Body Composition prediction equations for infants from Latin America and the Caribbean

Abstract Background Body composition assessment is essential for evaluating nutritional status and early growth, yet access to advanced measurement technologies remains limited in many Latin American and Caribbean (LAC) settings. Developing simple and accurate anthropometric prediction models offers a feasible alternative for estimating fat-free mass (FFM) in infants. Objective To develop and validate equations to predict FFM in infants from LAC countries with diverse socioeconomic and ethnic contexts. Methods A multicentre cross-sectional study was conducted in 559 infants aged 6–24 months from ten LAC countries. Anthropometric measurements and FFM (by deuterium dilution) were collected using standardized protocols. The sample was randomly divided into development and validation groups. Stepwise regression and multilevel modelling were used to develop predictive equations that incorporate individual- and country-level variability. Model performance was assessed using R², mean absolute error (MAE), root mean square error (RMSE), Bland–Altman analyses, and cross-validation. Results Weight, length, and head circumference were the strongest predictors of FFM. Six preliminary models were developed, and multilevel models substantially improved predictive accuracy. Three final models explained approximately 80–81% of the variance in the validation sample. Model 5 showed the best overall balance of accuracy, precision, and feasibility. Conclusions This study provides the first anthropometric equations for estimating FFM in infants aged 6–24 months from LAC countries. The selected model is based on routine anthropometric measurements, which supports its feasibility for primary health care and field research. External validation in other populations is recommended.