Diagnostic criteria for amyotrophic lateral sclerosis: A multicentre study of inter-rater variation and sensitivity

Abstract Objective This study assesses inter-rater agreement and sensitivity of diagnostic criteria for amyotrophic lateral sclerosis (ALS). Methods Clinical and electrophysiological data of 399 patients with suspected ALS were collected by eleven experienced physicians from ten different countries. Eight physicians classified patients independently and blinded according to the revised El Escorial Criteria (rEEC) and to the Awaji Criteria (AC). Inter-rater agreement was assessed by Kappa coefficients, sensitivity by majority diagnosis on 350 patients with follow-up data. Results Inter-rater agreement was generally low both for rEEC and AC. Agreement was best on the categories “Not-ALS”, “Definite”, and “Probable”, and poorest for “Possible” and “Probable Laboratory-supported”. Sensitivity was equal for rEEC (64%) and AC (63%), probably due to downgrading of “Probable Laboratory-supported” patients by AC. However, AC was significantly more effective in classifying patients as “ALS” versus “Not-ALS” (p  Conclusions Inter-rater variation is high both for rEEC and for AC probably due to a high complexity of the rEEC inherent in the AC. The gain of AC on diagnostic sensitivity is reduced by the omission of the “Probable Laboratory-supported” category. Significance The results highlight a need for initiatives to develop simpler and more reproducible diagnostic criteria for ALS in clinical practice and research.