Abstract
This retrospective cohort study focused on colposcopic accuracy for the diagnosis of cervical premalignant lesions using cytology and histology, as well as HPV data not included in current cervical screening practices in Kazakhstan. Colposcopy performance was assessed using the modified Reid index in women aged 18-63 years. In total, 1,129 colposcopic-HPV-cytology triple samples and 94 histology findings were collected. The sensitivity of colposcopy was 81.6% with specificity 72.6% for LSIL but fell to 56.6% with specificity 88.3% for CIN2+ vs. 89.6% and 74.5% for cytology at CIN2+, respectively. The ORs for high-grade lesion occurrence within each colposcopy group at viral load rising vs. ORs for HPV-negative women were 3.4; 5.3; and 39.7, respectively (p < 0.0001). Total attributive agreement between the colposcopy and histology findings reached 55.3%, κ 0.47 ± 0.06 vs. 0.62 ± 0.08 for cytology, and 0.34 ± 0.13 and 0.58 ± 0.1, for specialists, respectively. Outcomes obtained for colposcopy alone failed to show satisfactory reliability. Globally adopted primary HPV screening would be the best option despite the related costs.