HIV and severe wasting are associated with post-discharge mortality and hospital readmission among children with complicated severe acute malnutrition (SAM); however, the reasons remain unclear. We assessed body composition at hospital discharge, stratified by HIV and oedema status, in a cohort of children with complicated SAM in three hospitals in Zambia and Zimbabwe. We measured skinfold thicknesses and bioelectrical impedance analysis (BIA) to investigate whether fat and lean mass were independent predictors of time to death or readmission. Cox proportional hazards models were used to estimate the association between death/readmission and discharge body composition. Mixed effects models were fitted to compare longitudinal changes in body composition over 1 year. At discharge, 284 and 546 children had complete BIA and skinfold measurements, respectively. Low discharge lean and peripheral fat mass were independently associated with death/hospital readmission. Each unit Z-score increase in impedance index and triceps skinfolds was associated with 48 % (adjusted hazard ratio 0·52, 95 % CI (0·30, 0·90)) and 17 % (adjusted hazard ratio 0·83, 95 % CI (0·71, 0·96)) lower hazard of death/readmission, respectively. HIV-positive v. HIV-negative children had lower gains in sum of skinfolds (mean difference -1·49, 95 % CI (-2·01, -0·97)) and impedance index Z-scores (-0·13, 95 % CI (-0·24, -0·01)) over 52 weeks. Children with non-oedematous v. oedematous SAM had lower mean changes in the sum of skinfolds (-1·47, 95 % CI (-1·97, -0·97)) and impedance index Z-scores (-0·23, 95 % CI (-0·36, -0·09)). Risk stratification to identify children at risk for mortality or readmission, and interventions to increase lean and peripheral fat mass, should be considered in the post-discharge care of these children.
Keywords: Fat mass; HIV; Hospitalisation; Lean mass; Mortality; Severe acute malnutrition.