M probe comes first: Impact of initial probe choice on diagnostic performance of vibration controlled transient elastography.

Abstract

Probe choice (M or XL) in transient elastography can be made by the user's own measure of skin-to-liver-capsule distance (SCD) or with an automated tool (AUTO). We studied how AUTO depends on initial probe choice. Three fictive clinics were considered: The "M-first clinic" uses AUTO from the M probe, the "XL-first clinic" uses AUTO from the XL probe and a "reference clinic" measures SCD independently. Agreement and discrepancies to the reference clinic were measured. 200 patients with chronic liver disease were prospectively included (58% female, 56 years, BMI 28.1 kg/m²). Fleiss' kappa for agreement in probe selection was 0.11 (95% CI -0.09 to 0.31), but accuracy was above 0.8 for both. Probe failure occurred for 16 (M-first clinic), 4 (XL-first clinic) and 3 patients (reference clinic). Use of XL probe given M probe failure improved performance of the M-first approach. The odds ratio for discrepancy in the XL-first vs M-first clinic is 2.4 (95% CI 1.2 to 5.2, p=0.012) for liver fibrosis and 4.8 (95% CI 1.8 to 16.1, p < 0.001) for steatosis. Agreement in AUTO between M and XL probes is poor although each has acceptable accuracy. The M-first approach leads to fewer discrepancies and should be adopted as a standard.