Interobserver variability in the fine needle aspiration biopsy diagnosis of follicular lesions of the thyroid gland

Objective To study the degree of interobserver variability in the interpretation of fine needle aspiration (FNA) biopsies of the in the categorization of follicular lesions thyroid, specifically (FLs), and to examine the accuracy of FNA diagnosis of FLs with surgical follow-up. Study Design Fifty cases were chosen with surgical follow-up and a cytologic diagnosis of either FL (21) or follicular neoplasms (29). Representative slides were selected for each case and circulated to 4 pathologists for review. Interobserver variability was assessed using pair-wise kappa statistics. Accuracy of the cytologic diagnoses in predicting a nonneoplastic or neoplastic outcome was determined by measuring sensitivity and specificity. Likelihood ratios and receiver operator characteristic curves were calculated for each reviewer. Results Interobserver agreement between the 4 pathologists was fair to substantial (K scores,. 0.199-0.617). The accuracy of the 4 pathologists' cytologic diagnoses in predicting the surgical outcome was 77-90% for follicular neoplasms and 53-74% for nonneoplastic diagnoses. Conclusion FLs present diagnostic difficulties as to cytologic categorization. A wide range of interobserver agreement was found in this study of pathologists from the same institution. Some pathologists make greater use of intermediate categories, such as FL, favor nonneoplastic, or FL, favor neoplastic, whereas others show more definitive categorization into benign and neoplastic groups.