Predictive Model for the Risk of Severe Acute Malnutrition in Children

BACKGROUND: The nutritional status is the best indicator of the well-being of the child. Inadequate feeding practices are the main factors that affect physical growth and mental development. The aim of this study was to develop a predictive score of severe acute malnutrition (SAM) in children under 5 years of age. METHODS: It was a case-control study. The case group (n = 263) consisted of children aged 6 to 59 months admitted to hospital for SAM that was defined by a z-score weight/height < −3 SD or presence of edema of malnutrition. We performed a univariate and multivariate analysis. Discrimination score was assessed using the ROC curve and the calibration of the score by Hosmer–Lemeshow test. RESULTS: Low birth weight, history of recurrent or chronic diarrhea, daily meal's number less than 3, age of breastfeeding's cessation less than 6 months, age of introduction of complementary diets less than 6 months, maternal age below 25 years, parity less than 5, family history of malnutrition, and number of children under 5 over 2 were predictive factors of SAM. Presence of these nine criteria affects a certain number of points; a score <6 points defines children at low risk of SAM, a score between 6 and 8 points defines a moderate risk of SAM, and a score >8 points presents a high risk of SAM. The area under ROC curve of this score was 0.9685, its sensitivity was 93.5%, and its specificity was 93.1%. CONCLUSION: We propose a simple and efficient prediction model for the risk of occurrence of SAM in children under 5 years of age in developing countries. This predictive model of SAM would be a useful and simple clinical tool to identify people at risk, limit high rates of malnutrition, and reduce disease and child mortality registered in developing countries.