Vitamin A supplementation

What the evidence says

A 2022 Cochrane review pooled 19 trials with about 1.2 million children for its main mortality estimate. Supplementation cut deaths from all causes by 12 percent (risk ratio 0.88) and deaths from diarrhea by a similar share, and the reviewers rated both findings high-certainty. It also reduced night blindness and vitamin A deficiency, and it made vomiting more common in the two days after a dose.

The trials don’t all agree. The largest by far, DEVTA, followed about 1 million preschool children in north India. It found a 4 percent reduction in deaths that was not statistically significant, far below the 20 to 30 percent its authors had expected from earlier trials. DEVTA pulls the pooled estimate down: without it, the reduction is 23 percent. The Cochrane authors still judge the effect robust.

The harder question is whether old results hold today. GiveWell, which counts vitamin A among its longest-funded programs, notes that most supporting trials ran 30 to 40 years ago, when infections and deficiency were more common. It now estimates the effect in the places it models at a 1 to 11 percent lower risk of death, and it is exploring whether a new randomized trial is feasible. Charitology grades the evidence moderate for that reason: the historical trial base is strong, but the largest recent trial and today’s changed conditions both point to a smaller effect.

Where it sits on the depth scale

Vitamin A supplementation is prevention (depth 2). A dose protects a child for months, then the next round is needed. It corrects the deficiency in each child without changing why the deficiency exists, such as diet and food supply. Work that changes what a whole population eats would sit closer to structure.

Open questions

• Does supplementation still save lives where deficiency and child deaths have fallen? The Cochrane review found no clear benefit in trials from low-mortality countries, but there were only two, too few to settle it.
• What explains the gap between DEVTA and the earlier trials? The answer matters for every program built on decades-old evidence. See the replication rate.
• The Cochrane authors argue that new placebo-controlled trials would be unethical where deficiency is documented. Funders still need better evidence, which raises the question of what kind of study can fill the gap.