The Inter-rater Variability of Clinical Assessment in Post-anoxic Myoclonus

Background: Acute post-anoxic myoclonus (PAM) can be divided into an unfavorable (generalized/subcortical) and more favorable ((multi)focal/cortical) outcome group that could support prognostication in post-anoxic encephalopathy; however, the inter-rater variability of clinically assessing these PAM subtypes is unknown. Methods: We prospectively examined PAM patients using a standardized video protocol. Videos were rated by three neurologists who classified PAM phenotype (generalized/(multi)focal), stimulus sensitivity, localization (proximal/distal/both), and severity (Clinical Global Impression-Severity Scale (CGI-S) and Unified Myoclonus Rating Scale (UMRS)). Results: Poor inter-rater agreement was found for phenotype and stimulus sensitivity (κ = –0.05), moderate agreement for localization (κ = 0.46). Substantial agreement was obtained for the CGI-S (intraclass correlation coefficient (ICC) = 0.64) and almost perfect agreement for the UMRS (ICC = 0.82). Discussion: Clinical assessment of PAM is not reproducible between physicians, and should therefore not be used for prognostication. PAM severity measured by the UMRS appears to be reliable; however, the relation between PAM severity and outcome is unknown.