Efficacy of a remote screening model for oral potentially malignant disorders using a free messaging application: A diagnostic test for accuracy study

Authors:
Publisher website
Google Scholar
Abstract
Objective: To assess the feasibility of using a remote sensing model as a free messaging application tool in the preventive screening of oral potentially malignant disorders in a rural area of India.Design: An observational cross‐sectional study.Setting: Primary care setting in Udupi District, Karnataka, South India.Participants: One‐hundred and thirty‐one individuals with a mean (SD) age of 37.34 (11.31) years, of whom 64.1% and 35.9% were men and women, respectively.Interventions: Clinical oral examination followed by photo capture of five areas of the patients' mouth.Main outcome measures: Reliability measures for the use of a photo messaging service in diagnosing oral potentially malignant disorders, as compared to the clinical examination.Results: When lesions were categorised as normal and abnormal, the reliability (kappa) between the diagnoses, based on photo messaging and clinical oral examination, was 0.68 and 0.67 for Examiners 1 and 2, respectively. The sensitivity values for Examiners 1 and 2 were 98.5% and 99.04%, respectively, whereas the specificity was 72% and 64%, respectively. When the agreement between photo messaging and clinical oral examination for an exact diagnostic match was assessed, the reliability (kappa) was 0.59 and 0.55 for Examiners 1 and 2, respectively. The sensitivity values for Examiners 1 and 2 were 98.1% and 98.7%, respectively, whereas the specificity was 64% and 52% respectively.Conclusion: There was a substantial agreement between the diagnosis based on clinical examination and WhatsApp image for both the examiners, when the lesions were dichotomised as normal and abnormal, but slightly reduced when assessed for the exact diagnostic match. Screening for oral potentially malignant disorders using photo messaging can serve as an effective adjunct and a potential cost‐effective tool in a low‐resource setting.

This publication has 22 references indexed in Scilit: