Methods: We recruited 800 community-based participants from the Cox's Bazar Displaced Person's Camp in Bangladesh, split evenly between the rainy (July/August 2020) and dry (November/December 2020) periods. Participants were randomized evenly into either a standard survey asking carers if their child under five years old has had diarrhoea in the past fortnight, or a pictorial survey asking carers to pick from a pictorial chart which stools their child under five years old has had in the past fortnight. We collected stools from a random sub-sample of 120. Stools were examined visually, and tested for proteins associated with enteric infection and 16 enteric pathogens. We calculated sensitivities and specificities for each survey type, visual examination, and proteins with respect to enteric pathogen presence.
Findings: The sensitivity of the standard survey for enteric pathogen presence was 0.49[95%CI:0.32,0.66] and the specificity was 0.65[0.41,0.85]. Similar sensitivities and specificities were observed for pictorial survey, visual inspection, and proteins.
Interpretation: While diarrhoea is an important sign in clinical practice it appears that it is a poor proxy for enteric pathogen presence in stool in epidemiological surveys. When enteric infection is of interest, this should be measured directly.
Funding: The project was funded by the National Institutes for Health Research Global Health Research Unit on Improving Health in Slums (16/136/87) and by the University of Warwick.