Comparison between automatic and visual scorings of REM sleep without atonia for the diagnosis of REM sleep behavior disorder in Parkinson disease

Study Objectives: To compare three different methods, two visual and one automatic, for the quantification of rapid eye movement (REM) sleep without atonia (RSWA) in the diagnosis of REM sleep behavior disorder (RBD) in Parkinson’s disease (PD) patients. Methods: Sixty-two consecutive patients with idiopathic PD underwent video-polysomnographic recording and showed more than 5 minutes of REM sleep. The electromyogram during REM sleep was analyzed by means of two visual methods (Montréal and SINBAR) and one automatic analysis (REM Atonia Index or RAI). RBD was diagnosed according to standard criteria and a series of diagnostic accuracy measures were calculated for each method, as well as the agree- ment between them. Results: RBD was diagnosed in 59.7% of patients. The accuracy (85.5%), receiver operating characteristic (ROC) area (0.833) and Cohen’s K coefficient (0.688) obtained with RAI were similar to those of the visual parameters. Visual tonic parameters, alone or in combination with phasic activity, showed high val- ues of accuracy (93.5–95.2%), ROC area (0.92–0.94), and Cohen’s K (0.862–0.933). Similarly, the agreement between the two visual methods was very high, and the agreement between each visual methods and RAI was substantial. Visual phasic measures alone performed worse than all the other measures. Conclusion: The diagnostic accuracy of RSWA obtained with both visual and automatic methods was high and there was a general agreement between meth- ods. RAI may be used as the first line method to detect RSWA in the diagnosis of RBD in PD, together with the visual inspection of video-recorded behaviors, while the visual analysis of RSWA might be used in doubtful cases.