O‐10 THE ACCURACY OF CERVICAL CYTOLOGY: A COMPARISON BETWEEN THE THINPREP IMAGING SYSTEM AND CONVENTIONAL METHODS

Abstract

Douglass Hanly Moir is a large Australian laboratory which has recently introduced the ThinPrep Imaging System (TPI) for reading ThinPrep slides, which is still performed using a split‐sample technique. The Imager is a computerized system which identifies 22 fields for the cytologist to review using automated light microscopy. We compared the accuracy of TPI and conventional cytology (CC) during normal laboratory operation. The ThinPrep sample was prepared after taking a conventional Pap smear. TPI and CC reading was done without knowledge of the result of the other reading. The final cytology report issued to the referring doctor reflected the more severe of these two results. Histology results for all cases in which TPI and CC cytology results showed more than minimal disagreement were sought from the NSW Pap Test Register. Of 55 164 split sample pairs, 3.1% of CC of slides and 1.8% of TPI slides were unsatisfactory. There were 1758 women for whom there was more than minimal discrepancy between TPI and CC cytology results. TPI gave the more severe result in 1193 of the 1758 cases. In cases where only one of each pair of discrepant cytology results was CIN1 or higher grade, TPI detected 133 cases of high‐grade histology among 380 biopsies (35%), whereas CC detected 62 cases among 210 biopsies (29.5%). A repeat analysis based on reading of histology by one pathologist blinded to initial Pap smear result showed a similar result. Reading times were measured over 5 months for both TPI and CC for twenty cytologists who read both types of smears. On average, they read 13.3 TPI slides per hour and 6.1 CC slides per hour. This study provides evidence that cervical cytology read using the TPI detects more histological high‐grade disease than does CC. Further evidence shows that reading times are significantly reduced for cytologists using the TPI.