Accuracy of colposcopy in HIV seropositive and seronegative women with abnormal Pap tests

Objective. The aim of this study is to compare colposcopic findings and the accuracy of colposcopic impression in HIV seropositive and seronegative women with abnormal Pap tests. Methods. HIV seropositive and seronegative women in a national cohort study had Pap tests collected every six months, with colposcopy for any abnormal result. Prospectively collected colposcopy and histology findings were analyzed retrospectively using Pearson Chi-square, t-test and Wilcoxon two-sample tests, logistic regression models, and Kappa coefficients. Results. After adjusting for age and Pap result, 1618 eligible HIV seropositive women were more likely than 406 seronegative women to have inadequate colposcopic examinations, abnormal colposcopic findings, and large cervical lesions. However, among those with abnormal colposcopy, colposcopic characteristics and lesion size and number did not differ by HIV serostatus. Agreement between colposcopists' impressions and highest grade biopsy diagnoses was fair (kappa coefficient 0.35, 95% C.I. 0.31, 0.38). Agreement did not differ by HIV serostatus and did not improve with multiple biopsies (weighted kappa coefficient 0.35, 95% C.I. 0.32, 0.39) or after including all histology results over two years following colposcopy. Conclusion. Although HIV seropositive women with abnormal cytology are more likely to have colposcopic abnormality, the performance of colposcopy appears to be similar to that in HIV seronegative women. Biopsy is required to confirm colposcopic impression.