ISPOR 16th Annual International Meeting: WORKSHOP PRESENTATIONS
 

WORKSHOP PRESENTATIONS

WORKSHOPS - SESSION I Monday, May 23, 2011: 3:00 PM-4:00 PM

Clinical Outcomes Research

W1: CONDUCTING & INTERPRETING INDIRECT TREATMENT COMPARISON AND NETWORK META-ANALYSIS: LEARNING THE BASICS
Discussion Leaders: Jeroen P. Jansen, PhD, MSc, Research Director, Mapi Values, Boston, MA, USA; Neil Hawkins, PhD, Director, Oxford Outcomes Ltd, Oxford, UK; Joseph C. Cappelleri, PhD, MPH, Pfizer Inc, New London, CT, USA; Rachael Fleurence, PhD, Director, Oxford Outcomes, Bethesda, MD, USA
PURPOSE: During this workshop, the members of the ISPOR Task Force on Indirect Treatment Comparison Good Research Practices Leadership Group will present good research practices in conducting indirect treatment comparison and network meta-analysis studies and interpreting these studies. This workshop will include the following: a) an overview of how networks of randomized controlled trials allow for multiple treatment comparisons of competing interventions, b) key good research practices on the synthesis of evidence; and c) key statistical methodological issues. An illustrative example will be used to demonstrate practical issues, followed by an audience open discussion. The Report of the ISPOR Task Force on Indirect Treatment Comparison Good Research Practices Part I and Part II will be available for attendees.
DESCRIPTION: Evidence-based health care decision-making requires comparisons of all relevant competing interventions. In the absence of randomized controlled trials involving a direct comparison of all treatments of interest, indirect treatment comparisons and network meta-analysis provide useful evidence for judiciously selecting the best choice(s) of treatment. Mixed treatment comparisons, a special case of network meta-analysis, combine direct with indirect evidence for particular pairwise comparisons thereby synthesizing a greater share of the available evidence than traditional meta-analysis. The Report of the ISPOR Task Force on Indirect Treatment Comparison Good Research Practices (ITC GRP Report) provides guidance on the interpretation of indirect treatment comparisons and network meta-analysis to assist policy-makers & health-care professionals in using its findings for decision-making (ITC GRP Report - Part 1) and a guidance on conducting indirect treatment comparisons and network meta-analysis (ITC GRP Report – Part 2).

Economic Outcomes Research

W2: CHALLENGES AND SOLUTIONS IN CONDUCTING RETROSPECTIVE HEALTH ECONOMICS AND OUTCOMES RESEARCH STUDIES FOR ORPHAN DRUGS AND DISEASES
Discussion Leaders: Peter Sun, MD, PhD, Vice President, Health Economics & Outcomes Research, Kailo Research Group, Fishers, IN, USA; Zhimei Liu, PhD, Associate Director, Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA; Amy Guo, PhD, Senior Director, Health Economics & Outcomes Research, US CD&MA, Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA; Jie Zhang, PhD, Director, Novartis Pharmaceuticals Corporation Medical, East Hanover, NJ, USA
PURPOSE: 1) To examine major challenges to the use of retrospective health economics and outcomes research studies for orphan drugs and diseases; 2) To discuss potential solutions that researchers can use to overcome these challenges; and 3) To enable audience to embark on retrospective health economics and outcomes research studies for orphan drugs and diseases with adequate selection of empirical data bases, study approaches and analytical tools.
DESCRIPTION: As more and more payers seek for real-world information on comparative effectiveness of orphan drugs, the demand for retrospective health economics and outcomes research studies for orphan drugs or diseases has increased dramatically. However, such studies often face daunting challenges in data acquisition, study design and analytical methods. This workshop will examine these major challenges and pitfalls, so that audience can be well informed before they embark on a retrospective health economics and outcomes research study for an orphan drug or disease. Further, the workshop will discuss potential solutions to these challenges and recent methodological advance in this area. Finally the workshop will use a real-world case to demonstrate the challenges and solutions in conducting retrospective health economics and outcomes research studies for an orphan drug or disease. At the end of the workshop, audience will be offered an exercise opportunity to apply these solutions to a hypothetical research topic for an orphan drug or disease, so that they are ready to embark on their own retrospective data analysis of orphan drugs or diseases in the future.

Health Care Policy Development Using Outcomes Research

W3: PATIENT-CENTERED BENEFIT-RISK ANALYSIS: THE CASE FOR ANALYTIC HIERARCHY PROCESS
Discussion Leaders: Maarten J. IJzerman, PhD, Professor & Chair, Department of Health Technology & Services Research, University Twente, Enschede, The Netherlands; John F.P. Bridges, PhD, Assistant Professor, Department of Health Policy and Management, Johns Hopkins Bloomsberg School of Public Health, Baltimore, MD, USA; Sonal Singh, MD, MPH, Assistant Professor, Division of General Internal Medicine, Johns Hopkins School of Medicine, Baltimore, MD, USA
PURPOSE: The purpose of this interactive workshop is to demonstrate the relative merits of using Analytical Hierarchical Process (AHP) in health care, with a special emphasis on measuring patient preferences and informing regulatory decisions via benefit-risk analysis.  Participants will gain a deeper appreciation of the method through a hands-on exercise that will demonstrate how the method works and how some simple calculation can inform medical decision making. The session will be valuable to conference attendees interested in regulation, benefit-risk assessment, the measurement of preferences and decision making. 
DESCRIPTION: Regulators of medicines are facing increasing pressure to make their deliberations concerning the benefits and risk more transparent, scientific and patient-centric. Given that both benefits and risks are often measured via multiple (often interrelated) outcomes, any attempt to estimate preferences across benefits and risk must take into account the multiple criteria available. Analytic Hierarchy Process (AHP) is a technique that stems from operations research and decisions sciences. In AHP a decision problem is structured into a hierarchy of attributes followed by a series of pair-wise comparisons to obtain attribute weights directly from patients (and potentially any other stakeholder). AHP can also be applied using a group-decision mode, involving stakeholder panels. By using remote units, panel members provide their scores visible to other panel members. This may facilitate group discussion and supports consensus building. In recent years AHP is increasingly being applied to health care decisions, and health care regulators (e.g. FDA). HTA agencies (e.g. IQWiG and NICE) have expressed interest in evaluating and adopting the method to guide their benefit-risk and benefit-cost analyses respectively. AHP also has many other potential applications in assessing outcomes, including methods to inform shared decision making, hospital management decisions, and the development of new interventions (especially as part of early health technology assessment).

Patient-Reported Outcomes & Preference-based Research

W4: CHOOSING PRO STATISTICAL ENDPOINTS TO MAXIMIZE SUCCESSFUL OUTCOMES
Discussion Leaders: Dennis D. Gagnon, MA, MABE, Senior Director, Strategic Consulting, Thomson Reuters, Santa Barbara, CA, USA; Margaret Rothman, PhD, Senior Director, Worldwide Market Access, Worldwide Market Access, Johnson & Johnson Pharmaceutical Services, LLC, Raritan, NJ, USA
PURPOSE:   Patient-reported outcomes (PRO) researchers are frequently called upon to recommend statistical endpoints for clinical trials.  The importance of the relationship between PRO statistical endpoints used in clinical trials, study design and research hypotheses is frequently overlooked.  This workshop will focus on understanding this relationship to improve trial outcomes.
DESCRIPTION:  There has been increasing attention to the choice of the PRO concept to include in clinical studies which is often critical to demonstrating the treatment benefit of an intervention.  Yet not enough attention is given to specifying the PRO statistical endpoint for testing that treatment benefit.  Choosing the correct PRO statistical endpoint is as important as choosing the correct concept to be measured.  Often protocols specify that PRO change scores will be compared between treatment groups, where another statistical endpoint would be more appropriate.  This workshop will be broken into four sections.  Section I (15 minutes):  Several possible PRO statistical endpoints will be discussed beyond simple change scores including proportion of responders, time-to-response, area under the PRO curve, average follow-up, worst score during follow-up, etc.  Section II (15 minutes):  The appropriateness of the PRO statistical endpoints for various study designs and research hypotheses will be explored; with special attention to the fact that change scores are not always appropriate, and thus the US FDA’s suggestion for presentation of cumulative distribution of change scores may not always be appropriate.  Section III (15 minutes): Three case studies will be presented; two based upon personal experience of the presenters (involving clinical trials in oncology and anti-virals), and one based upon the published approval process for dalfampridine in multiple sclerosis.   Section IV (15 minutes): Three examples will be presented to the audience, where they choose the appropriate PRO statistical endpoint given the study designs and research hypotheses presented in the individual vignettes.

Use of Real World Data

W5: USING CMS MEDICARE CLAIMS PUBLIC USE FILES (PUFS) FOR COMPARATIVE EFFECTIVENESS RESEARCH (CER)
Discussion Leaders: Samuel C. "Chris" Haffer, PhD, Program Manager - Information & Methods Group, Office of Research, Development, & Information, Centers for Medicare and Medicaid Services (CMS), Baltimore, MD, USA; Craig G. Coelen, PhD, President, IMPAQ International, LLC, Columbia, MD, USA; Elizabeth C. Hair, PhD, Senior Research Scientist, Public Health, NORC at the University of Chicago, Bethesda, MD, USA; Jane Hyatt Thorpe, JD, Associate Research Professor, Department of Health Policy, School of Public Health and Health Services, George Washington University Medical Center, Washington, DC, USA
PURPOSE: To introduce and discuss the CMS comparative effectiveness research (CER) public use data pilot project, especially focus on the demonstration of Medicare claims public use files (PUFs).
DESCRIPTION: Recently, comparative effectiveness research (CER) has been getting more attention since Congress passed the American Recovery and Reinvestment Act of 2009. Studies using claims data offer the advantage of large sample sizes and actual performance in a real-world setting for CER. CMS Medicare serves 46 million beneficiaries and is a rich claims data source for CER. The goal of the CMS CER Public Use Data Pilot Project is to create, disseminate, and support privacy-protected Medicare claims PUFs that can be accessed to conduct research, without the costs and time delays in access and user fees/agreements that impede current researchers' efforts. Currently, Medicare claims level data are only available through data use agreements with CMS. All Medicare claims PUFs have personally identifying variables removed in compliance with the HIPAA Privacy Rule, but have also been tested to ensure that individuals cannot be re-identified. The discussion leaders in this workshop will discuss how to access, how to use, and the utility of the Basic Stand-Alone PUFs for calendar year 2008 Medicare claims using a 5% sample of beneficiaries, as well as a preliminary discussion of the Enhanced and Linked PUFs for 3 years of Medicare claims. In addition, the audience will learn about the legal, ethical, and IT challenges to producing and using PUFs of claims data. The workshop will end with an interactive moderated discussion with the audience focusing on feedback on the proposed claims PUFs and the needs of the research community when conducting claims level data analyses. This workshop is appropriate for physicians, policymakers, health care decision-makers, and others interested in conducting claims level research with PUF data.

W6: POWERFUL DATA, MEANINGFUL ANSWERS: INTRODUCTION TO THE HEALTHCARE COST AND UTILIZATION PROJECT
Discussion Leaders: Claudia Steiner, MD, MPH, Senior Research Physician, Center for Delivery, Organization and Markets (CDOM), Agency for Healthcare Research and Quality (AHRQ), Rockville, MD, USA; Elizabeth Stranges, MS, Research Leader, Thomson Reuters, Evanston, IL, USA
PURPOSE: This workshop will introduce health services researchers and health care decision-makers to Healthcare Cost and Utilization Project (HCUP) and provide them with the foundational resources to apply HCUP data to their research interests.
DESCRIPTION: This workshop will provide in-depth exposure to several Healthcare Cost and Utilization Project (HCUP) resources. HCUP is a family of health care databases, software tools, research publications, and support services created through a Federal-State-Industry partnership and sponsored by the U.S. Department of Health and Human Services (DHHS), Agency for Healthcare Research and Quality (AHRQ). HCUP captures information on 95 percent of all hospital stays in the United States, and is the largest collection of multi-year, all-payer, encounter-level data publicly available. Current databases include a mix of national and statewide inpatient, ambulatory surgery, and emergency department data. HCUP data and products support cutting-edge health services research and policy analyses. Hospital administrative data, such as HCUP, have been used in health economic and outcomes investigations because they contain large numbers of cases for specific conditions and procedures and because charge information is available. Faculty will provide an overview of HCUP databases, tools, and learning resources and then discuss attendee research ideas and how they could be investigated using HCUP products. At the end of this session, participants will have an improved understanding of the following: 1) How HCUP databases, tools, and learning resources can be used to enhance the quality of clinical and health services research; 2) How to access HCUP data via HCUPnet – a free on-line data query system – as well as through the HCUP Central Distributor; and 3) How to maximize the value of the HCUP databases by applying free software tools that help categorize, transform, and enhance existing HCUP data files.

WORKSHOPS - SESSION II Monday, May 23, 2011: 4:15 PM-5:15 PM

Clinical Outcomes Research

W7: NEW METHODS TO ADJUST FOR SELECTIVE CROSSOVER IN SURVIVAL ANALYSIS: IN ASSESSMENTS OF COST-EFFECTIVENESS OF CANCER THERAPIES
Discussion Leaders: Thomas E. Delea, MSIA, Senior Research Consultant, Policy Analysis Inc. (PAI), Brookline, MA, USA; Mei-Sheng Duh, MPH, ScD, Managing Principal, Analysis Group, Inc., Boston, MA, USA; Lee-Jen Wei, PhD, Professor, Biostatistics, Harvard University, Boston, MA, USA; James Robins, MD, Professor of Epidemiology, Harvard School of Public Health, Boston, MA, USA
PURPOSE: Selective crossover is common in trials of novel cancer therapies wherein ethical and/or practical considerations require that the option to change from control to experimental therapy is available upon disease progression or based on findings of benefit for experimental therapy in interim analyses. Also, patients in both control and active treatment arms may receive other non-study therapies following disease progression.  In such situations, the effect of treatment on survival estimated by intent-to-treat (ITT) analyses may not provide an accurate estimate of effectiveness in typical clinical practice where selective crossover/receipt of non-study treatment does not occur.  The UK National Institute for Health and Clinical Excellence (NICE) has recently provided positive recommendations for novel cancer therapies after consideration of survival estimates using the Rank Preserving Structural Failure Time (RPSFT) and/or Inverse Probability of Censoring Weight (IPCW) methods to control for crossover and receipt of non-study therapy.  This workshop will review advances in survival analyses adjusting for crossovers.
DESCRIPTION: The workshop will cover four main topics.  First, biases with traditional approaches for assessing treatment effects on survival in the presence of selective crossover/receipt of non-study therapy will be discussed, including ITT analysis with no adjustment for crossover/non-study treatment, exclusion of patients with crossover/non-study treatment, censoring of patients at crossover/receipt of non-study treatment, and inclusion of crossover/non-study treatment as a time-dependent covariate in Cox regression analysis. Second, new methods for adjusting for selective crossover and receipt of non-study therapy, RPSFT and IPCW, will be presented. Third, using recent assessments of novel cancer therapies for NICE as examples, the effects of using different methods on estimates of overall survival and cost-effectiveness will be demonstrated.  Fourth, unanswered questions, controversies and areas for future research will be discussed.  The audience will be invited to participate through discussion of case examples and of statistical methods employed.

Economic Outcomes Research

W8: BRICS & MORTAR: BUILDING EMERGING MARKETS INTO GLOBAL HEOR PROGRAMS
Discussion Leaders: David Thompson, PhD, Senior Vice President, Health Economics & Strategic Consulting, i3 Innovus, Medford, MA, USA; Gabriela Tannus, MSc, President, Axia.Bio, Sao Paulo, SP, Brazil; Jianwei Xuan, PhD, Senior Director/Team Leader, Pfizer, New York, NY, USA
PURPOSE:  Rapid economic growth in the so-called “BRIC countries” (Brazil, Russia, India, China) and neighboring regions has led pharmaceutical manufacturers to place greater priority on emerging markets.  The purpose of this workshop is to discuss issues that arise from incorporating emerging markets into global HEOR programs.  The workshop will be of interest to health economics managers in the pharmaceutical industry, applied researchers in academic and consulting environments, and policy makers.
DESCRIPTION: Historically, health technology assessment did not play as prominent a role in decisions related to the adoption, pricing, and reimbursement of new pharmaceutical products in emerging markets as it did in the major markets.  Manufacturers accordingly developed their global HEOR programs with an emphasis on meeting the needs of decision makers in North America and Europe.  Now, with the rising prominence of the BRIC countries and other markets in Latin America, Eastern Europe, and Asia, there is a growing need to address the HTA needs of countries beyond the major markets.  But doing so presents considerably more challenges than just expanding the latitude and longitude of global data collection efforts.  Differences across markets in HEOR expertise, data availability, population health priorities, and health spending constraints are just a partial listing of the issues to address in pursuing HEOR programs in emerging markets.  Despite these differences, some general fundamentals underpin successful design and implementation of global HEOR programs, including the need to “think globally, but act locally”, by planning and coordinating centrally but utilizing local health economic experts to collect data, conduct analyses, and prepare country-specific dossiers.  This workshop will use Brazil and China as case studies to illustrate commonalities and differences in HTA practices and associated implications for global HEOR programs.  Workshop participants will be encouraged to share their thoughts and experiences related to the design and conduct of HEOR in emerging markets.

Health Care Policy Development Using Outcomes Research

W9: USING COMPARATIVE EFFECTIVENESS RESEARCH (CER) TO SUPPORT A VALUE BASED HEALTH CARE SYSTEM, EXAMPLES FROM THE UNITED STATES AND EUROPE
Discussion Leaders: Rachael Fleurence, PhD, Director, Oxford Outcomes, Bethesda, MD, USA; Feng Pan, PhD, Senior Research Associate, Center for Health Economics and Science Policy, United BioSource Corporation, Bethesda, MD, USA; Corinna Sorenson, MPH, MHSA, Research Officer, European Health Policy, Health & Social Care, London School of Economics and European Health Technology Institute for Socio-Economic Research, London, UK
PURPOSE: To provide outcomes researchers with a in-depth understanding of how comparative effectiveness research (CER) can support the move towards a value based health care system in both the United States and Europe.
DESCRIPTION: CER aims to generate real-world comparative evidence that is meaningful to patients, providers and decision-makers. In this workshop, we focus on what it means in practical terms to use CER results to move towards a value based health care system. A value based health care system is one where we aim to achieve the highest value per dollar spent. This kind of system rewards performance based on outcomes that matter to the patient and aligns incentives accordingly. In the first part of this workshop, we present the concept of value based health care and review coverage and reimbursement strategies that can help translate CER results into quality improvements in the health care system and increased efficiency. We will review approaches such as value based insurance design (VBID) which aims to distinguish low value from high value care, value based pricing (VBP) and coverage with evidence development (CED). We will illustrate how CER studies may contribute to VBID by presenting some real world examples and we will demonstrate how this type of real world evidence is useful for supporting value based decisions. Finally, we will review how comparative effectiveness is used in Europe to support value-based health care and provide an overview of the recent developments in the area including the move towards “relative efficacy” and the creation of the Joint Action on HTA.  Participants will get a thorough understanding of how the results of CER can be used to support a value based health care system through examples of value based insurance design in the United States and value based health care initiatives in Europe.

W10: DEVELOPING BETTER EVIDENCE OF PRODUCT SAFETY
Discussion Leaders: Joshua S. Benner, PharmD, ScD, Research Director and Fellow, Engelberg Center for Health Care Reform, The Brookings Institution, Washington, DC, USA; Judy Racoosin, MD, MPH, Sentinel Initiative Scientific Lead, Office of Medical Policy, Center for Drug Evaluation and Research, U.S. Food and Drug Administration, Silver Spring, MD, USA; Jeffrey S. Brown, PhD, Assistant Professor, Department of Population Medicine, Harvard Medical School, Harvard Pilgrim Health Care Institute, Boston, MA, USA; Paul Stang, PhD, Senior Director of Epidemiology, Johnson & Johnson and Principal Investigator, Observational Medical Outcomes Partnership, Foundation for the National Institutes of Health, Philadelphia, PA, USA
PURPOSE: To engage the outcomes research community in a discussion of the FDA’s Sentinel Initiative and related initiatives to improve medical product safety.
DESCRIPTION: In 2007, Congress mandated the U.S. Food and Drug Administration (FDA) to develop a national electronic system to monitor the safety of marketed drugs. In response, FDA launched the Sentinel Initiative, which established a vision and objectives for an active surveillance system that maintains the privacy and security of patients’ protected health information. In less than three years, through its Mini-Sentinel initiative, FDA has established the ability to query the electronic health care data of over 70 million Americans to assess medical product safety.  Mini-Sentinel is FDA’s pilot distributed database system, coordinated by the Harvard Pilgrim Health Care Institute.  It was constructed to monitor the safety of approved medical products using electronic health information in administrative claims, inpatient and outpatient medical records, and registries.  Mini-Sentinel initially focused on developing the ability to use claims data, but clinical information will be incorporated in 2011. Mini-Sentinel will soon begin to actively monitor specific safety questions, including myocardial infarction among users of oral hypoglycemic agents. Methodological questions important to safety surveillance are also being informed by the work of the Observational Medical Outcomes Partnership (OMOP), a public-private partnership. OMOP has evaluated the performance of 12 active surveillance methods using known product-adverse event pairs across 10 disparate observational databases covering 120 million lives.  OMOP has developed a research infrastructure and suite of publicly available tools to advance methodology development and assessment. The governance, scientific operations, and data infrastructure of the eventual Sentinel System will be informed by Mini-Sentinel, OMOP, and other related projects. This workshop will provide an update on these activities and publicly available resources, featuring perspectives from scientists at FDA, Mini-Sentinel, OMOP, and the Brookings Institution. 

Patient-Reported Outcomes & Preference-based Research

W11: PATIENT PREFERENCES AND POLICIES: THE ROLE OF HEALTH-PREFERENCE DATA IN REGULATORY DECISION MAKING
Discussion Leaders: F. Reed Johnson, PhD, Distinguished Fellow and Principal Economist, Health Preference Assessment, RTI Health Solutions, Research Triangle Park, NC, USA; Axel Mühlbacher, PhD, Harkness Fellow in Health Care Policy and Practice, Duke Clinical Research Institute/ Fuqua School of Business, Duke University, Durham, NC, USA; Derek S. Brown, PhD, Research Health Economist, RTI International, Research Triangle Park, NC, USA
PURPOSE: The purpose of this workshop is to appraise the usefulness of patient-preference data in informing public-policy decision making.
DESCRIPTION: Benefit-cost analysis based on aggregating individual preference data is mandated by government policy in many countries and is commonly applied in virtually every other area of applied economics except health.  However, rising health-care costs and increasing pressures to reduce total government expenditures make it increasingly difficult to exempt health from standard societal valuation approaches. This workshop provides new data and three perspectives on this debate. All three applications obtain utility-theoretic preference measures using conjoint or discrete-choice experiment methods and discuss their applicability in specific public-policy contexts.  The discussion will include how a similar methodological impasse in the 1980s was resolved in environmental policy evaluation; recent developments in Germany to identify alternatives to cost per QALY for prioritizing and reimbursing new treatments; and how discrete-choice methods have been used to identify the relative importance and WTP for generic features of vaccines.  Participants will be asked to critique the vaccine instrument and evaluate its relevance for informing public-health decisions about federal vaccination policies.

Use of Real World Data

W12: INVERSE PROBABILITY-WEIGHTED LEAST SQUARES REGRESSION FOR ESTIMATING POPULATION MEAN COSTS OF ALTERNATIVE INTERVENTIONS USING OBSERVATIONAL DATA
Discussion Leaders: Anthony O'Hagan, PhD, Professor, Department of Probability and Statistics, University of Sheffield, Sheffield, UK; Michelle L. Gleeson, PhD, Senior Research Associate, Outcomes Insights, Inc., Westlake Village, CA, USA; Mark D. Danese, PhD, President, Outcomes Insights, Inc., Westlake Village, CA, USA; Robert I. Griffiths, ScD, Vice President, Outcomes Insights, Inc., Westlake Village, CA, USA
PURPOSE: There is great interest in comparing the population mean costs of alternative interventions using observational data.  However, complications arise due to censoring in the data and the need to account for the effects of covariates. The objectives of this workshop are as follows: provide an overview of inverse probability weighting (IPW) as an approach to account for censoring; discuss implementation of IPW least-squares regression to account for the effects of covariates when comparing the costs of alternative interventions; present a case study illustrating the application of IPW least-squares regression to compare the population costs of alternative interventions using Surveillance, Epidemiology, and End Results (SEER)-Medicare data; and describe alternative approaches for calculating confidence intervals around IPW regression coefficients, including a bootstrap approach.
DESCRIPTION: Dr. O’Hagan will provide an overview of estimating costs with censoring including: 1) why censoring is a problem; 2) IPW to correct for censoring; and 3) IPW regression to address the extended problem of comparing the population costs of alternative interventions adjusting for covariates.  Dr. Griffiths will present a case study illustrating the application of IPW partitioned least-squares regression in which clinical, administrative, and claims data from the SEER cancer registry linked to Medicare claims were used to examine associations between granulocyte-colony stimulating factor (G-CSF) use and medical costs after initial adjuvant chemotherapy in early-stage breast cancer.  Following the presentation of the case study, the audience will be asked to participate in an exercise led by Dr. Danese in which, using a small hypothetical cohort of patients, observation partitions are created, and IPWs are assigned to each patient in each partition depending on the vital and censor status of the patient.  Drs. Gleeson and O’Hagan will describe the alternative approaches to calculating confidence intervals for regression coefficients, highlighting their strengths and limitations. 

WORKSHOPS - SESSION III Tuesday, May 24, 2011: 4:00 PM-5:00 PM

Clinical Outcomes Research

W13: USE OF SIMULATION TO INFORM THE DESIGN OF PRAGMATIC COMPARATIVE EFFECTIVENESS TRIALS
Discussion Leaders: David Wilson, MA, Research Scientist, United BioSource Corporation, Lexington, MA, USA; J. Jaime Caro, MDCM, FRCPC, FAC, Senior Vice President of Health Economics, United BioSource Corporation, Lexington, MA, USA; K. Jack Ishak, PhD, MSc, Director, & Research Scientist Biostatistics, United BioSource Corporation, Dorval, QC, Canada; Myoung Kim, PhD, MA, MBA, Director, Health Economics & Outcomes Research, Ortho-McNeil Janssen Scientific Affairs, Raritan, NJ, USA
PURPOSE: Bringing a new drug to market presents great challenges, risks, and cost. Recently, important policy initiatives and discussion in the marketplace have emphasized the need to better understand the effectiveness of interventions in routine practice. The purpose of this workshop is to describe and illustrate how simulation techniques can help address trial design challenges, particularly for pragmatic comparative effectiveness trials.
DESCRIPTION: Trialists face many uncertainties in the planning and design of a new trial. Pragmatic trials present new and possibly greater sources of uncertainty due to the broad study populations and minimally-restrictive protocols they employ. While prior experience, published materials, and expert opinion can help mitigate specific uncertainties, it is impossible to ascertain their combined impact. Trial simulation offers a powerful means of exploring the performance of competing designs in terms of their likelihood of success and efficiency (e.g., power or required duration). Uncertainty elements become parameters in the simulation that can be varied simultaneously to create new potential designs, which can be tested by running multiple study replications. Discrete event simulation (DES) is well-suited for this purpose, as it provides the necessary flexibility to emulate the typical processes in the course of the conduct of a trial. The process of developing and applying a trial simulation model will be described in this workshop. The following topics will be covered: the specification of potential designs, specification of simulation parameters/inputs, identification of appropriate data sources, setting inputs in the absence of empirical evidence is available, measuring the performance of designs, and appropriate handling of uncertainty and variation. The latter is a key issue, as it directly affects measures of performance of the simulated designs. Simulation of a pragmatic clinical trial in the management of chronic low back pain will serve as the basis for discussion.

Economic Outcomes Research

W14: MICROECONOMIC TOOLS FOR UNDERSTANDING, MODELING, AND INFLUENCING HEALTHCARE DECISION MAKING
Discussion Leaders: F. Reed Johnson, PhD, Distinguished Fellow and Principal Economist, Health Preference Assessment, RTI Health Solutions, Research Triangle Park, NC, USA; Juan Marcos Gonzalez, PhD, Research Economist, RTI Health Solutions, Research Triangle Park, NC, USA; Deborah A. Marshall, MHSA, PhD, Associate Professor, Canada Research Chair, Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, ON, Canada
PURPOSE: The purpose of this hands-on workshop is to introduce participants to recent advances in using microeconomic models to understand the determinants of health care decision making.  Session participants will gain understanding of the practical implications of developing and using economic behavioral approaches to evaluating medical treatments and health-promoting interventions.
DESCRIPTION: There is increasing interest in identifying methods of quantifying net clinical benefits that incorporate patients’ behavioral responses to treatment options. Although microeconomic theory is well suited analyze decisions related to consumption of health care services, this approach has not been widely applied in the health care industry. As a result, the evaluation of health care decision making has been largely left in the hands of other disciplines. The workshop will include a discussion of how standard microeconomic models can be used to evaluate households’ health care decisions, including patients’ adherence to treatment, risk aversion and moral hazard in health insurance, and treatment uptake.  Discussion leaders also will respond to criticisms of standard economic models by behavioral economists. The workshop will present empirical examples showing applications of microeconomic models to real-world problems and the opportunities and challenges associated with their use. Participants will complete an exercise to construct a simple microeconomic model of household production and use it to analyze a hypothetical new treatment for a chronic disease that has better efficacy but also has a burdensome mode of administration.  Workshop participants will evaluate how factors related to nonmonetary therapy costs influence the perceived benefit of the medication, the allocation of household time and financial resources, and likely uptake and adherence.   The exercise will allow participants to specify assumptions that influence the degree to which treatment characteristics and behavioral responses affect observed health outcomes. They also will assess what data would be required to estimate the model and discuss possible uses of the model to support pharmaceutical decision making.

Health Care Policy Development Using Outcomes Research

W15: PRACTICAL EXPERIENCES WITH THE USE OF BEST-WORST SCALING IN ECONOMIC EVALUATION
Discussion Leaders: Terry Flynn, PhD, Head of Social Policy & Economic Evaluation, Centre for the Study of Choice (CenSoC), University of Technology, Sydney, Sydney, NSW, Australia; John F.P. Bridges, PhD, Assistant Professor, Department of Health Policy and Management, Johns Hopkins Bloomsberg School of Public Health, Baltimore, MD, USA; Christine Poulos, PhD, Senior Economist, Health Preference Assessment, RTI International, Research Triangle Park, NC, USA
PURPOSE: The purpose of the workshop is to discuss practical experience with the use of best-worst scaling (BWS), an emerging theory-driven evaluation tool, in economic evaluation. In additional to a brief overview of the types of BWS methods available, session participants will be invited to participate in several activities and to discuss the psychological processes behind the methods. The workshop aims to offer participants an overview of BWS methods and greater insight into their use in economic evaluation.
DESCRIPTION: Best-Worst Scaling (BWS) is a theory-driven method that is increasingly being used as an evaluation methodology in health services research and pharmacoeconomics. While there have been reviews of the theory and applications of BWS in economic evaluation, many practitioners and policy makers remain unfamiliar with the various types of BWS methods available and their relative pros and cons. Participants in this interactive session will gain a deeper understanding of 1) the three prominent types of BWS methods (cases 1, 2, and 3); 2) how study subjects respond to best and worst choice problems; and 3) the relatively simple analytical methods available to obtain individual utility scores . Practical experiences from three studies using different types of BWS study will also be presented and discussed. The first uses BWS case 1 in an early health technology assessment (horizon-scanning) study of emerging technologies. The second used Case 2 to estimate person-specific PRO and QALY valuations.  Finally, the use of case 3 BWS to augment the traditional conjoint analysis data to accommodate a larger number of study attributes is discussed.

W16: THE EVOLVING ROLE OF THE AGENCY FOR HEALTHCARE RESEARCH AND QUALITY (AHRQ) IN COMPARATIVE EFFECTIVENESS RESEARCH (CER)
Discussion Leaders: Jean Slutsky, PA, MSPH, Director, Center for Outcomes and Evidence, Agency for Healthcare Research and Quality, Rockville, MD, USA; Nina A. Thomas, MPH, Vice President, Clinical Affairs & Health Economics, Doctor Evidence, LLC, New York, NY, USA; Steven Blume, MS, Research Scientist, Center for Health Economics and Science Policy, United BioSource Corporation, Bethesda, MD, USA
PURPOSE: To describe the current and future CER activities of the Agency for Healthcare Research and Quality (AHRQ) and discuss the implications for manufacturers
DESCRIPTION: AHRQ is mandated to support development, synthesis, and dissemination of available scientific evidence of health care interventions and prevention, including research and analytic methods or systems for rating the strength of scientific evidence.  The President’s FY2011 budget requested a total of $600 million for AHRQ, $300 million of which is intended specifically for comparative effectiveness research, a 50% increase above FY2010.  An additional $175 million of HHS money will be administered by AHRQ.   AHRQ supports multiple evidence development and synthesis activities through the US Preventative Services Task Force and the Effective Health Care program (EHC).  First authorized in 2003, EHC includes multiple Evidence-based Practice Centers (EPCs) that conduct systematic literature reviews to determine the state and quality of evidence for specific clinical questions, as well as Centers for Education and Research on Therapeutics (CERTs) and the DEcIDE Network (Developing Evidence to Inform Decisions about Effectiveness).  An important aspect of the program is dissemination, tailoring findings for clinicians, decision makers, and patients.   More recently, using funds from the 2009 Recovery Act, AHRQ will be funding CER trial methods development, CER trials, registry studies, and data infrastructure.  The workshop will describe these activities as well as sample projects and the opportunities for industry input.  The evolving relationship of AHRQ efforts to other CER initiatives, especially the health reform-created Patient-Centered Outcomes Research Institute (PCORI), will be discussed.  The growing role of evidence reports and CER for U.S. market access will also be discussed from both government and industry perspectives.   Given the scope of AHRQ activities, it is expected that many audience members will have questions and comments and at least 20 minutes will be allotted for such discussion.

Patient-Reported Outcomes & Preference-based Research

W17: PATIENT-REPORTED OUTCOME (PRO) ASSESSMENTS IN CLINICAL TRIALS: NAVIGATING THE EMA AND FDA REGULATORY FRAMEWORK
Discussion Leaders: Ingela Wiklund, PhD, Senior Research Leader, Center for Health Outcomes Research, United BioSource Corporation, London, UK; Olivier Chassany, PhD, MD, Medical Manager, Department of Clinical Research and Development, Assistance Publique-Hopitaux de Paris, Paris, France; Kathleen W. Wyrwich, PhD, Senior Research Leader, Center for Health Outcomes Research, United BioSource Corporation, Bethesda, MD, USA
PURPOSE: The purpose of this workshop is to provide the requirements for a successful submission of patient-reported outcomes (PRO) endpoints to EMA and FDA in support of label claims as outlined in the respective FDA and EMA documents by navigating the communalities and differences in their respective clinical trials framework. 
DESCRIPTION: EMA published a Reflection paper on the regulatory guidance for the use of health related quality of life (HRQL) measures in the evaluation of medicinal products in 2005, while the FDA first published its Draft Guidance on Patient-Reported Outcome Measures in 2006 followed by the Final Guidance for Industry Patient-Reported Outcome Measures: Use in Medical Product Development to Support Labeling Claims in 2009. The development of these guidelines was driven by an increasing number of submissions based on PROs and HRQL measures used as primary or key secondary endpoints to support label claims. Since their release these documents have attracted considerable attention, including efforts to successfully harmonize their requirements for use of PROs in international trials. While there are differences between the two documents, the similarities are many, especially around trial design, implementation of patient-based data collection during the trial, and analysis of  PRO data. With increasing collection of patient-based data in global trials, the importance of navigating the requirements for instrument development history, psychometric documentation of PRO and HRQL data, the analytical approaches, and the format of respective the briefing books and dossier submissions have escalated. This workshop will detail the  similarities and differences between criteria outlined by EMA and FDA, and how to approach harmonization from the qualitative, quantitative, analytical, design and data-reporting perspectives, and potential implications of outstanding regulatory discrepancies will be examined. Finally, clarification of evidentiary requirements and approaches to data presentation to improve efficiency and relevance of PRO and HRQL endpoints will be discussed.

Use of Real World Data

W18: METHODS UTILIZING LONGITUDINAL DATA TO ESTIMATE OF CASUAL EFFECTS IN HEALTH ECONOMICS: ESTIMATION OF TREATMENT EFFECT USING OBSERVATIONAL DATA
Discussion Leaders: William Crown, PhD, President, i3 Innovus, Waltham, MA, USA; Nilay Shah, PhD, RPh, Associate Consultant, Mayo Clinic, Rochester, MN, USA; Henry J. Henk, PhD, Director, HEOR, i3 Innovus, Eden Prairie, MN, USA
PURPOSE: The purpose of this workshop is to 1) provide a brief review of the interpretation of treatment effects captured through typical analytic methods (e.g., cross sectional data analysis) for analyses of observational data and 2) a review of longitudinal (panel) data analysis methods for CER analyses of observational data. 
DESCRIPTION: Some estimates suggest that as much as one-third of all health care expenditures generate no clinical benefit (Fisher, Wennberg, Stukel, et al. 2003).  Hence, with the goal to stop treating patients with therapies that generate little, or no, clinical benefit it is not surprising that comparative effectiveness research has received considerable attention in the health reform discussions currently taking place in the United States. A variety of methods may be used to support CER including systematic reviews and meta analyses, head-to-head clinical trials, registries, and analyses of retrospective claims data, medical records, and survey data.   It is reasonable to expect, however, with the emphasis of CER on the real world effectiveness and safety of treatments, that increased emphasis will be placed on observational research methods (e.g., analysis of observational data).  The purpose of this workshop is to 1) provide a brief review of the interpretation of treatment effects captured through typical analytic methods (e.g., cross sectional data analysis) for analyses of observational data, and 2) a review of longitudinal (panel) data analysis methods for CER analyses of observational data.  Although these methods have not been commonly used for outcomes research, they offer the opportunity to better estimate causal treatment effects than the typical analytic methods that rely on cross sectional data. Methodological and all other issues associated with the use of these methods will be discussed. Audience participation will be encouraged through the use of a real-world example.

WORKSHOPS - SESSION IV Wednesday, May 25, 2011: 1:45 PM-2:45 PM

Clinical Outcomes Research

W19: IMPROVED INDIRECT TREATMENT COMPARISONS FOR COMPARATIVE EFFECTIVENESS RESEARCH
Discussion Leaders: James Signorovitch, PhD, Manager, Analysis Group Inc, Boston, MA, USA; Robert Navarro, PharmD, President, Navarro Pharma, LLC, Green Cove Springs, FL, USA; Keith Betts, PhD, Associate, Analysis Group Inc, Boston, MA, USA; Eric Q. Wu, PhD, Managing Principal, Analysis Group, Inc., Boston, MA, USA
PURPOSE: Indirect treatment comparisons play an essential role in comparative effectiveness research (CER) and health technology assessment (HTA). When treatments have not been directly compared to all clinically and economically relevant comparators in randomized trials, indirect comparisons may provide the only comparative evidence for decision makers. This workshop will provide a framework for critically evaluating indirect comparisons.  We also introduce new methods that can improve the reliability of indirect comparisons. 
DESCRIPTION: To provide reliable evidence for CER and HTA, indirect comparisons should thoroughly adjust for differences between treatment populations.  We provide a framework for critically evaluating indirect comparisons: 1) Given the available data, was appropriate adjustment methodology used; 2) Are results reported in sufficient detail for evaluation; 3) Given appropriate methodology and reporting, what can the data tell us about the reliability of the results?  When subjected to this framework, traditional methods for indirect comparison will be shown to often rely on assumptions that are unnecessarily strong and not generally accepted for comparisons of non-randomized treatment groups.  In particular, indirect comparisons that attempt to adjust for differences across trials by measuring treatment effects relative to a common comparator (e.g, odds ratios vs. placebo) can remain biased by cross-trial differences in the comparator arm outcomes.  We will describe a new approach to adjusted indirect comparison that avoids such biases by more thoroughly adjusting for cross-trial differences in comparator arm outcomes.  The new method contains the traditional adjusted indirect comparison as a special case, and can detect situations in which the traditional approach does not adequately fit the data.  Adjustment for information beyond comparator arms responses will also be discussed. Analyses of real and simulated data will show that the new approach can correctly detect a treatment’s superiority when a traditional adjusted indirect comparison incorrectly detects inferiority.

Economic Outcomes Research

W20: STOCHASTIC MODELING IN PHARMACOECONOMICS – COMMON MISTAKES AND HOW TO AVOID THEM
Discussion Leaders: Francisco J. Zagmutt, DVM, MPVM, Managing Partner, EpiX Analytics, Boulder, CO, USA; Huybert Groenendaal, PhD, MSc, MBA, Managing Partner, EpiX Analytics, Boulder, CO, USA; Jane Castelli-Haley, MBA, Director, Health Economics & Outcomes Research, Teva Neuroscience, Inc., Kansas City, MO, USA
PURPOSE: The purpose of this workshop is to discuss the most common mistakes that practitioners make when implementing stochastic models in pharmacoeconomics, and the good modeling practices that can be used to prevent them.
DESCRIPTION: Stochastic modeling is becoming an increasingly important technique in outcomes research, as it allows for the incorporation of uncertainty in parameters that otherwise have to be assumed as known and for the modeling of random or uncertain events. Implementing stochastic models, however, can be difficult and requires good knowledge of probability theory and statistics. Outcomes research practitioners are often first exposed to stochastic modeling by converting previously deterministic models into stochastic ones. This process can result in useful and insightful new models, but there are several traps and mistakes that are frequently made when building stochastic models which can make a model even less insightful than a deterministic model. Even the most seasoned economist or statistician can make these mistakes, as some of them are not easily apparent when using software such as spreadsheets or script-based modeling platforms. During this hands-on workshop, we will use real-life examples to demonstrate some of these principles and to start a dialog about ways to avoid these mistakes. Examples of correct approaches for the modeling mistakes we will discuss will include the multiplication of probability distributions, confusion of uncertainty and variability, incorrect use of distributions, and the incorrect combination of different sources of information. The audience will be encouraged to share their experiences in this subject and to propose their solutions to implement sound stochastic models. The target audience for this workshop is anyone with an interest in stochastic modeling, from people new to the discipline to seasoned modelers that would like to participate in an open debate about common modeling mistakes and correct approaches.

Patient-Reported Outcomes & Preference-based Research

W21: IMPLEMENTING EPRO IN A GLOBAL CLINICAL TRIAL ENVIRONMENT
Discussion Leaders: Jean Paty, PhD, Founder & Senior Vice President, Scientific, Quality & Regulatory Affairs, Invivodata, Inc., Pittsburgh, PA, USA; Bob Young, Invivodata, Inc., Pittsburgh, PA, USA; Felicia Bergstrom, Covance, Leeds, West Yorkshire, UK
PURPOSE: This workshop will focus on the issues surrounding implementation of electronic patient reported outcomes (ePRO) in global clinical trials.  The purpose of this workshop is to demonstrate that ePRO is now a viable, accepted, and sometimes preferred, method of capturing PRO data.  A second purpose is to clearly articulate the challenges and successes when implementing ePRO in a global environment.
DESCRIPTION: ePRO is quickly becoming the standard method of implementing PRO instruments in clinical trials.  While historically ePRO was used for take home diaries, it is also now being used to administer traditional PRO instruments at the investigative site.  The primary motivators in both contexts are improved data quality and efficiency (e.g., no data entry and less missing data).   The reach of ePRO has also extended to large global trials.  One should not underestimate the challenges of implementing ePRO in a global trial, however, even though there is significant benefit with respect to the data. In this workshop, the conceptual and practical issues in moving from paper to an ePRO platform for a global clinical trial will be articulated.  One presentation will describe how to ensure that moving from paper to electronic PRO does not change the nature of the data being collected.  A second presentation will discuss the pragmatic matters of implementing ePRO technology in many languages across a host of countries.  Finally, a third presentation will provide feedback on real-life experience from ePRO users of their experiences to date.

Use of Real World Data

W22: ELECTRONIC HEALTH RECORDS AND MEANINGFUL USE: OPPORTUNITIES FOR EVOLUTION IN COMPARATIVE EFFECTIVENESS RESEARCH
Discussion Leaders: Gregory W. Daniel, PhD, RPh, MPH, Vice President, Government and Academic Research, HealthCore, Inc., Wilmington, DE, USA; Jane Griffin, RPh, Director, Research Client Connect, Cerner Corporation, Kansas City, MO, USA; Eugene Rich, MD, Senior Fellow and Director, Center on Health Care Effectiveness, Mathematica Policy Research, Washington, DC, DC, USA; Rolin Wade, RPh, MS, Healthcare Executive and Principal Investigator, Cerner LifeSciences, Beverly Hills, CA, USA
PURPOSE: This workshop will explore the potential impact of the expansion of electronic health records (EHRs) and federal meaningful use requirements on comparative effectiveness research (CER).
DESCRIPTION:  CER, which includes comparative observational studies and retrospective database analyses, will likely become increasingly accepted, even mandated by clinicians and payers to support existing and novel treatments. Recent federal legislation provides $19.2 billion in incentives to providers to adopt certified EHR technology, promising to offer new opportunities for conducting CER. Currently, large amounts of comprehensive, high-quality clinical data for real-world CER studies can be challenging to obtain: data on potentially important elements describing some types of clinical innovation are not readily available in claims databases, while EHRs lack the broad view of resource utilization that claims offer. While its primary intent was to improve patient care via adoption of EHRs, the Health Information Technology for Economic and Clinical Health Act will facilitate the generation of “measures” to justify meaningful use, enhancing the capability of the EHR to capture detailed clinical information. During the workshop, experts in EHR design, EHR and claims data sets, and CER will explore opportunities and challenges in leveraging the clinical detail available in the medical record to answer outcomes research questions that may be out of reach through the use of traditional retrospective claims datasets alone. They will compare and contrast the utility for CER of claims data with EHR data (as well as personal health records). They will also consider how changes in the structure and content of EHRs over time might be helpful to future CER, and the potential role of meaningful use criteria in these changes. Workshop leaders will solicit examples of relevant projects from attendees and facilitate discussion of how advanced EHRs may impact the study design and choice of data sources.

W23: PRACTICAL APPROACHES FOR SYSTEMATIC ANALYSIS OF OBSERVATIONAL DATA; REAL WORLD CASE STUDIES FROM THE PHARMACEUTICAL INDUSTRY
Discussion Leaders: Stephanie Reisinger, Senior Director, United BioSource Corporation, Harrisburg, PA, USA; Gregory E. Powell, PharmD, MBA, Manager, GCSP, Research and Development, GlaxoSmithKline, Research Triangle Park, NC, USA; David Miller, ScD, SM, Director of Risk Management and Pharmacoepidemiology, Schwarz Bioscience Inc, Raleigh, NC, USA; Jonathan A. Morris, MD, Senior Vice President, United BioSource Corporation, Blue Bell, PA, USA
PURPOSE: To describe approaches for rapid, systematic analyses of disparate observational (claims, EMR) databases to increase the efficiency and transparency of outcomes research.
DESCRIPTION: Observational databases, containing longitudinal records of health care encounters, are rich sources of information for outcomes research.  Access to information contained within these databases is mainly accomplished by one-off analyses, a time-consuming and resource-intensive process requiring statistical programmers to write custom programs extracting information from individual databases for each analysis.  A more systematic approach, utilizing a library of standardized and validated analysis routines applied across disparate databases, enables rapid and efficient analyses to be performed and standardizes assumptions used across databases, making the comparison of results obtained from different databases more meaningful.  In the first part of this workshop we will describe a common data model developed to enable systematic observational analysis.  We will describe the results of transforming a claims and an EMR database into the model, including characteristics of the transformed data for each database to show the effects of the transformation process on the data itself.  We will then present two case studies to illustrate the results of executing standardized analyses using the transformed databases.  The first case study will show how descriptive analyses can be executed to understand the characteristics of databases, patient cohorts, and outcome definitions. These types of rapid analyses can be used for feasibility assessments and to inform study design.  In the second case study, we will describe a method developed to perform a retrospective cohort analysis using the common data model.  We will provide the results of executing this method on both the claims and EMR data.  Participants in this workshop should gain a good understanding of how systematic observational analyses can complement current practices, and learn about approaches that have been used to implement a framework for systematic observational analysis. 

WORKSHOPS - SESSION V Wednesday, May 25, 2011: 3:00 PM-4:00 PM
Clinical Outcomes Research

W24: GENERALIZED EVIDENCE SYNTHESIS IN COMPARATIVE EFFECTIVENESS RESEARCH: COULD THE EVIDENCE BASE BE BROADENED IN MIXED TREATMENT COMPARISONS?
Discussion Leaders: Agnes Benedict, MSc, MA, Research Scientist, United BioSource Corporation, Budapest, Hungary; Huseyin Naci, MHS, Research Associate III, United BioSource Corporation, London, UK; David Vanness, PhD, Assistant Professor, Department of Population Health Sciences, University of Wisconsin, Madison, WI, USA
PURPOSE: To review methodological considerations that arise in generalized evidence syntheses, particularly mixed treatment comparisons (MTCs), to support comparative effectiveness research (CER).
DESCRIPTION: The promise of CER is to provide evidence on the relative clinical value of multiple interventions in real-world settings. MTC, a meta-analytic method, has been increasingly used to synthesize evidence on the comparative effectiveness of multiple interventions. In this workshop we will first outline the objectives of evidence synthesis in CER and how MTCs can play an important role. We will then focus on study design considerations that come into sharp focus when conducting MTCs to address CER questions, including the evidence base and analytic approach. Currently, when evidence from randomized controlled trials (RCTs) is available, observational evidence is discounted on the grounds that only RCTs can provide unbiased estimates of treatment effects. Although this ensures high internal validity for the findings, their generalisability to populations in routine clinical practice may often be questionable. Statistically meaningful (or clinically relevant) conclusions may not be drawn when RCT evidence is inadequate. We will illustrate how selectively incorporating nonrandomized evidence would be expected to reduce uncertainty in relative treatment effects in Bayesian MTCs. With a case study evaluating the comparative effectiveness of treatments in rheumatoid arthritis we will demonstrate the flexibility of the Bayesian approach by incorporating nonrandomized evidence as ‘background’ information. Finally, we will review approaches to assess and address heterogeneity (e.g. meta-regression). We will caution participants about the potential dangers of a mechanistic combination of randomized and observational forms of evidence, without a careful exploration of possible sources of heterogeneity. Participants will gain a thorough understanding of how the results of evidence synthesis analyses can be used to support CER decisions through examples of MTCs.

Economic Outcomes Research

W26: APPLICATION AND USE OF DYNAMIC MODELS IN HEALTH ECONOMIC ANALYSES
Discussion Leaders: Sonya J. Snedecor, PhD, Director, Health Economics, Pharmerit North America LLC, Bethesda, MD, USA; Elamin H. Elbasha, PhD, Director, Scientific Staff, Merck & Co., Inc., North Wales, PA, USA; Erik Dasbach, PhD, Health Economic Statistics, Merck Research Laboratories, North Wales, PA, USA
PURPOSE: To introduce participants to dynamic models, its useful applications and illustrate differences in health economic outcomes resulting from dynamic and static models. 
DESCRIPTION: This workshop describes the utility and appropriate use of dynamic models, which have been used in a wide variety of settings.  Dynamic models differ from static models in that they can account for changes in model parameters resulting from intrinsic changes in model components.  These models are based on the mathematical representation of the interaction of demographic and epidemiologic factors associated with a given population and infectious agent.  In particular, dynamic models can represent interactions among individuals in a population and transmission of disease among them, thereby allowing for an analysis of changes in disease transmission over time after an event of interest (e.g., introduction of vaccine) as well as externalities where a health care intervention may have benefit to those other than the original recipient.  Additionally, because dynamic models can assess temporal changes in disease epidemiology, they allow for quantification of the effect of a health care intervention at different periods of time. The workshop begins with an overview and history introducing dynamic models to the general listener, describes their advantages over static models, and their useful applications.  We will provide illustrative examples, including an economic analysis of a vaccine demonstrating the differences in outcomes that would be observed when using a static and dynamic model.  Finally, we discuss some of the problems that may arise and potential barriers to the use of economic modeling.  Participants are invited to discuss their thoughts and perceptions of dynamic models.  The intended audience includes those interested in dynamic models and those looking to become more familiar with how to include these methods in their own research analyses.  No mathematical experience is necessary for the workshop.

Patient-Reported Outcomes & Preference-based Research

W27: PATIENT-CENTRIC OBSERVATIONAL RESEARCH: SUCCESSFULLY DESIGNING AND IMPLEMENTING STUDIES
Discussion Leaders: Elisa Cascade, MBA, Vice President, MediGuard.org, Rockville, MD, USA; Eric Gemmen, MA, Senior Director, Epidemiology & Outcomes Research, Late Phase, Quintiles, Rockville, MD, USA; Paul Wicks, PhD, Research Director, PatientsLikeMe, Cambridge, MA, USA
PURPOSE: To characterize the emergence of the patient as a critical stakeholder in health care decision making; discuss benefits and potential challenges with on-line patient-reported data; and present case examples of observational research using patient-centric data collection.
DESCRIPTION: Globally, payers are increasingly demanding real world outcomes data to appropriately manage care delivery and costs.  Historically, researchers have relied on two methods to collect this information: 1) claims and electronic medical records, and 2) traditional physician-centric registries (researchers recruit sites, sites recruit patients).  With the growth in patient empowerment and increased access to patients, a new, third option for collecting observational data has arisen, patient-centric studies (researchers recruit patients directly, without the need for physician sites).  This patient-centric approach offers the potential for collection of data much more rapidly and at a lower cost in comparison to physician-centric methods.  Although patient-centric studies initially focused on retrospective data mining for adverse events or prospective cross-sectional collection of health resource utilization, HRQoL, work productivity, and treatment satisfaction, successful proof of concept studies have been completed for both cross-sectional and longitudinal prospective studies that combine these patient-reported endpoints with medical record and laboratory data.  Because the landscape is shifting so rapidly, it is important for outcomes research experts to monitor trends in the rise and growth of patient-centric studies and understand the pros/cons of this approach relative to other methods.  The panelists are all skilled in the conduct of patient-centric studies and will be able to share examples of successful implementation as well as raise issues for the group to discuss such as: selection bias, validity, and retention for longitudinal studies.

W28: ADVANCED MISSING DATA TECHNIQUES IN OBSERVATIONAL RESEARCH: CASE STUDIES IN DATA LINKAGE AND IMPUTATIONS
Discussion Leaders: Christopher M. Blanchette, PhD, MS, MA, Director, Health Data Analytics, Applied Outcomes & Analysis, GlaxoSmithKline, Research Triangle Park, NC, USA; Alex K. Exuzides, PhD, Director, Statistical Analysis, Lifecycle Sciences Group, ICON Clinical Research, San Francisco, CA, USA; William B. Saunders, PhD, MPH, Healthcare Research Manager, GE Healthcare, Charlotte, NC, USA; Stephen Stemkowski, PhD, MHA, Associate Research Scientist, Division of Clinical and Outcomes Research, Lovelace Respiratory Research Institute, Kannapolis, NC, USA
PURPOSE: The changing health care environment increases the need for accountability and demand for rigorous outcomes research. Data for assessments require more sophisticated use of clinical, economic, and humanistic information. Current data sources providing all of these observations are limited, resulting in the need for methods to explore varying outcomes until such integrated data sources become available. This workshop will present a variety of case studies where missing data imputations and data linkage techniques were used to assess outcomes from different data sources.
DESCRIPTION: Four case studies that utilize novel methods in data linkage and missing data imputation techniques will be presented to provide a research roadmap when complete data are lacking for the assessment of clinical, economic, and humanistic observations. Case study 1 presents a data linkage example by which simulated cohorts were created to assess clinical and economic outcomes using two large proprietary databases.  Case study 2 presents an application of three imputation techniques to improve data availability from electronic medical records (EMRs) for patients with metabolic syndrome.  Case study 3 will review the application of missing data techniques to a large EMR database for a cohort of diabetic patients. Analyses using EMR data benefit from the availability of laboratory result and vital statistics data, but given the real-world nature of such data, creative application of missing data techniques is critical to interpreting results from retrospective cohort studies. Case study 4 presents the linkage of administrative claims data to a hospital episode dataset and its application in examining medication adherence and re-hospitalization in acute coronary syndrome patients. In this workshop we will present real-world case studies to showcase techniques to evaluate outcomes when substantial missing data exist.  Attendees will be engaged to provide examples of their approaches to evaluating outcomes in the presence of missing data.