External quality assurance in cervical/vaginal cytology: Interlaboratory agreement in the Emilia-Romagna region of Italy.

OBJECTIVE: To evaluate the diagnostic agreement between seven cervical/vaginal cytology laboratories participating in the first external quality assurance (EQA) scheme developed in Italy. STUDY DESIGN: Between 1991 and 1993, 110 cytologic smears were selected and classified by a committee and circulated and reported on by the laboratories according to the 1988 Bethesda System. Agreement was evaluated with the kappa statistic. Systematic disagreement was assessed by means of the Wilcoxon signed rank test. RESULTS: Interlaboratory kappa values varied between .01 and.29 (group score,.11) for sample adequacy and between.53 and.78 (group score,.67) for epithelial abnormalities. The lowest specific kappa values were observed for the three classes of sample adequacy (unsatisfactory,.07; less than optimal [LTO],.10; satisfactory [SAT],.14) and for the class of atypical cells of undetermined significance (ACUS), (.29). As compared with the study committee, 5/7 laboratories showed a systematic (P<.O1) tendency to undercall sample adequacy. Agreement on epithelial abnormalities was also analyzed according to the pattern of adequacy reported by paired laboratories (LTO/LTO, LTO/SAT, SAT/SAT). As compared with smears designated SAT/SAT, those classified as LTO/SAT were associated with lower specific kappa values for agreement on the presence of carcinoma and ACUS and with equal or greater values for agreement on the other classes, suggesting an arbitrary use of notations of LTO inversely related to the severity of epithelial lesions. CONCLUSION: EQA schemes, as applied to cervical/vaginal cytology, can shed light on major deficiencies in specific diagnostic areas.