Objective: The aim of this study was to assess the agreement between a web-based scoliosis screening tool and a standard screening procedure.
Methods: Sixty participants were selected (median age, 12 years; 75% were women) and separated into 2 groups: those with unknown spinal curvature status and those with confirmed scoliosis. Each participant was assessed by 2 blinded assessors, with one measuring the angle of trunk rotation using a scoliometer and the second using a web-based screening application. The app provided a relative risk score for having scoliosis based on a weighted algorithm. Those with an angle of trunk rotation ≥7° or risk score >2 were deemed as being at risk for having scoliosis.
Results: There was fair agreement (kappa = 0.34; 95% confidence interval [CI], 0.14-0.55; P < .001) between the app and the scoliometer among the unconfirmed cases. The McNemar test indicated a difference in the proportion of positive tests (P = .001), whereby the screening app produced a significantly higher number of positive tests (15/53 = 28.3%) compared to the standard screening procedure (4/53 = 7.5%) for unconfirmed cases. Among the confirmed cases, the app correctly identified 5 out of 7 (sensitivity: 71%; 95% CI, 29%-96%) participants, whereas the scoliometer correctly identified 6 out of 7 (sensitivity: 86%; 95% CI, 42%-100%) participants.
Conclusion: These findings indicate fair agreement between the app and the scoliometer, though it was not possible to precisely estimate the sensitivity of the app in this study.
Keywords: Adolescent; Diagnostic Techniques and Procedures; Reproducibility of Results; Scoliosis.
Copyright © 2022. Published by Elsevier Inc.