Does Churning in Medicaid Affect Health Care Use?

Abstract

Background: Transitions into and out of Medicaid, termed churning, may disrupt access to and continuity of care. Low-income, working adults who became eligible for Medicaid under the Affordable Care Act are particularly susceptible to income and employment changes that lead to churning.

Objective: To compare health care use among adults who do and do not churn into and out of Medicaid.

Data: Longitudinal data from 6 panels of the Medical Expenditure Panel Survey.

Methods: We used differences-in-differences regression to compare health care use when adults reenrolled in Medicaid following a loss of coverage, to utilization in a control group of continuously enrolled adults.

Outcome measures: Emergency department (ED) visits, ED visits resulting in an inpatient admission, and visits to office-based providers.

Results: During the study period, 264 adults churned into and out of Medicaid and 627 had continuous coverage. Churning adults had an average of approximately 0.05 Medicaid-covered office-based visits per month 4 months before reenrolling in Medicaid, significantly below the rate of approximately 0.20 visits in the control group. Visits to office-based providers did not reach the control group rate until several months after churning adults had resumed Medicaid coverage. Our comparisons found no evidence of significantly elevated ED and inpatient admission rates in the churning group following reenrollment.

Conclusions: Adults who lose Medicaid tend to defer their use of office-based care to periods when they are insured. Although this suggests that enrollment disruptions lead to suboptimal timing of care, we do not find evidence that adults reenroll in Medicaid with elevated acute care needs.

FIGURE 1

Illustration of churning and control group definitions. Arrows indicate full churning cycles, representing completed transitions from enrollment in Medicaid, to other insurance or no insurance, and a subsequent resumption of Medicaid coverage. Although we allow a baseline observation of Medicaid enrollment in the NHIS to contribute to our count of insurance transitions, adults must have experienced at least 1 observed disruption in Medicaid coverage, during a MEPS observation month, to contribute to our churning sample. *This second example illustrates a person with 2 full churning cycles during the observation period. In this paper, we analyze health care use only during the first episode of churning, due to the small number of adults who experience 2 or more churning cycles in our sample. MEPS indicates Medical Expenditure Panel Survey; NHIS, National Health Interview Survey.

FIGURE 2

Trends in health care utilization for churning and control groups where Medicaid is the primary payer. Time zero on the horizontal axis is defined as the month in which a churning adult resumed Medicaid enrollment (for the churning group) and is the 13th month of the MEPS survey for the continuously enrolled control group. The vertical axis shows the rate of use per person-month. Graphs are plotted using a kernel-weighted local polynomial smoothing function, with an Epanechnikov kernel. Graphs show utilization where Medicaid is the primary payer. We separately examined plots of utilization for all payers, and found very similar trends in the churning and control groups. A, Emergency department visits; B, Emergency department visits admitted inpatient; C, Office-based provider visits. CI indicates confidence interval.