Percentages of cervical cytologic diagnoses as a quality assurance method.

Objective: To demonstrate empirically that the efficiency of rescreening to discover false negative cytologic diagnoses is greatly enhanced by prospectively stratifying accessions according to risk level.
Study Design: We stratified accessions from 11 clinical sources and established the rate of diagnoses according to three categories: (1) "within normal limits"/"benign cellular changes" (WNL/BCC), (2) "atypical squamous/glandular cells of undetermined significance" (ASCUS/AGCUS) and (3) "squamous intraepithelial lesion/invasive carcinoma" (SIL/CA). We then prospectively rescreened all negative smears from sources with rates of positive diagnoses (ASCUS/AGCUS and SIL/CA) in excess of 20% and 5% of negative smears from sources with rates of positive diagnoses < 20%. We compared the detection rates of false negatives on rescreening target groups with random rescreening of 10% of all negative smears.
Results: The rates of SIL/CA, ASCUS/AGCUS and WNL/BCC varied from 0 to 43%, 4% to 14% and 46% to 94%, respectively. Rescreening 10% of all negative smears revealed a false negative fraction of 3%; rescreening target groups revealed a false negative fraction of 5.9%.
Conclusion: The yield of prospectively detected false negative diagnoses was significantly increased by targeting high-risk accession groups. When cytology laboratories serve diverse populations, stratifying accessions by risk to permit increased sampling from the proportionately higher risk categories is a simple and effective device to maximize the yield and benefit from rescreening.