LAUSR

OBJECTIVE The aim of the study was to assess whether lower proportions of cervical intraepithelial neoplasia 2+ diagnosed by random biopsy are from lower cut points defining an abnormal colposcopic impression or obtaining only one random biopsy when colposcopic impression is normal. METHODS We compared colposcopy experiences within Shanxi Province Cervical Cancer Screening Study (SPOCCS) (n = 1,383) and Shenzhen Cervical Cancer Screening Trial (SHENCCAST) (n = 631), which had instructive differences in the cut point defining an abnormal colposcopic impression. RESULTS The proportion of CIN 2+ diagnosed by random biopsy in SPOCCS (35.0%, 141/403) was higher than SHENCCAST (18.4%, 35/190, p < .001). Quadrant-specific receiver operating characteristic curves for diagnosis of CIN 2+ by colposcopic impression in SPOCCS and SHENCCAST were similar; a lower cut point for an abnormal colposcopic impression in SHENCCAST resulted in lower proportion of CIN 2+ diagnosed by random biopsy. Normal colposcopic impression was found in 85.1% (120/141) of cases of CIN 2+ diagnosed by random biopsy in SPOCCS and in 57.1% (20/35) of such cases in SHENCCAST. Of CIN 2+ diagnosed by random cervical biopsy with normal colposcopic impression, one cervical quadrant was involved with CIN 2+ in 66.7% (80/120) of colposcopies in SPOCCS and 80% (16/20) of colposcopies in SHENCCAST. CONCLUSIONS Colposcopy series with higher proportions of CIN 2+ diagnosed by random biopsy likely have more stringent cut points defining an abnormal colposcopic impression and have more random biopsies when the colposcopic impression is normal. At colposcopy, we advise multiple biopsies of all acetowhite epithelium or multiple random biopsies to increase the sensitivity of colposcopy.