Dietary staple type, protein quality and child wasting and stunting across 127 LMICs — an ecological cross-sectional study

Abstract Background Child stunting and wasting affect millions globally, with unexplained heterogeneity across LMICs. Staple foods — rice, wheat, maize, sorghum, millet and cassava — differ in protein quality (DIAAS) and zinc bioavailability. We hypothesised that these differences explain heterogeneity in undernutrition prevalence. Methods This ecological cross-sectional analysis included 127 LMICs (wasting) and 126 LMICs (stunting) using JME 2025 anthropometric data, GBD 2023 estimates and FAOSTAT 2019–2023 five-year mean food balances. Each staple was analysed individually using robust MM-estimation (primary) and quantile regression (secondary), with fixed covariates selected after circularity, correlation and multicollinearity assessment. Five sensitivity analyses tested robustness. Two secondary cross-reference analyses — WHZ, MUAC and oedema across 46 nationally representative surveys and undernutrition in infants under 6 months across 56 DHS datasets — were cross-referenced with dominant dietary staple supply. A post-hoc cross-reference analysis examined maternal short stature and low BMI by dominant dietary staple using DHS data. Findings Sorghum (β=0·361, p<0·001) and rice (β=0·086, p=0·004) were positively associated with wasting while Maize was negatively associated (β=−0·139, p=0·004). Cassava showed a positive, borderline association with stunting (β=0·975, p=0·092). HIV showed a negative wasting association reversing on exclusion of high-HIV countries. Secondary analyses confirmed MUAC-predominant wasting in cassava- and maize-dominant countries and WHZ-predominant wasting in sorghum- and millet-dominant countries. Interpretation The above findings are consistent with staple-specific growth phenotypes. Currently DHS and MICS surveys collect height and weight only — incorporating MUAC and individual staple dietary data appears imperative. Preventive measures may differ by staple context. Funding No funding was received. Research in context Evidence before this study We searched PubMed using the terms “dietary staple”, “protein quality”, “DIAAS”, “child wasting”, “child stunting”, “ecological”, “epidemiology” and combinations thereof, with no language or date restrictions. Frongillo et al (1997) examined country-level determinants of stunting and wasting using energy availability as the only dietary variable, identifying unexplained residuals for India and Bangladesh. Martorell and Young (2012) documented higher wasting in India than Guatemala despite similar stunting, pointing to maternal nutritional factors. A research prioritisation exercise involving 146 global nutrition experts (Frison et al, 2020) identified ten priority areas for wasting prevention; dietary protein quality featured in none of them. Garenne(2009) identified two populations in Niakhar(Senegal) and Bwamanda(DRC) consuming staples Sorghum/millet and Cassava/maize respectively having vastly different wasting profiles. Specifically they had higher Mid-upper-arm-circumference(MUAC) based wasting in Bwamanda and lower Weight for height Z(WHZ) scores in Niakhar with differing age profiles of malnourished children. No prior ecological study has examined whether the specific type of dietary staple — and its protein quality and zinc bioavailability profile — is associated with child wasting or stunting prevalence and differing anthropometric profiles across a representative sample of LMICs. Added value of this study This is the first ecological analysis to examine the independent associations of six individual dietary staples — wheat, rice, maize, millet, sorghum and cassava — with child wasting and stunting prevalence across 127 LMICs (126 for stunting), using robust MM-estimation, quantile regression and five pre-specified sensitivity analyses. Sorghum and rice protein supply independently predicted higher wasting; maize independently predicted lower wasting. Cassava showed a positive, borderline association with stunting (β=0·975, p=0·092). These findings are consistent with each staple driving a distinct growth phenotype determined by its protein quality (DIAAS) and zinc bioavailability profile, explaining the long-documented geographic dissociation between wasting and stunting that conventional socioeconomic and disease models have failed to account for. We have additionally integrated two large datasets of child wasting by MUAC and WHZ as well as wasting and stunting (WaSt) prevalence across countries, with our dataset, to demonstrate these differences between staples. Cross-referencing the Grellety-Golden dataset (46 surveys) with dominant dietary staple showed MUAC-predominant wasting in cassava- and maize-dominant countries and WHZ-predominant wasting in sorghum- and millet-dominant countries (Kruskal-Wallis p=0·005). Cross-referencing the Kerac et al dataset (56 DHS surveys) showed that WaSt prevalence in infants under 6 months was more than double in sorghum- and millet-dominant countries compared with cassava- and maize-dominant countries, suggesting these staple-specific growth phenotypes are established from the earliest weeks of life. We also undertook a post-hoc analysis cross-referencing maternal short stature and low BMI with staple type, establishing a statistically significant negative relationship of maize with maternal low BMI, corroborating our primary robust MM-regression finding of a negative association of maize with child wasting. Implications for all the available evidence These findings have direct implications for global nutritional surveillance — DHS and MICS surveys aggregate all cereals and all roots and tubers into single food groups, precluding individual-level testing of the staple-specific protein quality hypothesis documented here. Reclassification of dietary assessment modules to capture individual staple types and preparation methods is a research priority. The use of WHZ as the sole wasting metric introduces systematic bias — height in the denominator may overestimate wasting prevalence in sorghum- and millet-dominant Sahelian populations, while underestimating it in maize- and cassava-dominant populations where linear growth deficits artificially elevate WHZ. Incorporating MUAC alongside WHZ in surveillance and programme assessment is essential to capture the full burden of wasting across differing staple-linked growth phenotypes. For policy, targeted pulse or micronutrient supplementation in cereal or tuber dominant populations could help address undernutrition.