Colposcopy “Under the Loupe”: Diagnostic Accuracy of Colposcopic Examination of CIN2+ Lesions: Systematic Review and Meta-Analysis of 141,355 Cases

Background: Colposcopy is a widely used clinical and diagnostic tool for detecting cervical intraepithelial neoplasia grade 2 or higher (CIN2+). This paper aims to analyze the diagnostic accuracy, pooled sensitivity, specificity, likelihood ratios, and overall diagnostic performance of colposcopy. Methods: A systematic review of papers on diagnostic accuracy, preregistered in PROSPERO (CRD420251028776), was conducted following PRISMA-DTA guidelines. Statistical analyses were performed using SPSS 29 and METADISC 2.0. Study quality and risk of bias were assessed using QUADAS-2. Results: A total of 49 studies, including 141,355 colposcopic examinations, were included in the final analysis. The overall sensitivity was 0.745 (95% CI: 0.684–0.797) and specificity was 0.831 (95% CI: 0.770–0.879). The diagnostic odds ratio was 14.388 (95% CI: 9.673–21.400). The positive likelihood ratio was 4.419 (95% CI: 3.260–5.991). Conversely, the negative likelihood ratio was 0.307 (95% CI: 0.249–0.378). The area under the curve was 0.8569. Significant heterogeneity was observed across studies (I2 > 93.2%). Conclusions: Colposcopy demonstrates moderate-to-high diagnostic accuracy in detecting CIN2+ lesions, with a good balance between sensitivity and specificity. However, significant variability exists among studies, highlighting the need for the implementation of standardized colposcopic criteria, improved examiner training, and adjunctive diagnostic methods such as HPV testing and ancillary techniques to enhance accuracy and reduce false positives and false negatives. Future research should focus on minimizing interobserver variability and refining diagnostic algorithms to optimize colposcopic performance.