Is the Epworth sleepiness scale suitable for use in stroke?

ABSTRACT

BACKGROUND: The Epworth Sleepiness Scale (ESS-8) is frequently used in stroke but has never been validated for this condition. There is concern regarding the suitability of the driving item (item 8). A summed raw score of 10 or more (from a maximum of 24) signifies pathological sleepiness. OBJECTIVE: To determine the construct validity of the ESS-8 by Rasch analysis and in particular to determine whether omission of item 8 confounds the scale. METHOD: A pack containing the ESS-8 and questions regarding sleep and demographics was sent to 999 patients who had experienced a stroke within the past 4 years. Data were assessed for fit to the Rasch model. RESULTS: Analysis of 269 records revealed a unidimensional scale that was free from differential item functioning by age and sex with good overall fit to the Rasch model. Item 4 had disordered thresholds. Analysis of the ESS without item 8 (ESS-7) also revealed a valid scale. Equating person locations between the ESS-8 and ESS-7 showed no differences below a summed raw score of 18. CONCLUSION: The ESS-8 has good construct validity for use in stroke and is reliable at the cutpoint of 10. Summed raw scores below 18 will be unaffected if nondrivers either score as zero or simply omit item 8. The scale is therefore robust for detecting cases of pathological sleepiness in stroke but may not be suitable for measuring high levels of sleepiness in a sample containing both drivers and nondrivers. Instead, the ESS-7 could be used for this purpose.