S135. Reliability of ultrasound measurements of peripheral nerves: A systematic multicenter study in patients and controls

Introduction Nerve ultrasound is a relatively new diagnostic tool in peripheral neuropathies. Some studies investigated inter-observer variability of this technique, but had several limitations such as only including healthy controls, studying few sites or still images, or being single center. This multicenter study systematically investigated inter-observer variability of ultrasound at multiple sites in healthy and diseased nerves to gain more insight on its applicability in daily clinical practice. Methods We enrolled 20 chronic inflammatory neuropathy patients and 10 healthy controls in 3 different centers. All participants underwent an extensive nerve ultrasound protocol, including cross-sectional area measurements of median, ulnar, fibular, tibial, and sural nerves, and brachial plexus. Ultrasound was performed blinded by a local investigator and an investigator visiting all centers. In 5 patients ultrasound was performed on two different types of sonographic device. All investigators had at least one year of experience with nerve ultrasound. Results Overall, mean difference between investigators was −0.11 mm 2 (95%-CI −0.23 to 0.00 mm 2 ). The investigated nerve site had significant influence on the mean difference in a multilevel mixed model, but only at the ulnar sulcus there was significant deviation from 0 mm 2 (−0.46 mm 2 ). No significant effect of different hospitals or devices was found. Intraclass correlation coefficients (ICCs) were 0.48–0.96 in arm nerves compared to 0.46–0.61 in large leg nerves. Standard deviation (SD) for difference between investigators was 1.0–1.7 mm 2 in arm nerve sites compared to 1.5–3.1 mm 2 in brachial plexus, posterior tibial, and fibular nerve sites. Conclusion Sonographic measurement of nerve size was reliable with a low mean difference between investigators and high ICCs, especially in arm nerves. Still, solitary abnormal measurements at sites with a high variation in difference (high SD) should be interpreted with care. Different devices and a multicenter setting had no significant influence on the mean difference. This indicates that nerve ultrasound studies can reliably obtain nerve size data that may be pooled and generalized to daily clinical practice.