The Effect of Medicaid Disease Management Programs on Medicaid Expenditures

Thumbnail Image

Publication or External Link






Disease Management (DM) programs for Medicaid patients with chronic diseases have become very popular, with a majority of states having introduced some type of DM program in the last decade. These programs provide interventions designed to assist patients and their health care providers appropriately manage their chronic health condition(s) according to established clinical guidelines. Cost-containment has been a key justification for the creation of DM programs, despite mixed evidence that DM actually saves money for the Medicaid program or for society as a whole.

While most studies on the impact of DM focus on estimating the impact of a single DM program, Chapter 2 estimates the average, national impact of state Medicaid DM programs by linking a detailed survey of state Medicaid programs to the nationally representative Medical Panel Expenditure Survey. Difference-in-difference models are used to test the hypothesis that medical expenditures change after a DM program is implemented, exploiting variation in the timing at which state Medicaid programs implemented DM programs. DM coverage also varies within states over time due to variation in program eligibility by disease, insurance category, and/or county of residence. Although the models estimate the effect of DM imprecisely, point estimates are stable across multiple specifications and indicate that DM programs for common chronic diseases may decrease total medical expenditures, potentially by 10 percent or more.

Chapter 3 evaluates one DM program in the state of Georgia using a proprietary data set. By exploiting a natural experiment that delayed the introduction of high-intensity services for several thousand high and moderate risk patients, the research identifies the causal impacts of the program's interventions on total Medicaid expenditures, categories of health care utilization, and other indicators. These patients are observationally similar to those who received interventions at the beginning of the program. For example, I find the interventions lowered health costs and hospital utilization, after controlling for unobservable individual characteristics. Health expenditures were lowered about 4.4 percent for patients with positive expenditures. Heterogeneous treatment effect analysis indicates that the savings were largest at the most expensive tail of the distribution.