Abstract
Objective
To examine the day-to-day, bidirectional associations of accelerometer-derived sedentary behavior and physical activity (PA) with reported time-in-bed in a large cohort of older women.
Methods
Data are from 10,086 Women’s Health Study participants [aged 71.6 (SD): 5.7] years who agreed to wear an accelerometer and complete a diary for seven, consecutive days. Generalized linear (multi-level) models with repeated measures were used to examine the adjusted associations of: (1) reported time-in-bed with next-day accelerometer-determined counts and time spent sedentary and in light and moderate- to vigorous- intensity PA (MVPA) and (2) accelerometer estimates with reported time-in-bed that night, expressed as short (<7 hours), optimal (7-9 hours), and long (>9 hours) sleep.
Results
Across days, short sleep was associated with an average of 5,500 (SE: 1,352) higher accelerometer counts the following day, but was also related to higher average sedentary [46.5 (SE: 1.5) minutes] and light intensity PA [11.9 (SE: 1.2) minutes] than optimal sleep (all P<0.001). Long sleep was associated with lower accelerometer counts, time spent sedentary and in light intensity PA, and a reduced likelihood of engaging in ≥20 minutes of MVPA (all P<0.001) than optimal sleep. Higher PA during the day (higher accelerometer counts and ≥20 minutes of accumulated MVPA) was associated with a reduced likelihood of reporting short or long sleep that night (all P<0.001).
Conclusions
Findings support the bidirectional associations of accelerometer-determined sedentary behavior and PA with reported time-in-bed in older women. Future studies are needed to confirm findings with sleep actigraphy in older women.
Keywords: accelerometer, epidemiological methods, exercise, sleep, older women
INTRODUCTION
Changes in sleep patterns are thought to be part of the normal aging process.1 While sleep needs remain relatively constant across the adulthood, older adults have more difficulty initiating and maintaining sleep.2 Research has also shown that reduced sleep efficiency in older adults is associated with existing chronic diseases3 and the medications that are used to treat these conditions.4 There is growing evidence to suggest that both short (<6 hours) and long (>9 hours) sleep duration are associated with adverse health outcomes, including total mortality, cardiovascular disease, type 2 diabetes, obesity, and psychiatric illness.5,6
Regular physical activity (PA) has numerous health benefits including a reduced risk of premature death, coronary heart disease (CHD), stroke, hypertension, hyperlipidemia, type 2 diabetes, metabolic syndrome, depression, and maintenance of cognitive function in older adults.7 Yet, older adult women are one of the least active subgroups in the U.S.8 A reciprocal association of PA with sleep has also been hypothesized and evaluated.9-12 Once an association is established, prevention strategies targeting older adult women to improve sleep hygiene may lead to improvements in physical activity levels and vice versa. However, there is a lack of consensus in the literature regarding the day-to-night associations of waking and non-waking behaviors. More specifically, does sleep duration at night impact sedentary behavior and PA the next day and/or does sedentary behavior and PA during the day influence sleep duration that night?
Historically, to examine these associations in population-based research, daily records or accounts of sedentary behavior, PA, and sleep duration were required. However, with recent advancements in technology, examining these associations with accelerometers is now possible. Accelerometers provide direct estimates of waking behaviors, including sedentary behavior and PA. Several strategies have been used to concurrently assess sedentary behavior, PA, and sleep duration using accelerometers to describe 24-hour periods. The first strategy is to use accelerometers that are capable of complete 24-hour monitoring. However, this technology is relatively new and can be challenging given the optimal monitor placement varies by behavior (waist during waking hours; wrist during sleep). The second option is to request participants to wear two monitors during data collection, which increases participant burden and study costs. The third option is to ask participants to wear the monitor only during waking hours and complete a wear time diary where times in and out of bed are recorded. Yet, this option requires participants to record time in and out of bed (rather than simply reporting non-wear time), representing an abbreviated sleep diary for subjective quantification of time-in-bed.
Recent studies examining the bidirectional associations of sedentary behavior and PA with sleep duration reflects these different measurement strategies. Using a single monitor worn on the waist during waking hours and wrist during sleep, Kishida et al. (2016)13 found that an increase in daily physical activity was related to increases in total sleep time that night; however, there was a null association between MVPA and total sleep time among a sample of 103 midlife women. Higher total sleep time was also significantly related to lower MVPA the next day. Studies by Lambiase9 and Mitchell10 used two monitors, worn concurrently, to assess these behaviors. Lambiase et al.9 examined these associations using data collected from the ActiGraph GT1M accelerometer and Actiwatch-64m in older women [n=143; aged 73.3 (1.7) years] and found that a 10% increase in daily accelerometer counts and accumulated moderate to vigorous intensity PA (MVPA) was associated with a 0.04 and 0.02% reduction in sleep duration that night. While statistically significant, the impact of PA on sleep duration was negligible. Mitchell et al.10 used two ActiGraph GT3X+ monitors in 353 women, aged 21-75 years, and found no significant associations of sedentary behavior and MVPA with total sleep time that night or total sleep time with sedentary behavior or MVPA the following day. In a study by Baron et al.14, 11 women (61.3 ± 4.1 years) with insomnia wore an Actiwatch-64 during sleep and completed an exercise log where they reported start time, duration, perceived exertion, and exercise type. Results showed that reported exercise was not significantly associated with sleep time that night. Also, there was no association between total sleep time and reported exercise duration the following day. Given this conflicting research, additional studies are needed to confirm these findings in larger study samples while utilizing clinically meaningful cut-points that reflect the (1) well-established U-shaped distribution15,16 between sleep and mortality and chronic disease and (2) current physical activity guidelines for general health benefit to determine the impact of both short and long sleep on waking behaviors. Further, no previous studies have examined the bi-directional associations of light intensity physical activity and sleep and only the study by Mitchell and colleagues10 explored potential day-to-day relations with sedentary behavior.
The primary objective of this study is to examine the bi-directional, day-to-day associations between accelerometer-derived PA, sedentary behavior, and participant recorded time-in-bed in the Women’s Health Study (WHS), a large, population-based sample of older women. Similar to the assessment strategy employed by Baron et al.14, and because sleep actigraphy data were not collected in the WHS, reported accounts of time-in-bed were used in the current study to estimate sleep duration. Subjective report of time-in-bed from sleep diaries has been shown to be comparable to objective sleep duration measurements.17-20 Also in the current study, the longitudinal structure of the data collected over 7 consecutive days was retained and reported time-in-bed was used as an estimate of sleep duration18, which were further classified as short [<7 hours (h)·night−1], optimal (7-9 hours·night−1), and long (>9 hours·night−1) sleep to reflect National Sleep Foundation recommendations.21 Further, due to the well-established benefits of a PA dose reflecting 150 minutes of MVPA per week on health and well-being22, MVPA was further classified as <20 and ≥20 minutes per day to reflect 2008 Physical Activity Guidelines for Americans.22 These categorical estimates were used specifically to enhance the interpretation of study findings.
PARTICIPANTS AND METHODS
Design overview and study participants
Participants are from the WHS, a completed randomized trial (1994 to 2004) testing the benefits and risks of low-dose aspirin and vitamin E for primary prevention of cancer and cardiovascular disease (CVD) in 39,876 U.S. women aged ≥ 45 years.23,24 When the trial ended, women were invited to continue in an observational follow-up study, and 89% of those alive consented. In 2011, an ancillary study was initiated with the primary goal to examine the associations of accelerometer-determined PA and sedentary behavior on health outcomes.
As of January 1 2014, 26,978 women had been invited to participate in the accelerometer ancillary study and 23,934 (88.7%) responded. Of the 23,934, 1,204 (5.0%) were ineligible because they were unable to walk independently outside the home (eligibility criterion). Among the remaining 22,730 women, 16,689 (73.4%) agreed to participate. Participants provided written informed consent and the study was approved by the Brigham and Women’s Hospital’s institutional review board committee.
Participants included in the ancillary study were mailed an accelerometer and asked to wear it on the hip for 7 days during waking hours. They also completed a wear time diary, indicating which days the monitor was worn, and returned both the accelerometer and diary by mail. Women were then mailed a brief PA questionnaire; inquiring about leisure time PA during the time the monitor was worn. As of May 2014, 14,796 women had returned their accelerometers, which represent those potentially eligible for these analyses.
Data collection
Participant characteristics
WHS participants complete annual questionnaires that ask about socio-demographic characteristics, health behaviors, and medical history. Date of birth, race, and education were obtained from the baseline questionnaire at the start of the WHS trial, while information on weight, height, self-rated health, and smoking for the present analysis were from the 2011 follow-up questionnaire. Data on cancer, CVD, and diabetes were obtained throughout the WHS from self-reports and validated against medical records.
Accelerometer-determined PA and sedentary behavior
Women were mailed a tri-axial accelerometer (ActiGraph GT3X+) and were asked to wear it on the hip, secured by an adjustable elastic belt, for 7 consecutive days during all waking hours. Monitors were initialized to begin data collection one day prior to the estimated delivery date and continued to collect data until data were downloaded. Raw accelerometer data were sampled at 30 Hertz (Hz) and data were reintegrated and expressed as 60-second epochs. ActiGraphs have been extensively validated.25
Accelerometer data, using all three axes, were screened for periods of wear26 and summary estimates were derived using the vertical axis data. Total accelerometer counts (ct·d−1) were calculated using summed daily counts detected over wear periods. Time spent per day (min·d−1) in different intensity levels were estimated using established count threshold values [sedentary (0-99 ct·min−1), light- (100-1951 ct·min−1), moderate- (1952-5724 ct·min−1), and vigorous- (≥5725 ct·min−1) intensity].27 The MVPA estimate included every minute above the 1952 ct·min−1 threshold (accumulated MVPA). For multivariable models, total accelerometer counts were expressed per every 1,000 ct·d−1, sedentary time and light intensity PA estimates were expressed per every 10 minutes, and accumulated MVPA as the proportion of participants <20 or ≥20 min·d−1 as it represents a daily exercise dose that approximates 2008 PA Guidelines.22 Participants who wore the accelerometer <4 days of ≥10 hours of wear per day were excluded from further analysis.
Time-in-bed
Participants recorded the time in and out of bed for sleep in the wear time diary, which were used as our estimate of sleep duration (min·d−1). For multivariable models, categories of time-in-bed were created and defined as short, optimal, and long sleep (<7, 7-9, and >9 h·night−1, respectively) based on National Sleep Foundation recommendations.21
Data management and statistical methods
Two analytic datasets were created to capture the day-to-day, bidirectional associations of sedentary behavior and PA with time-in-bed. The first dataset included each day with the preceding night’s time-in-bed to examine if time-in-bed at night impacts sedentary behavior and PA the next day (i.e., reported time in bed on night i was aligned with sedentary behavior and PA on day i + 1). The second dataset included each day with time-in-bed that following night to evaluate if sedentary behavior or PA during the day influences time-in-bed that night (i.e., sedentary behavior and PA on day i aligned with time in bed on night i).
Descriptive univariate analyses were conducted on measured parameters and all continuous estimates were assessed for normality using histograms and Kolmogorov-Smirnov tests, for each dataset, separately. Normally distributed variables were reported as means and SDs and non-normal distributed variables as medians with 25th and 75th percentiles. Frequencies and percentages were noted for categorical variables. PA, sedentary behavior, and time-in-bed estimates were also described, stratified by day of the week (e.g., Monday). When results were stratified by day of observation of monitor wear (e.g., Day 1), findings were similar (data not shown).
Using each dataset, separate generalized linear models with repeated measures (also known as multi-level models) were constructed to examine the adjusted day-to-day associations of PA and sedentary behavior with categories reflecting the previous night time-in-bed (four models) and that night’s time-in-bed (four models). This modeling strategy incorporates the day-to-day variability within participants (level 1) as well as the variability between participants (level 2).
All models were adjusted for: age (years), education level (<bachelor’s degree, bachelor’s degree, and graduate school), body mass index (BMI; <18.5, 18.5 to 24.9, 24.9 to 29.9, and ≥30 kg/m2) and self-rated health (excellent, very good, good, fair, and poor), which were selected a priori based on established associations with the exposure and/or outcome variables. To examine if time-in-bed categories at night predicted total accelerometer counts, sedentary behavior, and light intensity PA the next day, models with normal link and an autoregressive covariance structure were used. For MVPA categories, a cumulative logit link was used. To determine if total accelerometer counts, sedentary behavior, light intensity PA or MVPA categories during the day predicted time-in-bed categories that night, models with cumulative logit link were also used. For all generalized repeated linear, multi-level models, optimal sleep served as the reference group. All statistical significance tests were two-sided with the type I error level set at P<0.05. All analyses were generated with SAS/STAT software, version 9.4 (Cary, NC).
RESULTS
Of the 14,796 women who returned their accelerometer, 245 were excluded due to technical error with the accelerometer; 1,850 women who did not return a diary, 2,197 women who did not return a PA questionnaire, 389 women who did not have ≥4 days of ≥10 hours of wear per day (convention for valid accelerometer data), and 29 had missing time-in-bed data were also excluded resulting in an analytic sample of 10,086 participants. The mean age and BMI was 71.6 [standard deviation (SD): 5.7] years and 26.1 (SD: 4.9) kg/m2, respectively, and the majority of participants were white (96.7%) (Table 1).
Table 1.
Descriptive Characteristics of Women’s Health Study (n=10,086).
| n | Mean (SD); Median (25th, 75th percentile), or n (%) |
|
|---|---|---|
| Participant Characteristics | ||
| Age | 10,086 | 71.6 ± 5.7 |
| Race, % | 10,013 | |
| White / Non-Hispanic | 96.7 | |
| Non-White | 3.3 | |
| Education, % | 9,917 | |
| <Bachelor’s Degree (e.g., LPN) | 50.2 | |
| Bachelor’s Degree | 26.0 | |
| Graduate Degree | 23.8 | |
| BMI, kg/m2 | 9,982 | 26.1 ± 4.9 |
| BMI, % | 9,982 | |
| Underweight (<18.5 kg/m2) | 2.0 | |
| Normal (18.5 to 24.9 kg/m2) | 45.0 | |
| Overweight (25 to 29.9 kg/m2) | 34.6 | |
| Obese (≥30 kg/m2) | 18.4 | |
| Self-rated health, % | 9,853 | |
| Excellent | 26.8 | |
| Very good | 52.5 | |
| Good | 18.5 | |
| Fair | 2.1 | |
| Poor | 0.1 | |
| Cigarette smoker, % | 10,024 | |
| Yes | 3.2 | |
| No | 96.8 | |
| Self-reported Physical Activity, MET·hr·wk−1 |
10,086 | 17.0 (5.6, 32.2) |
| Self-reported Physical Activity, min·wk−1 | 10,086 | 221.0 (79.0, 423.5) |
| CHD, % | 10,086 | |
| Yes | 2.9 | |
| No | 97.1 | |
| Cancer, % | 10,086 | |
| Yes | 11.0 | |
| No | 89.0 | |
| Type 2 Diabetes, % | 10,086 | |
| Yes | 8.7 | |
| No | 91.3 |
LPN = Licensed Practical Nurse; BMI = body mass index; CHD = coronary heart disease
Time-in-bed predicting PA and sedentary behavior the next day
Table 2 presents the median (25th, 75th percentile) daily and weekly summary PA, sedentary behavior, and reported time-in-bed, including times corresponding to reported time in and out of bed, using the first dataset that includes the previous night time-in-bed with PA and sedentary behavior the next day. Across the 7-days of observation, median recorded time in bed (hh:mm), time out of bed (hh:mm), and total time-in-bed (min·d−1) were consistent among study participants. However, total time in bed (min·d−1) appeared slightly longer on weekend days, which was likely due to getting out of bed a reported 10-15 minutes later than recorded on weekdays. Accelerometer data, including monitor wear time (min·d−1) and time spent sedentary and in light- moderate- and vigorous- intensity (min·d−1) were also consistent across observation days.
Table 2.
Descriptive Statistics for Daily and Weekly Summary Estimates Reflecting Reported Time-in-bed Predicting Physical Activity and Sedentary Behavior the Next Day.a
| Sleep | Sunday to Monday |
Monday to Tuesday |
Tuesday to Wednesday |
Wednesday to Thursday |
Thursday to Friday |
Friday to Saturday |
Saturday to Sunday |
Weekly Summary |
|---|---|---|---|---|---|---|---|---|
| n=8,525 | n=7,240 | n=8,219 | n=8,534 | n=8,059 | n=7,877 | n=7,864 | n=10,086 | |
| Median (25th, 75th %ile) |
Median (25th, 75th %ile) |
Median (25th, 75th %ile) |
Median (25th, 75th %ile) |
Median (25th, 75th %ile) |
Median (25th, 75th %ile) |
Median (25th, 75th %ile) |
Median (25th, 75th %ile) |
|
| Reported time in bed (at night), hh:mm |
22:30 (21:54,23:15) |
22:30 (21:55,23:15) |
22:30 (21:53,23:15) |
22:30 (21:55,23:15) |
22:15 (21:55,23:15) |
22:35 (22:00,23:21) |
22:32 (22:00,23:24) |
22:34 (21.58,23:11) |
| Reported time out of bed (in the morning) hh:mm |
7:00 (6:12,7:45) |
7:00 (6:06, 7:45) |
7:00 (6:05,7:40) |
7:00 (6:10, 7:40) |
7:00 (6:10, 7:45) |
7:15 (6:30,8:00) |
7:10 (6:30,8:00) |
7:01 (6:24,7:42) |
| Total Time-in- bed, min·d−1 |
503 (455, 550) |
500 (454, 545) |
500 (452, 550) |
501 (455, 550) |
505 (455, 550) |
515 (470, 568) |
510 (465, 560) |
506 (469, 545) |
|
Physical
Activity (next day) |
Monday | Tuesday | Wednesday | Thursday | Friday | Saturday | Sunday |
Weekly
Summary |
| Accelerometer Wear Time, min·d−1 |
905 (845, 960) |
908 (847, 963) |
909 (848, 963) |
908 (848, 965) |
907 (846, 964) |
890 (826, 947) |
888 (824, 946) |
900 (850, 942) |
| Total Counts (per 10,000), ct·d−1 |
19.0 (12.6, 27.4) |
18.2 (12.2, 26.4) |
18.2 (12.4, 26.6) |
18.2 (12.2, 26.3) |
18.4 (12.4, 26.8) |
18.1 (12.0, 27.0) |
15.4 (10.3, 23.1) |
18.9 (13.4, 25.6) |
| Sedentary, min·d−1 |
587 (505, 663) |
594 (512, 671) |
593 (512, 672) |
593 (511, 673) |
591 (506, 671) |
577 (492, 654) |
606 (521, 685) |
590 (526, 651) |
| Light intensity, min·d−1 |
291 (228, 362) |
288 (225, 358) |
291 (227, 357) |
288 (223, 359) |
291 (227, 362) |
287 (222, 360) |
259 (198, 327) |
288 (237, 343) |
| Moderate intensity, min·d−1 |
7 (2, 24) |
6 (1, 22) |
6 (2, 22) |
6 (1, 22) |
6 (1, 22) |
6 (1, 21) |
4 (1, 16) |
9 (3, 22) |
| Vigorous intensity, min·d−1 |
0 (0, 0) |
0 (0, 0) |
0 (0, 0) |
0 (0, 0) |
0 (0, 0) |
0 (0, 0) |
0 (0, 0) |
0 (0, 0) |
| Accumulated MVPAb, min·d−1 |
7 (2, 25) |
6 (1, 22) |
6 (2, 22) |
6 (1, 22) |
6 (1, 22) |
6 (1, 21) |
4 (1, 16) |
9 (3, 22) |
ct·d−1 = counts per day; MVPA = Moderate to vigorous intensity physical activity; min·d−1 = minutes per day; %ile = percentile
Reported time-in-bed obtained via recorded times corresponding to time in bed at night and out of bed the next morning; sedentary behavior and physical activity data were assessed via accelerometers.
Accumulated MVPA were computed as every minute ≥1952 counts per minute threshold.
Generalized linear (multi-level) models with repeated measures for time-in-bed predicting PA and sedentary behavior the next day are shown in Table 3. Across observation days, short sleep was associated with an average of 5,500 [standard error (SE): 1,352] higher accelerometer counts the next day compared to optimal sleep (P<0.001), after adjustment for covariates. Further, compared to optimal sleep, short sleep was associated with an average of 46.5 (SE: 1.5) and 11.9 (SE: 1.2) additional min·d−1 of time spent sedentary and in light intensity PA the next day, respectively. There was no difference in the odds of ≥20 min·d−1 of accumulated MVPA with short sleep compared to optimal sleep. After adjustment for covariates, long sleep was associated with an average of 12,680 (SE: 946) fewer total accelerometer counts, 46.5 (SE: 1.5) and 11.9 (SE: 1.2) additional min·d−1 of time spent sedentary and in light intensity PA, respectively, and a reduced odds of engaging in ≥20 min·d−1 of accumulated MVPA the following day (OR: 0.73 (95% CI: 0.67, 0.78) compared to optimal sleep (all P<0.001).
Table 3.
Multilevel, Generalized Linear Models with Repeated Measures for Reported Time-in-bed Predicting Physical Activity and Sedentary Behavior the Next Day.a,b,c
| Physical Activity and Sedentary Behavior | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Total accelerometer counts, per 1,000 ct·d−1 |
Sedentary, per 10 min·d−1 | Light Intensity, per 10 min·d−1 |
MVPA, % < 20 or ≥20 min·d−1 |
||||||||||
| Estimate | Standard Error |
p value | Estimate | Standard Error |
p value | Estimate | Standard Error |
p value | Estimate | Standard Error | p value | ||
| Intercept | 655.1 | 9.5 | <0.001 | 30.1 | 0.96 | <0.001 | 58.1 | 0.81 | <0.001 | 5.1 | 0.24 | <0.001 | |
| Time-in-bed | |||||||||||||
| Estimate | Standard Error |
p value | Estimate | Standard Error |
p value | Estimate | Standard Error |
p value | Odds Ratio |
95% CI | p value | ||
| <7 h·d−1 | 5.5 | 1.4 | <0.001 | 4.7 | 0.15 | <0.001 | 1.2 | 0.12 | <0.001 | 0.94 | 0.86 | 1.03 | 0.16 |
| >9 h·d−1 | −12.7 | 0.95 | <0.001 | −3.4 | 0.11 | <0.001 | −1.6 | 0.08 | <0.001 | 0.73 | 0.67 | 0.78 | <0.001 |
CI = confidence interval; ct·d−1 = counts per day; h·d−1= hours per day; min·d−1 = minutes per day; MVPA = Moderate to vigorous intensity physical activity
Reported time-in-bed obtained via recorded times corresponding to time in bed at night and out of bed the next morning; sedentary behavior and physical activity data were assessed via accelerometers.
Reference category is 7-9 hours of sleep per night.
Models adjusted for: age (years), education (<Bachelor’s Degree, Bachelor’s Degree, and Graduate School), body mass index (BMI) categories (underweight, overweight, obese), self-rated health status (poor, fair, good, very good).
Sedentary behavior and PA predicting time-in-bed that night
Using the second analytic dataset, the univariable results were similar to those previously reported (data not shown). When compared to optimal sleep, higher total accelerometer counts and time spent sedentary or in light intensity PA was associated with a reduced likelihood of short or long sleep that night (all P<0.05), after adjustment for covariates. After adjustment, participants classified in the ≥20 min·d−1 of accumulated MVPA category were 16% less likely to report short or long sleep that night when compared to optimal sleepers (Table 4).
Table 4.
Multi-level, Generalized Repeated Linear Regression Models for Physical Activity and Sedentary Behavior Predicting Reported Time-in-bed that Night.a,b,c
| Time-in-bed | |||||||
|---|---|---|---|---|---|---|---|
| <7 h·d−1 | >9 h·d−1 | ||||||
| Physical Activity and Sedentary Behavior | Estimate | Standard Error | p value | Estimate | Standard Error | p value | |
| Total accelerometer counts, per 1,000 ct·d−1 | Intercept | −1.34 | 0.18 | <0.001 | 0.37 | 0.18 | 0.04 |
| −0.001 | 0.0001 | <0.001 | −0.001 | 0.0001 | <0.001 | ||
| Sedentary, per 10 min·d−1 | Intercept | −1.66 | 0.17 | <0.001 | 0.05 | 0.17 | 0.79 |
| −0.001 | 0.001 | <0.001 | −0.001 | 0.001 | <0.001 | ||
| Light intensity, per 10 min·d−1 | Intercept | −1.50 | 0.18 | <0.001 | 0.20 | 0.18 | 0.26 |
| −0.006 | 0.001 | <0.001 | −0.006 | 0.001 | <0.001 | ||
| MVPA, % < 20 or ≥20 min·d−1 | Intercept | −1.61 | 0.17 | <0.001 | 0.09 | 0.17 | 0.58 |
| Odds Ratio | 95% CI | p value | Odds Ratio | 95% CI | p value | ||
| 0.84 | 0.79, 0.88 | <0.001 | 0.84 | 0.79, 0.88 | <0.001 | ||
CI: confidence interval; ct·d−1 = counts per day; h·d−1= hours per day; min·d−1 = minutes per day; MVPA = Moderate to vigorous intensity physical activity
Reported time-in-bed obtained via recorded times corresponding to time in bed at night and out of bed the next morning; sedentary behavior and physical activity data were assessed via accelerometers.
Reference category is 7-9 hours of sleep per night.
Models adjusted for: age (years), education (<Bachelor’s Degree, Bachelor’s Degree, and Graduate School), body mass index (BMI) categories (underweight, overweight, obese), self-rated health status (poor, fair, good, very good).
DISCUSSION
Findings support the bidirectional, day-to-day associations of accelerometer-determined sedentary time and physical activity with reported time-in-bed in older women. After adjustment for covariates, short sleep was associated with additional time spent sedentary and light intensity PA the next day, while long sleep decreased the likelihood of engaging in health promoting levels of accumulated MVPA the following day. Further, after adjustment for covariates, additional PA during the day (i.e., higher accelerometer counts and ≥20 minutes of accumulated MVPA) was associated with a reduced likelihood of reporting short or long sleep that night.
The technological capabilities of contemporary accelerometers and related methodologies provide unique opportunities to clarify the associations of habitual, lifestyle behaviors within the context of the participant’s home environment. Examining these associations is particularly intriguing and could influence strategies that promote PA and/or good sleep hygiene to improve health in older women. This study contributes to this burgeoning area of research by prospectively exploring the bidirectional, day-to-day associations between waking and non-waking behaviors using cut-points for reported time-in-bed, our proxy measure of sleep duration, that reflect national recommendations.21 Further, while the majority of previous research in this area has specifically focused on relations between MVPA and sleep duration, findings from the current study also provides compelling new information about bidirectional associations with time spent sedentary and in light intensity physical activity.
Does sleep duration at night impact sedentary behavior and PA the next day?
Conceptually, within the context of a 24-hour day, less time sleeping would result in additional waking time, the following day, where an individual can choose to either be inactive or active during more discretionary periods. The current study findings support this notion in that, after adjustment, short sleep was associated with higher total accelerometer counts the following day compared to those reporting optimal sleep. However, based on our findings, these additional accelerometer counts were likely spent sedentary or in light intensity PA, such as housework or self-care. While light intensity activity contributes to overall daily energy expenditure and the promotion of energy balance, the health benefits in relation to risk of mortality or development of chronic disease, are still unknown.7,28 Interestingly, short sleep was not associated with the likelihood of engaging in ≥20 min·d−1 of accumulated MVPA the following day. Therefore, while sleeping less at night results in additional waking time the following day, participants did not spend this extra time engaged in health promoting activity.
Conversely, additional time-in-bed (i.e., long sleep) would conceptually result in less waking time, and perhaps less discretionary time the next day; a notion supported by our findings. When compared to optimal sleep, long sleep was associated with fewer accelerometer counts and less time spent sedentary and in light intensity activity the following day. Long sleep was also associated with a reduced likelihood of ≥20 minutes of accumulated MVPA the following day. These findings suggest that there may be a threshold where the proposed restorative features of sleep (e.g., energy conservation, tissue repair and regeneration)29 does not increase the probability of engaging in health promoting PA the next day. The finding that long sleep at night was associated with less sedentary time is interesting as it contradicts previous hypotheses suggesting that long sleep and/or high sedentary time may be indicators of existing disease30-32, particularly in older adults. However, it is also quite possible that this observed association may have resulted from examining these bidirectional associations within the context of 24-hour cycles (i.e., more sleep = less waking hours with which to be sedentary).33
The findings from the current study contrast those previously reported by Lambiase et al9 and Mitchell et al.10 More specifically, both showed no association between sleep duration and MVPA the next day. However, a recent study by Kishida et al.13 found a similar association such that increased total sleep time was significantly associated with reduced time spent in MVPA the following day. In the current study, we found that short sleep was associated with higher sedentary time the following day; a finding that reflected those reported by Booth et al.34 However, in this study of 48 men and women (aged 21 to 40 years) at risk for type 2 diabetes, short sleep (<6 hours per night) was also statistically significantly associated with 27% fewer daily activity counts (P=0.04) and significantly less time spent in accumulated MVPA (−43 min·d−1); findings that were not observed in our study. In the paper by Mitchell et al.10, total sleep duration was not significantly related to sedentary time in their study sample of adult women (mean age: 55.5 ± 10.2 years).
Does sedentary behavior and PA during the day influence sleep duration that night?
The findings from the current study support the relations of time spent sedentary and active during the day with sleep that night. More specifically, increases in total accelerometer counts and time spent sedentary or in light intensity PA during the day were statistically significantly associated with a lower odds of being classified with short or long sleep that night. Further, if an individual engaged in ≥20 minutes of accumulated MVPA during the day, there were also less likely to report short or long sleep. Yet, it is difficult to discern whether these findings support or conflict with the results reported by Lambiase et al.9 and Kishida et al.13 given that continuous estimates of sleep duration were used in analysis rather than categories reflecting National Sleep Foundation recommendations. Interestingly, Mitchell and colleagues10 found no significant associations of sedentary behavior or MVPA on total sleep time. Among a small sample of women with insomnia (n=11), Baron et al14 also found no statistically significant association between reported exercise duration (i.e., time spent in MVPA) and sleep during the corresponding night; however, this may have been the result of a small sample size and limited statistical power. Interestingly, results of the current study suggested a significant association between higher sedentary time and reduced risk of short or long sleep. This finding lends support to the notion that lifestyle risk behaviors co-occur.35,36 Future studies should further explore if this particular behavioral profile predicts underlying or existing disease.
The strengths of the study included use of a large and well-characterized cohort of older women. Further, accelerometers are capable of detecting movement-based activity across a broad intensity spectrum, which provided the unique opportunity to additionally explore the reciprocal associations of sedentary and light intensity PA with sleep duration. Despite these strengths, there are limitations to consider. First, the analytic sample was predominantly white and participants were originally enrolled in an experimental study, which impacts generalizability. Second, sleep actigraphy measures of sleep duration and quality were not collected in the current study. Instead, times that correspond to time in and out of bed were used to estimate sleep duration. While not objectively-determined, reported time-in-bed has been shown to be an acceptable estimate of total sleep duration.18 Future studies are needed to confirm these findings with sleep actigraphy. Finally, the WHS did not include self-reported measures of sleep complaints (e.g., problems initiating or maintaining sleep), which may have provided additional context to these findings.
CONCLUSIONS
The National Sleep Foundation recommends that adults accumulate 7-9 hours of sleep per night. An estimated 50-70 million adults report chronic sleep or wakefulness disorders.2 Exercise is one of the top 10 strategies recommended by the National Sleep Foundation to promote good sleep hygiene; a recommendation that is largely supported by the findings from the present study.37 In this study, the acute, day-to-day effects were explored. However, PA may play a larger role in reducing the likelihood of chronic sleep deprivation (i.e., sleep debt), which may be more related to mortality and chronic disease. Given that accelerometers are now capable of collecting data over longer periods of observation, future studies should examine the effect of chronic short or long sleep on time spent sedentary and active.
ACKNOWLEDGMENTS
This research was supported by the University of Texas Health Science Center at Houston, School of Public Health Front of the Envelope award (to KPG). The Women’s Health Study supported by National Institutes of Health grants CA154647, CA047988, CA182913, HL043851, HL080467 and HL099355. EJS was supported in part by the Intramural Research Program at the National Institutes of Health, National Institute on Aging. We are grateful to the participants and staff of the Women’s Health Study, to Masamitsu Kamada for helping oversee accelerometer data collection, and to Eunjung Kim for conducting data analyses.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest/Financial Disclosure: No conflicts of interest or financial disclosures were reported by the authors of this paper.
REFERENCES
- 1.National Sleep Foundation [Accessed December 3, 2014];Aging and Sleep. 2014 http://sleepfoundation.org/sleep-topics/aging-and-sleep.
- 2.Institutes of Medicine [Accessed June 13, 2016];Sleep Disorders and Sleep Deprivation: An unmet public health problem. 2006 http://www.ncbi.nlm.nih.gov/books/NBK19960/pdf/Bookshelf_NBK19960.pdf.
- 3.Foley D, Ancoli-Israel S, Britz P, Walsh J. Sleep disturbances and chronic disease in older adults: results of the 2003 National Sleep Foundation Sleep in America Survey. J Psychosom Res. 2004 May;56(5):497–502. doi: 10.1016/j.jpsychores.2004.02.010. [DOI] [PubMed] [Google Scholar]
- 4.Wolkove N, Elkholy O, Baltzan M, Palayew M. Sleep and aging: 1. Sleep disorders commonly found in older people. CMAJ. 2007 Apr 24;176(9):1299–1304. doi: 10.1503/cmaj.060792. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Gallicchio L, Kalesan B. Sleep duration and mortality: a systematic review and meta-analysis. J Sleep Res. 2009 Jun;18(2):148–158. doi: 10.1111/j.1365-2869.2008.00732.x. [DOI] [PubMed] [Google Scholar]
- 6.Cappuccio FP, Cooper D, D’Elia L, Strazzullo P, Miller MA. Sleep duration predicts cardiovascular outcomes: a systematic review and meta-analysis of prospective studies. Eur Heart J. 2011 Jun;32(12):1484–1492. doi: 10.1093/eurheartj/ehr007. [DOI] [PubMed] [Google Scholar]
- 7.U.S. Department of Health and Human Services [Accessed March 19, 2011];Physical Activity Guidelines Advisory Committee Report. 2008 http://www.health.gov/paguidelines/Report/pdf/CommitteeReport.pdf.
- 8.U.S. Department of Health and Human Services. Centers for Disease Control and Prevention [Accessed January 24, 2015];Behavioral Risk Factor Surveillance System. 2013 http://www.cdc.gov/brfss/annual_data/annual_2013.html.
- 9.Lambiase MJ, Gabriel KP, Kuller LH, Matthews KA. Temporal relationships between physical activity and sleep in older women. Med Sci Sports Exerc. 2013 Dec;45(12):2362–2368. doi: 10.1249/MSS.0b013e31829e4cea. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Mitchell JA, Godbole S, Moran K, et al. No Evidence of Reciprocal Associations between Daily Sleep and Physical Activity. Med Sci Sports Exerc. 2016 Jun 9; doi: 10.1249/MSS.0000000000001000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Chennaoui M, Arnal PJ, Sauvet F, Leger D. Sleep and exercise: a reciprocal issue? Sleep Med Rev. 2015 Apr;20:59–72. doi: 10.1016/j.smrv.2014.06.008. [DOI] [PubMed] [Google Scholar]
- 12.Kline CE. The bidirectional relationship between exercise and sleep: Implications for exercise adherence and sleep improvement. Am J Lifestyle Med. 2014 Nov-Dec;8(6):375–379. doi: 10.1177/1559827614544437. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kishida M, Elavsky S. An intensive longitudinal examination of daily physical activity and sleep in midlife women. Sleep Health. 2016;2:42–48. doi: 10.1016/j.sleh.2015.12.001. [DOI] [PubMed] [Google Scholar]
- 14.Baron KG, Reid KJ, Zee PC. Exercise to improve sleep in insomnia: exploration of the bidirectional effects. J Clin Sleep Med. 2013 Aug 15;9(8):819–824. doi: 10.5664/jcsm.2930. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hublin C, Partinen M, Koskenvuo M, Kaprio J. Sleep and mortality: a population-based 22-year follow-up study. Sleep. 2007 Oct;30(10):1245–1253. doi: 10.1093/sleep/30.10.1245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ferrie JE, Shipley MJ, Cappuccio FP, et al. A prospective study of change in sleep duration: associations with mortality in the Whitehall II cohort. Sleep. 2007 Dec;30(12):1659–1666. doi: 10.1093/sleep/30.12.1659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.McCrae CS, Rowe MA, Tierney CG, Dautovich ND, Definis AL, McNamara JP. Sleep complaints, subjective and objective sleep patterns, health, psychological adjustment, and daytime functioning in community-dwelling older adults. J Gerontol B Psychol Sci Soc Sci. 2005 Jul;60(4):P182–189. doi: 10.1093/geronb/60.4.p182. [DOI] [PubMed] [Google Scholar]
- 18.Gubelmann C, Guessous I, Theler JM, Haba-Rubio J, Gaspoz JM, Marques-Vidal P. Trends and determinants of time in bed in Geneva, Switzerland. J Clin Sleep Med. 2014 Oct 15;10(10):1129–1135. doi: 10.5664/jcsm.4116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lockley SW, Skene DJ, Arendt J. Comparison between subjective and actigraphic measurement of sleep and sleep rhythms. J Sleep Res. 1999 Sep;8(3):175–183. doi: 10.1046/j.1365-2869.1999.00155.x. [DOI] [PubMed] [Google Scholar]
- 20.Monk TH, Reynolds CF, Kupfer DJ, et al. The Pittsburgh Sleep Diary. J Sleep Res. 1994 Jun;3(2):111–120. [PubMed] [Google Scholar]
- 21.Hirshkowitz M, Whiton K, Albert S, et al. National Sleep Foundation’s sleep time duration recommendations: methodology and results summary. Sleep Health. 2015;1:40–43. doi: 10.1016/j.sleh.2014.12.010. [DOI] [PubMed] [Google Scholar]
- 22.U.S. Department of Health and Human Services [Accessed October 10, 2008];Physical Activity Guidelines for Americans. 2008 www.health.gov/paguidelines.
- 23.Rexrode KM, Lee IM, Cook NR, Hennekens CH, Buring JE. Baseline characteristics of participants in the Women’s Health Study. J Womens Health Gend Based Med. 2005;9:19–27. doi: 10.1089/152460900318911. [DOI] [PubMed] [Google Scholar]
- 24.Ridker PM, Cook NR, Lee IM, et al. A randomized trial of low-dose aspirin in the primary prevention of cardiovascular disease in women. N Engl J Med. 2005 Mar 31;352(13):1293–1304. doi: 10.1056/NEJMoa050613. [DOI] [PubMed] [Google Scholar]
- 25.Nichols JF, Morgan CG, Chabot LE, Sallis JF, Calfas KJ. Assessment of physical activity with the Computer Science and Applications, Inc., accelerometer: laboratory versus field validation. Res Q Exerc Sport. 2000 Mar;71(1):36–43. doi: 10.1080/02701367.2000.10608878. [DOI] [PubMed] [Google Scholar]
- 26.Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357–364. doi: 10.1249/MSS.0b013e3181ed61a3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Freedson PS, Melanson E, Sirard J. Calibration of the Computer Science and Applications, Inc. accelerometer. Med Sci Sports Exerc. 1998 May;30(5):777–781. doi: 10.1097/00005768-199805000-00021. [DOI] [PubMed] [Google Scholar]
- 28.Lee IM, Shiroma EJ. Using accelerometers to measure physical activity in large-scale epidemiological studies: issues and challenges. Br J Sports Med. 2014 Feb;48(3):197–201. doi: 10.1136/bjsports-2013-093154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Youngstedt SD. Effects of exercise on sleep. Clin Sports Med. 2005 Apr;24(2):355–365. xi. doi: 10.1016/j.csm.2004.12.003. [DOI] [PubMed] [Google Scholar]
- 30.Reid KJ, Martinovich Z, Finkel S, et al. Sleep: a marker of physical and mental health in the elderly. Am J Geriatr Psychiatry. 2006 Oct;14(10):860–866. doi: 10.1097/01.JGP.0000206164.56404.ba. [DOI] [PubMed] [Google Scholar]
- 31.Biswas A, Oh PI, Faulkner GE, et al. Sedentary time and its association with risk for disease incidence, mortality, and hospitalization in adults: a systematic review and meta-analysis. Ann Intern Med. 2015 Jan 20;162(2):123–132. doi: 10.7326/M14-1651. [DOI] [PubMed] [Google Scholar]
- 32.Chau JY, Grunseit AC, Chey T, et al. Daily sitting time and all-cause mortality: a meta-analysis. PLoS One. 2013;8(11):e80000. doi: 10.1371/journal.pone.0080000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Rosenberger ME, Buman MP, Haskell WL, McConnell MV, Carstensen LL. Twenty-four Hours of Sleep, Sedentary Behavior, and Physical Activity with Nine Wearable Devices. Med Sci Sports Exerc. 2016 Mar;48(3):457–465. doi: 10.1249/MSS.0000000000000778. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Booth JN, Bromley LE, Darukhanavala AP, Whitmore HR, Imperial JG, Penev PD. Reduced physical activity in adults at risk for type 2 diabetes who curtail their sleep. Obesity (Silver Spring) 2012 Feb;20(2):278–284. doi: 10.1038/oby.2011.306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Fine LJ, Philogene GS, Gramling R, Coups EJ, Sinha S. Prevalence of multiple chronic disease risk factors. 2001 National Health Interview Survey. Am J Prev Med. 2004 Aug;27(2 Suppl):18–24. doi: 10.1016/j.amepre.2004.04.017. [DOI] [PubMed] [Google Scholar]
- 36.Pronk NP, Anderson LH, Crain AL, et al. Meeting recommendations for multiple healthy lifestyle factors. Prevalence, clustering, and predictors among adolescent, adult, and senior health plan members. Am J Prev Med. 2004 Aug;27(2 Suppl):25–33. doi: 10.1016/j.amepre.2004.04.022. [DOI] [PubMed] [Google Scholar]
- 37.National [Accessed November 11, 2015];Healthy Sleep Tips. 2015 https://sleepfoundation.org/sleep-tools-tips/healthy-sleep-tips.