Aims and Background: Thyroid carcinomas display several pathologic fea
tures, show different behavior and necessitate different treatment; th
us correct classification is mandatory. Methods: The kappa statistic w
as used as a measure of agreement in a panel of seven pathologists who
reviewed 200 cases of thyroid tumors. Results: Overall agreement was
83% (k = 68). Good agreement was found for anaplastic (k = 0.85) and p
apillary carcinomas (k = 0.81); agreement for medullary carcinoma was
acceptable (k = 0.80), suboptimal for other (k = 0.67), and poor for f
ollicular carcinoma (k = 0.54). Conclusions: Central pathology review
of thyroid carcinomas is recommended when clinical and epidemiologic t
rials are planned.