Background: There is a paucity of recent prospective data on the incidence of postpartum infections and associated risk factors in sub-Saharan Africa. Retrospective studies estimate that puerperal sepsis causes approximately 10% of maternal deaths in Africa.

Methods: We enrolled 4231 women presenting to a Ugandan regional referral hospital for delivery or postpartum care into a prospective cohort and measured vital signs postpartum. Women developing fever (> 38.0 °C) or hypothermia (< 36.0 °C) underwent symptom questionnaire, structured physical exam, malaria testing, blood, and urine cultures. Demographic, treatment, and post-discharge outcomes data were collected from febrile/hypothermic women and a random sample of 1708 normothermic women. The primary outcome was in-hospital postpartum infection. Multivariable logistic regression was used to determine factors independently associated with postpartum fever/ hypothermia and with confirmed infection.

Results: Overall, 4176/4231 (99%) had ≥1 temperature measured and 205/4231 (5%) were febrile or hypothermic. An additional 1708 normothermic women were randomly selected for additional data collection, for a total sample size of 1913 participants, 1730 (90%) of whom had complete data. The mean age was 25 years, 214 (12%) were HIV-infected, 874 (51%) delivered by cesarean and 662 (38%) were primigravidae. Among febrile/hypothermic participants, 174/205 (85%) underwent full clinical and microbiological evaluation for infection, and an additional 24 (12%) had a partial evaluation. Overall, 84/4231 (2%) of participants met criteria for one or more in-hospital postpartum infections. Endometritis was the most common, identified in 76/193 (39%) of women evaluated clinically. Twenty-five of 175 (14%) participants with urinalysis and urine culture results met criteria for urinary tract infection. Bloodstream infection was diagnosed in 5/185 (3%) participants with blood culture results. Another 5/186 (3%) tested positive for malaria. Cesarean delivery was independently associated with incident, in-hospital postpartum infection (aOR 3.9, 95% CI 1.5– 10.3, P = 0.006), while antenatal clinic attendance was associated with reduced odds (aOR 0.4, 95% CI 0.2–0.9, P = 0.02). There was no difference in in-hospital maternal deaths between the febrile/hypothermic (1, 0.5%) and normothermic groups (0, P = 0.11).