Objective To explore the consistency and difference between the thyroid imaging reporting and data system proposed by American College of Radiology (ACR-TIRAD) and European Thyroid Association (EU-TIRADS) in the diagnosis of thyroid nodule (TN).

Methods Patients with TN admitted to Lai'an Jianing Hospital (272 cases) and Nanjing Hospital Affiliated to Nanjing Medical University (10 cases) from June 2019 to January 2022 were selected as the study subjects. A total of 282 patients with TN diagnosed via fine needle aspiration biopsy or histopathological examination were recruited, and their age, sex, TN characteristics (size, number, structure, echo, shape, margin, and calcification), ACR-TIRADS and EU-TIRADS results were recorded. A total of 72 males and 210 females, aged (45.9±10.9) years were included. Measurement data were compared by utilizing two independent samples t-test, and count data were compared by using χ² test. McNemar paired χ² test was applied to compare the diagnostic accuracy of the two classification systems. Kappa test of multiclassification data was used to analyze consistency. The results of two kinds of system examination were consistent as group A, and not consistent as group B, and the ultrasonic diagnostic characteristics of the two groups were compared. Analysis of independent risk factors with inconsistent results of the two systems by using multi-factor Logistics regression prediction model. The accuracy of the model was evaluated by clinical decision curve.

Results The malignant risk of TN increased with the increase in ACR-TIRADS and EU-TIRADS classification. EU-TIRADS was significantly more sensitive than ACR-TIRADS (91.54% vs. 79.31%) , the difference between the two systems was is statistically significant (χ²=19.135, P<0.001). ACR-TIRADS had significantly higher specificity than EU-TIRADS (75.67% vs. 62.38%), the difference is statistically significant (χ²=12.900 P<0.001). Benign and malignant TN in ACR-TIRADS system 2–4 were consistent with those in EU-TIRADS system 2–4. Logistic regression analysis showed that TN measuring<20 mm (OR=1.196, 95%CI: 1.005–1.422, P<0.001), spongiform TN (OR=1.119, 95%CI: 1.022–1.226, P=0.016), low/middle/low echo (OR=1.627, 95%CI: 1.031–1.732, P=0.011), round shape (OR=1.072, 95%CI: 1.012–1.134, P=0.022), blurred edges (OR=1.567, 95%CI: 1.063–1.683, P<0.001), and differential lobe margins (OR=1.169, 95%CI: 1.051–1.301, P=0.004) were all independent risk factors for inconsistency between the results of the two examination systems.

Conclusion EU-TIRADS has higher sensitivity and lower specificity than ACR-TIRADS in the diagnosis of TN, and the test results of the two methods for type 2−4 TN are in good agreement.