Mortality rates are part of the story, but only a part, and for many, not even the most important part. Other measures include morbidity data, disability days, work loss days, and other indicators. Research available also measures the quality of life. The RAND Health Insurance Experiment (RHIE) is one of the largest randomly controlled economic experiments ever conducted. It was designed to test the effect of alternative health insurance policies on the demand for health care and on the health status of a large and closely observed group of people from all walks of life. RAND researchers discovered that the greater the portion of the health care bill that individuals are required to pay, the less health care they choose to purchase. While this should not have been surprising, what did surprise most health economists was how great the difference was; the fully insured purchased roughly 40 percent more health care than those who had to pay their own bills. This provides an opportunity to ask whether those with 40 percent more health care were also 40 percent more healthy. This was serendipitous from a research standpoint, but a perfectly valid way to test the real contribution of health care to people’s health within the context of a scientifically controlled experiment. Fortunately, RHIE analysts kept detailed records on each person, including a dozen or more measurements under each category of physical health, mental health, social health, and general health index. They also examined their subjects’ dental health, persistence of symptoms, health habits, and disability days. The results are easy to summarize. For dozens of items, virtually no differences were found between the groups studied; health care and health insurance did not seem to matter. A simple example from the RHIE illustrates the point. Table 5-4 provides detail on workloss days per employed person per year—a measure of health status and morbidity that some economic researchers like to use because it ties directly to both health and productivity. This table separates the RAND subjects into four groups, which differ by type of health insurance policy. Some subjects pay nothing out of pocket for their health care/health insurance package; some pay 25 percent to 50 percent of their bill themselves; others pay all of their health care bills up to a certain amount, called a deductible. The subject’s out-of-pocket cost ranges from zero (free) to about 95 percent of the bill. Newhouse et al. (1993) summarize: “Our results show that the 40 percent increase in services on the free-care plan had little or no effect on health status for the average adult.” The effects on children showed a somewhat similar pattern. Valdez et al. (1985) examined data for 1,844 children in the RAND study—children who differed primarily by the type of insurance plan their families obtained. Children under the cost-sharing plans consumed up to one-third less care. However, the reduction in care was not significantly related to health status measures. It may seem from the RAND results that public provision of health insurance to both adults and children might not be justifiable on the basis of benefits to health. However, as Jonathan Gruber (2008) points out, this conclusion does not follow. No one in the RAND Experiment was “uninsured,” completely without insurance, as are close to 50 million Americans as of this writing (the Patient Protection and Affordable Care Act begins to address them in 2014). The least insured individuals studied by RAND had full coverage for health expenditures above a deductible, which was $1,000. Studies of the truly uninsured began to appear showing significant health gains from the provision of public insurance (Currie and Gruber, 1996; Doyle, 2005; Hanratty, 1996). These studies report reductions in infant and neonate deaths of around 5 to 10 percent. The Institute of Medicine estimates suggest that even larger gains are possible; they claim that the uninsured face a 25 percent greater mortality risks. Gruber further explains why these studies do not conflict with RAND. He proposes that the marginal effectiveness of medical expenditures is quite high for the first expenditures but then drops off precipitously, a plausible pattern given diminishing marginal returns. Eventually, additional spending does no good and the effectiveness curve flattens out . . . This appears to be the case as we move from less to more generous coverage, as in the RAND Health Insurances Experiment (Gruber, 2008: 584).
No comments:
Post a Comment