Does deworming pay off decades later?

The long-run gains are real

A randomized program in Kenya found that children who got more years of deworming had higher consumption and hourly earnings as adults. Trials that treat children one by one miss the spillover benefits of treating whole schools, and short trials measure too little, too soon, to see effects that build over a lifetime.

The evidence is too thin to rely on

Dozens of trials, pooled in Cochrane and Campbell reviews, find that mass deworming does little for growth, cognition, school performance or survival. A large economic effect with no measurable health effect needs an explanation, and it rests mostly on one program whose reanalysis weakened some of its findings.

Act on expected value

Deworming may well do little, but it costs about 50 cents to a dollar per child treated. A modest chance of large lifelong gains can still make it a good bet, as long as the estimate is discounted heavily for the doubt.

The debate

In 2004 Miguel and Kremer reported that a school deworming program in Busia, Kenya, cut absenteeism in treated schools by about a quarter and helped untreated children nearby, though it did not raise test scores. Later follow-ups estimated about 14 percent higher consumption, pooling the 15- and 20-year surveys, and 13 percent higher hourly earnings, pooling the 10-, 15- and 20-year surveys, among people assigned two to three additional years of treatment. Both estimates meet the 10 percent significance threshold, but not the conventional 5 percent threshold.

Systematic reviews read the wider evidence differently. A 2019 Cochrane review of 51 trials concluded that mass deworming does not appear to improve height, hemoglobin, cognition, school performance or mortality, and it cautioned against basing programs on two older studies that found large weight gains. A 2017 review found little to no effect on weight, height, cognition, attendance or mortality, and judged the long-run evidence too weak to draw conclusions.

In 2015 independent researchers published two reanalyses of the Kenya data, setting off what David Roodman, writing on GiveWell’s blog, later called the “great worm wars.” After coding errors were corrected, they found little evidence that attendance rose in neighboring schools, and evidence of better attendance in treated schools that depended on the method used and carried a high risk of bias. The original authors replied that the corrected data strongly support their findings on spillovers.

GiveWell took the third position. In 2016 it wrote that most of its staff thought deworming more likely than not to have very little or no impact, but at $0.50 to $1 per person treated, it applied a 1 to 2 percent multiplier to the trial results and still found the expected value worth funding. Roodman judged the earnings result credible but was less sure it would hold in other times and places.

Where Charitology stands

Charitology grades deworming’s long-run evidence mixed: rigorous studies disagree, and the strongest positive evidence comes from one program. How much to stake on an uncertain chance of a large benefit is a value question, and Charitology doesn’t settle it. The evidence gap is clearer: long-run results from other times and places are what would move the debate. See the deworming entry, the replication rate and expected value.

The cruxes

What would have to be true for each side to change its mind:

Whether large economic gains can exist when short-term health gains are too small to detect.
Whether results from heavily infected children in Busia in 1998 carry over to the lighter infections common today.
Whether new long-run studies in other places would find what the Kenya studies found.
How much weight a funder should give a small chance of a large benefit compared with a near-certain small one.