Essays on the Production of Primary Health Care

Expanding access to primary health care is a longstanding objective of government policy. One persistent impediment to access is the relative scarcity of primary care physicians and other health care professionals in rural and low-income urban areas. In response, numerous federal programs have been administered over several decades to attract health professionals to regions where they are otherwise scarce. These programs have been expanded considerably in recent years. Areas to be targeted under the programs are designated as Health Professional Shortage Areas (HPSAs). The first three chapters focus on the effects of these programs on different measures of population health as well as health care access and quality. The fourth chapter is a coauthored paper on the relationship between health information technology adoption in the ambulatory care setting and the quality of ambulatory care.

In the first chapter, I describe the HPSA program and its associated benefits, and discuss my identification strategy. Because HPSA designation must be applied for through a burdensome application process, designation may be endogenous to unobserved attributes of the local health care system. Results from a fixed effects and paired event study specification indicate selection bias, so in my main specifications I estimate the effect of HPSA designation on the various outcomes by fuzzy regression discontinuity, exploiting a cutoff in the HPSA eligibility criteria.

In the second chapter, I estimate the effect of HPSA designation on the rate of ambulatory care sensitive hospitalizations (ACSH), a measure of the effectiveness of local primary care for which a lower value suggests better care. I calculate ACSH rates at the county level using data on all hospital stays of Medicare fee-for-service beneficiaries between 2003 and 2012. I find that designation substantially reduces hospitalizations for acute ACS conditions. Designation might reduce ACSH for chronic conditions as well, but those findings are statistically insignificant. I then test whether HPSA designation affects measures of primary care utilization, provider availability, Medicare reimbursements, and process of care. I find no effect of HPSA designation on any of these outcomes. This, together with a number of empirical challenges related to the institutional complexity of the HPSA program, argues for caution before using the large point estimates in cost-benefit analysis. Nevertheless, the analysis in this chapter is a first step in providing evidence of the effectiveness of the HPSA program.

In the third chapter, I use data from the National Vital Statistics System to study the relationship between HPSA designation and outcomes such as mortality, fertility, and prenatal care. I estimate these relationships by fuzzy regression discontinuity, as described in Chapter 1. The analysis yields no evidence of an effect of HPSA designation on mortality, prenatal care utilization, birth weight, or gestation. However, the results do suggest that designation reduces fertility among women aged 15-24. I then test whether HPSA designation affects abortion rates using data from nine states and find no effect. The pattern of results suggests that the interventions attached to HPSA designation may increase access to contraception.

The fourth chapter on the effects of health IT adoption in the ambulatory setting is coauthored with Carole Roan Gresenz, Amalia Miller, and Catherine Tucker. US government investments in health information technology (IT) have focused on giving incentives for digital health records in hospital settings and by individual physicians. We evaluate the omission of ambulatory care centers, by studying the effects of healthcare IT on ambulatory care quality, which we measure using the rate of hospital admissions for conditions identified as sensitive to ambulatory care quality, using data from Medicare and the Nationwide Inpatient Sample. Results from difference-in-differences models that control for location and time fixed effects, as well as observable factors related to healthcare quality and population demographics, indicate that increased ambulatory IT adoption lowers local area ambulatory care sensitive (ACS) hospitalizations, suggesting quality improvements. The magnitudes imply that a 45% increase in ambulatory IT adoption in a county (the average increase over our sample period 2003-2012) lowers the ACS admission rate in that county by about 1.6%.