Economic evaluation in health care: A modern day quagmire

Abstract

The report by Pulikottil-Jacob et al 1 provides us with an opportunity to reflect on various issues associated with the conceptualization and the conduct of economic evaluations in health care. In our opinion, too frequently such analyses, while important in concept, lack optimal execution. Cost-effectiveness analysis, as a means to make informed decisions concerning resource allocation, has a long and distinguished history. 2–4 These decisions have often been associated with government programs, including attempts to critically appraise medical technology. In this regard, from 1972 to 1995, the now defunct Office of Technology Assessment, formerly an arm of the U.S. Congress, routinely performed cost-effectiveness analyses in relationship to technological innovations in health care. 5 Despite its perceived utility, cost-effectiveness analysis has often been criticized as a means to ration health care services—a taboo topic in the United States. Thus, some federal agencies, including the Centers for Medicare and Medicaid Services (CMS), have excluded cost-effectiveness information from consideration when making coverage (what to pay for) and reimbursement (how much to pay) decisions. Meanwhile, in the United Kingdom, the National Institute for Health and Clinical Excellence (NICE) has routinely included the results of cost-effectiveness analyses in its decisions on how medical technology will be deployed within the National Health Service. 6 Many other countries have done likewise, with little concern about the rationing of health care resources. In effect, the results of costeffectiveness analyses are viewed as a means to better manage national single-payer health care systems. As we see it, there are 5 issues worthy of consideration when conducting an economic evaluation to establish relative cost-effectiveness. Pulikottil-Jacob et al 1 have addressed most of them, although some better than others. First, ideally, cost-effectiveness analyses are conducted alongside prospective, randomized, clinical trials. Admittedly, this gold standard is rarely achieved, and we compromise accordingly, as have Pulikottil-Jacob et al. Their cost-effectiveness analysis is based on non-randomized, observational data derived from an administrative database. In our opinion, such data in their study enhance the probability of bias for the following reasons: 1. The database excludes variables needed to risk stratify the patient groups subjected to analysis. 2. Many of the HeartMate II (Thoratec, Pleasanton, CA) patients received the device in an earlier period when less was known about patient selection and post-implant management, giving rise to an unacknowledged “period effect.” 3. The frequency of missing data is greater for the HeartMate II group than the HeartWare (HeartWare International, Inc. Framingham, MA) group.