Background: Interhospital transfer (IHT), which is often performed to provide patients with necessary procedural and specialized care, is a common occurrence in modern healthcare. Racial/ethnic inequities in IHT have been described using nationally representative data, however such inequities have not been characterized at a more local level, which can provide more granular data to adjust for possible confounding. The purpose of this study was to determine if there were racial/ethnic inequities in IHT for common medical conditions within our major academic healthcare system.

Methods: We performed a retrospective matched cohort study of patients admitted to general medicine services at community hospitals within our system. We included adult patients age  18 with all medical diagnoses admitted to these hospitals between June 2015 to December 2018 and excluded intensive care unit, surgical and obstetrical patients. The outcome was interhospital transfer to the tertiary care hospitals within our system. The primary predictor of interest was race/ethnicity, specifically Black race and Latinx ethnicity. We used a matched cohort study design in which one Black patient with an index admission to one of our system’s community hospitals was matched to 3 White patients based on their origin hospital, age within 5 years and similar electronic cardiac arrest risk triage (eCART) score (a predictor of cardiac arrest, ICU transfer, or death) on admission. The same design was then used for Latinx and White patients. Following this match, rates of transfer were compared between the groups. This was done using a series of conditional logistic regression models, including an unadjusted model and a model which adjusted for patient-level covariates including sex, age, primary insurance, zip code median income, primary language, marital status, education level, Elixhauser comorbidity index, admission season, number of admissions in previous month and having a PCP in our system. Analyses were considered significant at a 2-sided p-value of 0.05.

Results: Among the 72,113 admissions included in the cohort, 1,209 (2.1%) of White, 132 (2.3%) of Black and 138 (2.1%) of Latinx patients underwent IHT. After matching, patient characteristics that were not used for matching still differed by race; most notably, compared to White patients, Black and Latinx patient had significantly higher rates of Medicaid as primary insurance and of being in the lowest zip code median income quartile. There was a non-significant signal toward lower odds of IHT for Black compared to White patients in unadjusted (OR 0.86, 95% CI 0.70-1.05, p=0.14) and adjusted (OR 0.80, 95% CI 0.62-1.02; p=0.074) models. There was no significant difference for Latinx compared to White patients.

Conclusions: In this analysis of patients hospitalized on general medicine services at community hospitals within our health care system, we found that Black patients had a non-significant signal toward lower odds of IHT to tertiary hospitals within our system compared to White patients, and this persisted after adjusting for patient clinical and demographic variables. There are several potential explanations for these findings, including provider bias toward Black patients. Our findings emphasize the need for better understanding of transfer practices so that such inequities may be eliminated.