Background:  Assumptions about survival of HIV-infected children in Africa in the absence of antiretroviral therapy need to be updated and refined to inform ongoing UNAIDS modeling of HIV epidemiology among children. Precise estimations of infant survival according to the timing of HIV-infection (perinatally or through breastfeeding) and estimates on the effect of background mortality on survival post-infection of children by timing of infection are needed.

Methods:  A pooled analysis of the individual data of all available intervention cohorts and randomized trials aimed at the prevention of mother-to-child transmission of HIV in Africa was conducted. Data were right-censored at the time of infant antiretroviral initiation. Overall mortality as a rate per 1000 child-years of follow-up was calculated by selected maternal and infant characteristics. The Kaplan-Meier method was used to estimate survival curves according to children's HIV infection status. Random-effects Weibull regression models were computed to estimate mortality hazard ratios accounting for heterogeneity between trials and cohorts. Assumptions were also made to ignore all non-HIV causes of mortality to estimate survival as if HIV were the only cause of death.

Results:  Individual data from 12 studies was pooled together, accounting for 10,998 children born to HIV-infected women. Preliminary results show the mortality rate per 1000 child-years of follow-up was 33.3 and 387.9 among HIV-uninfected and HIV-infected children, respectively. By age 1 year, an estimated 4.3% uninfected, 15.9% infected through breastfeeding and 44.3% perinatally infected children will have died. Infant mortality was independently associated with maternal death (adjusted hazard ratio = 1.7, 95%CI 1.3 to 2.5), maternal CD4 <200 cells/mL (1.9, 1.5 to 2.3), HIV-infection through breastfeeding (4.8, 3.6 to 6.4) or peri-partum HIV-infection (14.1, 11.5 to 17.3). Background mortality had very little effect on the differences on survival post infection for both early and late infection.

Conclusions:  These results update previous work and will inform future UNAIDS modeling to provide estimates on survival of HIV-infected untreated African children according to the timing of infection. These findings highlight the urgent need for the early assessment of HIV infection in HIV-exposed children to allow the timely initiation of antiretroviral care and support for HIV-infected children.