Determinants of post-discharge stunting among diarrhoeal children aged 2-23 months in Bangladesh: findings from Antibiotics for Children with Severe Diarrhea (ABCD) trial.

Background: Despite the global burden of stunting, data regarding the determinants of stunting observed on follow-up after discharge from health care facilities among children treated for severe diarrhoea are limited. We investigated factors influencing stunting during post-discharge follow-up among under-2 children treated for severe diarrhoea. We developed a predictive model to estimate stunting prevalence and identify risk factors without post-discharge anthropometry. Methods: We analysed data from two sites in Bangladesh participating in the multi-country, double-blind, randomised clinical trial on Antibiotics for Children with Severe Diarrhoea among children aged 2-23 months, from 2017 to 2019. Severe diarrhoea was defined as a child who had acute diarrhoea with severe/some dehydration, or moderate wasting, or severe stunting on admission. Multiple linear regression was constructed to predict the independent factors associated with stunting among the enrolled children who were followed for 180 days after hospital discharge. We developed a predictive model for stunting using 75% of the available data for training, with the remaining 25% reserved as a test set to evaluate model performance. We then applied this predictive model to the children not enrolled in the trial but admitted to the hospital for diarrhoea to estimate the predictive prevalence and determinants of post-discharge stunting at 180 days. Results: Of the 1431 enrolled children, 589 (41.2%) were stunted at enrolment. By day 180, stunting increased to 698 (49.3%). Linear growth of diarrhoeal children over this follow-up period was positively associated with enrolment length-for-age z score (LAZ), maternal body mass index (BMI), maternal education, duration of diarrhoea, and breastfeeding. We applied the predictive model developed on the enrolled patients' data to the 33 341 children admitted for diarrhoea but not enrolled in the trial. The model estimated that 6142 children (18.4%) in this non-follow-up group would have been stunted at 180 days. Persistent stunting at 180 days for these not-followed children was predicted to be positively associated with dehydration on admission and a higher number of children in the household, while a negative association was predicted with formal parental education and increased maternal BMI. Conclusions: Our study highlights the need for targeted interventions to mitigate persistent stunting in this vulnerable group recovering from diarrhoea.