Automated screening for quality control using PAPNET: a study of 638 negative Pap smears.

To determine if the PAPNET screening system can be used for quality control to lower false-negative rates for Pap smears 638 manually screened, "negative" Pap smears were subjected retrospectively to the PAPNET screening system. Twenty-nine of the smears came from 18 patients who subsequently had biopsyproven high-grade squamous intraepithelial lesions (SIL). The remaining 609 negative smears were arbitrarily selected as controls. One hundred twenty-eight (128) of the retrospectively reviewed smears (20%) were selected by PAPNET for microscopic referral because of potential abnormalities. Abnormalities were confirmed on 14 of these smears upon microscopic evaluation. Five of these 14 smears were from smears obtained from four of the 18 women with high-grade SIL. The incidence of manually screened false-negatives detected by PAPNET rescreening was 14/638 or 2.2% for the entire patient population in this study and 5/29 or 17.24% for the targeted patients known to have subsequently developed highgrade lesions. The 2.2% decrease in the false-negative rate in this experiment may be partly artificial as this study group was seeded with false-negative cases at high risk for containing missed abnormalities. The implementation of the PAPNET system for quality control may lower false-negative rates for Pap smears.