OBJECTIVE: Sepsis is associated with significant morbidity and mortality in pediatric hematology, oncology, and transplant (PHOT) patients. This study characterized PHOT patients who developed hospital-onset sepsis more than 12 hours after admission and identified risk factors for 30-day sepsis-attributable (SA) mortality. PATIENTS AND METHODS: We analyzed an existing multicenter database of sepsis collected prospectively over 5 years (2017-2021) as part of the Improving Pediatric Sepsis Outcomes Collaborative. Sepsis was defined using operational elements documented in the health records based on International Classification of Diseases, Tenth Revision codes, treatment, diagnostic tests, and sepsis screen, huddle, or order set use. RESULTS: A total of 9604 sepsis episodes in PHOT patients from 49 hospitals were analyzed: 70.5% were identified in the emergency department (ED), 10.9% in inpatient settings less than or equal to 12 hours from admission, and 18.6% were hospital onset. Only 52.5% of patients with hospital-onset sepsis were identified using a sepsis recognition method compared with 87.2% in the ED (P <
.001). The overall 30-day SA mortality was 2.2%, with a higher rate (6.9%) among those with hospital-onset sepsis compared with those who developed sepsis at presentation or less than or equal to 12 hours (1.1%, P <
.001). CONCLUSIONS: Although the difference in SA mortality between hospitalized and nonhospitalized patients may be impacted by nonmeasurable confounders inherent to the type of patients presenting in the different care settings, we reported system-based improvements that may reduce mortality. The 30-day SA mortality was lower in hospitalized PHOT patients when sepsis was detected by early recognition methods, supporting the need for efforts to implement sepsis recognition tools in the inpatient setting.