Acute febrile illness in Kenya: Clinical characteristics and pathogens detected among patients hospitalized with fever, 2017-2019

Acute febrile illness (AFI) is a common reason for healthcare seeking and hospitalization in Sub-Saharan Africa and is often presumed to be malaria. However, a broad range of pathogens cause fever, and more comprehensive data on AFI etiology can improve clinical management, prevent unnecessary prescriptions, and guide public health interventions. We conducted surveillance for AFI (temperature >= 38.0 degrees C <14 days duration) among hospitalized patients of all ages at four sites in Kenya (Nairobi, Mombasa, Kakamega, and Kakuma). For cases of undifferentiated fever (UF), defined as AFI without diarrhea (>= 3 loose stools in 24 hours) or lower respiratory tract symptoms (cough/difficulty breathing plus oxygen saturation <90% or [in children <5 years] chest indrawing), we tested venous blood with real-time PCR-based TaqMan array cards (TAC) for 17 viral, 8 bacterial, and 3 protozoal fever-causing pathogens. From June 2017 to March 2019, we enrolled 3,232 AFI cases; 2,529 (78.2%) were aged <5 years. Among 3,021 with outcome data, 131 (4.3%) cases died while in hospital, including 106/2,369 (4.5%) among those <5 years. Among 1,735 (53.7%) UF cases, blood was collected from 1,340 (77.2%) of which 1,314 (98.1%) were tested by TAC; 715 (54.4%) had no pathogens detected, including 147/196 (75.0%) of those aged <12 months. The most common pathogen detected was Plasmodium, as a single pathogen in 471 (35.8%) cases and in combination with other pathogens in 38 (2.9%). HIV was detected in 51 (3.8%) UF cases tested by TAC and was most common in adults (25/236 [10.6%] ages 18-49, 4/40 [10.0%] ages >= 50 years). Chikungunya virus was found in 30 (2.3%) UF cases, detected only in the Mombasa site. Malaria prevention and control efforts are critical for reducing the burden of AFI, and improved diagnostic testing is needed to provide better insight into non-malarial causes of fever. The high case fatality of AFI underscores the need to optimize diagnosis and appropriate management of AFI to the local epidemiology.