Humphrey 2006.

Methods	Randomized, placebo‐controlled trial.
Participants	4495 mother‐infants pairs who were part of the Zimbabwe Vitamin A for Mothers and Babies (ZVITAMBO) trial in Harare Zimbabwe. Mother‐infant pairs were enrolled at 96 hours (or less) after delivery.
Interventions	A 2‐by‐2 factorial design with 4 treatment groups Aa, Ap, Pa, and Pp; where “A” was maternal vitamin A supplementation (400,000 IU), “P” was maternal placebo, “a” was infant vitamin A supplementation (50,000 IU), and “p” was infant placebo. In this review, we used only data from Ap (intervention) versus Pp (placebo).
Outcomes	Primary outcome: breastfeeding–associated MTCT and HIV‐free survival. Secondary outcome: adverse effects in HIV‐positive women or their infants.
Notes	All but 4 mothers initiated breastfeeding, no information on ART or cotrimoxazole prophylaxis.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The trial authors performed randomization using computer‐generated blocks of 12.
Allocation concealment (selection bias)	Low risk	Treatment assignment was concealed by pre‐packing study supplements in sequentially numbered series assigned to study identification numbers. Concealed envelopes with study number; link between number and treatment assignment kept at central location.
Blinding (performance bias and detection bias) All outcomes	Low risk	Participants, care providers, and outcome assessors were blinded to the treatment allocation. Mothers were assigned an original study identification number at enrolment and were given the next sequentially numbered opaque bottle with supplements.
Incomplete outcome data (attrition bias) All outcomes	Low risk	One hundred and forty‐three (3.2%) mother‐infant pairs were lost to follow‐up. We do not believe that the loss to follow‐up was related to the outcome.
Selective reporting (reporting bias)	Low risk	The trial authors reported outcomes that were prespecified in the protocol (NCT00198718).
Other bias	Low risk	We do not believe that other biases were introduced that could have affected the study findings.