Influence of participation on mortality in a randomized trial of vitamin A prophylaxis

Mortality of Sumatran children living in villages randomized to participate in a vitamin A capsule (200 000 IU) distribution program who received the capsule (n = 9776) was compared with those who did not (n = 2447) and with children living in villages randomized to serve as control subjects (n = 12 173). During the 4 mo after completion of the first distribution, mortality among preschool capsule recipients was <4% that of nonrecipients (p<0.001). Mortality among preschool nonrecipients was three times that of controls (p<0.05), suggesting strong selection bias. The potential biologic impact on childhood mortality attributable to vitamin A supplementation is estimated to exceed the 34% previously derived from the more conservative intent-to-treat analysis. One capsule every 6 mo may provide adequate protection for the vast majority of children. The single major limitation to maximum impact appears to be inadequate program coverage.