The Effect of Health Status of New Medicare Beneficiaries

For decades the advocates of universal health insurance have tried to say that having insurance is associated with better health and ultimately lower spending.  This notion has been totally dashed by sound research, both in regard to better health and lower spending.  In fact, as would be economically obvious, having insurance means people will use more services and cause more health spending, but not necessarily engage in better health behaviors.  Nevertheless, people will undoubtedly try to use a recent Government accounting Office report on the effect of health insurance on the health status of incoming Medicare beneficiaries to advance their perspective.   (GAO Report)  The report examined the effect of having or not having continuous health insurance in the 6 years before becoming covered by Medicare on spending in the first few years on Medicare.  The data source was self-reported health insurance status and health status from the Health and Retirement Study and Medicare claims for spending.  The end study sample was small, from about 1150 to 3200 people.  The differences found are often very small, for example, while 84% of people with continuous insurance report being in good health in the first years after being in Medicare, 79% of those without insurance also did.  Beneficiaries without prior insurance had 35% higher average  spending in their first year of enrollment, but as the report notes, this is likely the common effect of higher utilization after a person gets any form of insurance.  And the higher spending was limited to a few categories and did not persist across categories.  Higher spending on hospital outpatient basically disappeared in the third year.  Higher spending on physician services was gone after one year and reversed itself significantly by the fourth year of enrollment.  Most baffling is the complete absence of an analysis of hospital inpatient spending differences or drug spending.  No explanation is given for this.

The report has obvious flaws, which the usually thorough and cautious GAO fails to adequately address.  One is the reliance on self-reported survey data in regard to insurance and health status, which has been repeatedly shown to have major flaws.  And the definition of prior insurance would, for example, leave out people who had five and half years of coverage.  No attempt was made, it appears, to account for the presence or absence of Medigap coverage or being a dual eligible and enrollees in Medicare Advantage were excluded from the analysis, all of which would likely affect the outcomes.  But more importantly, GAO fails to adequately independently examine other characteristics of people that would be associated with having or not having insurance and/or with the level of health spending.  The most important of these is socio-economic and demographic factors which are strongly associated with poor health and poor health-care seeking behaviors and lack of insurance in the years before obtaining Medicare coverage.  If those had been fully taken into account, the real association would be between poorer health behaviors and higher Medicare spending.  GAO prides itself on being independent, but this report is done in response to three Democratic Senators’ request with a specific end in mind and it is hard not to think that GAO gave them what they wanted in this flawed analysis.