PAPNET-assisted rescreening of cervical smears: cost and accuracy compared with a 100% manual rescreening strategy.

CONTEXT The Food and Drug Administration has recently approved several devices that use computerized image analysis to rescreen Papanicolaou (Pap) smears that have already been examined by cytotechnologists. Physicians and laboratories must decide whether the utility of these devices justifies the cost. OBJECTIVE To determine the effectiveness and cost of PAPNET-assisted rescreening in identifying cervical abnormalities not identified by manual rescreening. DESIGN PAPNET-assisted rescreening of 5478 Pap smears obtained in 1994 and 1995 previously identified as "within normal limits" or "benign changes" on both initial and random screening. PATIENTS Female service members and dependents aged 12 to 88 years. SETTING Air Force clinics in the United States and Japan. INTERVENTION Rescreening of Pap smears by PAPNET, followed by reevaluation of abnormal smears by the consensus panel, consisting of 3 cytotechnologists and 3 pathologists. MAIN OUTCOME MEASURES Proportion of Pap smears initially screened as normal identified as abnormal by both PAPNET and consensus panel; costs of rescreening. RESULTS PAPNET screening identified 1614 (29%) slides requiring additional microscopic review. On further review, 448 (8% of total) had possibly abnormal cells. Ultimately, 11 of these cases were reviewed by the consensus panel for potentially atypical cells. Of these 11 cases, 5 were reclassified as atypical squamous cells of undetermined significance (ASCUS) and 1 as atypical glandular cells of undetermined significance (AGUS). No additional squamous intraepithelial neoplasia (SIL) was identified in these smears; the patient with a diagnosis of AGUS on rescreening was diagnosed as having a low-grade SIL (LSIL) on follow-up. Costs were $5825 to $33781 for each additional ASCUS or AGUS diagnosis. A cost of $17475 to $101343 is expected for each case of LSIL identified by PAPNET-assisted rescreening and not by traditional manual rescreening. CONCLUSIONS PAPNET-assisted rescreening identified a few more cases of ASCUS than did manual rescreening, but at a relatively high cost. The costs of rescreening should be carefully compared with the expected efficacy in reducing cervical cancer mortality.