Risk factors for fatal diarrhoea among dehydrated malnourished children in a Madagascar hospital

OBJECTIVE: To examine mortality risk factors during rehydration among 6-35 month malnourished children with diarrhoea. DESIGN: Data collected prospectively during a clinical trial comparing two oral rehydration solutions (ORS). SETTING: Paediatric ward. SUBJECTS: Study children had either a weight-for-age Z-score below -2 or a weight-for-height below 70% of NCHS median. All had diarrhoea for < 5 days. 150 were enrolled and two were excluded for intercurrent infection. INTERVENTION: Children were randomly allocated to receiving 100 ml/kg of standard or rice-based ORS during the 6h following admission. Then they received 420 kJ/kg/day of high energy milk, progressively increased to 840 kJ/kg/day. RESULTS: Mortality rate was 16% and with no difference by ORS group. In univariate analysis, the risk of dying (mean odds ratio; 95% confidence interval) was significantly higher among girls (3.5; 1.4-8.9), in non-breast-fed children (3.7; 1.4-9.6) and in children with a low weight-for-height (5.1; 1.9-14.1). Low weight, moderate or severe dehydration, low plasma specific gravity or total plasma protein and longer duration of diarrhoea before inclusion also were significant risk factors. In multivariate logistic analysis, only absence of breast-feeding was associated with a higher risk of dying among girls with a low weight-for-height. Among them, eight out of nine died, compared to 15 out of 139 for other children. CONCLUSION: Breast-feedings protected severely malnourished girls against death from diarrhoea even when dehydration was corrected. Mechanisms underlying this selective effect are poorly understood.