Abstract Introduction The Epworth Sleepiness Scale (ESS) measures subjectively reported behaviors of daytime sleepiness but has never been fully validated in women. Given documented sex differences in sleep symptoms, sleep physiology, and daytime function, questions have emerged about whether the ESS measures daytime sleepiness similarly in women and men. This study aimed to examine sex differences in the psychometric properties of the ESS, including an item-level analysis. Methods We conducted a retrospective analysis of the eight-item ESS collected from the electronic medical records of sleep study patients. Items measure the likelihood of falling asleep while 1) sitting and reading; 2) watching tv; 3) sitting in a public space; 4) a car passenger; 5) lying down in the afternoon; 6) sitting and talking; 7) sitting quietly after lunch; 8) stopped in traffic. We used confirmatory factor analysis to assess the ESS’s unidimensional structure and to compare congeneric (unequal item weights) and tau-equivalent (equal weights) models. Measurement invariance by sex was tested at the configural, metric, and scalar levels using Satorra–Bentler scaled chi-square tests. Sex differences in item responses were examined using Rao–Scott chi-square tests. Results The sample included 3,277 adults (women: n=1,730, 52.8%; men: n=1,547, 47.2%), was predominantly White (88.4%), with a mean age of 61.6 years (SD=14.7) and mean BMI of 33.4 kg/m2 (SD=8.0). Despite variation in item loadings, the ESS showed a clear unidimensional structure by sex. However, the findings indicated metric invariance (Δχ2 = 5.22, df = 7, p = 0.63) but not scalar invariance, suggesting structural model comparisons are valid, but individual ESS scores may not be directly comparable. Significant sex differences were found for items 1, 2, 3, 4, 6, and 8 (Rao–Scott χ2 range: 9.87–93.67; p = 0.032 to 0.001) but not items 5 and 7. Conclusion The ESS remained a unidimensional measure for both women and men, though several items showed sex-specific response differences that offset each other. Depending on the research question, investigators should consider sex and item-level variation when using the ESS to assess daytime sleepiness. Support (if any) University of Pittsburgh, School of Nursing, Hub for Sleep and Circadian Science; 5R01HL170675-02 (Morris/Mazzotti);
Morris et al. (Fri,) studied this question.