Histopathologic reproducibility of thyroid disease in an epidemiologic study.

An investigation of the long-term effects of childhood scalp irradiation demonstrated a significantly increased risk of thyroid tumors in the irradiated population. Because of the complexity of thyroid cancer diagnosis, a histopathologic slide review of 59 of the 68 patients (irradiated and nonirradiated) with thyroid disease was undertaken. The review revealed 90% agreement (kappa = +0.85, P less than 0.01) between the original and review diagnosis. Four of 27 cases previously diagnosed as malignant were reclassified as benign, yielding a cancer misdiagnosis rate of 14.8%. All four of the misdiagnosed cancers were of follicular or mixed papillary-follicular type. As a result of the histologic review, the ratio of malignant to benign tumors decreased from 2.55 to 1.75. Since disagreement in diagnosis was similar in the irradiated and nonirradiated groups, the relative risk of radiation-associated neoplasms did not change substantially. The histopathologic review shows that although there were some problems in diagnostic reproducibility, they were not statistically significant and did not alter our previous conclusions regarding radiation exposure. However, a 15% reduction in the number of malignancies might affect epidemiologic studies with an external comparison as well as geographic or temporal comparisons.