LAUSR

Despite the Epworth Sleepiness Score being widely used, there are limited studies of its reliability in clinical practice. The aim of this study was to assess the reliability of the Epworth Sleepiness Score in a clinical population. The study included patients referred to Middlemore Hospital sleep service between October and November 2014, aged over 17 years, with at least two Epworth Sleepiness Score measurements at up to three different points on the diagnostic pathway: on General Practitioner referral (GP Epworth Sleepiness Score); at overnight oximetry assessment (Oximetry Epworth Sleepiness Score); and at a specialist clinic (Specialist Epworth Sleepiness Score). No treatment was administered between scores. One‐hundred and thirty‐three patients were included in the study. There was a median of 91 days from GP Epworth Sleepiness Score to Oximetry Epworth Sleepiness Score, and 11 days from Oximetry Epworth Sleepiness Score to Specialist Epworth Sleepiness Score. There was poor test–retest reliability between GP Epworth Sleepiness Score and Specialist Epworth Sleepiness Score; 72.4% and 17.8% of patients had an absolute difference of more than 2 and 8 Epworth Sleepiness Score points, respectively. A Bland–Altman plot of mean Epworth Sleepiness Score versus the difference between GP Epworth Sleepiness Score and Specialist Epworth Sleepiness Score demonstrated a wide scatter of data and 95% confidence interval for the difference in Epworth Sleepiness Score for an individual patient of −14 to +10. There was similar variability between GP Epworth Sleepiness Score and Oximetry Epworth Sleepiness Score. The reliability of the Epworth Sleepiness Score is unproven in clinical settings. This study shows poor test–retest reliability of Epworth Sleepiness Score, particularly between primary and secondary care, arguing against the use of Epworth Sleepiness Score for clinical decision‐making or prioritisation of services without first assessing the reliability of the Epworth Sleepiness Score in the relevant clinical population.