HIV-exposed children account for more than half of 24-month mortality in Botswana

Background: The contribution of HIV-exposure to childhood mortality in a setting with widespread antiretroviral treatment (ART) availability has not been determined. Methods: From January 2012 to March 2013, mothers were enrolled within 48 h of delivery at 5 government postpartum wards in Botswana. Participants were followed by phone 1-3 monthly for 24 months. Risk factors for 24-month survival were assessed by Cox proportional hazards modeling. Results: Three thousand mothers (1499 HIV-infected) and their 3033 children (1515 HIV-exposed) were enrolled. During pregnancy 58 % received three-drug ART, 23 % received zidovudine alone, 11 % received no antiretrovirals (8 % unknown); 2.1 % of children were HIV-infected by 24 months. Vital status at 24 months was known for 3018 (99.5 %) children; 106 (3.5 %) died including 12 (38 %) HIV-infected, 70 (4.7 %) HIV-exposed uninfected, and 24 (1.6 %) HIV-unexposed. Risk factors for mortality were child HIV-infection (aHR 22.6, 95 % CI 10.7, 47.5 %), child HIV-exposure (aHR 2.7, 95 % CI 1.7, 4.5) and maternal death (aHR 8.9, 95 % CI 2.1, 37.1). Replacement feeding predicted mortality when modeled separately from HIV-exposure (aHR 2.3, 95 % CI 1.5, 3.6), but colinearity with HIV-exposure status precluded investigation of its independent effect. Applied at the population level (26 % maternal HIV prevalence), an estimated 52 % of child mortality occurs among HIV-exposed or HIV-infected children. Conclusions: In a programmatic setting with high maternal HIV prevalence and widespread maternal and child ART availability, HIV-exposed and HIV-infected children still account for most deaths at 24 months. Lack of breastfeeding was a likely contributor to excess mortality among HIV-exposed children.