Cost ineffective

Medicaid's impact on health is ambiguous, but its costs are real

By Geoffrey Lawrence
  • Thursday, August 9, 2012

Does Medicaid save lives?

A new study from the Harvard School of Public Health purports to find that it does.

Specifically, the study examines the impact on mortality rates of expanding Medicaid eligibility to include childless adults. 

Between 2000 and 2005, three states — Arizona, Maine and New York — enacted such changes to their state Medicaid programs. According to the study, the result was a 6 percent reduction in mortality relative to neighboring states that did not expand Medicaid.  Therefore, conclude the authors, such expansion of Medicaid eligibility could prevent 2,840 deaths per year for every 500,000 adults given Medicaid coverage.

The conclusions have been heralded as especially poignant because the federal Affordable Care Act (ACA) originally sought to compel all states to adopt the same type of Medicaid expansion.  That was before the U.S. Supreme Court declared such compulsion unconstitutional, leaving states the option of not expanding Medicaid

Now, with this new evidence in hand, commentators on the Left declare: "The question of whether the states will expand Medicaid, therefore, is not just a question of politics; it is a question of life, health and death."  Other left-oriented media outlets have seized upon the Harvard study to proclaim that failure of any state to expand Medicaid eligibility requirements as the ACA originally envisioned would be socially irresponsible and result in thousands of avoidable deaths.

But is that really true?

A closer look at the study, as well as a review of recent related studies, reveals that Medicaid's ability to improve health outcomes is, at best, ambiguous.

The Harvard study compares mortality rates in Arizona to those in neighboring control states New Mexico and Nevada while also comparing Maine to New Hampshire, and New York to Pennsylvania.  In only one of these comparisons — that between New York and Pennsylvania — did the study find a statistically significant decline in mortality rates due to Medicaid expansion.  Comparing Maine to New Hampshire, the study actually found that Medicaid expansion was associated with an increase in mortality.

The New York-to-Pennsylvania comparison yielded results that were strong enough to skew the average of all the case studies examined and produce a result that associated Medicaid expansion with slightly lower mortality rates. For a number of reasons, however, this particular case study is highly problematic. 

Demographically, New York is vastly different from Pennsylvania. The Empire State's poverty rate is substantially higher, with a much-larger proportion of immigrants, who are much more likely to be uninsured.  Minorities account for 38 percent of the state's population, with Hispanics and Asians alone accounting for 21 percent.  Those numbers for Pennsylvania are 16 percent and 5 percent, respectively. 

The other case studies looked at states with similar demographics and, tellingly, found no meaningful impact of Medicaid expansion on mortality rates.  In other words, the study's principal finding is anchored on a single case study that might rightly have been discarded as an unrepresentative outlier.

That's enough to encourage suspicions that the authors purposefully chose their data selections to produce a particular policy-relevant outcome — a suspicion that gains credibility upon learning that one of the authors has since left his Harvard post to work for the Obama administration.

The Harvard study also contravenes recent research into the impact of Medicaid.  A much more comprehensive 2010 study conducted at the University of Virginia examined 894,000 individual cases nationwide — more than 13 times as many observations as the 68,000 used for the Harvard study — and determined that objective health outcomes for those on Medicaid are worse than for the uninsured.

Examining a broad range of surgical outcomes and adjusting for age and risk factors, the Virginia study found that "surgical patients on Medicaid are 13 percent more likely to die than those with no insurance at all, and 97 percent more likely to die than those with private insurance."

Similarly, early results from the Oregon Health Insurance Experiment, which expanded Medicaid coverage to 10,000 new individuals based on a random lottery, have shown little improvement in objective health outcomes.

If one merely accepted all three of these major efforts at face value, the reasonable conclusion would be that Medicaid's impact on health outcomes is uncertain, with the largest and most comprehensive study showing Medicaid having negative impacts on health.

But any public health insurance program that costs American taxpayers $460 billion annually certainly should save lives. In fact, it should do far better than even the 6 percent reduction in mortality that Harvard researchers bent over backwards to concoct.

As any economist can tell you, it's the opportunity cost that matters. Medicaid has delivered far fewer beneficial results than enrolling Medicaid recipients into private insurance plans would have produced. 

After all, the real debate is about whether health care can be more efficiently delivered by government or private markets. It's disingenuous from the outset to compare individuals with no insurance to those who receive nearly half-a-trillion dollars in taxpayer-financed insurance each year.

And it's an outrage that, after making such comparisons, the results would be so unclear.

Despite all the rhetoric and political spin, Medicaid is just a poster-child for cost-ineffective government programs. That's why policymakers should focus on benefits reform and not eligibility expansion.

Geoffrey Lawrence is deputy policy director at the Nevada Policy Research Institute. For more visit

blog comments powered by Disqus