Daron G Ferris1, Mark S Litaker. 1. Departments of Family Medicine and Obstetrics and Gynecology, 1423 Harper St, HH-100, Medical College of Georgia, Augusta, GA 30912, USA. dferris@mail.mcg.edu
Abstract
OBJECTIVE: This study was undertaken to estimate the accuracy of colposcopy quality control review of subjects' digitized colposcopic images during the atypical squamous cells of undetermined significance/low-grade squamous intraepithelial lesion triage study (ALTS). STUDY DESIGN: After colposcopic examination and the acquisition of 2 digitized cervical images, colposcopists recorded colposcopic diagnoses in a standardized computer database. The images were transferred by modem to colposcopy quality control reviewers for a blinded interpretation, including assessment of image quality. Reviewers' and colposcopists' diagnoses were compared with histologic diagnoses using standard contingency table methods. RESULTS: Colposcopists and reviewers underdiagnosed 16.1% and 25.3% of subjects, and overdiagnosed 44.7% and 19.8% of subjects compared with histology, respectively. The sensitivity and specificity for detecting cervical intraepithelial neoplasia 2 or greater were 35.4% versus 23.2%, P < .0001, and 89.9% versus 95.0%, P < .0001, respectively, for colposcopists and reviewers. CONCLUSION: Colposcopy quality control by review of digitized colposcopic images in clinical trials warrants further evaluation if the accuracy can be improved.
OBJECTIVE: This study was undertaken to estimate the accuracy of colposcopy quality control review of subjects' digitized colposcopic images during the atypical squamous cells of undetermined significance/low-grade squamous intraepithelial lesion triage study (ALTS). STUDY DESIGN: After colposcopic examination and the acquisition of 2 digitized cervical images, colposcopists recorded colposcopic diagnoses in a standardized computer database. The images were transferred by modem to colposcopy quality control reviewers for a blinded interpretation, including assessment of image quality. Reviewers' and colposcopists' diagnoses were compared with histologic diagnoses using standard contingency table methods. RESULTS: Colposcopists and reviewers underdiagnosed 16.1% and 25.3% of subjects, and overdiagnosed 44.7% and 19.8% of subjects compared with histology, respectively. The sensitivity and specificity for detecting cervical intraepithelial neoplasia 2 or greater were 35.4% versus 23.2%, P < .0001, and 89.9% versus 95.0%, P < .0001, respectively, for colposcopists and reviewers. CONCLUSION: Colposcopy quality control by review of digitized colposcopic images in clinical trials warrants further evaluation if the accuracy can be improved.