Interlaboratory Reproducibility In Reporting Inadequate Cervical Smears—A Multicentre Multinational Study

Abstract

A set of 300 vaginal smears was interpreted by 13 cytologists from six European laboratories, who were requested to report inadequate and suboptimal smears. The set had been appropriately seeded to reach approximately 10% inadequate and 20% suboptimal smear frequency. According to the majority report, 230 smears were classified as adequate (76.7%), 43 as inadequate (14.3%), and 27 as suboptimal (9.0%). Agreement with the majority report ranged from 52% to 91% (average 78%). Kappa statistics for reporting inadequate smears showed a high level of agreement for five cytologists, and fair to good agreement for eight. In contrast, kappa statistics for reporting suboptimal smears showed fair to good agreement with the majority report only in five instances, whereas agreement was poor for eight cytologists. 'Inadequate smear' rates may be used to compare the quality of smears received in different laboratories, as there is a high level of agreement among cytologists as to what constitutes an inadequate smear. However, this is not true for "suboptimal smear" rates, and more precise reporting criteria must be defined and tested if an intermediate category is to be retained to report poor quality smears: more precise reporting criteria must be defined and tested if an intermediate category is to be retained to report poor quality smears.

Outcome of women with inadequate cervical smears followed up for five years

“…However, a European multicentre study showed a high level of agreement among cytologists as to what constitutes an inadequate smear. 13 Another limitation of our study was the inability to correlate the clinical signs and symptoms (if there are any) with colposcopic findings. This is because such findings are not consistently recorded or are not given to the cytology laboratory.…”

Risk of CIN2 in Women With a Pap Test Without Endocervical Cells vs. Those With a Negative Pap Test With Endocervical Cells

“…The order these parts of the diagnosis are presented in may influence the accuracy of reporting. Furthermore, it has been shown that the reproducibility of the ECC presence/scarcity/absence classification is poorly reproducible, 30,31 and few efforts have been made to improve reproducibility. As a consequence, this study measured the risk of a nonhomogeneous class, with strong variation in prevalence among centers that is not due to any biologic differences but most likely to differences in the cytologists' interpretations.…”

Reproducibility of Classification and Correction for Verification Bias as Determinants of Performance of Papanicolaou Smear Cytology in the Screening Setting

“…Most important, these include technical innovations in the late 1990s leading to the development of liquid-based cytology techniques, 9 cytology automation [1][2][3]12,13 and rigorous quality control measures to improve the reproducibility. 7,8,15,18,[21][22][23][24][25][26] As part of our systematic efforts to improve the quality of conventional Pap smear cytology, 21,22,25,27 we recently developed laboratory software (CONQUISTADOR) to assist this intraobserver and interlaboratory quality assurance, validated in a simulation study 28 and recently shown useful in a screening setting ongoing in Latin America (the LAMS study). 29 In addition to the discussed technical, classification and reproducibility issues, performance of Pap smear cytology in a screening setting also critically depends on the prevalence of the abnormalities (squamous intraepithelial lesion [SIL] and cervical cancer) in the target population.…”