A Nationally Representative Survey Assessing Restorative Sleep in US Adults

Restorative sleep is a commonly used term but a poorly defined construct. Few studies have assessed restorative sleep in nationally representative samples. We convened a panel of 7 expert physicians and researchers to evaluate and enhance available measures of restorative sleep. We then developed the revised Restorative Sleep Questionnaire (REST-Q), which comprises 9 items assessing feelings resulting from the prior sleep episode, each with 5-point Likert response scales. Finally, we assessed the prevalence of high, somewhat, and low REST-Q scores in a nationally representative sample of US adults (n= 1,055) and examined the relationship of REST-Q scores with other sleep and demographic characteristics. Pairwise correlations were performed between the REST-Q scores and other self-reported sleep measures. Weighted logistic regression analyses were conducted to compare scores on the REST-Q with demographic variables. The prevalence of higher REST-Q scores (4 or 5 on the Likert scale) was 28.1% in the nationally representative sample. REST-Q scores positively correlated with sleep quality (r=0.61) and sleep duration (r=0.32), and negatively correlated with both difficulty falling asleep (r=-0.40) and falling back asleep after waking (r=-0.41). Higher restorative sleep scores (indicating more feelings of restoration upon waking) were more common among those who were: ≥60 years of age (OR=4.20, 95%CI: 1.92-9.17); widowed (OR=2.35, 95%CI:1.01-5.42), and retired (OR=2.02, 95%CI:1.30-3.14). Higher restorative sleep scores were less frequent among those who were not working (OR=0.36, 95%CI: 0.10-1.00) and living in a household with two or more persons (OR=0.51,95%CI:0.29-0.87). Our findings suggest that the REST-Q may be useful for assessing restorative sleep.