Performance of Early-Warning Scores in Predicting Mortality in an HIV-Infected Population with Sepsis in Uganda

Abstract

Background Early-warning scores (EWS) have the potential to improve resource allocation and hasten care in sub-Saharan Africa (SSA). Despite the high prevalence of HIV infection in SSA, current EWS do not take into account HIV serostatus.Methods We conducted a retrospective study at Mbarara Regional Referral Hospital (MRRH) in Uganda to evaluate the performance of CRB-65, modified early-warning score (MEWS), quick sepsis-related organ failure assessment (qSOFA), rapid acute physiology score (RAPS), rapid emergency medicine score (REMS), South African triage scale (SATS), and shock index (SI) in predicting mortality among HIV-infected patients presenting with sepsis. We included patients admitted with sepsis to MRRH between January 2014 and December 2015 that had an HIV-positive serostatus and at least one valid heart rate, respiratory rate, systolic blood pressure, diastolic blood pressure, temperature, and oxygen saturation. Glasgow coma scale was imputed with the median. We used the area under the receiver operating curve (AUC) with tenfold cross-validation to assess the performance of each EWS.ResultsOf the 193 patients, the median (interquartile range) age was 34 (27, 42) years, 87 (45.0%) were female and 65 (44.6%) died. The AUC (95% confidence interval) was 0.53 (0.43, 0.62) for CRB65, 0.53 (0.44, 0.62) for MEWS, 0.57 (0.46, 0.68), for qSOFA, 0.60 (0.51, 0.69) for RAPS, 0.55 (0.46, 0.63) for REMS, 0.53 (0.45, 0.62) for SATS, and 0.54 (0.46, 0.63) for SI.Conclusion The ability of EWS to predict mortality in an HIV-infected patient population with sepsis in Uganda was poor. EWS used in SSA should be derived from African patient populations and adjust for HIV serostatus.Disclosures All authors: No reported disclosures.