The Medicaid Undercount: Real or Perceived Bias in Estimates of Coverage in General Population Surveys?

Publication Date: September 2008

Publisher: Robert Wood Johnson Foundation

Author(s): Bonnie.J. Austin

Research Area: Health

Type: Report

Abstract:

Policies to cover the uninsured rely in large part upon the accuracy of estimates of this population. However, estimates of those uninsured can vary by data source due to variations in the definition of coverage, time frame differences, and recall bias and confusion by individuals about their insurance status. Some research supports a general assumption that population surveys, a critical data source, may in fact undercount the number of individuals with Medicaid coverage, which in turn can bias estimates of uninsurance. Administrative data, considered to be the "gold standard" in terms of accuracy of enrollment counts, has limitations as well. These include the potential for duplicate enrollment in programs and the potential for ambiguity about enrollment status. The discrepancy between survey estimates and administrative counts is assumed to lead to an overestimate of the number of uninsured and, therefore, a disconnect between the real needs of a state’s residents and the allocation of resources to develop programs to address those needs. For state policy-makers who monitor the dynamics of health insurance coverage, estimate costs of public programs, and develop outreach and coverage initiatives, complete and accurate data are critical.

Kathleen Thiede Call, Ph.D., and colleagues at the University of Minnesota tested the undercount assumption and reported the following findings:

* Population surveys that collect point-in-time coverage information do a good job of measuring and determining those with and without health insurance.
* The small amount of bias created by an undercount among Medicaid enrollees does not greatly undermine the validity of estimates of the uninsured or the policies resulting from that information.

The researchers discovered that upwards of 80 percent of enrollees accurately reported their coverage in the Medicaid program, and many of those failing to report Medicaid coverage, reported that they had another form of public or private insurance.