Characteristics and outcomes of children 2-23 months of age with prolonged diarrhoea: A secondary analysis of data from the 'Antibiotics for Children with Diarrhea' trial

Abstract

Background: Approximately 12% of all diarrhoeal episodes last for 7-13 days. As such, they are termed prolonged diarrhoea, and are associated with over two-thirds of all diarrhoeal deaths. Due to a lack of robust data, we aimed to evaluate a comparative background characteristics of young children with acute and prolonged diarrhoea, and their outcomes at day 90 follow-up.

Methods: We performed a secondary analysis of data from the Antibiotics for Children with Diarrhea (ABCD) trial. Children aged 2-23 months were enrolled between July 2017 and July 2019 from seven Asian and sub-Saharan African countries. For this analysis, we divide diarrhoea into two categories: acute diarrhoea (duration <7 days) and prolonged diarrhoea (duration ≥7-13 days). We used logistic regression to observe baseline crude and adjusted associations and linear regression to compare post-discharge outcomes.

Results: We analysed data on 8266 children, of whom 756 (9%) had prolonged diarrhoea and 7510 (91%) had acute diarrhoea. Pakistan had the highest proportion of children with prolonged diarrhoea (n/N = 178/1132, 16%), while Tanzania had the lowest (n/N = 12/1200, 1%). From an analysis that adjusted for sex, breastfeeding, nutritional status, clinical presentation, housing, water supply, sanitation, and country, we observed that presentation at a health facility with prolonged diarrhoea was associated with low age (2-12 months) (adjusted odds ratio (aOR) = 1.25; 95% confidence interval (CI) = 1.02, 1.53; P = 0.028), presence of three or more under-five children in the family (aOR = 1.54; 95% CI = 1.26, 1.87; P < 0.001), maternal illiteracy (aOR = 1.45; 95% CI = 1.21, 1.74, P < 0.001), moderate underweight (aOR = 1.25; 95% CI = 1.01, 1.55; P = 0.042) and pathogen (Campylobacter) (aOR = 1.27; 95% CI = 1.12, 1.44; P < 0.001). At day 90 follow-up, children with prolonged diarrhoea had significantly lower weight-for-age z-score compared to children with acute diarrhoea (-1.62, standard deviation (SD) = 1.11 vs -1.52, SD = 1.20; P = 0.032), as well as significantly higher frequency of hospital admission (6.1% vs 4.5%; P = 0.042).

Conclusions: Prolonged diarrhoea was more common in children of younger age, those who were moderately underweight, those with Campylobacter in stool, those with three or more under-five children in a family, and those with illiterate mothers compared to those who had acute diarrhoea. Children with prolonged diarrhoea more often required hospitalisation during the three-month follow-up period compared to their counterparts.

Figure 1

Study flowchart.

Figure 2

Site-specific proportion of prolonged diarrhoea.

Figure 3

ROC curve for the final logistic regression model. AUC – area under the receiver operating characteristic curve, ROC – receiver operating characteristic curve.