RESUMO
INTRODUCTION: The Bethesda system is widely accepted for thyroid FNA diagnostics, but has scarcely been analysed in relation to clinical background data. Our aim was to analyse the thyroid FNA diagnostic process in view of clinical data, and to assess the validity of the Bethesda system during the first year of implementation. METHODS: There were 415 thyroid FNAs taken from 363 patients during October 2011-September 2012 in the Pirkanmaa Hospital District, Finland. The median age of the patients was 59 years, and the female-to-male ratio 4:1. Clinical data were collected from patient registries, and thyroid FNA and histopathological data from the pathology registry. RESULTS: The Bethesda categories were represented as follows: 94 non-diagnostic cases (26%); 177 benign (49%); 32 atypia of undetermined significance/follicular lesion of undetermined significance (9%); 31 follicular neoplasm (9%); 20 suspicious for malignancy (5%); and nine malignant cases (2%). Only 23 (24%) of the non-diagnostic samples and 18 (56%) of the atypia of undetermined significance/follicular lesion of undetermined significance led to repeat FNA. Thyroid cancer was histopathologically diagnosed in 28 cases (8%). When the categories requiring surgical treatment were considered true positive findings, the sensitivity of the Bethesda system was 90%, and specificity was 70%. Interobserver accuracy was 86%. CONCLUSIONS: Already during the first year of implementation, the Bethesda system proved reliable in evaluating the risk of thyroid malignancy. Nevertheless, the clinical judgement of the indication of ultrasound/FNA and management according to the FNA findings need improvement. The relatively high proportion of non-diagnostic FNAs could be diminished by obtaining the samples by radiologists experienced in ultrasound-guided FNA techniques.