Boris Dubrovsky Ph.D., CBSM

Psychology of Sleep

What Behaviors Predict Sleep Duration and Quality?

A new study elucidates the roles of activity and sedentary behaviors in sleep.

Posted Apr 25, 2019

Photo by Thu Anh on Unsplash
Source: Photo by Thu Anh on Unsplash

It has become a commonplace to note that sleep, together with exercise and diet, is an essential part of one’s health, performance, and longevity. But how does one improve sleep? It is possible to put more effort into exercise, to pay more attention to one’s food choices, but the very nature of sleep seems to shy away from contrived attention and effort. Sleep is one of the exceedingly few functions that get worse the harder one tries. The way to circumvent this apparent paradox is to view sleep as interconnected with other functions so that changes in certain waking behaviors can lead to changes in sleep. Behavioral sleep health research is making advances to define practical aspects of this synergistic approach to sleep.

On April 17 of this year, expansive research situating sleep within the context of other key lifestyle and health behaviors was published in Sleep, the flagship journal of the Sleep Research Society. Seth Creasy at the University of Colorado, with 11 colleagues from several educational and healthcare institutions, analyzed reports of 75,074 postmenopausal women enrolled in the Women’s Health Initiative Observational Study across 40 clinical centers nationwide. The aim of their analysis was to elucidate the relationship between sleep duration and quality on the one hand and sedentary time and physical activity on the other hand.

From the outset, three important aspects of this study should be noted. First, only postmenopausal women were included, as disturbed sleep and reduced activity are more prevalent after menopause. Thus, participants in this study were at a greater risk for poor sleep due to a specific physiological factor. Second, the authors selected sedentary time and physical activity as health-related “modifiable behaviors,” gently pointing with such terminology towards practical implications of their research. As will be seen from the study findings, treating sedentary time and physical activity as two distinct behaviors, and not just perfect complements of each other is a fruitful and informative method. And third, the very large sample size allowed the authors to control statistically for numerous demographic and health variables that might affect both sleep and activity but were outside of the study scope.

The authors were interested in predicting sleep duration and sleep quality based on the amount of time spent in physical activity and on the amount of sedentary time. Sleep duration was categorized as short (6 hours or less per day), long (10 hours or more), and “referent” or recommended (e.i. “normal”, 7-9 hours). Sleep quality was measured in two ways: 1) as a rating of restfulness categorized into restful (reference or “normal” category), average (e.i., less than restful), and restless; and 2) as a score on an insomnia symptom questionnaire falling below (reference category) or above a predetermined threshold. Reports of physical activity time and frequency, including walking, recreational activities (e.g., jogging, dancing) and housework, gave rise to 5 measures of activity: total, light, moderate, vigorous, and moderate-to-vigorous combined. Sedentary time was defined as the amount of time sitting or lying down for any reason (e.g., with the intention to sleep or to watch TV, etc.) excluding the amount of actual sleep. Just sitting time was also analyzed separately so as to avoid a degree of overlap between lying down awake and short sleep time.

The likelihood of each sleep measure being outside of a reference (“normal” or desirable) category was predicted based on the measures of physical activity and, separately, based on the amount of sedentary/sitting time. In models analyzing the relationship between sleep and activity, sedentary time was controlled statistically, and vice versa, models predicting sleep from a sedentary time controlled for physical activity. Such control procedures allowed the researchers to determine the unique predictive value of either activity or sedentary time as behaviors separate and distinct from each other.

The main statistically significant findings of the study can be summarized as follows. With regards to physical activity, women whose total activity ranged 7.5-17.5 hours per week were 11% less likely to sleep 6 hours or less, mainly due to light activity. Vigorous activity at any amount actually increased the likelihood of short sleep by 6%. Total physical activity in amounts exceeding 7.5 hours a week decreased the likelihood of less-than-restful (average) and restless sleep by approximately 7-15%, with contributions from light, moderate, and, to a lesser extent, vigorous activity. These results suggest that sleep duration and sleep quality relate differently to physical activity. More specifically, more-than-light activity is associated with better quality, but not with the quantity of sleep. Unexpectedly, no measures of activity were predictive of above- vs. below-threshold insomnia symptoms, undermining the often-held assumption of the beneficial role of exercise in combating insomnia as a condition.

With regards to sedentary behavior, a much more consistent pattern emerged. Total sedentary time in excess of 6 hours per day, and sitting time alone in excess of 10 hours per day were significantly related to the increased likelihood of short sleep. When sedentary time exceeded 8 hours, the likelihood of short sleep essentially doubled. Sedentary time above 6 hours also increased the likelihood of average sleep by 11-13% and the likelihood of restless sleep by 38-85%. Further, being sedentary for more than 6 hours and sitting for more than 5 hours per day increased the likelihood of above-threshold insomnia symptoms. For sedentary time, the chances of scoring above the threshold increased by 24-56% (greater likelihood associated with greater sedentary time), and for sitting time alone the increase was only modest 5-6%.

The authors conclude that “the most noteworthy finding of these analyses was the strong and consistent nature of the relationship between higher reported sedentary time and poor sleep health… The associations between sedentary time and sleep outcomes were strongest when calculating sedentary time as time spent lying and sitting compared to using sitting time only.”

Going over the study results and the authors’ conclusions, it appears that the discovered pattern suggests a graded relationship between sleep duration and quality on the one hand and the overall behavioral metabolic activation on the other. The reduction in metabolic activation appears to be a key aspect of sitting and lying down awake and occurs due to postural changes and possibly due to alertness changes. (On average, one is arguably more alert while working at a desk in the middle of the day than while watching a video in bed in the evening, and the brain is known to use up to 40% of the energy supply.) Consistently with the well-established physiology of the homeostatic sleep drive being strengthened by energy expenditure, the study indicates that the graded behavioral reduction in energy expenditure during wake is associated with the graded worsening of sleep. Regarding activity, the study appears to underscore a positive relationship between sleep and light physical activity, which is typically more protracted in time and may be also associated with other sleep-benefitting features, such as being outdoors. Thus, without negating other benefits of vigorous exercise, more consistent and prolonged light activity as a replacement for sedentary time may be more helpful for sleep in the long run.

In a less speculative and more cautious scientific tone, Dr. Creasy and his colleagues state that “it is possible, and likely, that physical activity and sedentary behavior have a bidirectional relationship with sleep quantity/quality.” This statement can be further contextualized within the notion that one cannot do anything directly to prolong or deepen sleep, but can, as the authors subtly suggest, modify activity levels and sedentary time. It is noteworthy that behavioral techniques developed specifically for insomnia rely primarily on taking wakefulness out of bed and regimenting the time in bed. Thus, the outcomes of this study not directly targeting insomnia patients are generally in agreement with the clinical approach.

In summary, in this very large sample of respondents, all of whom were already at a greater risk for poor sleep because of physiological changes related to menopause, better sleep duration and quality and fewer insomnia symptoms were associated with less sedentary time and more time spent in light physical activity, as opposed to sitting and especially reclining awake. This pattern was not due to demographic or health-related factors that were controlled.

The take-home message is that physical activity as well as sitting and reclining awake can be behaviorally modified to help promote healthy sleep. Being mindful of consistent physical engagement (e.g., walking and taking stairs whenever possible, considering a stand-up desk or taking brief walking/stretching breaks from deskwork) is one possible approach. Not using a bed for reclining with electronics and generally minimizing lying down awake, especially during times not allotted for sleep, is another potentially beneficial behavior change. Certainly, physical activity in the middle of the night should be avoided, but consistently scheduling a block of time during the day for fitness is also a welcome addition – not for tonight or tomorrow night, but for months and years of good sleep and good health.


Creasy, S. A., Crane, T. E., Garcia, D. O., Thomson, C. A., Kohler, L. N., Wertheim, B. C., Baker, L., Coday, M., Hale, L., Womack, C. R. Wright, K. P., Melanson, E. L. (2019). Higher amounts of Sedentary Time are Associated with Short Sleep Duration and Poor Sleep Quality in Postmenopausal Women, Sleep, zsz093,