223. Predicting Mortality Among Immunocompromised Patients Who Present With Bloodstream Infection

Abstract

Abstract Background The revised definition of sepsis (Sepsis-3) uses Sequential Organ Failure Assessment (SOFA) and quick SOFA (qSOFA) to identify patients with sepsis instead of systemic inflammatory response syndrome (SIRS) criteria. Subsequent studies revealed contradictory results pertaining to qSOFA, and limited data are available for immunocompromised patients. The objectives of this study were to (1) evaluate the performance of Sepsis -3 in a cohort of immunocompromised patients with microbiologically-proven sepsis, defined as having received antibiotics and having bloodstream infection (BSI); and (2) to compare its performance in the BSI cohort to its performance in immunocompromised patients who received antibiotics but did not have BSI (Non-BSI cohort). Methods Adult patients with hematologic malignancy or solid transplant recipients admitted to Michigan Medicine between 2012–2017 with suspected infection were included based on criteria used in the Sepsis-3 study: having both a body fluid culture and having received intravenous antibiotics. SOFA, qSOFA and SIRS components within 1 day of the index date (culture date or antibiotic date, whichever came first) were extracted from the medical record. For each group, a baseline risk model for mortality was created including age, gender, race, and Charlson comorbidity index. Each score (SOFA ≥ 2, qSOFA ≥ 2, SIRS ≥ 2) was added to the baseline risk model as a dichotomous variable and AUROC values were calculated. Results 2822 patients with a mean age of 56.8±15.6 were included. 349 (12.4%) had BSI. The most common immune compromising conditions were solid-organ transplantation (47%), lymphoma (21.3%) and acute leukemia (17%). 14% of patients in the BSI cohort died during hospitalization compared with 6.6% in the non-BSI cohort (P < 0.001). For the BSI cohort, when SOFA ≥ 2, qSOFA ≥ 2, SIRS ≥ 2 scores were added to the model, the AUROC values were less than those for the non-BSI cohort (table). The addition of SOFA ≥6 to the baseline risk model produced the highest AUROC values in both the BSI and non-BSI cohorts (figure). Conclusion Among immunocompromised patients, an SOFA score ≥6 was the strongest predictor of mortality. Surprisingly, sepsis scores performed better in the non-BSI cohort than in the BSI cohort. Disclosures All authors: No reported disclosures.