Quantifying and modeling the impact of mistimed and irregular sleep on health

• Applicant: Dr. Dorothee Fischer
• Subject Area: Public Health, Healthcare Research, Social and Occupational Medicine ; Epidemiology and Medical Biometry/Statistics
• Term: from 2018 to 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 423917010

• Final Report Year: 2020

Final Report Abstract

While the detrimental effects of short sleep are by now beyond dispute, emerging findings suggest that sleep timing and regularity may be just as important for health as sleep duration. Using two recent metrics – the 'Composite Phase Deviation’ (CPD) and the 'Sleep Regularity Index’ (SRI) – the research project aimed to (i) assess levels of exposure to mistimed and irregular sleep in a US cohort; (ii) quantify a dose-response relationship with health outcomes; and (iii) identify influencing factors and potential interventions using a physiology-based mathematical model. In Objectives I and II, CPD and SRI were calculated from 5-7d actimetry data of 2,121 participants of the “Sueño Ancillary Study”. The Sueño study is an ancillary study of the parent “Hispanic Community Health Study”, a prospective multi-center cohort to define the impact of poor sleep on health. Using complex survey logistic regression, a dose-response relationship was observed between irregular sleep and prevalent hypertension, with an almost two-fold increase in highly irregular sleepers, compared with highly regular sleepers. The relationship was modified by sex and employment status, showing stronger effects in males and unemployed. The findings suggest that irregular sleep may be a risk factor for hypertension, going beyond effects of sleep duration. In an additional project, the association between irregular sleep and mental healthrelated outcomes (i.e., well-being) was examined in a dataset of 223 US college students. Longitudinal growth models with time-varying covariates and hierarchical cluster analysis showed that (i) irregular sleep was associated with poorer average well-being, while short sleep was associated with poorer daily well-being, and (ii) the poorest well-being was reported by students for whom both sleep and event schedules were irregular. The findings suggest that interventions to stabilize sleep and/or scheduled events may help improve well-being. In Objective III, an established physiology-based mathematical model of human sleep and circadian rhythms was used to understand how an individual’s physiology and their social/work constraints interact to generate mistimed and irregular sleep. Schedules of weekly daytime work and rotating shift work were simulated by enforcing wakefulness during work (±commute). Results showed that the previously reported relationship between irregular sleep and a delayed circadian phase is not inherent to ‘late’ individuals but was the result of the interaction between environmental (i.e., self-selected light exposure) and endogenous features (i.e., circadian period). In the course of this fellowship, it became clear that many questions in the field of sleep regularity research were related to its assessment, e.g., how many days and participants are needed for an accurate estimate? In an additional three-part project, we conducted simulation studies, synthetically generating sleep-wake patterns over 2-28 days with daily random variation. Five metrics of sleep regularity (CPD, SRI, standard deviation (StDev), Inter-daily Stability (IS), and Social Jetlag (SJL)) were compared across sources of day-to-day variability (e.g., naps, awakenings, all-nighters, missing data). Results showed that (i) metrics measure sleep regularity on different time scales: global (StDev, IS, SJL) vs. circadian (CPD, SRI); (ii) only a subset of metrics (IS, SRI) can assess fragmented sleep patterns; (iii) simulating ‘all-nighters’ revealed that IS is fundamentally distinct from all other metrics; (iv) global metrics need relatively many days for an accurate estimate, whereas circadian metrics require larger samples; and (v) all metrics were highly sensitive to non-randomly missing data but remarkably stable for up to 50% randomly missing data. The findings suggest that the metrics all measure different aspects of sleep regularity and selecting the ‘right’ metric will depend on study length, sample size, and importantly, on what ‘kind’ of irregular sleep is expected to drive health effects.

Publications

• (2017). Quantitative relationships between novel metrics for assessing sleep regularity. Sleep Med 40:e98-e99
  Fischer, D, Klerman, EB, & Phillips, AJK
  
• (2018). Composite Phase Deviation (CPD) as a predictor of mood in college students. Sleep 41(suppl 1):A129
  Fischer, D, McHill, A, Sano A, Picard, RW, Barger, LK, Czeisler, CA, Klerman, EB, Phillips, AJK
  
• (2018). Determinants of irregular sleep on different work schedules using mathematical modeling. Sleep 41(suppl 1):A165
  Fischer, D, Klerman, EB, & Phillips, AJK
  
• (2019). Anwendung eines bio-mathematischen Modells zur Vorhersage von Schlaf und Fatigue in der Luftfahrt. DGLRM;W2:48-49
  Fischer, D, Aeschbach, D
  
• (2019). Assessing day-to-day regularity of sleep: theoretical and practical implications of available metrics. Somnologie 23(suppl 1):S26-27
  Fischer, D, Klerman, EB, & Phillips, AJK
  
• (2019). Irregular sleep and event schedules are associated with poorer self-reported well-being in US college students. Sleep;zsz300
  Fischer, D, McHill, A, Sano A, Picard, RW, Barger, LK, Czeisler, CA, Klerman, EB, Phillips, AJK