Essays in Health Economics

This dissertation examines how policies and behavioral choices influence various health outcomes across multiple dimensions. My first essay investigates whether media campaigns can alter sex-selective preferences by analyzing a novel policy intervention in India aimed at addressing skewed child sex ratios. In 2011, India's child sex ratio was observed to be significantly below natural levels due to strong preferences for male children. Responding to this crisis, the Indian government launched a comprehensive policy in 2015 designed to shift these preferences through targeted media campaigns. The intervention began in one hundred critical districts before expanding to additional areas nationwide. I employ a regression discontinuity design that exploits the government's systematic targeting of districts with child sex ratios below the national average, allowing me to estimate the intent-to-treat effects of the campaign. The analysis reveals minimal statistical evidence of improvement in child sex ratios, which contradicts findings from previous studies. However, since all districts received identical budgets regardless of population size, I examine the effects of per-capita investment intensity. This analysis uncovers meaningful positive impacts, particularly concentrated in districts with smaller populations that benefited from higher per-capita investment levels. These effects strengthen over time, suggesting that sustained exposure to the policy intervention may prove effective in changing deeply rooted preferences. My second essay studies potential racial and ethic bias that may exist in the healthcare community by examining the results of prostate-specific antigen (PSA) tests and the subsequent referral to prostate cancer screening for men of different ages and racial backgrounds. Using a novel, proprietary medical claims dataset, I built episodes of care for up to 180 days after the initial PSA test to study the rate of referral to either a MRI or a TRUS prostate biopsy, stratified by age and race. The outcome variable is a binary indicator of whether a patient underwent either a prostate biopsy or prostate MRI within 180 days after PSA test date. This outcome was defined both for all PSA testing dates and among the subset of PSA tests that revealed an elevated PSA level. I explore three PSA levels as thresholds for elevated PSA: 1.) PSA level exceeding 2.5 ng/mL, a lower threshold that is now being recognized for early detection of prostate cancer 2.) PSA level higher than 4 ng/mL, historically considered as an appropriate level to recommend a prostate biopsy 3.) PSA level greater than 10 ng/mL, a threshold previously associated with high rates of prostate biopsy and a high rate of prostate cancer Overall, I find that contrary to previous studies, disparities in follow-up PCa screening when simultaneously considering prostate biopsy or MRI are more consistent with the underlying risk of PCa by race. In my final essay, I use multiple data sources to compare and estimate low-dose computed tomography (LDCT) screening rates between populations of insured patients by different payer types, specifically, commercial plans, Medicare fee-for-service (i.e. Part A & B), and Medicare Advantage, and for several enrollee demographic and geographic characteristics. Under the Affordable Care Act, most insurance providers are required to provide lung cancer screenings to eligible individuals at no cost to the patients. However, these rates remain significantly low. Using proprietary claims data, data from 2017 Centers for Medicare & Medicaid Service’s (CMS) 5% Research Identifiable Files (RIF) and smoking rates from the University of Wisconsin’s County Health Ranking I estimate the county level LDCT screening rates. Overall, I find that eligible Medicare Advantage members undergo LDCT at a higher rate than Medicare FFS enrollees. These rates are substantially lower for individuals that come from a non-White ethnic and racial background. Additionally, rural areas tend to have lower screening rates as compared to urban areas, despite having a larger eligible population.

Date

2025-07-30

Resource Type

Text

Resource Subtype

Dissertation

Full item page