Randomized trial of vitamin supplements in relation to transmission of HIV-1 through breastfeeding and early child mortality

Randomized placebo-controlled trial in HIV-infected pregnant women attending public antenatal clinics in Dar es Salaam, Tanzania. Women were enrolled at 12 to 27 weeks' gestation between April 1995 and the next 2 years and followed through pregnancy, breastfeeding, and up to 24 months of child age. Active intervention arm sizes were 442 for vitamin A alone, 453 for multivitamins without vitamin A, and 456 for multivitamins with vitamin A.

Three active daily oral regimens were compared with identical placebo from enrollment through pregnancy and lactation: vitamin A alone (30 mg beta-carotene plus 5000 IU preformed vitamin A), multivitamins without vitamin A (B1 20 mg, B2 20 mg, B6 25 mg, niacin 100 mg, B12 50 µg, vitamin C 500 mg, vitamin E 30 mg, folic acid 0.8 mg), or the same multivitamins plus vitamin A. At delivery, women in the vitamin A arms also received 200,000 IU vitamin A orally; all regimens were oral and given daily.

Vitamin A supplementation was harmful, increasing HIV transmission through breastfeeding and raising total infant HIV infection by 24 months, while multivitamins without vitamin A did not improve overall transmission or survival but showed benefit in some higher-risk subgroups. In the vitamin A arm, total HIV infection by 24 months was higher (RR 1.38, 1.09-1.76; P=0.009), with no mortality benefit (RR 1.03, 0.66-1.62; P=0.89). In the multivitamins without vitamin A arm, overall total HIV infection was unchanged (RR 1.04, 0.82-1.32; P=0.76) and breastfeeding transmission was not significantly reduced (RR 0.85, 0.61-1.19; P=0.34), although mortality trended lower (RR 0.82, 0.66-1.02; P=0.08) and transmission fell in the lowest total lymphocyte quartile (RR 0.37, 0.16-0.85; P=0.02). The multivitamin-plus-vitamin A arm did not show a clear overall advantage.

The multivitamin findings were driven largely by subgroup and time-varying analyses rather than the primary overall comparisons, increasing the risk of chance findings from multiple testing. Follow-up was limited to 24 months, and the vitamin A-containing multivitamin arm was not fully separated for every outcome in the extracted summary, which limits precision for arm-specific interpretation.