Digital Stool Colour Card for Biliary Atresia Screening in North India: A Prospective Feasibility and Reliability Study.

BACKGROUND/AIMS: The aim of this study was to evaluate the feasibility, uptake, and inter-rater reliability of a smartphone-delivered digital stool colour card (dSCC) for community-based biliary atresia (BA) screening in North India, specifically assessing agreement between caregiver interpretation and healthcare provider assessment. METHODS: In this prospective observational study, consecutive newborns delivered at a tertiary centre were enrolled if caregivers had smartphone access. Participants received a dSCC via a messaging app with standardized counselling. Caregivers reported stool colour at 6- and 10-week immunization visits. These reports were compared against blinded assessments by two senior pediatric residents and three paramedics. The primary outcome was the weighted kappa (κ) for dSCC interpretation reliability. Secondary outcomes included enrolment, response, and utilization rates. RESULTS: Of 8347 newborns, 7529 (90.2%) were enrolled. The dSCC response rate was 85.8% (6459/7529) at 6 weeks and 65.0% (4896/7529) at 10 weeks. Weighted kappa (quadratic) for caregiver-doctor agreement improved from 0.28 (95% confidence interval [CI]: 0.25-0.31) at 6 weeks to 0.74 (95% CI: 0.71-0.77) at 10 weeks. Caregiver-paramedic agreement improved from 0.29 (95% CI: 0.26-0.32) to 0.69 (95% CI: 0.66-0.72). A longer total counselling duration (≥25 min) was significantly associated with improved agreement (Spearman's ρ = 0.41, P < 0.01). One BA case was identified by providers (though initially misreported by the family), leading to timely Kasai portoenterostomy at 59 days. Conservative and optimistic dSCC utilization rates were 90.6% and 95.9%, respectively. CONCLUSIONS: Integrating dSCC screening with routine immunization is a feasible and reliable strategy for engaging caregivers in BA screening in low-resource settings, with substantial caregiver-provider agreement after adequate counselling (weighted κ = 0.74). This remains a feasibility study, and further studies to evaluate its diagnostic accuracy are required.