Perspective

Our Flawed but Beneficial Medicaid Program

Austin Frakt, Ph.D., Aaron E. Carroll, M.D., Harold A. Pollack, Ph.D., and Uwe Reinhardt, Ph.D.

N Engl J Med 2011; 364:e31April 21, 2011DOI: 10.1056/NEJMp1103168

Article

Many U.S. governors are proposing or implementing deep cuts to their states' Medicaid programs to address budget shortfalls, and some are calling for Medicaid to be converted to a block-grant program. Some commentators defend such policy retrenchment by claiming that Medicaid coverage fails to improve health outcomes — indeed, that its beneficiaries may have worse health outcomes than patients with no insurance at all.1 To make that case, these commentators have creatively interpreted observational studies that have examined clinical outcomes associated with particular medical interventions according to the patient's insurance type — Medicaid, Medicare, private coverage, or none — and found that outcomes for Medicaid patients are worse than those for uninsured patients.2

Do these studies really show that Medicaid causes worse health outcomes than having no insurance? This inference's validity can be questioned at two levels: the studies lack an articulated causal model explaining the observed data, and they're beset by analytic problems.

Statistical studies should begin with an explicit model embodying the hypothesis being tested. In this case, that model should articulate the causal pathway by which a patient's insurance status translates into clinical outcomes from particular medical procedures. Specifically, what is being assumed about the behavior and skills of physicians and hospital executives or about the groups of patients undergoing the procedures?

The commentators arguing that Medicaid causes poor outcomes anticipate some objections by noting that the cited studies include some variables to address socioeconomic and cultural factors that can negatively influence the health of poorer Medicaid patients. Their interpretation of the results, then, must be that Medicaid patients have worse clinical outcomes than uninsured patients with the same socioeconomic and cultural characteristics, including, presumably, health-related behavior before and after a given procedure.

If so, the problem must lie with the physicians and hospitals (many of them academic medical centers) providing care for Medicaid patients. Are these commentators assuming that poor, uninsured patients, who in principle may qualify for Medicaid, actually have the resources to pay doctors and hospitals more than Medicaid would and that providers therefore give these patients better care and attention, leading to better outcomes? Or is the assumption that only less technically proficient doctors and health care facilities accept Medicaid patients, and the associated lack of skill and resources results in poor clinical outcomes?

The studies' statistical methods are also problematic, since their authors cannot completely control for confounders that may affect health outcomes. Bias introduced by omitting or censoring certain variables inevitably remains, calling into question any causal inference that Medicaid is harmful. In fact, one study used the same methods to show that Medicare is associated with worse health outcomes than having no insurance.3 Since Medicare is not generally subject to Medicaid's socioeconomic dynamics or provider restrictions, this result casts further doubt on any theoretically causal link between Medicaid and poor health.

It's far more likely that such results are driven by selection bias. Medicaid enrollees (including dual-eligible recipients of both Medicaid and Medicare) tend to be sicker than uninsured patients and to have lower socioeconomic status, poorer nutrition, and fewer community and family resources. Medical and social service providers may also help the sickest or neediest patients to enroll in Medicaid — a more direct cause of selection bias. Few of these potential confounders can be completely addressed using commonly available clinical or population data.

Health economists use an alternative approach in analyzing Medicaid's outcomes that seeks to eliminate selection bias related to unobservable factors affecting enrollment and health outcomes. By exploiting the variation in Medicaid eligibility rules or other program characteristics influencing states' enrollment rates, scholars have consistently found that Medicaid coverage leads to health improvements.4,5 The assumption behind these “instrumental variables” approaches is that Medicaid enrollment depends on state-level eligibility rules but patients' health status does not.

This approach is certainly not equivalent to random assignment to Medicaid coverage or uninsured status, but for causal inference, it's far more reliable than observational studies controlling for only some of the factors leading people to enroll in Medicaid. It relies on the “natural” randomness of state Medicaid policy, whereas observational studies presume to have achieved, through covariates available to researchers, what only randomization can: the equivalence of patients assigned to the treatment and control groups.

If Medicaid actually harmed health, instrumental variables studies would show that; they don't. Other complementary research, such as the RAND Health Insurance Experiment and studies of patients 65 years of age or older who were uninsured before entering Medicare, support the belief that basic public health insurance coverage improves health.

Medicaid has many problems, low reimbursement rates being arguably the most serious. If Medicaid's critics were seeking to raise its reimbursement rates and increase spending on the program, we would join their chorus. But they are using the invalid argument that Medicaid coverage is worse than no coverage at all to support proposals to cut back the program. Such an attack further damages this highly challenged program by undermining the political case for additional resources.

Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.

This article (10.1056/NEJMp1103168) was published on April 6, 2011, at NEJM.org.

Source Information

From Health Care Financing and Economics, VA Boston Healthcare System, and Boston University — both in Boston (A.F.); the Center for Health Policy and Professionalism Research, Indiana University School of Medicine, Indianapolis (A.E.C.); the University of Chicago School of Social Service Administration, Chicago (H.A.P.); and the Woodrow Wilson School of Public and International Affairs, Princeton University, Princeton, NJ (U.R.).

References

References

1. 1

   Gottlieb S. Medicaid is worse than no coverage at all. Wall Street Journal. March 10, 2011.
   

2. 2

   Gaglia MA Jr, Torquson R, Xue Z, et al. Effect of insurance type on adverse cardiac events after percutaneous coronary intervention. Am J Cardiol 2011;107:675-680
   CrossRef | Web of Science | Medline

3. 3

   LaPar DJ, Bhamdipati CM, Mev CM, et al. Primary payer status affects mortality for major surgical operations. Ann Surg 2010;252:544-550
   Web of Science | Medline

4. 4

   Bhattachary J, Goldman D, Sood N. The link between public and private insurance and HIV-related mortality. J Health Econ 2003;22:1105-1122
   CrossRef | Web of Science | Medline

5. 5

   Currie J, Gruber J. Saving babies: the efficacy and cost of recent changes in the Medicaid eligibility of pregnant women. J Polit Econ 1996;104:1263-1296
   CrossRef | Web of Science

Citing Articles (1)

Citing Articles

1. 1

   Sommers , Benjamin D. , Baicker , Katherine , Epstein , Arnold M. , . (2012) Mortality and Access to Care among Adults after State Medicaid Expansions. New England Journal of Medicine 367:11, 1025-1034
   Free Full Text