Diagnostic Performance of Four Ultrasound Risk Stratification Systems: A Systematic Review and Meta-Analysis.

Background: Several ultrasound (US)-based risk stratification systems have been increasingly used for the optimal management of thyroid nodules. However, there are considerable discrepancies across these systems. This study aimed to summarize and compare the category-based diagnostic performance in the detection of thyroid cancer of different US-based risk stratification systems from four societies: the American College of Radiology-Thyroid Imaging Reporting and Data System (ACR-TIRADS), the American Thyroid Association (ATA), the Korean Thyroid Association/Korean Society of Thyroid Radiology (KTA/KSThR; K-TIRADS), and the European Thyroid Association (EU-TIRADS). Methods: MEDLINE/PubMed and EMBASE databases were searched to identify original articles investigating the category-based diagnostic performance according to at least one of the following guidelines: ACR-TIRADS, ATA, K-TIRADS, and EU-TIRADS. Pooled sensitivity and specificity were calculated using a bivariate random-effects model. A subgroup analysis on nodules of 1 cm or larger and a meta-regression analysis to identify factors associated with the diagnostic performance were performed. Results: A total of 29 articles including 33,748 thyroid nodules met the eligibility criteria and were included in the analysis. For ACR-TIRADS, the pooled sensitivity and specificity were, respectively, 66% and 91% for category 5 and 95% and 55% for category 4 or 5. For ATA, the pooled sensitivity and specificity were, respectively, 74% and 88% for category 5 and 91% and 64% for category 4 or 5. For K-TIRADS, the pooled sensitivity and specificity were, respectively, 55% and 95% for category 5 and 89% and 64% for category 4 or 5. For EU-TIRADS, the pooled sensitivity and specificity were, respectively, 82% and 90% for category 5 and 96% and 52% for category 4 or 5. Study location, proportion of female patients and malignant nodules, and study design were associated with study heterogeneity. Conclusions: The overall diagnostic performance of the four US-based risk stratification systems was comparable.