Factors Associated With Death at 30 Days and Evaluation of Clinical Risk Scores Among Patients With Cancer Admitted With Postchemotherapy Infection in Uganda: A Prospective Cohort Study

Abstract

Background Little is known about outcomes from cancer chemotherapy-associated infections in sub-Saharan Africa. Accordingly, among cancer patients admitted with post-chemotherapy infection in Mbarara, Uganda, we aimed to determine 1) the 30-day case fatality rate, 2) factors associated with mortality, and 3) clinical risk score performance. Methods We enrolled participants aged 18 years or older if they 1) received cancer chemotherapy within the past 30 days, 2) were admitted to the oncology ward, and 3) were prescribed intravenous antibiotics. We used Cox proportional hazards regression to determine predictors of 30-day mortality and calculated the area under the receiver operating characteristic curve (AUC) for each clinical risk score. Results Among 150 participants, 67 (45%) were female and the median (IQR) age was 56 (43-66) years. Esophageal cancer (18%) and pneumonia (42%) were the most common malignancy and infection, respectively. Death by 30 days occurred in 63 (42%) participants. qSOFA ≥2 (adjusted hazard ratio [aHR] 2.51, 95% confidence interval [CI] 1.42-4.44, p=0.001), and UVA score >4 (aHR 2.13, 95% CI 1.08-4.18, p=0.03) were independently associated with 30-day mortality. An ECOG score ≥3 was similarly independently associated with 30-day mortality in the qSOFA and UVA models. The AUCs for qSOFA and UVA were 0.70 (95% CI 0.63-0.79) and 0.72 (95% CI 0.64-0.80), respectively. Conclusions In participants with post-chemotherapy infection in Mbarara, Uganda, the case fatality rate was high. ECOG, qSOFA, and UVA scores were associated with 30-day mortality.