Background: Optimum management of childhood diarrhoea in low-resource settings has been hampered by insufficient data on aetiology, burden, and associated clinical characteristics. We used quantitative diagnostic methods to reassess and refine estimates of diarrhoea aetiology from the Etiology, Risk Factors, and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development (MAL-ED) cohort study. Methods: We re-analysed stool specimens from the multisite MAL-ED cohort study of children aged 0–2 years done at eight locations (Dhaka, Bangladesh; Vellore, India; Bhaktapur, Nepal; Naushero Feroze, Pakistan; Venda, South Africa; Haydom, Tanzania; Fortaleza, Brazil; and Loreto, Peru), which included active surveillance for diarrhoea and routine non-diarrhoeal stool collection. We used quantitative PCR to test for 29 enteropathogens, calculated population-level pathogen-specific attributable burdens, derived stringent quantitative cutoffs to identify aetiology for individual episodes, and created aetiology prediction scores using clinical characteristics. Findings: We analysed 6625 diarrhoeal and 30 968 non-diarrhoeal surveillance stools from 1715 children. Overall, 64·9% of diarrhoea episodes (95% CI 62·6–71·2) could be attributed to an aetiology by quantitative PCR compared with 32·8% (30·8–38·7) using the original study microbiology. Viral diarrhoea (36·4% of overall incidence, 95% CI 33·6–39·5) was more common than bacterial (25·0%, 23·4–28·4) and parasitic diarrhoea (3·5%, 3·0–5·2). Ten pathogens accounted for 95·7% of attributable diarrhoea: Shigella (26·1 attributable episodes per 100 child-years, 95% CI 23·8–29·9), sapovirus (22·8, 18·9–27·5), rotavirus (20·7, 18·8–23·0), adenovirus 40/41 (19·0, 16·8–23·0), enterotoxigenic Escherichia coli (18·8, 16·5–23·8), norovirus (15·4, 13·5–20·1), astrovirus (15·0, 12·0–19·5), Campylobacter jejuni or C coli (12·1, 8·5–17·2), Cryptosporidium (5·8, 4·3–8·3), and typical enteropathogenic E coli (5·4, 2·8–9·3). 86·2% of the attributable incidence for Shigella was non-dysenteric. A prediction score for shigellosis was more accurate (sensitivity 50·4% [95% CI 46·7–54·1], specificity 84·0% [83·0–84·9]) than current guidelines, which recommend treatment only of bloody diarrhoea to cover Shigella (sensitivity 14·5% [95% CI 12·1–17·3], specificity 96·5% [96·0–97·0]). Interpretation: Quantitative molecular diagnostics improved estimates of pathogen-specific burdens of childhood diarrhoea in the community setting. Viral causes predominated, including a substantial burden of sapovirus; however, Shigella had the highest overall burden with a high incidence in the second year of life. These data could improve the management of diarrhoea in these low-resource settings. Funding: Bill & Melinda Gates Foundation. © 2018 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license
