Abstract
BACKGROUND: Transfusion of packed red blood cells (pRBCs) below a hemoglobin concentration of 7.0 g/dL is common, but it is unclear if the timing of transfusion during hospitalization modifies transfusion effectiveness. We sought to determine if effects of pRBC transfusion are heterogenous based on time from hospital admission. METHODS: Multicenter retrospective cohort study including hospitals in the Premier Inc. AI Healthcare Database between 2016 and 2022. Hospital encounters for adults with at least one hemoglobin concentration measured between hospital Days 1-7 were included. For each day, the lowest hemoglobin concentration was identified; patient-days were then separated into seven analytic cohorts based on the day in which a hemoglobin concentration was recorded (index day). We used regression discontinuity to quantify the effect of index day pRBC transfusion versus no transfusion on an outcome of hospital mortality or discharge to hospice (risk difference [RD]) at a hemoglobin concentration threshold of 7.0 g/dL in each cohort. RESULTS: A total of 2,293,021 index days across 997,277 inpatient encounters were included. The association between pRBC transfusion and hospital mortality or discharge to hospice differed based on days from admission, shifting from benefit on Day 1 (RD: -2.9 [95% CI: -5.9, -0.04] %) to harm on index Days 6 and 7 (Day 6, RD: 3.3 [95% CI: 0.4, 6.1] %; Day 7, RD: 4.1 [95% CI: 0.8, 7.3] %, p-interaction < .0001). CONCLUSIONS: Transfusion during hospitalization was associated with benefit early in hospitalization and harm at later time points.