Rural Safety Net Provision and Hospital Care in 11 States
Study Design: We use hospital discharge data on total and ACS charges, admissions, and patient days by payer type from 11 states. We link hospital records to multiple data sources: American Hospital Association (AHA) data, Primary Care Service Area (PCSA) and Area Resource File (ARF) data for population controls and the availability of primary care safety net providers (clinics) within the local service area, and Small Area Health Insurance Estimates (SAHIE) of the number of uninsured within the service area. We examine the share of admissions that are ACS, the admission and ACS admission rates for the population and by payer type, and average length of stay by payer type.
Findings: The sample of 1133 hospitals in 11 states is nationally representative of hospitals in terms of size and ownership characteristics. We use rural residents as a share of PCSA population to measure the rurality of the hospital service area. Hospitals serving the more rural populations are smaller, less likely to be privately owned, and serve a higher share of Medicare patients than hospitals serving more urban populations. Furthermore, as the service area population becomes more rural, an increasing share of admissions is classified as ACS admissions. Rural-urban differences in ACS admission rates appear to be determined at least in part by differences in bypass behavior for ACS versus other admissions. Thus, hospital specific ACS admission rates may be a weak outcome measure of the effect of policy interventions designed to expand primary care access.
Based on descriptive data we find that in communities with less than 50 percent rural populations, the presence of a primary care safety net provider is unrelated to the share of hospitalizations that are ACS admissions. However, in service areas with majority rural populations a positive relationship exists between the share of admissions that are ACS and safety net provider presence, opposite the hypothesized direction. If clinic location is endogenous to the share of the population most prone to having ACS admissions, multivariate analysis must be used to better measure the effect of safety net presence on population adjusted ACS admission rates.
Based on multivariate analysis we find no measurable effect of the presence of safety net providers on the proportion of all admissions that are ACS or on the share of uncompensated care. Furthermore, we find no effect of clinic presence on average length of stay. However, we do find strong evidence that the presence of primary care safety net providers within the PCSA reduces the rate of all admissions and ACS admissions per 1000 residents, and particularly the rate of ACS admissions per 1000 public beneficiaries. Moreover, these effects are more pronounced in service areas that are majority rural than in less rural or urban areas. The results suggest that in the most rural service areas, between 3 and 5 percent of all admissions and about 6 percent of preventable hospitalizations can be avoided through each additional primary care safety net clinic.
Conclusions: The presence of primary care safety net providers reduces ACS admissions per 1000 residents and publicly funded ACS admissions per 1000 beneficiaries, particularly in rural communities, suggesting improved access to primary care among low-income patients.
Implications for Policy:
Public funding for primary care safety net providers results in an offsetting savings for hospital care among the public beneficiaries who gain access to services. These gains may reduce health disparities.
There may be products related to this project; please contact the lead researcher for more information.