Reproducibility and Interobserver Agreement of Different Thyroid Imaging and Reporting Data Systems (TIRADS)

A number of classification systems (TIRADS) have been developed to estimate the likelihood of malignancy in thyroid nodules, but their reproducibility is yet to be assessed. We evaluated the interobserver variability and diagnostic performance of the TIRADS in Kwak's modification (Kw-TIRADS) and European TIRADS (EU-TIRADS).

Two independent specialists, blinded concerning the morphology of the nodules, evaluated ultrasound images of 153 thyroid nodules identified in 149 patients at multiple time points.

The interobserver agreement (Cohen's κ) was 0.52 and 0.67 for Kw-TIRADS and EU-TIRADS, respectively, and rated as substantial. There were strong correlations between Kw-TIRADS and EU-TIRADS for the two observers with Spearman's coefficients of 0.731 (p = 0.00025) and 0.661 (p = 0.0012), respectively. Sensitivity of Kw-TIRADS for the diagnosis of thyroid cancer was 95-92.31% and that of EU-TIRADS was 92.31-89.74%, with specificity of about 60% for both TIRADS.

Despite the wide variability in the description of single ultrasonographic features, both Kw-TIRADS and EU-TIRADS may be a useful diagnostic tool in clinical practice.