Background & aims: Change in hydration is common in children with severe acute malnutrition (SAM) including during treatment, but is difficult to assess. We investigated the utility of bio-electrical impedance vector analysis (BIVA), a quick non-invasive method, for indexing hydration during treatment.

Methods: We studied 350 children 0·5-14 years of age with SAM (mid-upper arm circumference <11·0 cm or weight-for-height <70% of median, and/or nutritional oedema) admitted to a hospital nutrition unit, but excluded medically unstable patients. Weight, height (H), resistance (R), reactance (Xc) and phase angle (PA) were measured and oedema assessed. Similar data were collected from 120 healthy infants and preschool/school children for comparison. Means of height-adjusted vectors (R/H, Xc/H) from SAM children were interpreted using tolerance and confidence ellipses of corresponding parameters from the healthy children.

Results: SAM children with oedema were less wasted than those without (p < 0·001), but had BIVA parameters that differed more from those of healthy children (P < 0·05) than those non-oedematous. Initially, both oedematous and non-oedematous SAM children had mean vectors outside the reference 95% tolerance ellipse. During treatment, mean vectors migrated differently in the two SAM groups, indicating fluid loss in oedematous patients, and tissue accretion in non-oedematous patients. At admission, R/H was lower (oedematous) or higher (non-oedematous) among children who died than those who exited the hospital alive.

Conclusions: BIVA can be used in children with SAM to distinguish tissue-vs. hydration-related weight changes during treatment, and also identify children at high risk of death enabling early clinical interventions.