

Who Would Benefit: This workshop is intended for individuals who want to learn more about the methods of analyzing registry data. Individuals responsible for design and implementation of registries would also benefit from this workshop.

Workshop Description: Procedural Challenges: Analysis of observational studies has different challenges than analysis of randomized clinical trials. Obtaining accurate data on practice patterns, resource utilization, and patient-reported outcomes requires a naturalistic study design. As a result, treatment groups will not be randomized, sites that do not regularly perform certain diagnostic tests may have a great deal of missing data, and unexpected practice patterns are almost guaranteed. From a procedural standpoint, this means that a certain amount of directed exploratory analysis is usually necessary before a comprehensive analysis plan can be developed. Methodological Challenges: The fact that the analysis of observational data tends to be less driven by a priori hypotheses leads to some specific methodological challenges. The absence of pre-specified hypotheses does not mean that the analysis is strictly descriptive; comparisons can still be made among groups even though those groups probably have not been pre-defined or randomized. However, there are special statistical techniques that need to be used to account for the likely differences between groups. Patients undergoing the various treatments or no treatment are likely to differ in material ways from each other. The workshop compares several statistical approaches to overcoming this problem, including using propensity scores and adjusting using covariates.
 

Compliance issues (CMP)

W2: METHODS FOR MEDICATION COMPLIANCE STUDIES: THE IMPORTANCE OF STATISTICAL DISTRIBUTIONS
Nichol MB¹, Gwadry-Sridhar F², Benner J³, Cramer J^4
1University of Southern California, Los Angeles, CA, USA; ²University of Western Ontario, London, ON, Canada; ³ValueMedics Research, Oakton, VA, USA; ⁴Yale University, West Haven, CT, USA

Workshop Purpose: This workshop will demonstrate the statistical properties of various measures of medication compliance, and discuss implications for future research in this field.

Who Would Benefit: Researchers, practitioners, and policy makers interested in the field of compliance would benefit from the discussion at this workshop.

Workshop Description: Compliance with medications has become an increasingly important area of research as policy makers have perceived the extent of noncompliance and the effect of this problem on important patient outcomes. One of the areas that can benefit from clarity is the methodology involved in measuring and defining compliance. Differences in distributional characteristics of the measures contribute to the difficulty researcher’s encounter analyzing compliance. This workshop will illustrate the distributional characteristics of compliance measures commonly used in the literature, explore the implications of these characteristics, and recommend specific descriptive analyses that should be included in articles that report measures of compliance. Examples will be drawn from published and ongoing projects measuring compliance in a variety of medications and diseases. Workshop participants will be encouraged to offer their perspectives and methodological recommendations for future studies of medication compliance.
 

W9: METHODS FOR MEDICATION COMPLIANCE STUDIES: AN OVERVIEW OF THE ISPOR MEDICATION COMPLIANCE SIG GUIDELINES
Nau DP¹, Peterson A², Cramer J^3
1University of Michigan, Ann Arbor, MI, USA; ²University of the Sciences in Philadelphia, Philadelphia, PA, USA; ³Yale University, New Haven, CT, USA

Workshop Purpose: Participants will gain an understanding of newly developed guidelines for conducting medication compliance studies using retrospective databases. The guidelines can assist researchers in their development of appropriate methods for measuring and analyzing medication compliance data.

Who Would Benefit: Researchers, practitioners, and policy makers interested in the field of compliance would benefit from the discussion at this workshop.

Workshop Description: Compliance with medications has become an increasingly important area of research as policy makers have perceived the extent and effect of the problem. Unfortunately, the wide range of measures and analytic methods has led to difficulty in evaluating and comparing research in this field. The purpose of the workshop is to share guidelines developed by members of the ISPOR Medication Compliance SIG Analysis Standards Working Group that will assist researchers in conducting appropriate analyses and evaluating the literature. The guidelines focus on methodological issues that should be addressed in compliance research using administrative data (i.e., claims data). Members of the Analysis Methods Working Group will present and facilitate the discussion on the guidelines, and workshop participants will be encouraged to offer their perspectives and recommendations.
 

W16: ESTIMATING MEDICATION PERSISTENCY USING ADMINISTRATIVE CLAIMS DATA
Sikka R¹, Aubert R^2
1Boston University School of Medicine, Boston, MA, USA; ²Medco Health Solutions, Franklin Lakes, NJ, USA

Workshop Purpose: Participants will gain an understanding of: 1) the conceptual definition of medication refill persistency; 2) the advantages and limitations of different methodologies to measure medication refill persistency from pharmacy claims data; 3) the practical application of methodologies for persistency measurement through an interactive practicum.

Who Would Benefit: Health services researchers, policy makers and payers concerned with medication compliance issues and their impact on quality and costs.

Workshop Description: Pharmacy claims data has become a common tool in the assessment of medication compliance. Although these large population databases afford access to a vast amount of information regarding medication dosing and refilling patterns, the challenge remains to convert these large quantities of claims data into intuitive and meaningful surrogate measures of medication compliance. This workshop will review one particular dimension of the measurement of medication compliance, refill persistency. Participants will be introduced to a proposed definition for medication persistency and will review three, literature-based techniques to measure persistency from pharmacy claims data. The advantages, limitations and implementation aspects of each technique will be reviewed. In addition to this discussion of concepts and methods, participants will independently work through a case study to determine refill persistency from a cohort of individuals with simulated claims. Participants will de-brief the case study and will be encouraged to share their own assessment of the techniques to estimate medication persistency.

Cost study methodology issues (CS)

W3: BAYESIAN NETWORKS, INFLUENCE DIAGRAMS AND THEORETIC DECISION MODELS IN HEALTH ECONOMICS
Baio G¹, Jansen J², Van Genugten ML^2
1University of Florence, Florence, Italy; ²MAPI Values, Houten, The Netherlands

Workshop Purpose: We aim at presenting the most interesting features of Bayesian Networks as a tool for decision theoretic analysis. These will be illustrated by means of a set of examples in the health economics framework. Participants will be gently introduced to the main features of Bayesian Networks models. The approach will be highly quantitative, and several examples will be discussed, with particular focus to the applications of such extended decision models to problem of health economic evaluations. The workshop is intended to provide the participants with insight on the advantages produced by the use of such models in terms of understanding the problem, simplicity of representation and power of analysis.

Who Would Benefit: Health services researchers in academia, industry or regulatory bodies and decision-makers at all levels of clinical practice.

Workshop Description: The workshop is presented by members of the statistical community, with long experience in health economics, as well as by specialist in health economics and pharmacoeconomics consultants. They will present Bayesian Networks theory and applications in different settings, including clinical practice as well as clinical trials and economic evaluation. The workshop examines the versatility of the instrument as an extremely powerful modelling tool. Examples are drawn from several different countries. Participants will be able to access expert opinion on Bayesian Networks methodology and will be encouraged to share their own assessment of its practical utility.

W10: CODING PRACTICES AND PROBLEMS THAT CAN ALTER A COST ESTIMATE
O'Brien JA
Caro Research Institute, Concord, MA, USA

Workshop Purpose: To provide an understanding of how coding practices and database idiosyncrasies can impact a cost estimate, and potential solutions for dealing with these situations.

Who Would Benefit: Researchers who use coding classification systems such as ICD-9 and ICD-10 to identify cases in datasets for the purpose of developing resource use profiles and/or cost estimates.

Workshop Description: Not everyone using diagnosis and procedure codes is an expert coder. In fact, most code users have never taken coding classes nor have a solid understanding of the fundamental rules of coding. Yet, numeric and alphanumeric codes are the principle means of identifying cases in databases. Even those who are coding experts may encounter dataset structures that do not adhere to coding rules. Or, a research may be familiar with one coding system (e.g., ICD-9), but have to identify similar cases using another (e.g., ICD-10) because of the datasets employed for the analysis. Overcoming these potential problems requires adjusting case identification strategies. Misidentification of cases can lead to analyzing incorrect data which can result in a cost estimate or resource use profile that does not reflect the condition, therapy or population intended for evaluation. This workshop will discuss coding principals relevant to developing cost estimates, along with highlighting potential cost-related problems such as truncation of digits, differences in ICD-9 and ICD-10 codes, and cross-walking of different coding systems. Examples of how these problems can impact a cost estimate will be provided, along with practical suggestions for dealing with them. Participants will be encouraged to provide examples of problems they have encountered in identifying data using various coding systems and how they dealt with them. The examples provided and these shared experiences should help participants avoid, or at least understand the potential consequences, of miscoding and practices associated with some datasets.

W11: PROS AND CONS OF DIFFERENT MODELING TECHNIQUES FROM THE END-USER PERSPECTIVE
Mesrobian X¹, Van Genugten ML^2
1Mapi Values, Boston, MA, USA; ²Mapi Values, Houten, The Netherlands

Workshop Purpose: The objectives of this workshop are to discuss the relevance and caveats of different pharmacoeconomic models (i.e., decision tree, Markov, discreet event simulation) and to examine how the development of a model can influence the understanding and the use of a model by its audience.

Who Would Benefit: Healthcare decision-makers and researchers in charge of designing, assessing and utilizing pharmacoeconomic models.

Workshop Description: Researchers are eager to adapt the latest economic modeling techniques to health economics; therefore, they are utilizing advanced modeling techniques more and more frequently in this field. Model end-users, however, hold very diverse backgrounds and do not all hold a clear understanding of model techniques. Consequently, it is the model designer role to maximize the access to information of model end-users and help decision making process. This workshop will review the pros and cons of different modeling techniques from an end user standpoint rather than a modeler standpoint. The workshop will demonstrate that the situation to be analyzed and the expected outputs of the model (budget impact, cost-effectiveness) are key factors for the selection of model design and sophistication. The workshop will establish that the level of sophistication of models is function of the problem to solve, not of the model designer or the end-user. Furthermore, the workshop will show how specific model designs induce special requirements regarding the need for information, programming time and specific analyses. Finally, the workshop will detail how decision-makers, end users and model builders would benefit from more effective communication about models.

W17: PROPENSITY SCORE METHODS AND OTHER SAMPLE SELECTION MODELS: AN APPLIED EXAMPLE USING RETROSPECTIVE ADMINISTRATIVE CLAIMS DATA
Schultz J¹, Riedel AA¹, Crown WH², Harley C^1
1Ingenix Pharmaceutical Services, Eden Prairie, MN, USA; ²Ingenix, Auburndale, MA, USA

Workshop Purpose: Participants will gain an understanding of multiple sample selection models, including variants of propensity score methods, instrumental variables analysis, and the difference-in-differences method. An applied example using retrospective data will be used to illustrate these techniques.

Who Would Benefit: Health services researchers in academia, industry, regulatory bodies, and government institutions; decision makers at all levels of clinical practice; individuals who conduct or rely on research with non-randomized populations.

Workshop Description: One of the primary hurdles to be overcome in the use of retrospective data to compare pharmaceutical treatments is the presence of selection (or channeling) bias in treatment decisions. One technique gaining in popularity is the use of propensity score matching on baseline characteristics to control for selection bias. This workshop will review the use of standard regression and propensity score matching techniques to address sample selection, as well as four additional methods, including: the Heckman 2-stage model with instrumental variables, stratification with propensity scores, regression with propensity scores, and the differences in differences (DID) method. Participants will be engaged in a critical comparison of each method and its impact on study results through the use of an applied example. Discussion and questions will be solicited throughout the presentation (rather than limited to a question and answer period at the end of the session). Retrospective claims data for patients enrolled in a large, commercial, US managed care plan who use asthma controllers for the treatment of mild to severe asthma will be used to demonstrate and compare each method. The prescription of asthma controller therapy is known to be associated with disease severity and other patient factors, complicating head-to-head drug comparison and providing fertile ground for the critical evaluation of sample selection models.

W21: THE ANALYSIS OF INCOMPLETE (COST) DATA – THE APPROPRIATENESS OF AVAILABLE METHODS IN DIFFERENT SITUATIONS
Oostenbrink JB, Al MJ
Erasmus MC, Rotterdam, The Netherlands

Workshop Purpose: Participants will gain understanding of the methods that are available to analyze incomplete data caused by patients who withdraw (drop out) from a study before they have reached the scheduled end data. They will be learned how to investigate the distribution of the data, the underlying pattern of the missingness mechanism and how this information can be used to choose between available methods.

Who Would Benefit: Researchers and decision-makers in the field of HTA who perform or assess economic evaluations.

Workshop Description: Incomplete data occur in nearly every prospective economic evaluation and the methods that are used to deal with the incomplete data have been shown to have a large impact on the outcomes of the analysis. During the workshop, methods that have been applied by participants to analyze incomplete data will be evaluated. We will show additional naïve and principled (i.e. multiple imputation, expectation maximization, product-limit estimator of Lin et al.) methods that are available to analyze incomplete data. Using data from a large simulation study, it will be shown how these methods perform in data sets with different distributions and different dropout mechanisms. The examples are based on various types of cost data, but may also apply to other types of data like quality of life or clinical outcomes. At the end of the workshop, participants will be able to determine which methods can be applied to different data sets, given the dropout pattern and the distribution of the data.

Formulary development research issues (FS)

W4: THE ART AND SCIENCE OF BUDGETARY IMPACT MODELING FOR FORMULARY SUBMISSIONS
Taylor DC, Thompson D
Innovus Research, Inc, Medford, MA, USA

Workshop Purpose: To share tips and techniques for building scientifically rigorous budgetary impact models while maintaining transparency and a user-friendly interface.

Who Would Benefit: Applied researchers or research sponsors interested in including budgetary impact models into dossiers for formulary submissions, as well as potential end users of such models.

Workshop Description: Budgetary impact models (BIM) are increasingly used in formulary submissions to help decision-makers assess the potential financial implications of adding a new therapy to their formulary. These models must incorporate complex marketplace dynamics such as product substitution effects and market expansion, but need to be flexible enough for the users to modify the model parameters to fit their own institution’s practice patterns, and costs. The art of creating a successful BIM lies in making it transparent enough so that the user can understand all of the underlying assumptions and calculations, simple enough to be used with little or no training, yet complex enough to capture the cost and science of the disease and its treatment options in a world with and without the treatment being modeled. The objective of this workshop is to introduce the participants to ways of balancing these sometimes conflicting goals of a BIM. In true workshop fashion, participants will be asked to share their experiences on a variety of relevant issues including population selection, event rates, treatment regimens, and cost determination. Discussions will also focus on how methodologic and interface design decisions may affect the accuracy, usability, and perception of the model. Particular attention will be paid to the utility of sensitivity analysis and the many implications of incorporating the element of time into a BIM, including the need to estimate screening costs, discount rates, discontinuation rates, and plan turnover rates.

W22: COMMUNICATING THE VALUE: DEVELOPING MESSAGES FOR PAYERS
Lucero M, Neighbors D
RTI Health Solutions, Research Triangle Park, NC, USA

Workshop Purpose: Because value is so integral to pricing and reimbursement, it is critical for pharmacoeconomics and outcomes researchers to be able to communicate the relevance of their work in the marketplace. Participants will gain an understanding of how to design clear, concise, and impactful value messages for use with payers and those other healthcare decision-makers who influence them.

Who Would Benefit: Pharmacoeconomics and outcomes researchers in academia, consulting firms, or industry. Health care decision-makers.

Workshop Description: The workshop will provide an interactive, practical overview of how to craft a meaningful value message, how to gauge its relevance, and how to assure it is supported by evidence. Participants will learn how to create messages that perform the following functions: tie scientific product information (e.g., mechanism of action, class effects) to ultimate health outcomes; communicate abstract concepts (e.g., QALYs, NNT, ICERs) into terms that are tangible to payers; and help payers recognize the relevance of patient reported outcomes, e.g., health-related quality of life. A method for creating and evaluating value messages that are simple, relevant, and substantiated will be introduced. After a brief presentation that includes examples of messages developed through use of this method (i.e., “before- and after-method” messages), participants will be given additional “before” example messages and guided through exercises to translate them into “after” messages. Participants will be shown the value of asking “So what?” and putting themselves “in the shoes” of the consumers of their research. Examples are drawn from various therapeutic areas including pain, mental health, women’s health, and critical care.
 

Health care policy development issues using outcomes research (HP)

W5: EQUITY AND THE IMPACT OF POLICY ON PATIENT ACCESS TO MEDICINES: WHAT IS ACCEPTABLE VARIATION?
Kerrigan J, Costello S
Heron Evidence Development Ltd, Letchworth Garden City, Hertfordshire, UK

Workshop Purpose: The purpose of this workshop is: 1) to learn that considerable local variation in patient access to a pharmaceutical intervention (medicine) occurs within a country; 2) to learn how to quantify the variation in this local access; 3) to learn how to evaluate the variation in local policy position towards a medicine; and 4) to learn how to identify the local policy variables that are correlated with local patient access to the medicine.

Who Would Benefit: Anyone concerned about the large variation in local uptake of national clinical practice guidelines. E.g. Health service or industry researchers, health care resource allocation decision makers.

Workshop Description: The original researchers will present the workshop and will provide a unique insight into the methodologies developed to quantify the impact of local health policy on the prescribing of a medicine. The presenters will open the workshop by setting the context for equity of access to medicines (i.e. ‘zip-code prescribing’). The presenters will then describe the methodology used to measure local patient access to medicines before encouraging the participants to suggest any data adjustments that may be required. Participants will be grouped into countries and, aided by the presenters, will create a list of the policy factors that can affect prescribing choice in their country. Ideas from nominated groups will then be pooled, presented and discussed. The presenters will then describe how to measure the correlation between local policy and patient access to medicines. The workshop will conclude with the presenters’ discussing caveats, practical applications of the approach and by answering the question, ‘what is acceptable variation?
 

W6: DISCRETE EVENT SIMULATION TO IMPROVE DECISION MAKING
Hout van BA¹, Heeg BM², Botteman M^2
1PharMerit, Capelle a/d IJssel, Zuid Holland, The Netherlands; ²PharMerit, Bethesda, MD, USA 

Workshop Purpose: To provide an introduction to discrete event simulation, an increasingly popular and very flexible modelling technique used to conduct burden-of-illness and cost-effectiveness analyses, and discuss when and how to use, implement, simulate, and interpret discrete event models.

Who Would Benefit: Policy makers and researchers who want to understand the value of discrete event simulation in guiding health care decisions

Workshop Description: Discrete event simulation is a type of modelling technique that allows researchers and decision makers to depict flexibly the natural history of diseases and the impact of interventions. In discrete event simulation, the disease experiences or paths of individual patients are simulated one by one and repeated for a predetermined number of patients. At the onset of each patient’s individual simulation, he/she can be assigned a set of relevant socio-demographic and clinical characteristics. Some of these characteristics can be modified over time to create for each patient a cumulative disease profile. Over time, patients can experience important disease events, according to probability distributions that can either be fixed or vary according to probability distributions and can be affected by the accumulated profile of the patient. The timing of events does not have to occur at fixed intervals, unlike in Markov models. As suggested above, discrete event simulation provides a significantly more flexible approach than traditional modelling methods such as modelling or decision trees to represent the progression of disease over time. Some of the disadvantages of discrete event simulation include more demanding parameter specification, including more complex probabilistic sensitivity analyses. This workshop will illustrate with the help of a simplified example and interactive exercises how information critical to the decision making process can more accurately and flexibly be integrated into discrete event simulations than in more traditional models.
 

W12: GEP (GOOD EPIDEMIOLOGICAL PRACTICE)- COMING TO YOUR OFFICE NEXT
Parkinson J, Davey P
University of Dundee, Dundee, UK

Workshop Purpose: To ensure that all those who use observational data, within any pharmacoepidemiological sphere, understand that the rules of GEP and what this means to the future of the provision of data for research, the research itself and publication.

Who Would Benefit: Health services researchers in academia, industry or regulatory bodies who either use observational data in their research or who rely upon the results of such.

Workshop Description: The workshop is presented by two senior members of the Tayside/MEMO database and academic research group. Because of the very local geographic nature of the Tayside data as well as UK and local governance issues they have had to, not only keep pace with, but ahead of that which although not mandatory at this time, should be a requirement for the use of all observational data. Participants will learn that GEP is not a bureaucratic time waster rather a way of working that will add strength to their study and increase the validity of their findings so enhancing the impact of the work and publication. Audience participation: All aspects of GEP will be detailed however the audience will be invited to participate in a way that ensures the workshop fulfils their requirements of learning and understanding the areas most applicable to their own work. Examples will be given of where published studies show weakness because of flaws that would/should have been prevented by GEP.

W13: CAPTURING THE VALUE OF PHARMACEUTICALS: THE ROLE OF OUTCOMES RESEARCH IN PRICING DECISIONS
Tierce J¹, Negrini C², Lloyd A³, Grueger J^4
1ValueMedics Research, LLC, Arlington, VA, USA; ²PBE Consulting, s.r.l, Milano, Italy; ³Fourth Hurdle Consulting, Ltd, London, UK; ⁴Novartis Pharma AG, Basel, Switzerland

Workshop Purpose: This workshop helps outcomes researchers assume an active role in pharmaceutical pricing. At the conclusion, attendees will understand the principals for integrating pricing and reimbursement with outcomes research and health economics to determine, demonstrate and capture product value.

Who Would Benefit: This session is for those engaged in pricing, reimbursement, outcomes research and health economics. Anyone with an interest in integrating these disciplines will benefit.

Workshop Description: As medical technologies cost more to develop and launch, health budgets are receiving greater scrutiny for the value for money of each component. While outcomes researchers traditionally assess value of pharmaceuticals, they rarely apply their skills to product pricing. However, outcomes researchers involved in drug development know the most about product value. This sets up a disconnect—or worse, a “train wreck”—between business planning, expectations, and the value that can be credibly demonstrated. Capturing product value within today’s global market is most effectively accomplished by applying the skills of outcomes research and health economics to pricing and reimbursement throughout the product development and commercialization processes. In this workshop, attendees will learn pricing strategies from the perspective of: a) value determination, including market assessment, reimbursement and product/portfolio revenue and profit forecasts; b) value demonstration, including studies, information considerations, models and pricing research; and c) value capture, including outcomes research in value-based pricing research and the development of a product pricing strategy, qualitative and quantitative techniques in value-based pricing research, market segment differences and dynamics in pricing and reimbursement and applying the knowledge gained through participating in that process to reimbursement applications and contracting negotiations with payers. The workshop format will include case studies where participants will discuss the appropriate application of outcomes research to pricing decisions. Attendees will also have the opportunity to ask questions and share experiences with integrating these disciplines.
 

W18: USE AND INTEGRATION OF FREELY AVAILABLE U.S. PUBLIC USE FILES TO ANSWER PHARMACOECONOMIC QUESTIONS: DECIPHERING THE ALPHABET SOUP
Cisternas M, Noe L
Ovation Research Group, Highland Park, IL, USA

Workshop Purpose: Participants will: 1) learn about the various public use file (PUF) data sources available in the United States for pharmacoeconomic research, in general, and cost-of-illness studies, in particular; 2) evaluate their use for creating prevalence, resource utilization, and cost estimates; 3) identify which PUF sources are available and appropriate to answer disease-specific questions; and 4) receive guidelines for integrating these sources into pharmacoeconomic studies.

Who Would Benefit: Individuals in health economics, outcomes research, marketing, and academic departments interested in utilizing publicly available data as primary or secondary sources for pharmacoeconomic research.

Workshop Description: Public use files can be valuable data sources for conducting pharmacoeconomic research. Databases include information based on patient surveys, medical records, outpatient visits, and inpatient stays, and can represent various patient groups and settings such as Medicare, Medicaid, VA, long-term care, elderly patients, pediatric patients, and more. Information can be used to develop research hypotheses, as well as to help establish the incidence and prevalence of a disease, cost-of-illness, treatment patterns and costs, resource utilization, productivity and work loss, and variables for sensitivity analysis. While many pharmacoeconomic studies can be strengthened through the analysis of PUF data, others can be based solely on such data. This workshop will begin with a guided discussion concerning the questions that can be answered through PUF data. We will then present the various PUF databases available to answer these questions, along with brief examples of their use from the authors' own work and other published articles. The pitfalls of using PUF data and how to mitigate them will also be described, and finally, guidelines for combining estimates from various sources into a single study will be discussed. Materials summarizing the various data sources, as well as URLs for data download or purchase from federal agencies will be provided.

W23: NEW CONCEPTS OF CAUSAL INFERENCE IN MEDICAL DECISION MAKING AND OUTCOMES RESEARCH
Siebert U
Harvard Medical School, Boston, MA, USA

Workshop Purpose: 1) Define causal interventions and use causal graphs to distinguish causal from non-causal statistical associations; 2) Choose appropriate statistical methods to derive causal effect parameters; and 3) Use causal graphs to estimate the direction of bias in "non-causal" models.

Who Would Benefit: Decision modelers, outcome researchers, health economists, epidemiologists, biostatisticians, philosophers.

Workshop Description: One of the most important tasks of outcome researchers and decision makers is to derive causal interpretations, both on the level of decision modeling and the level of statistical analyses of original datasets. Usually, an intervention or risk factor is modeled to have a "causal effect" on the outcome of interest. Therefore, we must check: 1) when effect estimates have causal interpretations and when they do not; 2) which are appropriate methods to derive causal effects instead of mere statistical associations; and 3) what are “methodological traps” when controlling for confounding. This workshop has 2 parts. Part I comprises a brief introduction into the theoretical concepts of: a) statistical associations vs. causal relations; b) causal graphs (directed acyclic graphs, DAGs); and c) a new, graphically oriented definition of confounding. Part II consists of interactive exercises. The audience will use DAGs to solve problems qualitatively assess the direction of bias in published studies. This part involves no statistical calculations, the solutions can be derived by simply drawing “causal arrows” between variables and applying the learned rules. Exercises include controlling for compliance in randomized clinical trials (when both "intention to treat" and "per protocol" methods yield biased effect estimates), the fallibility of estimating direct effects (i.e., controlling for intermediate steps), and controlling for time-dependent confounding in observational studies (i.e., the confounder simultaneously acts as intermediate step). In the latter situation, traditional regression analysis fails and "causal methods" such as marginal structural models or g-estimation must be used. Examples include cardiovascular diseases, HIV, nutrition, and obstetrics. We will discuss potential applications of causal methods, their implication for outcomes research and medical decision making, as well as their strengths and limitations. There will be time for individual questions from participants. Requirements: None. Material: Handouts with selected papers.

W24: INFECTIOUS-DISEASES-RELATED BIOTERRORISM AND PANDEMIC THREATS: ASSESSING PHARMACOECONOMIC IMPACTS AND COST-EFFECTIVENESS OF RESPONSES
Bos JM¹, Versmoren DR², Mesrobian X³, Van Genugten ML⁴, Postma MJ^2
1The Netherlands Vaccine Institute, Bilthoven, Utrecht, The Netherlands; ²University of Groningen, Groningen, The Netherlands; ³MAPI values, Boston, MA, USA; ⁴MAPI values, Houten, Utrecht, The Netherlands

Workshop Purpose: Participants will gain an understanding of the nature and potential (economic) impact of epidemics through bioterrorism (for example, deliberate release of anthrax or smallpox) and natural pandemics (for instance SARS and infuenza). Potential responses will be identified and the (macro)-economic and health impact of several different threats will be assessed, inclusive cost-effectiveness of large- and small-scale responses.

Who Would Benefit: Decision-makers at government or corporate level. Health services researchers interested in infectious diseases.

Workshop Description: The workshop will present examples of impact studies on both bioterrorism (anthrax and smallpox) and pandemics (SARS and influenza), inclusive consequences for GP-visits, hospital bed needs and isolation beds at various stages of the epidemics. In this workshop, also the challenges and limitations of current frameworks for assessing economic impacts will be discussed, such as modelling and scenario-analysis. As such, this workshop intends to give participants more insight in the potential devastating economic effects of acts of bioterrorism and pandemics.

Patient registry development issues (PR)

W7: UTILIZING PATIENT REGISTRIES TO SUPPORT OUTCOMES RESEARCH: INTEGRATING OBSERVATIONAL DATA WITHIN ECONOMIC ANALYSES, MODELS, AND OTHER APPLICATIONS
Larson LR¹, Noe L¹, Mathias SD², Waller HD¹, Becker RV^1
1Ovation Research Group, Highland Park, IL, USA; ²Ovation Research Group, San Francisco, CA, USA

Workshop Purpose: Participants will gain an understanding of: 1) patient registry design and strategy; and 2) approaches to generating observational data to enhance the content and increase the value of economic models and other outcomes initiatives.

Who Would Benefit: Outcomes researchers interested in exploring the design and application of patient registries to meet outcomes and commercial objectives. Those with responsibility for implementing peri-approval programs to meet the growing demand for outcomes data would also benefit.

Workshop Description: Worldwide, the demand for observational data is rapidly increasing; patient registries provide the opportunity to respond effectively and efficiently to this need. The growing demands from regulators, including the EMEA and the FDA, as well as from global commercial markets, are accelerating the need to better document the real-world effects of new and marketed products, contributing to the necessity for – and proliferation of – patient registries. When designed appropriately, these registries represent a valuable new tool allowing clinical, economic, and outcomes researchers alike to collaboratively and efficiently meet a broad range of organizational objectives. Patient registries offer outcomes researchers the opportunity to create and access unique datasets. These datasets can enhance evaluation of a disease’s natural history, cost of illness, impact on quality of life, and a product’s clinical effectiveness. Moreover, their observational design may allow for collection of resource utilization data to evaluate a product’s economic impact, as well as a venue for validating patient-reported outcomes instruments. Patient registry data can also be used to populate economic models, providing a more real-world structure and evaluation.

This interactive workshop will review registry strategy and design, presenting several case studies considering the integration of economic and outcomes endpoints within patient registries.

Quality of life study methodology issues including patient reported outcomes (QOL)

W14: SCREENERS, SYMPTOM AND DISABILITY SCALES FOR USE IN CLINICAL PRACTICE: THEIR CONTRIBUTION TO IMPROVE THE USE OF DRUGS
Arnould B¹, De la Loge C¹, Abetz L², Regnault A³, Duru G^4
1Mapi Values, Lyon, France; ²Mapi Values Ltd, Macclesfield, Cheshire, UK; ³Université Lyon 1, Villeurbanne, France; ⁴Université C. Bernard Lyon I, Villeurbanne, France

Workshop Purpose: The objectives of the workshop are: 1) to demonstrate to the participants how designing a specific scale for use in clinical practice can add value to the drug development process; and 2) to provide them with a clear understanding of the methodological and analytical issues to be addressed during the quantitative steps of the patient scale development process.

Who Would Benefit: Researchers and analysts involved in the development and validation of PRO instruments.

Workshop Description: A direct consequence of the maturity of patient-reported outcome (PRO) methodology is the increasing demand by physicians for specific tools providing a controlled, standardised and interpretable measurement of the patient perspective. This workshop will focus on tools specifically adapted for application in clinical practice to improve disease awareness, help in patient management, and optimise treatment efficiency. The development and validation process of different scales according to their specific objective (screeners for eligibility to a specific treatment, predictors of treatment response, or instruments to monitor symptom severity) will be presented and compared. Specific attention will be given to the design of the studies to be conducted for a successful scale development, from initial data collection to final validation. The differences to traditional PRO evaluation questionnaires will be highlighted and discussed. The example of various instruments recently developed in upper gastro-intestinal disorders will illustrate the workshop. An active contribution of the participants, based on their personal experience, will be encouraged, and time will be kept available for discussion.
 

Risk assessment/risk management issues (RK)

W19: TOWARDS A MORE EFFICIENT CLINICAL DEVELOPMENT PROCESS WITH SCENARIO ANALYSES
Van Loon J¹, Van Genugten ML¹, Piercy J², Postma MJ^3
1Mapi Values, Houten, The Netherlands; ²Adelphi Group, Bollington, UK; ³University of Groningen, Groningen, The Netherlands

Workshop Purpose: Participants will be introduced into scenario analyses as a technique supportive in (re) defining the clinical development and strategic marketing of a new drug. Clinical development is characterised by uncertainties. To deal with these uncertainties the clinical development team has to be provided with overviews of past and future alternative developments including anticipated developments in competitor products. Scenario analysis can enhance long-term planning by constructing different views of the future in order to gain insights into costs and benefits related to certain choices. The multidisciplinary approach, quantification of the problem, the translation of the problem in terms of a computer model and construction of future scenario’s are essential parts of the methodology. An introduction in scenario analyses with focus on how to construct scenarios will be given. The differences between Markov modelling and scenario analysis and the use of scenario analysis in early phase development will be explained. The scenarios will also feed into the GAP analyses related to the key value messages necessary for market access. The workshop is intended to provide the participants with insight in scenario analyses and the advantages of early phase scenario analyses for drug development.

Who Would Benefit: Health Economics managers in industry or regulatory bodies. And decision-makers at all level of clinical practice.

Workshop Description: The workshop is led by members from Mapi Values, Adelphi An introduction to scenario analyses with focus on how to construct scenarios will be given. The differences between Markov modelling and scenario analysis and the use of scenario analysis in early phase development will be explained. The workshop is led by members from Mapi Values, Adelphi and academics with experience in scenario analysis performed in different settings. The workshop will provide a unique insight in scenario analyses and how this can enable a more efficient drug development process. Many examples based on experience will be presented. The audience will be encouraged to share their ideas on scenario analyses in the drug development process.