BACKGROUND: HIV causes substantial mortality among African children but there is limited data on how this is influenced by maternal or infant infection status and timing. METHODS: Children enrolled in the ZVITAMBO trial were divided into 5 groups: those born to HIV-negative mothers (NE, n = 9510), those born to HIV-positive mothers but noninfected (NI, n = 3135), those infected in utero (IU, n = 381), those infected intrapartum (IP, n = 508), and those infected postnatally (PN, n = 258). Their mortality was estimated. RESULTS: Two-year mortality was 2.9% (NE infants), 9.2% (NI), 67.5% (IU), 65.1% (IP), and 33.2% (PN). Between 8 weeks and 6 months, mortality in IU infants quintupled (from 309 to 1686/1000 c-y). The median time from infection to death was 208, 380, and >500 days for IU, IP, and PN infants, respectively. Among NI children, advanced maternal disease was predictive of mortality. Acute respiratory infection was the major cause of death. CONCLUSIONS: Perinatally infected infants are at particular risk of death between 2 and 6 months: cotrimoxazole prophylaxis and early pediatric HAART should be scaled up. Uninfected infants of infected mothers have at least twice the mortality risk of infants born to uninfected mothers: all HIV-exposed infants should be targeted with child survival interventions. HIV-positive mothers with more advanced disease are not only more likely to infect their infants, but their infants are more likely to die, whether infected or not: provision of antiretroviral treatment to pregnant and lactating women is an urgent need for both mothers and their children. © 2007 Lippincott Williams & Wilkins, Inc.
