Three Essays in Preventive Health Economics.

Abstract

This dissertation examines the oftentimes complicated nature of preventive care utilization decisions to help inform preventive health policy. The first paper examines how patient experiences with two seemingly disparate types of preventive care—mammography and medication adherence—can significantly affect each other. Using Marketscan Medicaid and commercial claims data and a difference-in-difference approach, I show that a false-positive mammogram leads to improved cholesterol medication adherence for the Medicaid insured population, with suggestive evidence showing that the improvement may be due to increased interaction with the health care system for a population with access problems. However, I find reduced adherence for the commercially insured, possibly resulting from reduced trust in the health care system following the negative experience of a false-positive. The second paper provides evidence that a false-positive mammogram can lead to significant increases in the probability of initiating depression or anxiety medication utilization compared to women who have a true-negative mammogram for both the Medicaid and commercially insured populations. In terms of magnitude the effect is 4 to 6 times smaller than for women diagnosed with breast cancer (true-positives). I also find several factors that appear to increase the risk of initiating depression or anxiety medication for women experiencing a false-positive mammogram—the first experienced false-positive for women with Medicaid; and invasiveness of the test and greater time to resolution for the commercially insured. The final paper analyzes how anticipated unemployment and uninsurance affect preventive health utilization using MEPS data. To find more exogenous sources of variation in unemployment and uninsurance, my primary analyses restrict analysis to individuals in the 75th percentile or greater of predicted unemployment or uninsurance and separate unemployment into layoffs and non-layoffs. Across both methods, I find evidence of significant stocking up for a variety of preventive care services in anticipation of future unemployment but evidence of significant delaying for individuals who anticipate becoming uninsured. Particularly strong delays for cancer screenings, suggest the delays may result from lower expected net benefits of screening for people who anticipate becoming uninsured and therefore facing high out-of-pocket costs for any treatments.