Evidence-based evaluation of the risks of malignancy predicted by thyroid fine-needle aspiration biopsies.

A National Cancer Institute (NCI) "Thyroid Fine-Needle Aspiration (FNA) State of the Science Conference" recently proposed standardized nomenclature and "risks of malignancies" associated with various diagnostic categories. We evaluated the evidence levels of the data used by NCI to predict malignancy risks and whether those estimates had clinical validity in our patient population.Eight hundred seventy-nine patients underwent thyroid FNA during 2006. FNA diagnoses were translated into NCI diagnostic categories, and 2-year follow-up retrospective information was obtained. Four percentages of malignancies were calculated for each diagnostic category using follow-up information from FNA, thyroidectomy, both, and all patients as denominators. 95% confidence intervals (CI) were estimated for all proportions, and results were analyzed with chi-square statistics. "Relative risk" calculations were performed using the percentage of malignancies in the entire population under study as a denominator.Most of the studies cited by the NCI provided incomplete and variable level III evidence based mainly on surgical follow-up. Among our patients, the percentages of malignancies calculated with follow-up data from all patients as the denominator were similar to the "risk estimates" proposed by the NCI, but estimates based on surgical follow-up overestimated the probability of thyroid malignancy for patients with FNA diagnosis of "benign" and "follicular lesions of undetermined significance" (FLUS). Relative risk and 95% CI calculations suggested that the NCI classification could be simplified into three categories: "benign," "FLUS + neoplasm," and "suspicious + malignant."