|
The first chapter, co-authored with Season Majors, Christopher Connolly, Mary Ann Friesen and Hassan Ahmed, studies how electronic blood glucose monitoring impacts physician and patient behavior. Recent technological development has led to increased availability of patient generated health data, which has the potential to influence medical treatment and health outcomes. However, it is not well understood how to most effectively integrate this new technology and data into large health systems. We conducted an experimental evaluation of multiple approaches to increase utilization of electronic blood glucose monitoring, among 7,052 patients with diabetes at 20 primary care practices. A physician education intervention successfully increased provider take-up of an online blood glucose monitoring tool by 64 percentage points relative to control, while a comparison of patient-focused reminder interventions revealed that emphasizing accountability to the provider was most successful at encouraging patients to actively track their blood glucose online. An assessment of downstream outcomes also revealed impacts of the interventions on prescribing behavior and A1c testing frequency. We interpret these results in the context of a conceptual framework in which patient generated data can affect patient behavior directly, and may also influence physician treatment decisions by acting as a complement or substitute for traditional health data sources. In the second chapter, I study the effects of Medicaid and other means-tested benefits on immigrants' health outcomes, health care utilization, financial outcomes, and remittance behavior. The Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996 bars most legal immigrants from receiving social benefits such as Medicaid and SNAP for their first five years of residency in the United States. I exploit this discontinuity in benefit eligibility to estimate the causal impact of means-tested benefits using regression discontinuity and difference-in-differences approaches. I find evidence for decreased savings and increased use of the emergency department as a result of gaining eligibility for Medicaid and SNAP.In the third chapter, co-authored with Benjamin R. Handel, we conduct a randomized evaluation of strategies to facilitate advance directive (AD) completion among 4,850 patients aged 65 and over. Despite the significant economic and personal implications of end-of-life healthcare decisions, many fail to document their wishes or to select a representative who can make medical decisions on their behalf. We evaluate the effects of (i) an in-person drive to facilitate AD completion and (ii) electronic distribution of an informational video discussing advanced care planning. Among patients to whom communication was sent via email, we find no effect of in-person AD drives or of the informational video on AD upload rates. However, we estimate a 4.5 percentage point increase in AD uploads for patients who were contacted via letter about the AD drive, relative to patients who were sent a reminder letter only. This suggests that in-person drives may be impactful for increasing AD completion, but only if effectively advertised to patients. We also leverage surveys and granular data on patient health to understand how information frictions and hassle costs may influence advance care planning decisions.
|
