Does Medicare Save Lives?
Medicare expenditures topped $400 billion in 2006, accounting for roughly one-fifth of total health care spending in the U.S. Despite average expenditures of nearly $9,500 per beneficiary, there is very little evidence that Medicare improves health. Mortality rates and self-reported measures of health display no sharp break at age 65, the age of Medicare eligibility, and there is no evidence that the introduction of Medicare in 1965 affected aggregate mortality rates.
In "Does Medicare Save Lives?" (NBER Working Paper 13668), researchers David Card, Carlos Dobkin, and Nicole Maestas revisit the health effects of Medicare.
Like previous studies, the authors' basic approach is to compare the health outcomes of patients treated just be-fore and after their 65th birthday. In general, this approach is subject to the concern that some patients may delay treatment until they reach age 65 to take advantage of their access to Medicare at that age. Since healthier patients can more easily delay treatment, differences in health outcomes for patients treated before and after age 65 may be due in part to differences in their underlying health status and not solely to the effect of Medicare.
The authors surmount this problem by focusing on a subset of patients who are admitted through the emergency room (ER) for relatively severe conditions that require immediate hospitalization. These conditions are defined as those with similar ER admissions rates on weekdays and weekends. The authors show that there is no jump in ad-missions for these conditions at age 65, suggesting that the underlying health status of patients treated for these conditions just before and after age 65 is likely to be quite similar. A second advantage of their focus on patients with relatively severe conditions is that these patients are sick enough that the provision of extra health care services through Medicare could plausibly affect their short run mortality.
For their analysis, the authors use a very large dataset of hospital records for patients discharged from California hospitals between 1992 and 2002. The authors first identify the most common conditions that result in "non-deferrable" admissions (those with similar weekday and weekend admission rates). These conditions include chronic bronchitis with emphysema (common among smokers and ex-smokers), respiratory failure, and acute myocardial infarction (heart attack).
Next, the authors document that there are significant changes in insurance status at age 65 for patients admitted with non-deferrable conditions. The fraction of patients with no insurance falls by about 10 percentage points when patients turn 65, while the fraction with Medicare as their primary insurer rises by nearly 50 percentage points. Thus, patients have much different insurance coverage just after age 65 than just before.
The authors then turn to examine whether there are distinct changes in health care services at age 65. They find that the average length of stay jumps by 0.4 days or 5 percent, while the average number of procedures performed rises by 0.1 or 4 percent. There is an even larger increase in the latter for "procedure intensive" diagnoses like acute myocardial infarction, where patients are more likely to receive important diagnostic procedures like cardiac catheterization after age 65. The authors also find increases in hospital list charges and the likelihood of transfer to other care units in the hospital at age 65.
Finally, the authors look at their key outcome of interest, mortality rates. They estimate that Medicare eligibility is associated with a 20 percent reduction in 7-day mortality, a 9 percent reduction in 28-day mortality, and a 3-4 per-cent reduction in 1-year mortality, relative to mortality rates among 64 year olds with similar conditions at admission. As the authors note, "the fact that the effect emerges within 7 days and persists for two years suggests that the extra services or changes in the quality of services provided to Medicare-eligible admittees have an immediate life-saving effect, and lead to a significant gain in the duration of life."
The authors point out that the estimated effect of Medicare on mortality is too large to be driven solely by the drop in the number of people without insurance at age 65. Rather, their findings suggest that there is also an "insurance generosity" channel - patients insured through Medicare (who may also have supplemental insurance) receive more services or a more timely delivery of services than typical insured 64-year-old patients. Finally, the authors note that the estimated reductions in mortality are achieved with only a modest (4 percent) increase in hospital list charges. While a full cost-benefit analysis of Medicare is beyond the scope of the current analysis, the authors note that with respect to covering expenses associated with non-deferrable conditions, the cost-benefit analysis appears to be "very favorable."
This research was supported by a grant from the National Institute on Aging.
