Objectives: The objectives of this study were to compare the sensitivity of colposcopically directed biopsy (biopsy of cervical quadrants with colposcopic impressions of human papillomavirus, cervical intraepithelial neoplasia [CIN], or cancer) for CIN 3 or cancer (CIN 3+) among physicians and to estimate the increase in yield of CIN 3+ per colposcopy associated with "random" biopsies and/or endocervical curettage (ECC).
Methods: Two studies in which 7 physicians performed 1,383 colposcopic examinations on women with abnormal cervical cytology were reviewed. At colposcopy, the cervix was divided into quadrants by lines from the 12- to the 6-o'clock and from the 3- to the 9-o'clock positions. Each quadrant was assigned a colposcopic impression of normal, human papillomavirus or CIN 1, CIN 2 or CIN 3, or cancer. Each quadrant had biopsy of colposcopically detected lesions or "random" biopsy at the squamocolumnar junction if the colposcopic impression was normal. Lastly, ECC was obtained.
Results: Among the physicians, the sensitivity of colposcopically directed biopsy for CIN 3+ varied from 28.6% to 92.9% (p < .001). With logistic regression, the sensitivity of colposcopically directed biopsy for CIN 3+ increased as the number of cervical quadrants with CIN 3+ increased (p < .001); once corrected for the number of quadrants with CIN 3+, differences between the physician's sensitivities remained significant (p = .01). For 6 of 7 physicians, the yield of CIN 3+ per colposcopy was greater when colposcopically directed biopsy was augmented by up to 4 "random" biopsies plus ECC (p = .03 to p < .001).
Conclusions: The sensitivity of colposcopy for CIN 3+ varies widely. Performing up to 4 "random" biopsies plus ECC increases the yield of CIN 3+.