The Stability of Sleep Duration and Insomnia Measured Using Self-Report Questionnaires, Sleep Diaries, and Actigraphy

Abstract

Self-report retrospective sleep questionnaires, prospective sleep diaries, and actigraphy are often used in epidemiological research interested in examining temporal changes in sleep in relation to temporal changes in other disease states of interest (e.g., cardiovascular disease). However, few studies have investigated the long-term or day-to-day stability of these sleep measures. This study examines the 2-year and 2-day stability of retrospective sleep questionnaires, prospective sleep diaries, and actigraphy compared to retrospective psychosocial questionnaires. Three-hundred diverse, community participants (50% female, age range 21-70 years) completed (baseline and 2 years later) a psychosocial survey which included the Insomnia Severity Index-Modified (ISI-M) and a habitual sleep duration questionnaire, then 2-days (Thursday and Friday night) of prospective sleep diaries and actigraphy. Moderate 2-year stability (ICCs = 0.508-0.695) was observed for ISI-M scores, self-reported retrospective habitual sleep duration, and 2-night average actigraphy-derived total sleep time and sleep efficiency, which was similar to the stability for the retrospective psychosocial measures (ICCs = 0.641-0.752). Poor 2-year stability (ICCs < 0.50) was observed for 2-night average prospective sleep diary-derived total sleep time and sleep efficiency. Poor 2-day stability was observed for prospective sleep diaries and actigraphy at both baseline and follow-up. Our findings suggest that retrospective sleep questionnaires and 2-days of actigraphy-measured sleep may have long-term (i.e., ≥ 2 years) stability in adult populations, however, more than 2-days of prospective sleep diaries may be needed for long-term stability. Moreover, our findings suggest a high level of variability for prospective sleep diaries and actigraphy between weekends versus weekdays.