Background: Childhood diarrhoea can be caused by many pathogens that are difficult to assay in the laboratory. Molecular diagnostic techniques provide a uniform method to detect and quantify candidate enteropathogens. We aimed to develop and assess molecular tests for identification of enteropathogens and their association with disease.

Methods: We developed and assessed molecular diagnostic tests for 15 enteropathogens across three platforms—PCR-Luminex, multiplex real-time PCR, and TaqMan array card—at five laboratories worldwide. We judged the analytical and clinical performance of these molecular techniques against comparator methods (bacterial culture, ELISA, and PCR) using 867 diarrhoeal and 619 non-diarrhoeal stool specimens. We also measured molecular quantities of pathogens to predict the association with diarrhoea, by univariate logistic regression analysis.

Findings: The molecular tests showed very good analytical and clinical performance at all five laboratories. Comparator methods had limited sensitivity compared with the molecular techniques (20–85% depending on the target) but good specificity (median 97·3%, IQR 96·5–98·9; mean 95·2%, SD 9·1). Positive samples by comparator methods usually had higher molecular quantities of pathogens than did negative samples, across almost all platforms and for most pathogens (p<0·05). The odds ratio for diarrhoea at a given quantity (measured by quantification cycle, Cq) showed that for most pathogens associated with diarrhoea—including Campylobacter jejuni and Campylobacter coli, Cryptosporidium spp, enteropathogenic Escherichia coli, heat-stable enterotoxigenic E coli, rotavirus, Shigella spp and enteroinvasive E coli, and Vibrio cholerae—the strength of association with diarrhoea increased at higher pathogen loads. For example, Shigella spp at a Cq range of 15–20 had an odds ratio of 8·0 (p<0·0001), but at a Cq range of 25–30 the odds ratio fell to 1·7 (p=0·043).

Interpretation: Molecular diagnostic tests can be implemented successfully and with fidelity across laboratories around the world. In the case of diarrhoea, these techniques can detect pathogens with high sensitivity and ascribe diarrhoeal associations based on quantification, including in mixed infections, providing rich and unprecedented measurements of infectious causes.