Colposcopic proficiency-disease spectrum in a single family practice colposcopists' clinic.

PURPOSE We sought to assess colposcopic proficiency in a family practice teaching clinic. METHODS Subjects were a prospective cohort of women age 13 to 68 who were colposcopy clinic attendees from 1991 to 2002. Data recorded on each subject included demographic variables, sexual history, history of sexually transmitted diseases, reason for referral to colposcopy, Pap smear results, colposcopic impression, colposcopic biopsy results, and diagnoses. The Kappa statistic was used to measure agreement between clinical colposcopic assessment and biopsy results. RESULTS Eight hundred twenty-six patients were enrolled. Compared to biopsy, colposcopic impression overall correctly predicted normal cervical biopsy in 55.8% (95% CI: 45.8%, 65.8%) of cases, and predicted abnormal biopsy 84.9% (95% CI: 81.6%, 88.1%) of the time. Colposcopic impression of low-grade squamous intraepithelial lesion (LSIL) correctly predicted LSIL on biopsy in 64.6% of cases, and correctly predicted the absence of LSIL 74.2% of the time. Colposcopic impression of high-grade squamous intraepithelial lesion (HSIL) correctly predicted biopsy results of HSIL in 70.05% of cases (Kappa = 0.544, P<.0001). There was a 12.7% error rate in discriminating normal from LSIL (Kappa -.258, with P<.0001). CONCLUSION Family physicians perform colposcopy with good correlation between colposcopic impression and subsequent histology.