[Reproducibility of cytologic diagnosis: study of CRISAP Ile-de-France].

The cervical Pap smear is the method of choice for early detection of precancerous cervical lesions. However, the sensitivity of the Pap smear is not perfect, nor is the specificity, in particular for minor atypia; quality control is the best method to improve the efficiency of the test. The "Centre de regroupement informatique et statistique des données en anatomie pathologique" (CRISAP) Ile-de-France has initiated an external quality control with a protocol aimed at assessing diagnosis reproducibility among observers. Thirteen pathologists agreed to participate on a voluntary basis fur this protocol, which consisted of the rereading of 650 slides chosen at random. Each participant reread 50 cases and sent 50 cases to be reread anonymously. The diagnosis was given according to the Bethesda classification. The reproducibility was assessed using the percentage of agreement and kappa statistics. The percentage of agreement for the whole group was 65% and the weighted kappa 0.66. When normal and unsatisfactory cases were combined as negative and atypical squamous cells of undetermined significance (ASCUS), low-grade squamous intraepithelial lesion (LGSIL) and high-grade squamous intraepithelial lesion (HGSIL) as positive, the percentage of agreement was 83% and Kappa 0.66. When normal, unsatisfactory and ASCUS were combined as negative and LGSIL and HGSIL as positive, the percentage agreement raised to 90% and kappa to 0.76. This attempt at assessing the reproducibility of cytologic diagnosis using an informal method had led to good participation among participants. A fairly good agreement for overall cytologic diagnosis was found, though some variability remained concerning the diagnosis of unsatisfactory and those labelled as ASCUS. These results should lead the participants to reflect upon and eventually reassess their criteria on these specific cases.