Messages from the ISPOR Presidents
William H. Crown, PhD
2013-2014 ISPOR President
and Chief Scientific Officer, Optum Labs,
Cambridge, MA, USA
It is a great honor to be the incoming President of ISPOR. It's really quite a remarkable feeling; the last time that I ran for elected office (and won) was in high school in a small town in rural Vermont. But something tells me that I am not in Vermont anymore. ISPOR is the pre-eminent organization for promoting the science of pharmacoeconomics and outcomes research in the world. Equally important, ISPOR members focus on the translation of this research into useful information that can guide health care decision makers.
The ISPOR 18th Annual International Meeting had over 1,600 presentations, including short courses, workshops, issues panels and forums, research podium presentations, research poster presentations, plenaries, and educational symposia. Needless to say, putting together a meeting of this magnitude was an enormous undertaking. I would like to thank the Program Co-Chairs, Laurie Burke and Donald Patrick, along with the research, workshop and issue panel co-chairs for putting together a fabulous program. I would also like to extend thanks also to all of you who participated in the peer-review process. And on the execution side, the ISPOR staff once again did a marvelous job.
We are extremely fortunate that ISPOR has an outstanding Board of Directors. Being on the Board of Directors this past year has been a rewarding experience and an excellent opportunity to observe my predecessors, Deborah Marshall and Mark Sculpher, in action. I thank them both for their terrific leadership and hope that they will not mind if I call on them for advice from time to time. I welcome Adrian Towse, the President-Elect, to the fold and look forward to working with him. I’d also like to congratulate and welcome our three new Board members Maarten Ijzerman, W.G (Wim) Goettsch, and Adrian Griffan; I look forward to working with all of you and the rest of the Board in the coming year. Finally, we all owe a great debt of thanks to ISPOR Executive Director, Marilyn Dix Smith, who has guided and nurtured the organization from the very beginning-–building an ISPOR staff that is absolutely top notch.
When I think about how ISPOR has evolved over time, I am particularly struck by its emergence as a truly global organization. ISPOR has experienced remarkable growth since it was established in 1995—from its initial 35 founders to more than 7,000 members from more than 100 countries today. The reach of ISPOR is extended by its 64 regional chapters that constitute more than 6,000 additional members (Fig.1).
It is clear that ISPOR has grown dramatically over its first 18 years but where are we going? There are a couple ways of answering this question. First, I hope that many of you will be hopping on planes to Argentina, Ireland, and Canada in the coming 12 months. We have an exciting year ahead of us with the Latin American meeting in Buenos Aires (September 12-14), the European Congress in Dublin (November 2-6), and next year’s International Meeting in Montreal (May 31-June 4). There are many people hard at work putting together programs that will rival the quality of the recent ISPOR 18th Annual International Meeting.
Apart from our upcoming travel itineraries, it is also important to recognize how ISPOR’s evolution to this point has positioned the organization for the future. It is clear from the 1,600 presentations at the ISPOR 18th International Meeting that we have a lot to share with one another. One of the advantages of the organization being so large is that we are also extremely diverse. I am certain that ISPOR’s international perspective is one of its greatest strengths. Every country is facing the challenge of how to provide the best health care to the most people in the most efficient and effective way possible. The models across countries vary widely so, in effect, we have large numbers of natural experiments underway around the world. The challenge, of course, is that very little of the evidence being generated is from randomized experiments and so teasing out the actual program policy impacts that would allow us to compare health outcomes across countries is very difficult. This places ISPOR in an incredibly important position to help with the generation and translation of evidence. For my entire career, I have worked on various problems in the analysis of observational data. We are on the cusp of something big. We live in a time of exploding availability of data both in terms of volume and richness—particularly, the potential for linking clinical and health economic data for substantial patient populations in a de-identified manner. Such datasets will offer researchers ever expanding opportunities to draw improved inferences from observational data.
I am especially encouraged by the possibility that linking data may substantially reduce missing variable bias—moving us closer to the answers that we would have gotten had we conducted a randomized clinical trial (RCT) to answer the same question with the same patient population.
Some data are more difficult to integrate, however. Notably, the voice of the patient—patient reported outcomes (PROs), patient reports regarding their satisfaction with care, etc., are missing from almost all administrative claims and electronic health record datasets. I believe that we will need to push to get such assessments included in the process of care (e.g., annual health plan enrollment, part of the physician visit) before they will become widely and systematically available.
Finally, we need to continue to advance the use of correct methods to analyze the data. One example of this is the phenomenon that economists refer to as the identification problem. In the 1920s agricultural economists realized that the market clearing prices and quantities observed in the data told them little about the underlying supply and demand curves that led to these values. They discovered that using additional variables influencing supply (e.g., rainfall) traced out the supply curve and, conversely, using additional variables influencing demand (e.g., income) traced out the demand curve. We face this same identification problem in virtually any database where the data reflects a mix of both patient and physician behaviors. More data and appropriate statistical methods can help us disentangle the puzzle.
More data alone, however, does not solve the bias problem. A focus on “big data” may sometimes lead to misleading conclusions. For example, pure data mining approaches may be very valuable for describing variations in care along geographic or sociodemographic dimensions, but we still need multivariate methods from epidemiology, health econometrics, and health services research to identify clinical or programmatic treatment effects. Without multivariate methods to identify the incremental effects of an intervention, we get very precise, but biased, estimates of treatment effects! In fact, one of the greatest potentials of big data is that it may facilitate linkage across datasets to create richer data. Ultimately, it is richer data that will lead us to better (less biased) estimates of treatment effects because these types of data will facilitate efforts by researchers to control for confounders directly.
There has never been a better time to be in the field that we have all chosen. Collectively, we have an opportunity to make an enormous difference in both the generation and translation of evidence. Let’s get going; we have a lot of work to do!
William H. Crown