Abstract
BACKGROUND: Little is known about the microbiology and outcomes of chemotherapy-associated febrile illness among patients in sub-Saharan Africa. Understanding the microbiology of febrile illness could improve antibiotic selection and infection-related outcomes. METHODS: From September 2019 through June 2022, we prospectively enrolled adult inpatients at the Uganda Cancer Institute who had solid tumors and developed fever within 30 days of receiving chemotherapy. Evaluation included blood cultures, malaria rapid diagnostic tests, and urinary lipoarabinomannan testing for tuberculosis. Serum cryptococcal antigen was evaluated in participants with human immunodeficiency virus (HIV). The primary outcome was the mortality rate 40 days after fever onset, which we estimated using Cox proportional hazards models. RESULTS: A total of 104 febrile episodes occurred among 99 participants. Thirty febrile episodes (29%) had ≥1 positive microbiologic result. The most frequently identified causes of infection were tuberculosis (19%) and bacteremia (12%). The prevalence of tuberculosis did not differ by HIV status. The 40-day case fatality ratio was 25%. There was no difference in all-cause mortality based on HIV serostatus, presence of neutropenia, or positive microbiologic results. A universal vital assessment score of >4 was associated with all-cause mortality (hazard ratio, 14.5 [95% confidence interval, 5-42.7]). CONCLUSIONS: The 40-day mortality rate among Ugandan patients with solid tumors who developed chemotherapy-associated febrile illness was high, and few had an identified source of infection. Tuberculosis and bacterial bloodstream infections were the leading diagnoses associated with fever. Tuberculosis should be included in the differential diagnosis for patients who develop fever after receiving chemotherapy in tuberculosis-endemic settings, regardless of HIV serostatus.