Thyroid fine-needle aspiration biopsy: variability in reporting.

BACKGROUND The low incidence of thyroid cancer despite the high prevalence of thyroid nodules necessitates a screening tool to determine which patients require surgical management. The utility of fine-needle aspiration biopsy (FNAB) for this purpose requires a low false-negative (FN) rate and an acceptable sensitivity and specificity for the detection of malignancy. While reviewing our institution's experience with thyroid FNAB, we found significant discrepancies in how statistics of thyroid FNAB were tabulated and reported in the literature. Here we examine the sources of these discrepancies by evaluating large series of thyroid FNAB with regard to cytopathologic reporting and statistical calculation. METHODS Published series of thyroid FNAB with >200 FNAB and available histological data with sufficient raw data to recalculate statistics were analyzed. Considering indeterminate and malignant results to be positive FNAB results, since, in a four-tier system, both lead to surgical management, specificity, sensitivity, accuracy, positive predictive value, negative predictive value, FN, and false-positive (FP) rates were recalculated. Differences between reported and recalculated statistics were then evaluated for significance. RESULTS Nineteen studies and 20 series were identified. The following are reported and recalculated means, respectively: for sensitivity, 81% and 86%; for specificity, 81% and 62%; for accuracy, 77% and 71%; for positive predictive value, 65% and 50%; for negative predictive value, 84% and 93%; for FN rates, 13% and 14%; for FP rates, 10% and 38%. FP rates had a mean of 1.4% when recalculated considering only malignant FNAB as positive tests. Specificity and FP rates had statistically significant differences between the means of reported and recalculated values. CONCLUSIONS Thyroid FNAB remains the screening tool of choice in the evaluation of thyroid nodules. However, the variability in the calculation of reported thyroid FNAB statistics highlights the need for uniformity in statistical reporting for accurate understanding of thyroid FNAB's clinical utility.