857: Subphenotypes of Infection Site and Organ Failure Predict Mortality in Adults With Sepsis in Africa

Introduction: Africa has the highest global incidence of sepsis and associated mortality. Yet, sepsis in Africa remains poorly understood. We aimed to identify associations between infection site, organ dysfunction, and 30-day mortality in adults with sepsis in Africa. Methods: We analyzed pooled data from patients at the Mbarara Regional Referral Hospital in Uganda, University Hospitals of Kigali and Butare in Rwanda, and JFK Hospital in Liberia (2017-23). We included adults ≥18 years with ≥2 qSOFA criteria. We used logistic regression and latent class analysis to determine clinical subphenotypes associated with 30-day mortality. Results: Of the 473 patients in the pooled analysis, 306 (65%) were from Uganda, 135 (29%) were from Rwanda, and 32 (7%) were from Liberia; 257 (54%) were female, the median (interquartile) age was 52 (17-95) years, and 127 (27%) were living with HIV. Death by 30-days occurred in 128 (27%). In the univariate analysis, subphenotype pairs that predicted increased mortality included pulmonary/central nervous system (CNS) infection (odds ratio [OR] 3.13, 95% confidence interval [Cl] 1.15-8.57), CNS infection/respiratory failure (OR 2.83, 95%CI 1.32–6.08), CNS infection/cardiac failure (OR 2.77, 95%Cl 1.41-5.45), CNS infection/multiorgan dysfunction (OR 2.74, 95%Cl 1.42-5.29), CNS infection/altered mental status (AMS) (OR 2.31, 95%Cl 1.24-4.29), pulmonary infection/AMS (OR 2.04, 95%Cl 1.22-3.41), and multiple infection sites/AMS (OR 1.82, 95%Cl 1.01-3.28). Subphenotype pairs that predicted decreased mortality included soft tissue infection/multiorgan dysfunction (OR 0.28, 95%Cl 0.08-0.96) and soft tissue infection/cardiac failure (0.28, 95%Cl 0.08-0.99). After adjusting for age, sex, and country, subphenotype pairs that predicted increased mortality included pulmonary infection/CNS infection (adjusted OR [AOR] 3.39, 95%CI 1.23-9.30) and pulmonary infection/AMS (AOR 3.26, 95%CI 1.35-7.87). Latent class analysis identified a subphenotype of AMS, cardiac failure, and respiratory failure that was associated with increased mortality (N=138; AOR 2.36, 95%Cl 1.28-4.37). Conclusions: High-risk subphenotypes of sepsis in Africa included AMS, cardiac failure, and respiratory failure, whereas soft tissue infection was included in low-risk subphenotypes.