Laura Faden Garabedian*
*Harvard PhD Program in Health Policy Alumna & Faculty Member
Dissertation Title: "Quasi-Experimental Health Policy Research: Evaluation of Universal Health Insurance and Methods for Comparative Effectiveness Research"
This dissertation consists of two empirical papers and one methods paper. The first two papers use quasi-experimental methods to evaluate the impact of universal health insurance reform in Massachusetts (MA) and Thailand and the third paper evaluates the validity of a quasi-experimental method used in comparative effectiveness research (CER). My first paper uses interrupted time series with data from IMS Health to evaluate the impact of Thailand’s universal health insurance and physician payment reform on utilization of medicines for three non-communicable diseases: cancer, cardiovascular disease and diabetes. Expanding health insurance coverage with a medicines benefit to the entire Thai population increased access to medicines in primary care. But, there is evidence of potential unintended consequences of the reform - the universal coverage scheme did not increase use of medicines for diseases that are typically treated in secondary or tertiary care settings, or increase market penetration for generic drugs.
My second paper evaluates the impact of the MA health insurance reform on short-term enrollment and adverse selection in the unsubsidized individual insurance market. This project employs interrupted time series and pre-post survival analytic methods with claims data from Harvard Pilgrim Health Care (HPHC). Contrary to previous unpublished reports, we find that short-term enrollment decreased after the reform. And, post-reform members had lower rates of inpatient stays and emergency department visits, which suggests that the MA reform, as intended, actually reduced adverse selection in the overall individual market. However, there was a post-reform increase in use of infertility treatments, which are expensive elective procedures.
My third paper evaluates the validity of instrumental variable (IV) methods in CER. We perform a systematic review of the health and economic literature to identify IVs used in CER, evaluate trends in the use of IVs in published CER studies, and identify the existence and potential impact of IV-outcome confounders for commonly used IVs. We find that IV analysis is an increasingly popular method for CER. There is overwhelming evidence of IV-outcome confounders of the four most popular IVs that call into question the trustworthiness of the results of IV CER studies.