The relative contribution of HIV-exposure and HIV-infection to childhood mortality in a programmatic setting with widespread antiretroviral treatment (ART) use in pregnancy is not well described.
From March 2012-March 2013, women and their children were enrolled within 48 hours of delivery in 5 geographically diverse government-run postpartum wards in Botswana. Maternal obstetric and medical history, demographics and socioeconomic status were collected at enrollment. Women were interviewed every 1-3 months for 24 months by phone (or home visit if not reachable) to assess child HIV status, feeding modality, food security and child and maternal mortality. Mothers and children received all care, including ART, through the government health system. Child feeding was according to Botswana guidelines. Risk factors for 24-month survival were assessed by Cox proportional hazard modeling.
A total of 3000 women (1499 HIV-infected) and their 3033 children (1515 HIV-exposed) were enrolled. Among HIV-infected women, 58% were on ART, 23% on zidovudine (ZDV) alone, 11% received no ARVs during pregnancy, and 8% were unknown. Incidence of Mother-to-child HIV transmission (MTCT) was 2.1% through 24 months. Vital status at 24 months was known for 3018 (99.5%) children and there were 106 (3.5%) deaths overall. Mortality differed by HIV exposure status (Figure 1): 12 (38%) among HIV-infected, 70 (4.7%) among HIV-exposed uninfected (HEU), and 24 (1.6%) among HIV-unexposed uninfected (HUU). The only independent risk factors for mortality in adjusted analysis 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 7.7, 95%CI 1.9,31.9). Although the risk of replacement feeding was significant when modeled separately from HIV-exposure status (aHR 2.3, 95% CI 1.5, 3.6), feeding and HIV-exposure status were co-linear (99.7% of HIV-unexposed but only 13% of HIV-exposed children breastfed) and could not be modeled together. Given 26% HIV prevalence among pregnant women in Botswana, more than half of all 24-month child mortality is accounted for by HEU (46%) plus HIV-infected (8%) children.
In a programmatic setting with widespread maternal and child ART availability, HIV-exposed children contribute approximately half the overall deaths at 24 months. Lack of breastfeeding could not be assessed independently of HIV-exposure status, but was a likely contributor to excess mortality among HEUs.