The Impact of State-Varying Components of the Affordable Care Act on Prescription Fills for Diabetes and Hypertension

University Of Chicago, Chicago IL

Investigators

Linked publications, trials & patents

Abstract

? DESCRIPTION (provided by applicant): Policy-makers and the American public need to know whether the Affordable Care Act (ACA) is increasing access to effective health care, and which specific resources are working. The ACA was implemented through a cluster of federal grants to the states that varied in topic and size, making it difficult to assess the impact of individual funding components across states. For example, targeted funds to establish state-based exchanges support Medicaid expansions, and assist consumers post-enrollment were accepted in some states, but not in others. Decoupling the effects of these components is critical for policymaking. To make decisions about how to prioritize spending in the future, policy-makers need to know what has worked in the past. One important effect of the ACA could be increasing use of effective medications for diabetes and hypertension, two chronic conditions which affect millions of Americans and can be controlled through medications. As such, the overall aim of this project is to measure the impact of four state-varying components of the ACA - Medicaid expansions, state-based marketplace implementation, and funding for pre-enrollment and post-enrollment assistance programs - on prescription fills for diabetes and hypertension, using geographic variation to identify causal effects. By using two methods, I can reach more precise conclusions about how the components build on one another. Specifically, I will: Specific Aim 1. Use a parametric (regression) approach to estimate the causal impact of state-varying components of the ACA on prescription fills for diabetes and hypertension. Because the state-varying components of the ACA were not randomly assigned to states, I will correct for time-varying political, economic and health-related determinants of state-level decisions as well as time-invariant factors, and implement additional corrections if tests suggest it is necessary. To distinguish changes in prescription use due to improved access to care from changes in disease prevalence, I will correct for the need for prescriptions using estimates of self-reported diabetes from CDC and estimates of total hypertension from my previous research. Specific Aim 2. Use a semi-parametric (matching) approach to conduct an extensive robustness check of the estimates. Some counties at state borders are similar in many ways but subjected to vastly different ACA implementations. Using these counties, I can calculate the same causal effects as in Specific Aim 1 but using a matched-pairs difference-in-differences approach. Although this method is less standard in the health literature, it uses fewer assumptions about how the state-varying components of the ACA build on one another. Policy-makers need to know which resources related to the ACA had impact on access to effective health care. The goal of this project is to provide precisely this information. By employing two methods, I will be able to reach robust conclusions about the incremental impact of four state-varying components of the ACA.