Addressing publication bias

Vitamin A supplementation is one of the programs GiveWell has supported the longest, and we’re currently funding it in many African countries. The program has an unusually strong evidence base for reducing child mortality, with multiple randomized controlled trials. Yet, as is the case for most global health programs, the evidence for vitamin A supplementation has complex, unresolved questions, such as how well findings from decades-old trials apply today and the extent to which existing research has been influenced by publication bias. As GiveWell’s research team has grown over the last several years, we have expanded our capacity to carefully research these questions. 

In this episode, GiveWell CEO and co-founder Elie Hassenfeld speaks with Senior Researcher Stephan Guyenet about the evidence base for vitamin A supplementation, the complications in applying that evidence to our funding decisions, and how GiveWell has improved our cost-effectiveness estimates for the program.

Elie and Stephan discuss:

• The evidence base for vitamin A supplementation: Evidence from rigorous trials shows that vitamin A supplementation can significantly reduce child mortality. However, most of these trials were conducted 30 to 40 years ago, when infectious disease rates and vitamin A deficiency were more prevalent, and a more recent large trial in India found a much smaller effect. As a result, we worked to resolve our uncertainties about the effect of vitamin A supplementation in the contexts where we’re supporting it today. 
• How GiveWell worked through the complications: To address our questions about the existing evidence, we engaged in further research. This included an analysis focused on the specific diseases for which vitamin A supplementation reduces mortality, a novel dosing-frequency analysis conducted in consultation with an outside statistician, and an assessment of possible publication bias. 
• What all of this means for grantmaking: While the trials report that vitamin A supplementation reduces the risk of death by 19% on average in children 6 to 59 months old, we estimate that the impact in the modern settings we model is smaller: a 1% to 11% lower risk of death. Nevertheless, we still think vitamin A supplementation can be highly cost-effective. Using our updated cost-effectiveness analysis, we estimate our most recent grant for vitamin A supplementation to Helen Keller Intl is 25 times more cost-effective than our benchmark. This change reflects our more precise, location-specific analyses that allows us to direct funding to places where vitamin A supplementation is likely to be the most cost-effective. 

GiveWell continues to scrutinize the programs we fund, including those we have supported for years. In this case, years of rigorous research have largely held up the case for vitamin A supplementation. We will continue to review the program, funding new research to address remaining uncertainties and exploring whether a new randomized trial might be feasible. By doing so, we’ll continue to increase our confidence and refine our funding decisions to target the most cost-effective locations and do the most good we can with donors’ funds.

Visit our All Grants Fund page to learn more about how you can support this work, and listen or subscribe to our podcast for our latest updates.

This episode was recorded on March 25, 2026 and represents our best understanding at that time.