Assessing the accuracy of thyroid nodule risk-stratification tools in Hispanic and non-Hispanic patients

Ultrasound risk categories (thyroid imaging, reporting, and data systems, TI-RADS), fine-needle aspiration cytology (Bethesda), and molecular classifiers guide thyroid nodule management. However, these tools may perform differently across populations due to differences in thyroid cancer characteristics among Hispanic patients.

In this study, we compared the predictive performance of these tools in Hispanic vs non-Hispanic patients using surgical pathology as the gold standard.

We performed a retrospective study of adult patients (≥18 years) evaluated for thyroid nodules from January 2023 to June 2024 at a tertiary academic medical center. Nodules were stratified using TI-RADS, Bethesda, and molecular testing. Malignancy was confirmed by final surgical pathology, and positive predictive value was assessed within diagnostic strata among resected nodules.

A total of 308 patients (421 nodules) were included in the analytic cohort, with 29.2% being Hispanic. Mean age was 54.3 ± 15.4 years in Hispanic patients and 58.8 ± 16.1 years in non-Hispanic patients (P = .02). TI-RADS and Bethesda distributions were similar by ethnicity. Among TI-RADS4 nodules undergoing surgery, malignancy risk was higher in Hispanic patients (32.7% vs 12.0%; odds ratio [OR] 3.57, 95% CI 1.63-7.78; P = .001). Among Bethesda III/IV nodules with suspicious molecular results, malignancy was higher in Hispanic patients (78.6% vs 34.8%; adjusted OR 8.72, 95% CI 1.35-96.0; P = .022). Overall, Hispanic patients had higher odds of malignancy (OR 2.14, 95% CI 1.26-3.64; P = .0047).

Despite similar use of diagnostic tools, Hispanic patients had higher malignancy rates overall and within select diagnostic strata. These findings suggest variation in risk estimate performance across populations and support multicenter validation in diverse cohorts.