Determinants of stunting in Indonesian childrenevidence from a cross-sectional survey indicate a prominent role for the water, sanitation and hygiene sector in stunting reduction

Nearly one in three children under two in this study was already too short for their age. Not slightly behind — clinically stunted, by World Health Organization standards. In a country where more than a third of all children under five carry that same marker, the question of why is not academic. It is a question about whether millions of kids will reach their cognitive potential, stay in school, and earn a living wage. Height, it turns out, is measuring all of that. Stunting is the clinical sign of chronic undernutrition in early life. It refers to low height-for-age, specifically a length or height-for-age z-score more than two standard deviations below the World Health Organization median. Severe stunting is defined as being more than three standard deviations below. These cutoffs are not arbitrary. Torlesse and colleagues stress that stunting is not a cosmetic growth deficit; it increases the risk of child death, impairs cognitive and motor development, reduces school performance, and raises the risk of noncommunicable diseases in adulthood. In Africa and Asia combined, the economic burden has been estimated at up to eleven percent of gross national product. Indonesia sits squarely inside this burden. Nationally, thirty-seven percent of children under five were stunted in 2013, with eighteen percent severely stunted, and more than fifteen of thirty-three provinces exceeded forty percent prevalence. It was against that backdrop that Torlesse and colleagues set out to ask a more specific question: what is actually driving stunting among the youngest children — those in the first twenty-three months of life, the window when linear growth faltering typically begins and its developmental consequences run deepest? The study used baseline data from the Maternal and Young Child Nutrition Security Initiative survey, collected in late 2011 across three Indonesian districts: Klaten, Sikka, and Jayawijaya. The analytical sample consisted of one thousand three hundred sixty-six children. Households were eligible if they contained a child aged up to thirty-five months and the primary caregiver was available for interview, though the main analysis focused on the zero-to-twenty-three-month group. The team measured length and weight, collected feeding histories, and documented maternal characteristics — including education, antenatal care, and participation in household decisions. They also recorded a suite of water, sanitation, and hygiene indicators using UNICEF and World Health Organization Joint Monitoring Programme definitions for improved water sources, water treatment practices, and improved sanitation. A household wealth index was built from asset and housing variables and split into five quintiles. Statistical analysis utilized multiple logistic regression with generalized estimating equations to account for cluster sampling, reporting adjusted odds ratios — a measure of how much more likely an outcome is given a particular exposure, after holding other variables constant. In that sample, the prevalence of stunting was twenty-eight point four percent. Severe stunting was six point seven percent. Thus, roughly one in three toddlers was stunted, and one in fifteen was severely stunted. Those are the numbers the analysis was built to explain. Here is where the paper delivers something genuinely unexpected. The biggest finding is not that poor sanitation raises the risk of stunting, or that untreated water does; it is that the two interact. The effect of one depends critically on the presence of the other. Torlesse and colleagues found a statistically significant interaction between household sanitary facility and household water treatment, with a p-value for interaction of zero point zero zero seven. That is not a marginal result. Here is what the interaction actually looks like. In households that drank untreated water and used an unimproved latrine — meaning a latrine without a septic tank — the adjusted odds of stunting were three point four seven times higher than in households with improved conditions. The confidence interval ran from one point seven three to seven point two eight, and the p-value was below zero point zero zero one. A clear, large, robust association. However, in households that treated their drinking water, the adjusted odds ratio for using an unimproved latrine dropped to one point two seven, with a confidence interval of zero point nine nine to one point six three, and a p-value of zero point zero six. Statistically, that is not significant. The harm effectively disappears. Let that land for a moment. It means that water treatment, even in the presence of poor sanitation, appears to substantially blunt the risk to a child's linear growth. The authors frame this through fecal-oral contamination pathways — the routes by which bacteria and parasites travel from human waste into a child's gut. Poor sanitation and untreated water together multiply those routes. But when water is treated, one major pathway is closed. Children still face environmental exposure, but the burden is reduced enough that the growth penalty shrinks to statistical insignificance in this sample. Treating water does not fix bad sanitation, but it changes what bad sanitation costs a child's body. The provincial-level data reinforce the finding. The proportion of households with an improved latrine, plotted against provincial stunting rates across Indonesia, shows an inverse correlation with an R-squared of sixty-five point seven percent. Nearly two-thirds of the variation in provincial stunting is explained by latrine access alone. That is a striking ecological association. Water, sanitation, and hygiene — often grouped under the acronym WASH — is not typically where nutrition programs focus their energy. The instinct is usually to go straight for food. This study argues that is incomplete. The other risk factors in the multivariate model fill out the picture without stealing the spotlight. Male sex, older age within the zero-to-twenty-three-month window, and lower household wealth all independently predicted stunting. Boys had one point four five times the adjusted odds of stunting compared to girls. Children aged twelve to twenty-three months had four point four zero times the odds of being stunted compared to those aged zero to five months — a dramatic accumulation of risk across the first two years. And children in the lowest wealth quintile had two point three zero times the odds of stunting relative to the highest quintile. Severe stunting sharpened those gradients considerably. The wealth gradient was especially stark: the lowest quintile carried an adjusted odds ratio of eight point four eight for severe stunting compared to the highest, with a confidence interval of three point one nine to twenty-two point five seven. For severe stunting specifically, two additional factors emerged. Children whose mothers had not attended antenatal care at a health facility had more than twice the adjusted odds of being severely stunted. Additionally, mothers' participation in decisions about what food to cook was associated with higher unadjusted odds of severe stunting — an odds ratio of three point zero nine compared to mothers who did not make those decisions. The authors interpret this carefully; the direction of the association is counterintuitive, and the cross-sectional design makes it difficult to explain. It likely reflects something about the socioeconomic or household contexts in which women have or do not have that decision-making role, rather than food decisions themselves causing harm. Now, what does this evidence actually support in terms of action, and where does it fall short? Torlesse and colleagues clearly argue that programs targeting stunting in Indonesia must integrate water, sanitation, and hygiene interventions alongside nutrition actions. The Sustainable Development Goals create a structural opportunity here: Goal 2 targets nutrition, Goal 6 targets clean water and sanitation, and there is room — and reason — for those programs to work together rather than in parallel silos. The limitations are real and the authors state them plainly. This is a cross-sectional study. You can see association; you cannot establish causation. Household water and sanitation practices were based on maternal recall and may carry reporting bias. Maternal height and body mass index were not collected, leaving out a potentially important pathway. Birth weight data existed for only eight hundred twenty-three of the one thousand three hundred sixty-six children. And the three districts, however informative, are not a nationally representative sample. What the study does accomplish is to give a specific, testable hypothesis a strong evidentiary foundation: that untreated water and unimproved sanitation together are a particularly dangerous combination for early child growth, and that water treatment may partially compensate when sanitation is difficult to improve quickly. Torlesse and colleagues call for operational research to determine which specific water, sanitation, and hygiene interventions — and which combinations with nutrition programs — produce the greatest reductions in stunting in practice. That is the honest close of a careful study. Not a prescription, but a direction. One in three children stunted in the critical first two years of life, and a substantial share of that burden linked not to food alone, but to the water a child drinks and the latrine a family uses. The measurement is height. But what it is measuring is much more than growth. 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.