The Effect of India's Total Sanitation Campaign on Defecation Behaviors and Child Health in Rural Madhya PradeshA Cluster Randomized Controlled Trial

A field researcher walks between villages in rural Madhya Pradesh in the spring of 2009, counting latrines, asking families where they defecate, and collecting stool samples from toddlers. The work is methodical, almost unglamorous. But the study it belongs to is historic: the first randomized controlled trial of a large-scale government sanitation program anywhere in the world. Everything policymakers believed about latrines and child health had rested on observational data. Patil and colleagues were about to find out if the assumption held. The assumption had real weight behind it. India alone accounted for a third of the global sanitation deficit, with eight hundred fourteen million people lacking improved facilities, and nearly sixty percent of all open defecation worldwide. The country also bore a quarter of global deaths from diarrheal disease in children under five. The causal logic motivating sanitation investment is straightforward: open defecation loads the environment with fecal pathogens, which reach children through contaminated water, soil, food, and flies. Provide latrines, shift behavior, break that transmission route, and you should see less diarrhea, fewer parasitic infections, less anemia, and better growth. The chain of reasoning is sound. The question Patil and colleagues set out to answer was whether it actually worked at program scale. To test it, they designed something ambitious. Eighty rural villages in two neighboring districts of Madhya Pradesh, Dhar and Khargone, were randomized to intervention or control. The intervention was India's Total Sanitation Campaign, or TSC, delivered with capacity-building support from the World Bank's Water and Sanitation Program. That meant supply-side technical support, rural sanitary marts, social mobilization using community-led triggering tools to shift norms around defecation, promotion of a specific offset two-pit latrine design, and real money. Below-poverty-line households received at least four thousand two hundred rupees, which is about eighty-four dollars, combining TSC subsidies with funds from a concurrent employment scheme. Villages, blocks, and districts could also earn community awards ranging up to five million rupees for achieving open-defecation-free status. The program used community-led total sanitation tools, but Patil and colleagues are careful to note this was not a classical community-led total sanitation program because household hardware subsidies were included throughout. The enrolled population was large: five thousand two hundred nine children under five from three thousand thirty-nine households, with baseline data collected between May and July 2009 and follow-up fieldwork twenty-one months later, in early 2011. A subsample of one thousand one hundred fifty children under twenty-four months at enrollment also provided stool specimens, tested for soil-transmitted helminths and protozoan infections. Randomization successfully balanced intervention and control groups, and the analysis was conducted as an intention to treat. This means villages were analyzed according to their assigned group regardless of what actually happened on the ground. What happened on the ground is where the story gets interesting. The TSC moved the hardware numbers clearly. At follow-up, the percentage of households in intervention villages with World Health Organization and United Nations Children's Fund Joint Monitoring Programme-defined improved sanitation facilities had risen to forty-one percent, compared to twenty-two percent in control villages — a nineteen percentage point increase with a confidence interval running from twelve to twenty-six percent. Any type of individual household latrine showed a similar jump, from twenty-four percent in control villages to forty-four percent in intervention villages. But latrine construction and latrine use are not the same thing. Open defecation among adults fell by only ten percentage points. Seventy-three percent of adults in intervention villages were still defecating in the open, compared to eighty-four percent in control villages. Adult men in intervention villages reported daily open defecation at seventy-five percent versus eighty-four percent in controls. Women: seventy-three percent versus eighty-three percent. Even among the six hundred thirty intervention households that had built latrines meeting the Joint Monitoring Programme definition, forty-one percent still reported adult daily open defecation. Field teams heard the same explanations repeatedly: habit, cultural preference, lack of water at the latrine. There was also contamination in the control arm. Program activities reached at least ten of the forty control villages, and in those contaminated villages, Joint Monitoring Programme-defined sanitation coverage rose from seventeen percent at baseline to forty-one percent by follow-up — nearly identical to the intervention effect. That compression of the difference between arms matters for interpreting what came next. Because what came next was a wall of null results. Despite the gains in coverage and the modest reductions in open defecation, the intervention produced no statistically significant improvements in any of the child health outcomes the trial measured. Diarrhea prevalence at follow-up was seven point four percent in intervention villages versus seven point seven percent in control villages — a difference indistinguishable from chance, with a p-value of zero point sixty-nine. Highly credible gastrointestinal illness, a stricter outcome requiring more specific symptom criteria, ran at eleven point five percent versus twelve point zero percent. No difference. The stool subsample showed helminth prevalence at five point seven percent overall, with no significant difference between groups. Anemia showed no significant improvement. Weight-for-age and height-for-age Z-scores — the anthropometric measures that would reflect reduced growth faltering over time — showed no statistically significant gains in the intervention arm either. Let that land for a moment. This was not a small underpowered pilot. It was eighty villages, more than five thousand children, twenty-one months of follow-up, and a program backed by government infrastructure and World Bank support. The trial found nothing on the health outcomes it was designed to detect. Patil and colleagues are systematic about explaining where the causal chain broke. First, coverage was real but limited: intervention villages averaged forty percent household latrine coverage, ranging from five to seventy-nine percent across individual villages. That variation matters. Second, behavior change lagged hardware — even households that had latrines didn't always use them. The intervention raised child feces disposal to twenty-seven percent, up from eighteen percent in controls. That still means nearly three-quarters of households were not safely disposing of child feces. Third, the contamination in control villages compressed the measurable contrast between arms, making it harder to detect any health signal even if one existed. This brings Patil and colleagues to the concept they call a coverage threshold. Fecal-oral transmission may only be disrupted once a sufficiently high fraction of an entire community stops defecating in the open. At forty percent average latrine coverage and seventy-three percent residual open defecation in intervention villages, the program had not reached that threshold. The environment was still heavily contaminated. Children were still exposed. The study acknowledges two further limitations that likely matter. Twenty-one months may simply be too short to see effects on anemia and growth, which accumulate over years of repeated infection rather than a single episode. And the below-poverty-line subgroup analysis offers a hint of what more intensive implementation might achieve. In those households, Joint Monitoring Programme-defined coverage rose by thirty percentage points and open defecation among women fell by seventeen percentage points — larger shifts than in the full sample. The program reached further where subsidies were concentrated. The honest policy conclusion Patil and colleagues draw is that building latrines is necessary but not sufficient. The TSC model as implemented in these districts of Madhya Pradesh raised infrastructure, nudged behavior, and spent real money — and still produced no detectable improvement in child diarrhea, parasitic infections, anemia, or growth at twenty-one months. The implication is not that sanitation programs are hopeless. It is that achieving the health gains the causal model predicts may require reaching a community-wide level of consistent latrine use that this implementation did not attain. Half a village defecating in latrines may not be enough. The pathogens adapt to wherever the other half still goes. For the policymakers, program designers, and development economists who had been scaling sanitation initiatives on the assumption that coverage translates to health outcomes, this trial was a significant finding. It didn't close the question. But it forced an honest reckoning with the gap between building something and changing the behavior that makes it work. And it demanded that future programs measure not just how many latrines were constructed, but how many were actually used, by how many people, and whether that fraction crossed whatever threshold it takes to finally clean up the environment a child grows up in. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.