Boys are more stunted than girls in Sub-Saharan Africaa meta-analysis of 16 demographic and health surveys

If you want a single number that wraps a child's early life into something you can track, it's height for age. Stunting — being more than two standard deviations below the World Health Organization reference for height at a given age — is a quiet ledger of what's happened to a child's body. Not just food, but infection, care, play, and the safety of the home. And there's a puzzle people have noticed across low-income settings: boys seem to struggle more than girls. The question is, does that pattern really hold when you look across countries with comparable data, and can household circumstances blunt or widen that gap? Wamani and colleagues took that on by going big and going comparable. They drew on sixteen Demographic and Health Surveys—nationally representative snapshots—from ten countries across sub-Saharan Africa, all between the mid-1990s and early 2000s. Think tens of thousands of kids under five measured the same way, using the same growth standard and asked the same questions about the home. About sixty-four thousand children made it into the analysis. Stunting was defined in the standard way, a height-for-age z-score below minus two. Roughly a quarter of the original anthropometry was missing, mostly because age or measurements weren't recorded, so the estimates rest on complete cases. To make socioeconomic status more than a hand-wave, they split it into two pieces that matter for children in distinct ways. One is material living standard — the things a household owns and the quality of the dwelling — rolled up into an asset index. That index wasn't guessed; they used principal components analysis, which finds the combination of items that best separates households into a gradient, then sorted families into five groups from poorest to least poor. In two surveys where the items were too tightly correlated, they used four groups instead. The other piece is mothers' education, recoded into no schooling, primary, and secondary or higher. Education and assets don't move in lockstep, and that's the point — they capture different routes by which advantage or disadvantage gets under the skin. The statistical playbook was straightforward and transparent. Within each country, they compared boys and girls on average height-for-age and on the share who were stunted, using standard tests at the five percent level. Then they modeled the odds of being stunted in a logistic regression, controlling for the child's age and for which survey the child came from, and pooled results across studies in a fixed-effects framework. They checked how similar the studies were using Cochran's Q, and they tested an interaction — a product term — between sex and each socioeconomic measure to see if the sex gap widened or narrowed in richer versus poorer homes or by mothers' schooling. Analyses were run in the Statistical Package for the Social Sciences and Stata, but the important thing is the comparisons were apples to apples. Here's the headline in plain language: across these countries, boys were more likely to be short for their age than girls. Put numbers to it, and the picture sharpens. On average, boys had a height-for-age score of minus 1.59, compared with minus 1.46 for girls. That difference was highly significant. In prevalence terms, forty percent of boys were stunted versus thirty-six percent of girls. That gap translates to a crude odds ratio of 1.16 — boys had sixteen percent higher odds of being stunted — and after adjusting for age and study, the odds ticked up slightly to 1.18. Not a trivial edge case. A region-wide signal. How steady is that signal when you zoom in country by country? Pretty steady, though not perfectly uniform. In three quarters of the surveys, the average score difference between boys and girls was statistically significant. In just under three quarters, the prevalence gap cleared the significance bar. What that means is we don't have one or two outlier countries pulling the average; most places point in the same direction. But there is texture underneath, which matters for policy and for understanding mechanisms. Start with the socioeconomic gradient itself. Regardless of sex, stunting climbed as you moved from the least poor to the poorest households, and as mothers' education fell. In several studies, those trends were textbook clear. Zimbabwe's 1999 survey, for example, showed strong step-wise increases in stunting across asset groups in both boys and girls, with a p-value of 0.007 for boys and 0.001 for girls. That's the dose-response pattern you hope to see if the socioeconomic measures are meaningful. Now overlay sex on top of that. In some countries, the male disadvantage was largest in the poorest homes or among children of mothers with no or only primary education. Nigeria's 2003 data are a good illustration: poorer households and lower maternal schooling tended to line up with wider gaps between boys and girls, even if the exact size of the gap jumped around between groups. But when Wamani and colleagues tested that idea formally — does socioeconomic status consistently magnify the boy-girl difference? — the interaction didn't hold up across the whole dataset. The sex-by-status term wasn't statistically significant in the pooled model, and within individual surveys, the pattern was inconsistent. So, two things are true at once. There is a clear socioeconomic gradient in stunting for both boys and girls, and in several settings, the gap between them looks wider at the bottom. Yet, across countries as a whole, you can't count on wealth or maternal schooling to systematically change the size of the sex gap. What about pooling across countries — is that even fair if contexts are so different? The team did the standard checks. One test of between-study homogeneity gave a p-value of 0.15, which they read as the studies being similar enough to justify a pooled estimate. At the same time, the Cochran's Q statistic was 85.5 with a p-value below 0.001, flagging real variability in the size of the gap across surveys. In practice, fixed and random-effects models told the same story: boys faced higher odds of stunting essentially everywhere in this set of countries, even if the exact margin varied. It's worth pausing on what those odds ratios do and don't say. An adjusted odds ratio of 1.18 doesn't mean every boy is destined to be shorter than every girl or that sex alone determines outcomes. It means that, after lining up kids by age and accounting for which survey they're in, the dice are a little more loaded against boys. In public health terms, a four percentage point higher prevalence across a region translates into a lot of individual children. That's the quiet power of a small, consistent effect. Any large, pooled study like this comes with caveats, and the authors don't shy away from them. About a quarter of potential child records were missing height or reliable age and had to be dropped; if that missingness isn't random — say, if sicker children are harder to measure — it could bias the estimates. The asset index and maternal education are proxies, and their meaning shifts across contexts; they aren't interchangeable or exhaustive measures of socioeconomic position. The analysis focused on children zero to fifty-nine months as a group, and while they looked at age bands in some checks, not every age-specific comparison yielded a clear sex difference. Measurement followed the older National Center for Health Statistics and World Health Organization reference; sex-specific standards are built in, but all standards have quirks. And with many tests across many subgroups, there's always a chance a few significant results appear by luck alone. So what do we do with a male disadvantage that shows up this reliably? One temptation is to reach immediately for mechanism. Maybe biology makes boys more vulnerable to infections early in life. Maybe care practices differ in ways that matter for growth. Wamani and colleagues are careful here: they frame the pattern as an equity signal and a prompt to dig deeper, not as a verdict on causes. The next steps they call for are concrete. Unpack the socioeconomic bundle — which components of material living standard or parental education track most closely with the sex gap, and in which countries? Test biological and social pathways head-to-head rather than assuming one or the other. And follow children forward in time to see how early height shortfalls play out in schooling, adult health, and economic outcomes. There's a broader lesson in their approach. When you split socioeconomic status into distinct, interpretable parts and insist on comparable measures across countries, you can see patterns that anecdotes blur. Boys' higher odds of being stunted is one of those. It isn't an iron law, and it isn't destiny. But across ten very different countries, measured the same way, it's there. And it tells us something quietly powerful about how early life risks line up, who's most exposed, and how much room there is to move those odds.