APPRAISAL OF THE MODALITIES USED TO EVALUATE AN INITIAL ABNORMAL PAPANICOLAOU SMEAR

Objective: To examine the efficacy of the different modalities used to evaluate an initial abnormal Papanicolaou smear. Methods: The study p opulation comprised 214 nonpregnant women referred with a Papanicolaou smear diagnosis of cervical intraepithelial neoplasia. Each patient w as evaluated by a repeat Papanicolaou smear, colposcopy, and colposcop ically directed cervical biopsies. Immediate loop excision was perform ed to remove the entire transformation zone in all patients. Kappa sta tistics were calculated to determine agreement among the modalities, a nd logistic regression was used for determining relative risks (RR). R esults: There was 53% agreement between the initial and repeat Papanic olaou smears. When low-grade squamous intraepithelial lesion (SIL) was diagnosed by Papanicolaou smear, there was 89% agreement with the col poscopic impression. However, a colposcopic impression of high-grade S IL was found in only one-third of the women diagnosed with high-grade SIL by Papanicolaou smear. A comparison of the histology of the cervic al biopsy and the loop specimen revealed 57% agreement. Univariate ana lysis indicated that each modality was able to predict the RR of high- grade dysplasia in the loop specimen at a statistical significance lev el of .05. A colposcopic impression of high-grade dysplasia conferred an RR of 7.43 (95% confidence interval [CI] 2.17-25.49) for high-grade dysplasia in the loop specimen. An initial Papanicolaou smear diagnos is of high-grade SIL did not contribute to the multivariate model for calculating the risk of high-grade dysplasia, as the RR was 1.6 (95% C I 0.68-3.81). Conclusion: Patients with an initial Papanicolaou smear showing low-grade SIL would benefit from a repeat Papanicolaou smear a nd colposcopically directed biopsies before proceeding with loop diath ermy. In contrast, patients with a colposcopic impression of high-grad e dysplasia combined with high-grade SIL on Papanicolaou smear appear to be candidates for immediate loop excision.