ABSTRACT
Cancer of uterine cervix is the second cause of death in women in the world and the most common cause in developing countries. Because the majority of women with invasive cervical cancer of the uterine have not previously undergone screening, many clinicians assume that Pap smear has a high degree of accuracy; but problems such as false positive and false negative interpretations, as well as interobserver variability have questioned its validity. We retrieved 162 positive cervical smears that had been originally interpreted as ASC-US, ASC-H, LSIL, HSIL, SCC, AGC and adenocarcinoma from the cytology archives of Women's Hospital in Tehran, Iran. The slides were rescreened by an experienced pathologist and reclassified in the mentioned categories. All the 162 slides were reviewed by three more pathologists in a blind study using interpretative criteria utilized in their daily routine to evaluate interobserver reproducibility. To increase the level of interobserver agreement, the diagnostic categories were reduced to squamous Vs. glandular abnormalities and invasive [SCC and adenocarcinoma] Vs. non-invasive abnormalities. The results obtained in this study indicated slight interobserver agreement [k=0.26]. The most reproducible category was the invasive category [SCC in addition to adenocarcinoma] and the least agreement was seen for HSIL [k=0.19]. This study showed that reproducibility of cytological interpretation of conventional Pap smears varies among interpretive categories and the overall interobserver agreement is slight. Since convening on the reduction of interobserver discrepancy in Pap smear interpretations necessitates more reliable information of interpretative variability, larger studies need to be undertaken