|
HMO Performance: The Debate Goes OnTo the Editor:In their July/Aug 02 review of studies comparing HMO to non-HMO care, Robert Miller and Harold Luft take issue with my criticism of their earlier literature reviews.1 Space constraints prevent a full discussion here. I will only note that the authors argue not only with me but with themselves. They have previously made statements indicating that they substantially agree with me. In 1999 I argued that studies that compared HMO patients to un- or underinsured fee-for-service (FFS) patients were more likely to find that HMO care was equivalent or superior to FFS care than studies that held coverage constant.2 I quoted several analysts who agreed, including Luft. In 1980 Luft observed, "Studies that involve a comparison between HMO enrollees and people with conventional insurance coverage...are actually testing two variables: (1) an HMO...effect, and (2) the differential financial coverages for preventive care." When "the insurance variable...is held constant," wrote Luft, there was "little or no HMO...effect."3 In a letter commenting on my 1999 article, Miller and Luft stated, "It is reasonable to exclude some studies...where FFS plans may not cover a service at all."4 Now the authors have changed their minds. They indicate that they will not reject a study simply because the non-HMO patients are uninsured for a particular service (I assume that they would reject a study that includes uninsured FFS patients, but they don’t say so explicitly). They base their argument on the fact that "plans are selected on the basis of a ‘bundle’ of characteristics." This statement is true, but it was true in 2000 when Miller and Luft agreed to reject studies in which FFS patients have no coverage at all for "a service," and it was true in 1980 when Luft urged analysts to distinguish between the "HMO effect" and coverage differences.
Minnesota Physician-Patient Alliance, Minneapolis, Minnesota
The authors respond: Kip Sullivan’s argument focuses on our statement in a 2000 reply that we would exclude a quality study if the service compared wasn’t covered at all, plus another statement from a 1980 article.1 Putting our 2000 reply in context, what we had in mind were prevention services comparisons (we gave the example of immunizations), where some FFS plans did not cover certain services at all and HMO plans likely did, and where the authors based the comparison specifically on whether or not that particular service was provided. Obviously, if some FFS plans do not cover mammograms and all HMO plans do, and researchers compare the rate of mammograms provided, HMOs would look better. Thus, we excluded comparisons of preventive services performance from quality-of-care performance—to avoid skewing the quality-of-care results simply based on superior HMO preventive services performance. In fact, we moved preventive services comparisons to a separate section in our 1994 and 2002 articles (we did not report on them in 1997).2 When selectively quoting from the 1980 reference, Sullivan neglects to point out that the article was focusing on preventive services, where coverage differences alone plausibly determined the differences in results. With respect to treatment or care of a particular disease or condition, we do not agree with Sullivan that studies should be excluded from a quality comparison simply because there isn’t completely comparable coverage for that disease or condition. In contrast to some preventive services, FFS plans do provide coverage for a wide range of treatments—albeit not with identical coverage for every specific service—and the comparisons that researchers make do not focus on measures of a completely noncovered service. In looking at acute and chronic care, we start with the position that people choose health benefit packages; some people trade off more choice for less coverage, presumably hoping that the ability to choose providers and have fewer nonfinancial access-to-care restrictions will overcome any quality "hit" resulting from the financial barriers imposed by less coverage. FFS plans offer treatment coverage for a variety of cancers and heart-related problems, childbirth, pregnancy, posthospital care, and stroke—that is, for all of the diseases and conditions reported on in various studies and summarized in our review article. Might differential coverage for preventive services or differential out-of-pocket costs for the above-mentioned curative services influence the outcomes of those conditions and diseases? Yes, in some cases, but so might differences among plans regarding choice of providers, nonfinancial barriers to care, integrated information systems, and quality improvement programs. We prefer to leave to the consumer the assessment of the relative value of each feature or package of features rather than to make the arbitrary assumption that financial benefits are the only thing that matters.
University of California, San Francisco, San Francisco, California NOTES
|
