Agreement and reliability of transient elastography in patients with chronic hepatitis C - a cross-sectional test-retest study

Abstract

Abstract Background Transient elastography (TE) has largely replaced liver biopsy to evaluate fibrosis stage and cirrhosis in chronic hepatitis C. Previous studies report excellent reliability of TE but agreement metrics have not been reported. This study aimed to assess inter-rater agreement and reliability of repeated TE measurements. Methods Two operators performed TE independently, directly after each other. The primary outcome was disagreement, defined as a difference in TE results between operators of ≥ 33%, as well as the smallest detectable change, SDC95 (i.e., the difference between measurements needed to state with 95% certainty that there is a difference in underlying stiffness). Secondary outcomes included reliability, measured as intra-class correlation, and patient and examination characteristics associated with agreement. Results In total, 65 patients were included, with a mean liver stiffness of 9.7 kPa. Of these, 21 (32%) had a disagreement in TE results of ≥ 33% between the two operators. The SDC95 on the log scale was 1.97, indicating that an almost twofold increase or decrease in liver stiffness would be required to confidently represent a change in the underlying fibrosis. Reliability, estimated using the intra-class correlation, was acceptable at 0.86. In a post-hoc analysis, fasting less than 5 hours before TE was associated with a higher degree of disagreement (48% vs 19%, p = 0.03). Conclusions In our clinical setting, agreement in directly repeated TE measurements was surprisingly low. It is essential to further investigate the reliability and agreement of TE to determine its validity and usefulness.