Workshop Purpose: This workshop has two main objectives, 1) to illustrate the need and use of proper statistical models for the meta-analysis of complex data such as that from mixed comparisons; and 2) to show the use of common statistical software code to estimate such models via case studies.
Workshop Description: The systematic review of the best available clinical evidence is a key component of evidence-based medicine. Meta-analyses of such evidence offers a quantitative summary of effects which can then be used as starting point for development of medical guidelines, decision analytic models and economic evaluations of interventions. Meta-analytic methods have evolved concurrently with the maturity of more sophisticated statistical models, for example those based on random effects, which simultaneously handle complex outcomes measures, comparisons, correlations and study designs. Examples of such complex scenarios are meta-analyses of studies assessing multiple agents/procedures for the same condition and whose evaluation is enriched by the combination of all available evidence, despite direct comparisons among all of the interventions are not present in the cohort of studies. Mixed effects models, containing fixed and random components, are a flexible framework which handles all those features under a common umbrella. In this workshop the audience will be introduced to the use of the mixed model framework for meta-analyses from a practical point of view, as well as to the value and caveats of direct and indirect comparisons. Data from recent meta-analyses will be used to illustrate the methods, and code from common statistical software package will be used to show their applicability. Interaction from audience is sought via a case study approach and frequent questions and answers sessions.


PATIENT-REPORTED OUTCOMES METHODS (INCLUDING COMPLIANCE & PREFERENCE STUDIES)

W6: VALUING EQ-5D HEALTH STATES : PAST PRACTICE AND FUTURE POSSIBILITIES
DISCUSSION LEADERS: Paul Kind, Principal Investigator, University of York, York, United Kingdom; Frank De Charro, Executive Director, EuroQoL Group, Rotterdam, Netherlands; BA Van Hout, PhD, Professor, University Medical Center Utrecht, Utrecht, Netherlands

Workshop Purpose: This workshop has several objectives - to provide participants with a detailed understanding of the basis on which values for EQ-5D have been generated; to describe the current status of advice provided when selecting a value set from amongst those currently available; to provide an opportunity for potential users to influence future valuation research and to help guide the choice to be made regarding new "standard" valuation methods.
Workshop Description: EQ-5D is a widely used generic index measure of health-related quality of life. It has achieved worldwide exposure, notably but not exclusively, as a means of measuring health outcomes in economic evaluation. For nearly 20 years visual analogue scale (VAS) rating has been the standard method of valuation for EQ-5D health states. However, guidance on technology appraisal issued by NICE flatly rejects such values, preferring either Standard Gamble (SG) or Time Trade-Off (TTO) utilities. A national UK survey of health state values conducted in 1993 included a TTO procedure which yielded weights that were, for a time regarded as default values for EQ-5D in economic analysis. Since then there has been continuous investigation of other valuation topics but the fundamental question remains unaddressed - namely which method(s) should be used to generate weights for EQ-5D when it is applied in economic evaluation? Revisiting this issue raises several questions? Firstly, what are the requirements if such weights are to be used in cost-utility analysis to compute QALYs? Are utility weights always necessary and if so, then how acceptable are TTO weights as substitutes for SG weights? Are “riskless” utility weights acceptable to decision-makers? Under what circumstances would non-utility weights be acceptable as a substitute quality-adjustment factor in computing QALYs? How might such weights be obtained in any future valuation surveys? These are fundamental issues of importance to all concerned with the conduct of economic analysis. This workshop has two primary objectives. Firstly to provide participants with a detailed understanding of the basis on which values for EQ-5D have been generated and to describe the current status of advice provided to end-users in selecting a value set from those currently available; secondly, to provide a rare opportunity for potential users to influence the nature of future valuation research based on EQ-5D.


W7: MEASUREMENT GUIDED MEDICATION MANAGEMENT IN IMPROVING ADHERENCE WITH PRESCRIBED DRUG DOSING REGIMENS IN AMBULATORY CARE
DISCUSSION LEADERS: John Urquhart MD, Professor, UCSF, Palo Alto, CA, USA; Bernard Vrijens, PhD, Chief Scientist, Pharmionic Systems, Vise, Liege, Belgium

Workshop Purpose: The aim of this workshop is to illustrate Measurement Guided Medication Management (MGMM) as an effective technique for improving patient compliance and persistence with prescribed dosing regimens, and to understand some key economic benefits of using this method in clinical practice.
Workshop Description: Successful techniques for improving compliance have revealed the requirement for two key elements: 1) a reliable, objective drug dosing history, including day by day dosing times, presented in a format that can be easily interpreted by most patients; 2) periodic review with the patient, by the prescribing physician, pharmacist, or nurse, of the patient's dosing history since the last visit. This approach, called Measurement-Guided Medication Management (MGMM), is a clear break with past attempts to improve compliance based on unsatisfactory methods of measurement that depend on the patient's recall or timely diary entries and afford the patient the easy ability to exaggerate his/her compliance. This workshop will include presentation of recent experience with MGMM, and a discussion of its strengths and limits. Participants will be asked to evaluate the compliance of some hypothetical patients using different assessment techniques. A key point for discussion will be the implications of complete vs incomplete dosing histories for improving compliance and extending persistence with prescribed drug dosing regimens.


Workshops Session II
Sunday, 29 October, 2006, 17:00-18:00

ECONOMIC STUDY METHODS

W8: DESIGN ISSUES FOR HEALTH ECONOMICS IN POST-LAUNCH REAL WORLD STUDIES
DISCUSSION LEADERS: Mike Drummond, DPhil, Professor of Health Economics, University of York, York, Heslington, United Kingdom; Deborah Marshall, PhD, Vice President, Global Health Economics and Outcomes, i3 Innovus, Burlington, ON, Canada; Daniela Belovich, MSc, Project Manager, i3 Innovus, Burlington, ON, Canada

Workshop Purpose: The aim of this workshop is to discuss and illustrate the key design issues surrounding economic evaluations in post-launch real world studies.
Workshop Description: There is increased interest in post-launch economic studies as more jurisdictions require economic data as part of the formal decision process on the pricing and reimbursement of drugs. ISPOR has recognized this through the creation of a Task Force on the Use of Real World Data in Coverage and Reimbursement Decisions to develop a framework to assist health care decision-makers in dealing with real world data. Much of the data requested by reimbursement agencies cannot be obtained before the drug is marketed. Pragmatic clinical trials and prospective observational cohort studies (registries) are two approaches for collecting health economic data (e.g. health care resource use and quality of life data) in post-launch real world studies. Although pragmatic clinical trials are considered the gold standard because of randomization to minimize bias in the estimation of treatment effects, the opportunities for undertaking them are limited. Registry studies have inherent limitations of observational studies and care must be taken in their analysis and interpretation. This workshop will discuss the opportunities and challenges of design issues for health economics in these post-launch real world studies. Differences will be highlighted using case examples. Discussion and questions will be solicited throughout the presentation with brief exercises. The last component of the session will be an interactive exercise for the audience to apply the contents of the workshop to decide on and develop a post-launch real world study design for a specific example.


W9: DEVELOPING ECONOMIC MODELS: PREDICTING CHANGE OVER TIME WITH FRACTIONAL POLYNOMIALS
DISCUSSION LEADERS: K Jack Ishak, PhD, Statistician, Caro Research Institute, Montreal, QC, Canada; J. Jaime Caro, MDCM, FRCPC, FAC, President & Scientific Director, Caro Research Institute, Concord, MA, USA; Irina Proskorovsky, BSc, Statistician, Caro Research Institute, Dorval, QC, Canada

Workshop Purpose: The purpose of this workshop is to describe and illustrate a modeling strategy that uses fractional polynomials, a flexible technique, to develop prediction equations of change over time.
Workshop Description: Economic models often require predictions of measures of patients' condition over time (e.g., blood pressure). The data used to develop the prediction equations are collected in longitudinal studies that involve repeated measurements of the variables of interest. The challenge in deriving these equations lies in adequately capturing the pattern of the change in measures over time, which may not always be described by simple (e.g., linear or quadratic) functions. Furthermore, finding a single parameterization that works for all or most patients in the sample may not be possible. We present a two step approach in which patients are first classified into groups based on the general direction of the pattern of changes in the measure. Each group is then modeled separately using fractional polynomials to find the best-fitting parameterization of time from a set of pairs of powers of time. This is done in the context of linear hierarchical models which provide several advantages: 1) they take into account correlations between observations collected from the same patient; 2) they can incorporate covariates with time-varying values; 3) they can be extended to allow variability of parameters across individuals (e.g., each subject may have a different rate of change). We describe the model fitting process and illustrate the implementation of the method with actual data.


HEALTHCARE POLICY

W10: THE IMPACT OF ADMINISTRATIVE COSTS RESULTING FROM PRESCRIPTION RESTRICTIONS ON THE COST-EFFECTIVENESS OF INNOVATIVE PHARMACEUTICALS
DISCUSSION LEADERS: Mark JC Nuijten, MD, PhD, MBA, Consultant, Ars Accessus Medica/Erasmus University Rotterdam, Amsterdam, Netherlands; Ken Redekop, PhD, Scientist, IMTA, Rotterdam, Netherlands

Workshop Purpose: The aim of this workshop is to illustrate that the health economic consequences of prescription restrictions, which often lead to administrative burden, are currently not considered in the current reimbursement decisions considering cost-effectiveness data for a new product.
Workshop Description: While traditionally reimbursement decisions applied to the officially registered indication, which was usually a broad indication, authorities have recently been imposing restrictions on the claim made for a new drug. These restrictions usually relate to follow a treatment protocol, to limit the prescribers (e.g. only specialists) or to limit the range of indications in order to limit the budgetary impact of the new drug. The narrowing of the indication especially depends on the clinical benefit, but also the results of the health economic analysis. However implementation of prescription restrictions is associated with additional administrative burden on the physicians, but also health insurers. In this workshop we will illustrate the economic impact of prescriptions restrictions using two cases: prescriptions restrictions for biologicals and cardiac drugs. The key point for discussion will be the appropriate handling of these administrative costs from a health economic perspective.


CLINICAL STUDY METHODS/RISK ASSESSMENT

W11: NEW DEVELOPMENTS IN DATA MINING: DIGGING INTO OBSERVATIONAL DATABASES AND ITS USE IN DRUG SAFETY SURVEILLANCE
DISCUSSION LEADERS: Charles M Gerrits, MS, PharmD, PhD, Head - Global Pharmacoepidemiology & Outcomes Research, Takeda Global Research & Development, Lincolnshire, IL, USA; Andrew Bate, MA, PHD, Manager - Research Methodologies, WHO Collaborating Centre for International Drug Monitoring, Uppsala, Sweden; Manfred Hauben, MD, MPH, Medical Director - Risk Management Strategy, Pfizer Inc, New York, NY, USA

Workshop Purpose: This session will focus on the use of data mining algorithms (DMA) on pharmacoepidemiological databases and will evaluate how this new application of DMAs contribute to pharmacovigilance practices. Learning objectives: 1. Review new techniques developed in data mining and signal detection. 2. Discuss potential application of data mining algorithms in pharmacoepidemiological and observational longitudinal databases as a tool in the proactive identification of potential drug safety issues
Workshop Description: In the last few years, with the ever-increasing volume of post-marketing spontaneous reporting system (SRS) data, considerable research is being devoted to computer-assisted data mining algorithms (DMA) to screen extremely large SRS databases for potential drug safety issues. So far, these DMAs have been applied primarily to large SRS databases (e.g. FDA-AERS, WHO etc). It is recognized that improvements are required particularly with regards to identifying signs and syndromes, drug-drug interactions, as well as long latency adverse reactions. Pharmacoepidemiological and population-based longitudinal databases ordinarily contain quality-audited anonymous longitudinal medical records, which includes information regarding diagnoses, signs & symptoms, long-latency events, prescriptions, hospital admissions and laboratory results. Hence, it is tempting to speculate whether pharmacoepidemiological and other longitudinal databases can be used in pharmacovigilance for signal detection purposes. Compared to SRS databases, the population-based databases include information regarding signs and symptoms as well as long latency events. However, they do not contain reporter-defined drug-event/symptom pairs. Whether the differences between SRS and pharmacoepidemiological databases present opportunities or challenges remains to be tested. This session will focus on the use of DMAs in pharmacoepidemiological and observational longitudinal databases as a tool in the identification of potential drug safety issues. This session will discuss the potential application of various commonly used DMAs on pooled automated databases for the early identification of potential drug safety signals. Other speakers will share their experience in applying DMAs in various settings and comment on the utility of these new techniques in signal detection. In addition, current signal detection practices and experiences with the daily application of DMAs in pharmacovigilance practices will be discussed.


W12: COMMONALITIES AND DISSIMILARITIES BETWEEN EUROPEAN OBSERVATIONAL DATASETS
DISCUSSION LEADERS: John Parkinson, PhD, Head GPRD, MHRA, MHRA, London, UK, United Kingdom; Miriam Sturkenboom, PhD, Associate Professor, Erasmus University Medical Centre, Rotterdam, Netherlands; Tjeerd P Van Staa, MD, PhD, Head of Research, General Practice Research Database, London, United Kingdom

Workshop Purpose: The aims of this workshop are two fold 1) to illustrate the way that data from different European countries can and cannot be used for pan European observational studies. 2) To further improve the level of cross country understanding of the available datasets.
Workshop Description: Healthcare is delivered across the EU in many different ways- centralised as in the UK-NHS, insurance/tax based as in France and the multi sickness fund approach as in Germany. Additionally various countries are at different states of computerisation, the use of a centralised healthcare number for record linkage is variable, coding systems vary, as do attitudes towards data privacy- despite all coming under EU law. Thus the opportunity to conduct pan European observational research which is important in this era of “risk management” is not made easy. This workshop will discuss the commonalities and dissimilarities of various European datasets as they are now and how they may look over the next 2-3 years; a timeframe in which IT developments may enable key changes to datasets and data linkages. The need for Risk Management Tools that can track both safety and effectiveness of new marketed products (indications and dose) will be used as examples of why and how datasets can and should be used together. Participants will be encouraged to discuss how their knowledge of their own countries datasets and the changes expected over the next years can contribute to better use of data on a European wide basis



PATIENT-REPORTED OUTCOMES METHODS (INCLUDING COMPLIANCE & PREFERENCE STUDIES)

W13: QUANTIFYING PHYSICIAN AND PATIENT WILLINGNESS TO ACCEPT RISK-BENEFIT TRADEOFFS: HOW TO COMPARE APPLES AND ORANGES
DISCUSSION LEADERS: F. Reed Johnson, PhD, Senior Fellow and Principal Economist, RTI International, Research Triangle Park, NC, USA; Brett Hauber, PhD, Senior Director, Health Economics, RTI International, Research Triangle Park, NC, USA; Josephine Mauskopf, PhD, Vice President, RTI Health Solutions, RTP, NC, USA

Workshop Purpose: To provide an understanding of the advantages and disadvantages of stated-preference (SP) methods such as standard gamble and discrete-choice experiments to quantify risk-benefit tradeoff preferences. To evaluate the possible effects of preference assumptions, elicitation methods, survey design, and interpretation of results on the validity and reliability of SP estimates through review and discussion of recent empirical studies and a survey completed by workshop attendees.
Workshop Description: Workshop attendees will obtain an overview of alternative methods for risk-benefit evaluation and their potential use in risk management. A tutorial will be given on using SP methods to directly measure patient and physician risk-benefit trade-offs. Specific topics will include a) the role of preferences in solving the problem of incommensurability of benefit and risk measures, b) the conceptual and empirical relationship between standard gamble utilities and maximum acceptable risk, c) sources of potential bias in using standard-gamble estimates to measure risk-benefit tradeoffs for clinically relevant risks and for nonlinear preferences, d) eliciting valid and reliable risk-benefit tradeoff preference estimates, e) deriving maximum acceptable risk, generalized QALY, and net benefit estimates from tradeoff data, and f) review of recent studies in menopause, Crohn's disease, multiple sclerosis, osteoarthritis, and irritable bowel syndrome to illustrate the key elements of risk-benefit tradeoff studies. Participants will complete an actual risk-benefit SP survey, critique the instrument, and assess the face validity of the results, which will be tabulated in real time. The discussion leaders will discuss the relevance and acceptability of these methods and specific results for regulatory and clinical decision making, including recent FDA decisions to reintroduce pharmaceuticals that had previously been withdrawn because of safety concerns.

W14: QUANTIFICATION OF HEALTH STATES WITH MULTIDIMENSIONAL SCALING
DISCUSSION LEADERS: Paul Krabbe, PhD, Senior Researcher, Radboud University Medical Centre Nijmegen, Nijmegen, Netherlands

Workshop Purpose: The aim of this workshop is to illustrate a basic scaling model, multidimensional scaling, as an alternative approach to the existing valuation techniques to quantify health states.
Workshop Description: The most controversial issue in the field of health-state valuation concerns selecting an appropriate valuation technique. Several valuation techniques are used for the valuation of health states (Standard Gamble, Time Trade-Off, Visual Analogue Scale, Person Trade-Off), each one drawing on a specific conceptual approach, methodological considerations, and scientific pedigree. Although considerable progress has been made over the past two decades in deriving health-state values these techniques are still debated. Much of the bias and distortion observed in these standard valuation techniques seems to result from the complexity and ambiguity of the cognitive tasks given to the respondents. Therefore, simplifying the valuation task for the respondent would be highly desirable. A range of approaches from psychometrics, in particular scaling models, are less prone to measurement biases than the traditional valuation techniques. Scaling models allow the investigator to transform data collected from individuals based on such simple judgmental tasks (ranking, choices) into group data with interval or cardinal measurement properties (values). We will introduce and demonstrate one of the most interesting scaling models, namely (rank-based) multidimensional scaling (MDS). The course will introduce the conceptual ideas behind and the usefulness of MDS in valuing health states. A demonstration of this model based on the input of the audience will be given and the pros and cons of the MDS scaling model will be discussed.


Workshops Session III
Tuesday, 31 October, 2006, 14:45-15:45

ECONOMIC STUDY METHODS

W15: PRICE, AVAILABILITY AND AFFORDABILITY OF CHRONIC DISEASE MEDICINES
DISCUSSION LEADERS: Richard Laing, MBCHB, MSc, MD, Medical Officer, WHO, Geneva, Switzerland; Margaret Ewen, Dip, Pharm, Principal, Global Projects, Health Action International Europe, Amsterdam, Netherlands; Susanne FAM. Gelders, MSc, Technical Officer, World Health Organization, Geneva, Switzerland

Workshop Purpose: The price, availability and affordability of medicines are major determinants of access to treatment for patients with chronic diseases. The purpose of the workshop is to illustrate high medicine prices, low availability, unaffordable treatments and excessive mark-ups, and discuss policy options to reduce prices and improve access.
Workshop Description: Chronic diseases are a serious public health issue, particularly because they require long-term therapy. Ensuring access to medicines for treating chronic disease, however, remains neglected. At the World Health Assembly in May 2006, Health Action International (HAI) and the WHO launched a report describing the prices, availability and affordability of 14 medicines for treating patients with hypertension, diabetes, asthma, epilepsy, and psychiatric disorders using data from 30 surveys undertaken around the world (mostly in developing and transitional countries) using the WHO/HAI price measurement methodology. The analysis shows that treatments are often unaffordable – as much as 50 days wages needed to purchase 30 days treatment. While governments are generally procuring efficiently, some are paying high prices for individual medicines. Availability in the public sector is often very poor. The private sector generally has better availability of medicines but prices are much higher ranging from three times the international reference price to peaks of one hundred times. In many countries, the manufacturer's price is the major contributor to the final patient price, however, in some countries numerous taxes, duties, and mark-ups can double the price from manufacturer to patient. It is important to identify and assess the contributing causes of high prices and low availability before embarking on policy changes. Numerous policy options are available to reduce prices. Once new policies are implemented, monitoring is vital to ensure access to treatment improves.

W16: SHOULD THE LOWER COST OF GENERICS OR PARALLEL IMPORTS BE APPLIED TO ECONOMIC EVALUATIONS?
DISCUSSION LEADERS: Donald Macarthur, BSc, Senior Consultant, PriceSpective Limited, Buntingford, Hertfordshire, United Kingdom; Keiron Sparrowhawk, MSc,  MBA, Principal, PriceSpective Limited, Buntingford, Hertfordshire, United Kingdom; Stephen Beard, MSc, Head of Health Economics, RTI Health Solutions, Manchester, Greater Manchest, United Kingdom

Workshop Purpose: To present the impact of generics and PIs on the “net” prices of drugs and to determine the potential effects of these prices on health technology assessments conducted or reviewed by (importing) funding authorities. Furthermore, to determine what level of macroeconomic savings is possible from generics and PIs.
Workshop Description: In some of the biggest markets in the world generics account for 50% of all prescriptions and in other markets they are the fastest growing segment. Equally, in some European markets, parallel imports (PIs) can account for over 50% of the sales of the top selling drugs. And yet, rarely does either figure prominently in economic evaluations, despite our desire to use the most “true” data available, including drug prices. Countries where the impact of PIs and generics are highest (the UK, Germany, Netherlands, Sweden) are also those where economic evidence plays a crucial role in reimbursement, access and uptake of new medicines. There is also a contentious macroeconomic debate as to whether PIs save money for the healthcare system or the patient. Anonymous data will be used to support the workshop participant's discussion and workings.


HEALTHCARE POLICY

W17: THE IMPLICATIONS OF REQUIREMENTS FOR SUBPOPULATION ANALYSES: THE DUTCH EXPERIENCE
DISCUSSION LEADERS: Mark JC Nuijten, MD, PhD, MBA, Consultant, Ars Accessus Medica/Erasmus University Rotterdam, Amsterdam, Netherlands; Carin Uyl-de Groot, PhD, Professor, IMTA, Rotterdam, Netherlands; Frans Rutten, PhD, Professor, IMTA, Rotterdam, Netherlands

Workshop Purpose: The aim of this workshop is to illustrate the methodological and practical constraints of requirements for subpopulation analysis in cost-effectiveness analysis at time of launch.
Workshop Description: While traditionally reimbursement decisions applied to the officially registered indication, which was usually a broad indication, authorities have recently been imposing restrictions on the claim made for a new drug. These restrictions usually relate to follow a treatment protocol, to limit the prescribers or to limit the range of indications in order to limit the budgetary impact of the new drug. In countries with formal requirements for cost-effectiveness data, health authorities are consequently also increasingly requesting cost-effectiveness analysis at subpopulation level. In this workshop we address the methodological and procedural constraints, which are illustrated using actual recent cases from the Dutch reimbursement process. Methodological constraints result from the fact the registration trials were not powered for analysis at subpopulation level. Because the cost-effectiveness analysis was performed for the broad indication, disaggregated values at subpopulation level for utilities and resource utilisation may probably not be available. A practical constraint is the rational of the definition of the subpopulations by the health authorities, e.g. the defined subpopulations may not correspond with existing or forthcoming clinical treatment guidelines. The key point for discussion will be the question, if the current constraints with cost-effectiveness analyses for subpopulations at time of launch justify their use for reimbursement decisions. Alternative solutions for dealing with subpopulations will be presented


CLINICAL STUDY METHODS/RISK ASSESSMENT

W18: ACCOUNTING FOR MULTIPLE HEALTH STATES IN COMPETING RISK CALCULATIONS IN ECONOMIC EVALUATION: A CASE STUDY IN ATHEROTHROMBOTIC DISEASE
DISCUSSION LEADERS: Kristen Migliaccio-Walle, BS, Senior Researcher, Caro Research Institute, Concord, MA, USA; H. Baris Deniz, MSc, Researcher, Caro Research Institute, Concord, MA, USA; J. Jaime Caro, MDCM, FRCPC, FAC, President & Scientific Director, Caro Research Institute, Concord, MA, USA

Workshop Purpose: To compare and contrast current modeling techniques used to account for the effect of complex risk profiles related to clinical history and patient characteristics in predicting patient outcomes.
Workshop Description: Economic models often must implement complex competing risks for patient cohorts (e.g., death, new MI, stroke, side effects). Although the interrelated effects of individual characteristics on patients' risk are well accepted, most economic models fail to adequately account for this. One common approach is to oversimplify by ignoring these relationships; another is to create numerous health states to reflect multiple combinations of key factors; a third is to risk adjust the inputs by carrying this out external to the model. The first approach does not reflect reality. The second leads to overly complex models that ultimately require simplifying assumptions of the inputs due to data constraints. The third does not provide flexibility for adjusting model inputs and running different scenarios. In this workshop, we will present a fourth technique that combines the estimation of risk equations with time-dependent sequential implementation to address the occurrence and development of competing risks on an ongoing basis. This technique will be compared to and discussed vis á vis commonly implemented methods. It is capable of reflecting the interrelated effects of the relevant clinical history and characteristics of a patient on the risks. We will briefly discuss the estimation of these equations and illustrate the implementation of this modeling method using discrete event simulation. AUDIENCE INTERACTION: Attendees will be asked to participate by presenting their own experiences and challenges relating to this modeling dilemma for discussion.


W19: TESTING AND CORRECTING FOR ENDOGENEITY IN ESTIMATING TREATMENT EFFECTS
DISCUSSION LEADERS: Henry Joe Henk, PhD, Researcher, i3 Magnifi, An Ingenix Company, Eden Prairie, MN, USA; William Crown, PhD, President, i3 Innovus, AUBURNDALE, MA, USA

Workshop Purpose: To describe the concept of endogeneity and its implications for bias in treatment effect estimates, as well as methods to test and correct for the problem.
Workshop Description: Economists define endogeneity as a correlation between an explanatory variable and the residuals of the outcome equation. Such a correlation can be caused by many factors but, regardless of its source, endogeneity introduces bias into the parameter estimate for the variable of interest. In the outcomes research literature, much attention has been placed on the problem of sample selection bias. This is an example of endogeneity bias introduced by unobserved variables that are correlated both with treatment selection and the outcome variable. Failure to correct for these unobserved variables introduces a correlation between the treatment variable and the residuals of the outcome equation—in turn, introducing bias into the treatment effect. Much less attention has been placed on other potential sources of endogeneity bias such as measurement error, incorrect functional form, and simultaneous equations bias. In this workshop, we describe the concept of endogeneity and the method of instrumental variables—the most widely used method to test and correct for the problem. In addition, we explore the sensitivity of instrumental variable to the key assumptions on which it is based. Although endogeneity is mainly an issue in observational studies, we also consider situations in clinical trials where endogeneity is a concern (or should be).


PATIENT-REPORTED OUTCOMES METHODS (INCLUDING COMPLIANCE & PREFERENCE STUDIES

W20: HEALTH OUTCOMES STRATEGIES TO BRIDGE CLINICAL AND MARKETING OBJECTIVES
DISCUSSION LEADERS: Meryl Brod, PhD, President, The BROD GROUP, Mill Valley, CA, USA; Torsten Christensen, MSc, Senior Health Economist, Novo Nordisk, Bagsvaerd, Denmark

Workshop Purpose: The purpose of this workshop is to present the full range of options for identifying and incorporating health outcomes into all stages of the drug development process. Attendees will be provided with a paradigm to identify the most appropriate health outcome for a given drug development goal and the optimal research designs to incorporate health outcomes at each phase of the drug development process as well as how to maximize health outcomes to bridge clinical and product positioning objectives.
Workshop Description: A well thought out health outcomes strategy can generate powerful data that can bridge clinical and commercialization objectives and optimally position a product for multiple purposes including regulatory review or publication in peer reviewed journals. Successfully incorporating health outcomes at each phase of the drug development process requires an understanding of not only the wide range of options for health outcomes but also issues of study design, objectives, drug and disease characteristics and the intended audiences. This workshop will provide a paradigm for addressing all of the issues required for a successful health outcomes strategy. The wide range of health outcomes, including quality of life, treatment satisfaction, productivity, functioning and economics will be discussed and each of the factors in the paradigm will be examined in detail. The industry perspective on how best to use health outcomes and case studies that will illustrate the advantages of a well thought out health outcomes strategy will be presented.


W21: THE WHAT AND WHO OF HEALTH OUTCOMES MEASUREMENT: INTERSECTION OF TYPES OF OUTCOMES AND SOURCES OF REPORTS
DISCUSSION LEADERS: Judith T Barr, ScD, Director, NERCOA, Northeastern University, Boston, MA, USA; Pennifer Erikson, PhD, President, OLGA, Pennsylvania State University, State College, PA, USA

Workshop Purpose: To assist participants in analyzing types of health outcomes and sources of reports incorporated into clinical studies and product labels, an organizing matrix will be developed. After reviewing the matrix in the light of recent EMEA and FDA guidances, workshop participants will utilize the matrix to determine the type and reporting source of outcome measures used for product labels and selected May 2006 ISPOR posters.
Workshop Description: Given the 2005 EMEA Reflection on health-related quality of life (HRQL) measures and 2006 Draft FDA Guidance on patient-reported outcomes (PRO), the types of health outcomes and sources of these reports are receiving regulatory scrutiny. However, there is a need to clarify that not all HRQL outcome studies are PROs, and conversely, not all PROs studies incorporate HRQL measures. In this workshop, we will present a conceptual overview examining the “What” and “Who” of health outcomes. The “What” are types of health outcomes based on an integration of two conceptual models of health outcomes: the 2000 Patrick/Chiang model and the 1995 Wilson/Cleary causal model linking clinical, symptom, functional status, health perceptions, and quality of life aspects of health and disease. Additional PRO measures such as patient satisfaction and adherence/compliance will be linked to our expansion of the integrated Patrick/Chiang/Wilson/Cleary model. The “Who” are the reporting sources of health outcomes based on a classification system proposed by the ISPOR Ad Hoc Task Force on PRO Harmonization. The contribution of clinician and proxy measures in our model also will be examined. The “What” (types of outcomes) and the “Who” (sources of reports) will be combined into a matrix which will serve as the reference tool for participant assessment of product labels and ISPOR abstracts. In the last third of the workshop, participants will work in small groups to examine cases based on abstracts/posters from ISPOR and other sources, place each study's outcomes into our framework, and determine whether or not the measures used are PRO and/or HRQL. We will conclude the workshop with a discussion of issues raised in these cases.



Workshops Session IV
Tuesday, 31 October, 2006, 16:00-17:00

ECONOMIC STUDY METHODS

W22: ADJUSTING FOR BASELINE HEALTH CARE RESOURCE USE AND HEALTH RELATED QUALITY OF LIFE IN PROSPECTIVE COST-EFFECTIVENESS STUDIES – WHEN AND HOW?
DISCUSSION LEADERS: Deborah Marshall, PhD, Vice President, Global Health Economics and Outcomes, i3 Innovus, Burlington, ON, Canada; Dan Pericak, MMath, Senior Manager, i3 Innovus, Burlington, ON, Canada; William Crown, PhD, President, i3 Innovus, Auburndale, MA, USA

Workshop Purpose: The objective of this workshop is to discuss current practice and explore alternative methods for adjustment of baseline health care resource use and health related quality of life (HRQOL) in the context of prospective cost-effectiveness studies.
Workshop Description: Guidelines for pharmacoeconomic evaluation fall short of addressing when and how to adjust for baseline costs and HRQOL for the estimation of incremental cost-effectiveness ratios. Although randomization should generate baseline balance on average, not all prospective evaluations involve randomization or are balanced on all covariates. In addition, because of the skewness typical in health care expenditure distributions, it is very likely that randomization will fail to balance patients with extreme health care costs across treatment cohorts--particularly, in trials with smaller sample sizes. Analyses involving covariate adjustment may enhance the credibility of the trial results and should be pre-specified in an analysis plan a priori. A question often asked is which analysis should be considered primary. In this workshop alternative methods for baseline adjustment including direct adjustment, regression modeling examining least square means and propensity scores will be examined to determine the effect of baseline adjustment on estimates of cost-effectiveness from multiple analytic viewpoints using a case example. Discussion will be solicited throughout the presentation. Participants will be asked to provide input on current practice and how they have accounted for treatment group differences on important baseline predictors of cost and effectiveness. Finally, we will describe potential biases that may arise from sample selection bias (in the case of non-randomized, observational studies), and non-random sample attrition in clinical trials.


W23: STRUCTURAL SENSITIVITY ANALYSIS WITH DES
DISCUSSION LEADERS: Jorgen Moller, MSc, Arena Specialist, Caro Research Institute, Concord, MA, USA; H. Baris Deniz, MSc, Researcher, Caro Research Institute, Concord, MA, USA; J. Jaime Caro, MDCM, FRCPC, FAC, President & Scientific Director, Caro Research Institute, Concord, MA, USA

Workshop Purpose: To show that the ISPOR Task Force on Good Research Practices-Modeling Studies which emphasizes the necessity for sensitivity analyses using alternative model structures. This is a cumbersome exercise using traditional modeling techniques, and is therefore not being done to the extent that it should be. The goal of this workshop is to show various approaches to structural sensitivity analysis that can easily be done while modeling with Discrete Event Simulation.
Workshop Description: In their discussion of model structure, the ISPOR Task Force on Good Research Practices –Modeling Studies emphasizes the necessity for sensitivity analyses using alternative model structures. This is a cumbersome exercise using traditional modeling techniques, and is therefore not being done to the extent that it should be. The goal of this workshop is to show various approaches to structural sensitivity analysis that can easily be done while modeling with Discrete Event Simulation. The workshop will assume participants have a basic knowledge of the DES technique. Using a DES example of a pharmacoeconomic problem, the group will interactively experience the difficulties and advantages of three different approaches to structural sensitivity analysis. Specifically, these are: sequencing (letting the user select the structure, very flexible, requires the most adapted programming), flagging (gives typically a few different flows) and the manual technique (reprogramming the model). The workshop will lead to insight in how DES allows for structural sensitivity analysis, and different ways to design the model to make it simple and straight forward to do. Ideally, to make certain that a model is not inherently biased by the way it is structured various models should be built by different teams and compared. As this is not realistic, the next best thing is to fulfill the ISPOR guidelines and build the model in a way that lends itself to modify the structure and thereby the flow of the model.

HEALTHCARE POLICY

W24: REIMBURSEMENT RESTRICTIONS IN GERMANY DUE TO REFERENCE PRICES
DISCUSSION LEADERS: Olaf Pirk, MD, Principal, Fricke & Pirk GmbH - Member of the IMS Health Group, Nuernberg, Germany; Martin Voelkl, MSc, Senior Consultant, Fricke & Pirk GmbH - Member of the IMS Health Group, Nuernberg, Germany

Workshop Purpose: The aim of this workshop is to demonstrate how the reference price system is set up in Germany. The influencing factors which determine reference price and reimbursement level will be presented. Alternative options for setting reference prices will be developed in a common discussion.
Workshop Description: Reference prices are setting the price which the Statutory Health Insurance (SHI) reimburses for a pharmaceutical form. The German reference price system which came into effect in 1989 is very complex and widely unknown. With implementing several changes in the Social Security Code during the recent years, even patented drugs can be taken up in a reference price group. The process of building reference prices is only revealed to a certain extent by the self-governing body that is responsible for introducing reference prices. The workshop demonstrates the various stages of the process beginning from building reference price groups up to setting the reference price level for the specific pharmaceutical form. The effects of the annual reference price adjustments on reimbursement will be demonstrated. Several case-studies will be presented. Participants will be asked to determine the reference price level for a pre-defined reference-price group. Problems of the current system with regard to different reimbursement levels of equivalent therapy options will be outlined. An essential point for the discussion will be the development of alternative options for calculating reference prices.

W25: EARLY ASSESSMENT OF NEW AND EMERGING PHARMACEUTICALS TO SUPPORT HEALTH CARE DECISION-MAKING
DISCUSSION LEADERS: Karla Douw, MSc, HTA consultant, University of Southern Denmark, Odense, Denmark; Sue Simpson, PhD, Research Fellow, University of Birmingham, Birmingham, United Kingdom; Birgitte Bonnevie, MSc, HTA consultant, National Board of Health, Copenhagen, Denmark

Workshop Purpose: The aim of this workshop is to introduce the audience, health care decision makers and industry, to the work of government funded Early Warning/Horizon Scanning Systems and the early assessment of new and emerging pharmaceuticals. We will discuss how these early assessments feed into national, regional, and local decision making on the uptake of new pharmaceuticals, and discuss the methods involved in the early identification and assessment of new pharmaceuticals.
Workshop Description: A number of countries worldwide have established Early Warning systems to identify and assess new and emerging health technologies, including pharmaceuticals. These systems provide early notice to health care decision makers on the likely clinical, financial, organisational, social and ethical impact that the technology will have if it is introduced in the health care system. In other countries, early assessment of new pharmaceuticals is organised in the context of existing Health Technology Assessment agencies. The information is either provided ahead of launch to facilitate planning or just after launch to inform reimbursement or adoption decisions. In all cases the assessment is made at an early stage when evidence is still scarce and uncertain. This workshop will provide an overview of characteristics of these early identification and assessment practices in Early Warning Systems/HTA agencies, and will include specific experiences of two programmes. One is that of the English National Horizon Scanning Centre which provides input into the selection of technologies to be considered by the National Institute of Health and Clinical Excellence (NICE) and the other is that of the Danish Centre for HTA that exclusively assesses new anticancer drugs, and informs a National Cancer Steering Committee. Key questions that will be addressed in the workshop are: Which pharmaceuticals are selected for early assessment? How to assess impact at an early stage when the evidence is scarce and uncertain? What information sources are generally used and what role does industry play in providing information? And finally, how do these early assessments feed into decision making processes? Participants are provided with some recent examples of early assessments of new pharmaceuticals and will be asked to discuss the above mentioned methodological issues


CLINICAL STUDY METHODS/RISK

W26: NUMBER NEEDED TO TREAT: MISLEADING, MISUNDERSTOOD, MISUSED?
DISCUSSION LEADERS: Jørgen Nexøe, PhD, General Practitioner, Senior Researcher, University of Southern Denmark, Odense, Denmark; Peder Halvorsen, PhD-student, General Practitioner, University of Southern Denmark, Odense, Denmark; Palle Mark Christensen, PhD, Chlinical Pharmacologist, University of Southern Denmark, Odense, Denmark; Ivar S Kristiansen, MD, PhD, MPH, Professor, University of Oslo, Oslo, Norway

Workshop Purpose: To explore important limitations and possible alternatives to the NNT for risk communication and medical decision making.
Workshop Description: Number-needed-to-treat ("NNT") has been advocated as the preferred effect measure when communicating effectiveness of drug therapies for chronic diseases such as diabetes or hypertension. Some research, however, indicates that NNT may be difficult to understand, and NNT has important limitations as it is based on one single measurement and does not capture aggregate effects over time. This aspect makes NNT a problematic measure of the effectiveness of interventions against chronic disease processes because here time to adverse event (e.g. heart attack, death) is an essential issue. NNT may leave the impression that adverse events are completely avoided by intervention when in fact they are postponed. Furthermore, NNT may be misused in cost-effectiveness analyses. When such analyses are used to guide reimbursement decisions, the use off NNT could lead to misguided resource allocation and potentially prevent patients from receiving optimal care. Additionally, the statistical properties of NNT are unfavourable. In randomised trials, lay-persons' preferences for therapies have been shown to be sensitive to type of event prevented, type of side effect and costs, but not to effect size when presented as NNT. This indicates that lay-persons simply do not understand NNT. Doctors and health authorities may do well in using NNT with great caution. Postponement of adverse events (death, heart attack, etc.) may be an alternative format to communicate treatment effects. In the workshop, participants will be asked to state preferences for measure of effectiveness of drug therapies for chronic diseases. Results from this survey and previous empirical studies will be presented and discussed with the audience.


PATIENT-REPORTED OUTCOMES METHODS (INCLUDING COMPLIANCE & PREFERENCE STUDIES)

W27: THE COSTS AND BENEFITS OF COMPLIANCE MEASUREMENT FOR ALL STAKEHOLDERS IN HEALTH CARE
DISCUSSION LEADERS: Mark JC Nuijten, MD, PhD, MBA, Consultant, Ars Accessus Medica/Erasmus University Rotterdam, Amsterdam, Netherlands

Workshop Purpose: The aim of this workshop is to illustrate that the health economic consequences of improving compliance may have substantial consequences for all stakeholders.
Workshop Description: We will describe the critical decision criteria for the various stakeholders in the reimbursement process of new innovative pharmaceuticals, the end-users (patients and physicians), and the pharmaceutical industry. The introduction of a new innovative drug with a premium price over standard treatment, usually leads often to a conflicting interest between the various stakeholders: an increase of health care costs, and especially drug costs versus a clinically superior treatment for the patient. The incorporation of the impact of compliance may overcome these contradictory interests. We will present a health economic model for depression, which shows the impact of compliance for all stakeholders. The key outcome of the model is that the investment in compliance programs for innovative pharmaceuticals may lead to a win- win situation for all stakeholders. The improvement of compliance increases total sales for the new pharmaceutical, which may be offset by reduction in other health care costs, including drug costs, and indirect costs and gain in QALY's and effectiveness.


W28: USING REAL WORLD DATA TO FOLLOW UP REIMBURSEMENT DECISIONS
Discussion Leaders: Andreas Engstrom, MSc, Health Economist, LFN Pharmaceutical Benefits Board, Solna, Sweden; Gepke Delwel, PhD, Policy Advisor, CVZ Healthcare Insurance Board, Diemen, Netherlands; Johan Brun, MD, Medical Director, Pfizer AB, Sollentuna , Sweden

Workshop Purpose: To discuss the uncertainties decision makers face and what type of follow up data they want to see and to give examples of how pharmaceutical companies have attempted to solve the problem of collecting and analysing the information requested by the decision maker.
Workshop Description: A large number of reimbursement agencies and managed care groups now require pharmaceutical companies to submit health economic evaluations as part of the reimbursement process. The information on the effect of the treatment available when the reimbursement application and decision is made is however often limited to data from the clinical trials used for regulatory approval. The reimbursement decision maker may feel that the information from the trial setting does not accurately reflect how the product will be used and work in actual clinical practice. In addition assumptions often have to be made about the economic variables since these were not gathered in the clinical trial or based on other studies. Since the health economic models are based on these data or assumptions about them, they too often have large uncertainties. Decision makers have to make decisions on whether to reimburse or not, based on less information than they would have preferred. One solution for a decision maker is to let a product onto the market and given reimbursement status and then review the reimbursement decision as more information becomes available. The pharmaceutical company may be required to present additional data or to update the economic model later on as more information on the costs and effects of the product in clinical practice becomes available. This new data can take many forms , from post launch clinical studies to registry studies and analysis of sales data. Participants are invited to share their own experiences with these issues and discuss how payers and the pharmaceutical industry can cooperate to develop the necessary data to show whether a product is cost-effective or not.
 

9th Annual European Congress Main Page