Background The World Health Organization (WHO) guidelines on cervical cancer screening recommends colposcopy as one of the methods to triage human papillomavirus (HPV)-positive women. We aimed to assess the diagnostic accuracy of colposcopy in this context. Methods In this systematic review and meta-analysis, we searched for articles reporting the diagnostic accuracy of colposcopy in women with a positive HPV screening test, published in PubMed, Embase or the Cochrane library up to Oct 20, 2025. We included cross-sectional or longitudinal studies if the number of true and false positive and negatives could be derived or obtained after contacting the authors. Studies were excluded if the study population was not representative of a screening population. We collected data on colposcopy triage (index) together with data on cytology triage (comparator), if available, and assessed risk of bias using an adapted Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2 checklist. The pooled sensitivity and specificity to detect cervical intraepithelial neoplasia grade 2 or worse (CIN2+) and grade 3 or worse (CIN3+) were calculated, as well as the relative sensitivity and specificity of colposcopy triage compared to cytology triage, using a bivariate logistic random effects model. Heterogeneity was measured using τ 2 and I 2 . Sample size effects were assessed by Deeks' funnel plots and regression test. Certainty of evidence was assessed using the GRADE tool. This study was registered in PROSPERO, CRD42023389772. Findings The literature search identified 5774 records. After screening, we included 11 studies (13,311 participants) in the review, but restricted the primary analysis to 8 studies (10,632 participants) that had complete follow-up after colposcopy triage. The pooled sensitivity for CIN2+ and CIN3+ using colposcopy (low-grade colposcopic impression) were 84.4% (95% CI: 79.0%–88.6%) and 86.9% (95% CI: 81.7–90.8%), respectively. The pooled specificity for < CIN2 was 64.3% (95% CI: 57.4%–70.7%). While heterogeneity between studies was high, with I 2 up to 95.5%, most studies reported a similar or higher sensitivity and lower specificity compared to cytology (at the lowest cut-off): the pooled relative sensitivity for CIN2+ and CIN3+ were 1.74 (95% CI: 1.53–1.97) and 1.58 (95% CI: 1.35–1.85). The relative specificity for < CIN2 was 0.67 (95% CI 0.61–0.72). Certainty of evidence was considered low to very low due to subjectivity in the assessment of colposcopy, cytology and the reference standard, as well as the high inter-study heterogeneity and the risk of reference standard misclassification. Interpretation Studies consistently reported good sensitivity but low specificity of colposcopy to triage women with a positive HPV screening test, showing colposcopy may be an acceptable triage test. While our findings are consistent with current WHO guidelines, low certainty of evidence and risk of over-treatment should be carefully considered. Funding World Health Organization; Horizon 2020 Framework Programme for Research and Innovation of the European Commission the European Joint Action EUCanScreen and the European Commission Initiative on Cervical Cancer.
Dhollander et al. (Mon,) studied this question.