Objective: To determine interobserver variability of nerve ultrasound in peripheral neuropathy in a prospective, systematic, multicenter study.
Methods: We enrolled 20 patients with an acquired chronic demyelinating or axonal polyneuropathy 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. Real-time image acquisition was performed blind by a local and a visiting investigator (reference). Five patients were investigated using different types of sonographic devices. Intraclass correlation coefficients were calculated, and a random-effects model was fitted to identify factors with significant effect on interobserver variability.
Results: Systematic differences between measurements made by different investigators were small (mean difference 0.11 mm2 [95% confidence interval 0.00-0.23 mm2]). Intraclass correlation coefficients were generally higher in arm nerves (0.48-0.96) than leg nerves (0.46-0.61). The hospital site and sonographic device did not contribute significantly to interobserver variability in the random-effects model.
Conclusions: Interobserver variability of nerve ultrasound in peripheral neuropathy is generally limited, especially in arm nerves. Different devices and a multicenter setting have no effect on interobserver variability. Therefore, nerve ultrasound is a reproducible tool for diagnostics in routine clinical practice and (multicenter) research.
© 2018 American Academy of Neurology.