Interrater and intrarater reliability in multiple sleep latency test

Multiple sleep latency test (MSLT) remains the golden standard for the clinical diagnosis and management of excessive daytime sleepiness. However, there was limited data on the reliability measurement of MSLT. Forty-four (M/F ratio: 26/18, mean age of 43.4 +/- 13.9 years) subjects of an ongoing family study of narcolepsy underwent a standard polysomnogram and MSLT. Three trained staff of various level of experiences independently and blindly scored each nap (n = 219). To test for intrarater reliability, 100 naps (n = 20 subjects) were re-scored half a year later. The interrater reliability for the mean sleep latency of 5 naps, sleep latency, and rapid eye movement (REM) sleep latency of individual nap, presence and numbers of sleep onset REM periods varied from the range of 0.668 to 0.964. The mean interrater reliability for the clinical diagnosis of narcolepsy was 0.883 (range, 0.824-0.938), whereas it was 0.750 (range, 0.674-0.858) for the diagnosis of narcolepsy spectrum disorder (shortened mean sleep latency and/or the presence of sleep onset REM periods) and 0.796 (range, 0.697-0.846) for normal cases. The intrarater reliability for the sleep latency and sleep onset REM periods varied from the range of 0.625 to 0.991. Our study demonstrated excellent inter- and intrarater reliability in scoring the sleep latency and sleep onset REM periods of MSLT. They also had excellent agreement on the diagnosis of narcolepsy and an "excellent to good" agreement on the diagnosis of narcolepsy spectrum disorder and normal cases. Our study lent further support that MSLT was an objective measurement with high inter- and intrarater reliability across various sleep disorders and healthy controls among different sleep centers.