Comparative Performance of Visual Inspection with Acetic Acid and Colposcopy for Detection of Cervical Precancer in Women with High-Risk Human Papillomavirus Infection: A Cross-Sectional Study
Introduction
The World Health Organization recommends 4 triage strategies for women with high-risk human papillomavirus infection (hrHPV). These include visual inspection with acetic acid (VIA), colposcopy, reflex cytology, and HPV16/18 partial genotyping. However, in many low-resource settings, access to colposcopy remains limited. This study aimed to compare the diagnostic accuracy of visual inspection vs colposcopy for detecting cervical intraepithelial neoplasia grade 2 or worse (CIN2 + ).
Methods
Women who tested positive for hrHPV and were referred for colposcopy, with cytology results available as part of routine clinical care, underwent visual inspection with 3% acetic acid immediately before colposcopy. Colposcopic impressions were recorded, and images were scored using a modified Reid colposcopic index and a modified Swede score without iodine staining. We compared diagnostic performance for CIN2 + across visual inspection, colposcopic impression, modified Reid index (score ≥4), and modified Swede score (score ≥5). Statistical analysis used IBM SPSS Statistics and the Cochran Q test, with significance set at P < .05.
Results
Among 450 women, the median age was 38.0 years. A single hrHPV type was detected in 70.4% of cases; types 16, 52, and 18 were most common. Histopathological confirmation of CIN2 + occurred in 97 women (21.6%). Diagnostic accuracy for predicting CIN2 + was 78.2% with VIA and 77.5% with colposcopic impression. Accuracy was 78.4% for the modified Reid index ≥4 and 78.2% for the modified Swede score ≥5. No significant differences were observed among the 4 methods ( P = .941).
Conclusions
VIA demonstrates diagnostic accuracy comparable to colposcopy-based assessments in hrHPV-positive women evaluated within a cytology-informed clinical pathway, supporting its potential role in resource-limited settings.