P034 Inter-scorer concordance impacts MSLT results

Abstract

Abstract Introduction A retrospective study on the effect of inter-scorer concordance and impact of analysing polysomnography (PSG) data prior to the Multiple Sleep Latency Test (MSLT) on clinical interpretation of Narcolepsy (N) and Idiopathic Hypersomnolence (IH). Methods Data of four individuals was randomly selected from a cohort of patients that participated in MSLT studies. De-identified MSLT fragments from four nap periods (n=16) were scored in two groups: analysis of PSG conducted prior to the respective MSLT fragments, and analysis without access to prior PSG. Individual scorers were compared to a master score set, by consensus from two experienced sleep scientists. Spearman correlation and percentage agreement statistics were applied to calculate the inter-scorer concordance in sleep latency and REM latency. Mann-Whitney test was utilised to assess differences between the two groups. A positive result was assigned as: mean (n=4) sleep latency of <10min (IH), and mean (n=4) sleep latency of <8min including (n=2) SOREMs (N). Results From 16 sets of data, four false positive results were identified when PSG was not analysed prior to scoring the MSLT fragments. Additionally, statistically significant differences were present when PSG analysis was conducted prior to scoring MSLT sleep latency and REM latency data. Discussion These results support a recommendation that PSG analysis (sleep and REM latency) should be encouraged prior to MSLT studies and performed by the same sleep scientist. Furthermore, including MSLT data in intra-lab concordance activities is important, particularly in relation to medical interpretation and practice.