Diarrheal disease kills hundreds of thousands of children a year, mostly from water that carries what it should not. The old public-health assumption was that clean water reduces diarrhea but does not much change death rates; individual trials were too small to show mortality effects.
The finding
A 2022 meta-analysis led by Nobel laureate Michael Kremer pooled the trials that did record deaths. Across them, water treatment, mostly chlorination, was associated with a 24 to 27% reduction in the odds of death among children under five. The authors’ expected effect for a new study was about 23%. On their cost model, the result put chlorination among the cheapest ways to save a life anywhere.
Evidence Action, which runs a Dispensers for Safe Water program in Kenya, Uganda and Malawi (a chlorine dispenser next to the village water point, refilled by the program), reported reaching more than four million people.
Why GiveWell’s number is lower
GiveWell did its own meta-analysis and arrived at a mortality reduction of about 14%, later revised to about 12%. The gap comes from judgment calls: which trials to include, how much to trust trials that were not designed to measure deaths, how to adjust for the fact that the trials were mostly small and their results noisy, and how much of the effect carries over when a program runs at scale with lower uptake than a trial.
Independent reviewers at The Unjournal reached a similar cautious view: the direction is convincing, the size is uncertain.
Here is the point that matters for a donor. Even at 12%, GiveWell judged the program cost-effective enough to fund at scale. In 2022 it granted about $64 million to Evidence Action to expand dispensers and pilot in-line chlorination, which treats water at the source so no household action is needed. It has continued funding water programs since.
Why the program is cheap
- A chlorine dispenser serves a whole village for a few dollars per person per year.
- In-line chlorination, where piped or shared supplies exist, costs less and removes the behavior step.
- The benefit accrues to everyone who drinks the water, not only the children being counted.
Why it was overlooked
Trials measured diarrhea episodes, not deaths, because deaths are rare and require huge samples. It took pooling many trials, decades after they ran, to see the mortality effect. That is a general lesson: a program can look mediocre on the outcome trials were designed to measure and excellent on the one that matters.
What to make of the two numbers
If Kremer’s group is right, chlorination is roughly as cost-effective as the very best programs in the world. If GiveWell is right, it is merely very good and worth funding. There is no reading of the evidence on which it is a poor use of money, and it addresses a problem that malaria and nutrition programs do not touch.
Method notes
Effect sizes are as reported by the authors and by GiveWell; the 12 to 14% figures are GiveWell’s successive estimates. The $64 million grant is GiveWell’s January 2022 grant to Evidence Action as described in its program page and blog. Per-person cost figures are rounded from program reporting.
Sources
- Kremer, Luby, Maertens, Tan, Więcek: Water treatment and child mortality, a meta-analysis and cost-effectiveness analysis (NBER working paper 30835)
- GiveWell: Water quality interventions
- GiveWell blog: A major update in our assessment of water quality interventions (April 6, 2022)
- Evidence Action: Emerging evidence suggests water treatment is even more impactful and cost-effective in averting child mortality
- The Unjournal: Evaluation summary of Water treatment and child mortality
Figures come from the cited sources as of the dates shown. Cost-effectiveness estimates change as programs and evidence change; check the source before giving. We are not financial advisers and receive nothing from any charity named here.