On this page
In 1998, a charity in western Kenya began giving free deworming pills in primary schools, a few schools at a time. Economists Edward Miguel and Michael Kremer used that rollout to ask: if children are cured of intestinal worms, do they come to school more? Their answer appeared in Econometrica in 2004. A 2015 re-analysis started an argument nicknamed the "worm wars".
What did the Kenya deworming study test?
The study covered 75 rural primary schools in southern Busia district, with more than 30,000 pupils aged 6 to 18 (Miguel and Kremer, 2004). Schools were sorted by area and other aid programmes, listed alphabetically, and every third one went to a group: three groups of 25. Group 1 was treated from 1998, Group 2 from 1999 and Group 3 from 2001, so each year some schools were the comparison.
Treated schools received deworming drugs and health lessons. Attendance was not taken from school records, which the authors say are often poorly kept; field workers made unannounced visits, about 3.8 per school per year, and counted who was present.
The key choice was to randomise whole schools, not children. Worms spread through soil and water, so treating one child can protect classmates, a spillover that child-level randomisation would miss. Later re-analysts called the assignment quasi-randomised.
What did it find about school attendance?
The headline: the programme "reduced school absenteeism in treatment schools by one-quarter" (Miguel and Kremer, 2004). School participation in treated schools was 7.5 percentage points higher over 1998 and 1999. Counting spillovers, they estimated each extra year of school participation cost only $3.50.
They also reported benefits for untreated children in treated and neighbouring schools. They found no evidence that deworming improved test scores.
Why is it called the "worm wars"?
In 2015, epidemiologists at the London School of Hygiene and Tropical Medicine re-ran the data in two papers. The first, a "pure replication" by Aiken, Davey, Hargreaves and Hayes, used the original methods and found coding errors. After fixing them, there was little evidence of spillovers on attendance in nearby schools, and no evidence of benefit on anaemia. The combined attendance gain among treated children, reported as 7.5% originally, fell to 3.9% and was no longer statistically significant. Effects inside treated schools were largely unchanged.
The second, by Davey and colleagues, used their own methods. Year by year, one analysis found no clear evidence of better attendance and another found some; combining both years showed a strong effect. They concluded the study gives "some evidence, but with high risk of bias" that attendance improved, and none for exam performance.
The original authors replied in the same journal. Hicks, Kremer and Miguel accepted the coding errors but said the corrected results were "extremely similar" for direct effects and spillovers within 3 km. They argued that adding a noisy 3 to 6 km estimate made the total look insignificant. World Bank economist Berk Özler found the original more robust than before. Splitting the data by calendar year, as the second paper did, counts many untreated early-1999 comparison observations (20%, citing the authors' reply) as treated.
The wider evidence is mixed. A 2019 Cochrane review of 51 trials found routine deworming does not appear to improve height, haemoglobin, cognition or school performance. On attendance it said the evidence was "inconsistent, and at risk of bias".
What happened to the children 20 years later?
A 2021 PNAS follow-up by Hamory, Miguel, Walker, Kremer and Baird had an effective tracking rate of 84%. People who got two to three more years of childhood deworming had 14% higher consumption spending and 13% higher hourly earnings as adults. It is the original team's work, not an independent check, and declares that Kremer works with USAID, which supports deworming, and formerly sat on Deworm the World's board.
What can Indian students and families take from it?
This part is our inference, not a finding. Treat the Kenya result as one well-designed trial, in a place with heavy worm infection, whose attendance effect careful people still dispute.
Use the worm wars to read study news:
What was randomised: pupils or schools?
What was measured, and how?
Has someone re-run the data, and did the numbers hold?
Does the result survive other reasonable analyses?
Do the authors declare links to the cause?
This post is not health advice. For deworming in India, see the official National Deworming Day handout from the National Health Mission, which says the tablet is given free to children aged 1 to 19 in schools and anganwadis.
If helminth life cycles come up in Class 12 biology, the Fermi Concept Library covers every concept in biology, each with its own lesson. Read also our explainer on what a randomised controlled trial is, another Kenyan school study on tracking pupils by ability, and our guide to correlation versus causation in study news.



