Colposcopic accuracy of obstetrics and gynecology residents

Objective. The objective of our study was to determine the accuracy of gynecology residents' colposcopic impressions. Methods. A retrospective review of colposcopic examinations was performed. Colposcopic impressions were compared to cervical biopsy and the results stratified by level of residency training. kappa Statistics were calculated to determine the strength of correlation between impression and biopsy results. Results. Agreement within one-step between cervical histology and the colposcopic impression was found in 351 (77%) of the subjects. Histology impression agreement occurred in 92% of the nurse practitioner procedures, 77% of the second year resident (R2) cases, 75% of R3 colposcopies and 73% of the R4 procedures. The association between cervical biopsy and impression was highly significant (P < 0.0001). However, the strength of the correlation was only slight (kappa=0.197). The kappa value was highest for the nurse practitioners (0.376, fair correlation) and lowest for the R3 residents (0.110, slight correlation). The positive predictive value for the association of any colposcopically detected abnormality with any histologic abnormality was 64.1%. The overall PPV was highest for the nurse practitioners (79.3%) and lowest for the R2 residents (58.7%). Conclusions. While the colposcopic impressions of gynecology residents were accurate, there was little difference in the accuracy of colposcopic assessment based upon the level of resident training. (c) 2006 Elsevier Inc. All rights reserved.