Research Objective. The 340B drug discount program was introduced in 1992, allowing some safety net hospitals to purchase Medicare-approved outpatient drugs at a discounted price. Hospitals are eligible for the 340B program if their Medicare Disproportionate Share Hospital (DSH) percentage are above the required threshold (11.75% for most hospitals). Although the program was intended to provide relief for those safety net hospitals in most dire financial need, over half of nonprofit hospitals now participate. Growing hospital participation has invoked criticism that participating hospitals serve more affluent communities, contradictory to the original intent of the program. Advocates of the program argue that greater financial flexibility generated from the program allows hospitals to subsidize services to low-income patients, however, there is little evidence to support these claims. This study examined whether hospitals use 340B program savings to increase inpatient utilization for uninsured and Medicaid patients, and unprofitable service types. 

Methods. We used 2011-2017 data on short-term, general, hospitals from the Healthcare Cost and Utilization Project (HCUP) State Inpatient Database (SID), the American Hospital Association (AHA) Survey, and CMS Hospital Cost Reports. HCUP-SID data was available for 11 states and used to identify insurance status and service lines by ICD-10 Major Diagnostic Categories. We considered psychiatric, substance use, burn care, trauma, emergency, and obstetric services to be unprofitable service lines. We used a difference-in-differences approach comparing 116 hospitals that began participating in 340B from 2011-2016 to 319 hospitals that did not participate or always participated in the program during the sample period. We estimated Poisson regression models with the total number of annual hospital admissions as the exposure term. Analyses controlled for hospital- and market-level factors and used bootstrapped clustered standard errors. Sub-analyses examined utilization by hospital ownership (e.g., nonprofit versus public) and rurality. We also performed a Goodman-Bacon decomposition to examine variation in treatment effect based on treatment timing and an event study to evaluate pre- and post-trends. 

Principal Findings. Among all hospitals, admissions by Uninsured and Medicaid patients declined by 3.6% following 340B participation (p=0.012). Among other payers, there was no change in private pay admissions, however, there was a marginally significant increase in Medicare admissions by 3.5% (p=0.063) in urban nonprofit hospitals. Unprofitable services line utilization was also largely unchanged following 340B participation. Psychiatric and substance use admissions declined, although not statistically significant. We found that urban, nonprofit hospitals were responsible for the majority of the declines in uninsured and Medicaid admissions. Rural and public hospitals were not impacted. 

Conclusions. Hospital participation in the 340B drug discount program was associated with declines in Medicaid and uninsured inpatient utilization. These results suggest that the 340B program may unintentionally disincentivize hospitals to expand safety net services beyond the target DSH percentage threshold. On the other hand, the program may allow rural and public hospitals to sustain safety net services. Increased program transparency, such as reporting revenue and safety net engagement activities, may be necessary to hold hospitals accountable for using 340B savings as intended.