Optimal anthropometric discharge criteria from treatment of wasting: meta-analysis of individual patient data from 34 studies

Background: Community-based treatment of acute malnutrition saves lives, but recovered children remain at risk of relapse postdischarge. Strategies to reduce this risk may include modification of anthropometric discharge criteria. Objectives: This study aims to compare the diagnostic accuracy of anthropometric indices to reduce postdischarge relapse risk. Methods: We searched PubMed from inception to June 2022. We included studies that enrolled children aged 0–59 mo successfully treated for severe or moderate acute malnutrition (SAM or MAM), assessed anthropometry at discharge, and had ≥1 follow-up assessment ≤6 mo postdischarge. Pooled sensitivity and specificity for anthropometric indices at discharge over multiple cutoffs were calculated using a bivariate mixed-effects model. Area under the pooled receiver operating curve (AUC) was estimated to measure diagnostic accuracy. “Pragmatic” cutoffs were defined as those maximizing AUC given both pooled sensitivity and pooled specificity ≥0.75. Primary outcomes were SAM relapse (SAM episode after successful SAM treatment: weight-for-height Z-score (WHZ) < −3, mid-upper arm circumference (MUAC) < 11.5 cm and/or edema) and MAM relapse (MAM episode after successful MAM treatment: −3 ≤ WHZ < −2 or 11.5 cm ≤ MUAC < 12.5 cm). Exposures were WHZ, MUAC, and weight-for-age Z-score (WAZ) at discharge. Results: We included 34 studies from 16 countries contributing 21,989 children. WHZ at discharge had a higher AUC in predicting lower SAM and MAM relapse risk than MUAC or WAZ at discharge. None of the cutoffs examined met the study definition of “pragmatic.” The closest “pragmatic” cutoffs suggested that WHZ cutoffs of −1.4 and −1.8 or MUAC of 12.6 and 12.7 cm had the highest sensitivity and specificity in predicting lower SAM and MAM relapse risk. Conclusions: Relapse risk is high after successful MAM/SAM treatment. Future research can consider optimization of anthropometric discharge criteria as a strategy to reduce postdischarge relapse risk, weighing the operational and financial tradeoffs associated with any modification. This trial was registered at PROSPERO as CRD42022342009.