Objective: To study the diagnostic correlation between cervical cytology specimens and corresponding surgical biopsies.
Design and setting: College of American Pathologists Q-Probes laboratory quality improvement study in 348 laboratories.
Main outcome measures: Sensitivity, specificity, and positive predictive value of cervicovaginal cytology diagnosis.
Results: Statistical analysis of 22 439 paired cervicovaginal cytology--cervical biopsy specimens reveals a sensitivity of 89.4%, specificity of 64.8%, and predictive value of a positive cytology of 88.9%. The majority of discrepancies were attributed to cytology or biopsy sampling errors. Routinely providing the patient's recent cervical cytology report to the surgical pathologist at the time the biopsy was examined resulted in improved sensitivity. Correlations for cytology specimens obtained at the time of biopsy revealed lower sensitivity and higher specificity than for those obtained at a time prior to the biopsy.
Conclusions: We have defined current statistical expectations for cervical cytology-biopsy correlation, reasons for noncorrelation, and have provided recommendations for quality improvement.