Background:  All children, irrespective of HIV-infection status, of HIV-infected mothers may be at increased risk of death; in infected children, age at infection acquisition may be associated with mortality risk. Using individual data from 7 randomized clinical MTCT intervention trials we estimated mortality in children born to HIV-infected mothers and investigated the effect of maternal health, infant’s HIV infection, feeding practices, and timing of infection acquisition.

 Methods: Feeding status was classified as ever/never breastfed. Early HIV infection was defined as a single positive RNA or DNA PCR test before 4 weeks of age. Late postnatal transmission was defined by a negative HIV-1 diagnostic test result(s) at or after 4 weeks of age, followed by a positive test result(s). Mortality was expressed as a rate per 1000 child-years and correlates assessed with Cox proportional hazards models.

Results:   A total of 378 (of 3468; 10.9%) children died. By age 1 year, an estimated 36.8% infected and 3.6% uninfected children died; by 2 years 54.6% and 5.5%. Mortality risk was independently associated with living in West Africa (AOR 1.73), maternal death (AOR 2.23), CD4 cell counts < 200 (AOR 1.66), and infant HIV infection (AOR 11.5); OR were similar in stratified analyses by infection status. Breastfeeding was not significantly associated with mortality in either infected or uninfected children. In multivariate analyses of survival from birth for infected children, there was evidence of increased mortality risk for children infected early (AOR 1.67, p = 0.001). This effect disappeared, however, when mortality rates were calculated from the time of acquiring infection (birth for early infection, and age at postnatal acquisition for late) (AOR 1.21, p = 0.21), while the effect of the other correlates remained the same.  Of children with early infection, 35.7% died within 12 months of acquiring infection, and of those with late infection,32.9% did.

Conclusions:   HIV-infected children have a very high mortality, and there is an urgent need to improve diagnosis and care of infected children in Africa. Maternal health influences survival of both HIV-infected and uninfected children , which highlights the need to prioritise maternal health for infected women and preventing primary HIV infection in young women. The lack of effect of timing of infection on survival may suggest that the effect of HIV infection on mortality dominates at the ages where non-HIV mortality is lower.

Keywords: infant mortality; breastfeeding; timing of acquisition