INTRODUCTION
Lack of sleep is associated with poor health outcomes, such as diabetes, obesity, and cardiovascular disease (CVD)1-5. Despite the importance of sleep, many people in the US report that sleepiness interferes with daily activities6 and fewer than half report awaking feeling rested7. To offset the damaging effects of sleep deficit, researchers have examined improving sleep quality. Contributors to better sleep include a daily routine, physical activity, and exposure to green spaces8,9. Living in a greener neighborhood has been suggested to lower the risk of short sleep duration10 and exposure to nature provides protection from insufficient sleep11. However, less is known regarding overall time outdoors and sleep quality.
Spending time outside is associated with positive health outcomes, including decreased risk of diabetes, obesity, and depression12,13. Some investigations indicate a relationship between time outside and sleep quality. Murray et al.8 reported that the interaction between increased time outdoors and increased physical activity had a positive association with total sleep time8. In addition, time outside in the morning can improve college students’ sleep quality14 and less exposure to daylight influences sleep deficiencies for the elderly15.
The purpose of this study was to assess the association between total time spent outdoors per week and sleep normality. We hypothesize that an increase in time outdoors will increase the odds of normal sleep patterns.
METHODS
Sample
During the summers of 2018 and 2019, data were collected from 735 participants (aged 25–70 years) in Green Heart Louisville’s health study (i.e., Health, Environment, and Action in Louisville—HEAL). HEAL is a non-randomized clinical trial to assess how an intervention of added greenery may affect health, especially risks for and incidence of CVD. Participants completed questionnaires on health status, health behavior, neighborhood characteristics, and demographic information (Supplementary file Figure 1 gives the study design). Due to missing data, the analytic sample for crude analysis and for the regression was 728 and 709, respectively. This study was approved by the University of Louisville’s Institutional Review Board.
Measures
Sleep
Participants reported sleep habits using Patient Health Questionnaire-9 (PHQ-9)16 item 3: ‘Over the past 2 weeks, how often have you been bothered by any of the following problems – trouble falling asleep, staying asleep or sleeping too much?’, with responses ‘not at all’, ‘several days’, ‘more than half the days’ or ‘nearly every day’. This item does not differentiate between insomnia and hypersomnia, but rather includes both as sleeping problems. Therefore, responses were dichotomized as ‘normal sleep’ and ‘non-normal sleep’, where normal included ‘not at all’ responses, and non-normal included all other responses.
Time spent outdoors
Time outdoors per week was reported from participant answers to the question: ‘How much time per week do you spend outdoors in nature?’. Response options were 13 categories that ranged from ‘less than 1 hour’ to ‘more than 16 hours’. Due to small sample sizes in some response areas, the categories were collapsed to five, so that time outdoors responses were analyzed as: ≤4, >4 – ≤8, >8 – ≤12, >12 – ≤16, and >16 hours.
Demographic and other variables
Participants were categorized by several demographic characteristics: age, gender, and race. These characteristics, as well as other variables related to perceptions and behavior, were considered as potential confounders (Table 1). Participants reported self-assessments of overall health, level of bodily pain, feeling safe walking in their neighborhood, regular exercise (over 10 minutes duration), smoking status, typical work area, depression status (PHQ- 9)16, and stress level (Perceived Stress Scale, PSS)17. Lastly, the Perceived Benefits of Nature (PBN)18 questionnaire assessed participant views of nature with higher scores indicating greater perceived benefits.
Table 1
Participant characteristics stratified by sleep status, Louisville, Kentucky, 2018–2019 (N=728)
| Characteristics | Normal sleep (n=296) n (%) or median (IQR) | Non-normal sleep (n=432) n (%) or median (IQR) | p* | Missing |
|---|---|---|---|---|
| Age (years) | 51.0 (37.4–61.1) | 50.9 (38.2–60.3) | 0.788 | 0 |
| Gender | 0.008 | 0 | ||
| Female | 163 (55.1) | 280 (64.8) | ||
| Male | 133 (44.9) | 152 (35.2) | ||
| Race | 0.836 | 7 | ||
| White | 230 (78.8) | 333 (77.6) | ||
| Black | 51 (17.5) | 76 (17.7) | ||
| Other | 11 (3.8) | 20 (4.7) | ||
| Income (US$) | 0.292 | 40 | ||
| <20000 | 61 (22.2) | 109 (26.4) | ||
| 20000–44999 | 75 (27.3) | 130 (31.5) | ||
| 45000–64999 | 71 (25.8) | 85 (20.6) | ||
| 65000–89999 | 36 (13.1) | 51 (12.4) | ||
| 90000–124999 | 23 (8.4) | 31 (7.5) | ||
| ≥125000 | 9 (3.3) | 7 (1.7) | ||
| Hours spent outdoors per week | 0.361 | 3 | ||
| ≤4 | 118 (40.1) | 190 (44.1) | ||
| >4 – ≤8 | 77 (26.2) | 110 (25.5) | ||
| >8 – ≤12 | 42 (14.3) | 41 (9.5) | ||
| >12 – ≤16 | 22 (7.5) | 34 (7.9) | ||
| >16 | 35 (11.9) | 56 (13.0) | ||
| Health | <0.0001 | 1 | ||
| Excellent | 22 (7.4) | 14 (3.25) | ||
| Very good | 128 (43.2) | 96 (22.3) | ||
| Good | 117 (39.5) | 203 (47.1) | ||
| Fair | 23 (7.8) | 98 (22.74) | ||
| Poor | 6 (2.0) | 20 (4.7) | ||
| Bodily pain in past 4 weeks | <0.0001 | 2 | ||
| None | 95 (32.3) | 35 (8.2) | ||
| Very mild | 88 (29.9) | 106 (24.7) | ||
| Mild | 54 (18.4) | 91 (21.2) | ||
| Moderate | 38 (12.9) | 111 (25.9) | ||
| Severe/very severe | 19 (6.5) | 86 (20.1) | ||
| Feel safe to walk neighborhood, day or night | 0.001 | 6 | ||
| Strongly agree | 45 (15.5) | 36 (8.4) | ||
| Agree | 115 (39.5) | 144 (33.4) | ||
| Neither agree nor disagree | 36 (12.4) | 71 (16.5) | ||
| Disagree | 65 (22.3) | 103 (23.9) | ||
| Strongly disagree | 30 (10.3) | 77 (17.9) | ||
| Regular exercise | 0.008 | 8 | ||
| Yes | 194 (66.0) | 239 (56.1) | ||
| No | 100 (34.0) | 187 (43.9) | ||
| Smoker | 0.005 | 39 | ||
| Never | 155 (55.9) | 180 (44.3) | ||
| Former | 56 (18.9) | 78 (19.2) | ||
| Current | 71 (25.2) | 149 (36.5) | ||
| Work area | 0.0021 | 81 | ||
| Mainly outdoors | 32 (11.9) | 45 (11.9) | ||
| Travel to different buildings/sites | 15 (5.6) | 17 (4.5) | ||
| In a motor vehicle | 4 (1.5) | 8 (2.1) | ||
| Mainly indoors | 180 (66.9) | 206 (54.5) | ||
| Unemployed | 38 (14.1) | 102 (27.0) | ||
| Depression | <0.0001 | 0 | ||
| None or minimal | 259 (87.5) | 153 (35.4) | ||
| Mild | 31 (10.5) | 144 (33.3) | ||
| Moderate | 1 (0.3) | 74 (17.1) | ||
| Moderately severe | 4 (1.4) | 38 (8.8) | ||
| Severe | 1 (0.3) | 23 (5.3) | ||
| Stress | <0.0001 | 0 | ||
| Low | 189 (63.8) | 143 (33.1) | ||
| Moderate | 98 (33.1) | 240 (55.6) | ||
| High | 9 (3.0) | 49 (11.3) | ||
| Perceived benefits of nature | 57.0 (49.0–66.0) | 57.0 (47.0–65.0) | 0.972 | 0 |
Statistical analysis
Crude associations were assessed between all variables and the dichotomized sleep variable for 728 participants. When comparing categorical variables with sleep, chi-squared or Fisher’s exact test were conducted depending on the distribution of the variable. When assessing continuous variables with sleep, Mann-Whitney U tests were conducted due to the non-normal distribution of continuous variables.
To determine the association of time spent outdoors per week and sleep, logistic regression was utilized (n=709). Although not all variables were found to be statistically significant in crude analysis, an initial model included each variable. However, variables were retained in the model only if they were identified as confounders using backward elimination. Variables not retained were: age, gender, race, income, smoker status, and work area. Reported results include adjusted odds ratios for normal sleep, with 95% confidence intervals, of categorized time spent outdoors per week, with the reference being ≤4 hours. Associations were considered significant for p≤0.05. Analyses were conducted using SAS (version 9.4).
RESULTS
Participant characteristics stratified by sleep status are shown in Table 1 (see Supplementary file Table 1 for more detailed characteristics). The median age of normal sleepers was 51 years. In crude analysis, normal sleep was more prevalent in females (55.1%) and Whites (78.8%) than in males and Non-Whites. Further, normal sleep was more prevalent in those who spent less time outside per week; perceived better general health, less pain, and a higher sense of safety when walking in their neighborhood; engaged in regular exercise, no smoking, and mainly indoor work; and experienced less severe depression and less stress.
The results of the final logistic regression model produced after backward elimination are shown in Table 2. General health was retained in the model but later removed due to multicollinearity. The main predictor of this analysis was time spent outdoors per week. In the adjusted model, as time spent outdoors increased from ≤4 hours to >4 – ≤8 hours (OR=1.04; 95% CI: 0.65–1.64) and >8 – ≤12 hours (OR=1.17; 95% CI: 0.63–2.17), the odds of normal sleep increased; however, those who spent >12 – ≤16 hours (OR=0.63; 95% CI: 0.31–1.27) or >16 hours (OR=0.83; 95% CI: 0.45–1.53) outdoors had a lower likelihood of normal sleep. No estimates of association between time outdoors and sleep were significant. Likewise, there was no significant trend (p-trend=0.374).
Table 2
Adjusted odds ratios for normal sleep, Louisville, Kentucky, 2018–2019 (N=709)
Compared to those who experienced no bodily pain in the past four weeks, those with any pain had lower likelihood of normal sleep. In fact, those that experience very mild (OR=0.26; 95% CI: 0.15–0.46), mild (OR=0.27; 95% CI: 0.15–0.50), moderate (OR=0.27; 95% CI: 0.14–0.50) or severe/very severe (OR=0.30; 95% CI: 0.13–0.67) pain had significantly reduced odds of normal sleep. There was a significant trend of less pain being associated with normal sleep (p-trend<0.001).
Additionally, compared to those with no or minimal depression, those with mild (OR=0.13; 95% CI: 0.08–0.22), moderate (OR=0.01; 95% CI: 0.01–0.52), moderately severe (OR=0.04; 95% CI: 0.01–0.17) or severe (OR=0.02; 95% CI: 0.01–0.14) depression had significantly lower likelihood of normal sleep. There was a significant trend in the association of depression and sleep, where the odds of normal sleep decreased as depression severity increased (p-trend<0.001).
DISCUSSION
Sufficient sleep has been shown to reduce the risk of adverse health outcomes, including CVD, diabetes, and obesity10. Therefore, exploring predictors of sleep quality is pertinent for improving individual and community health. Consistent with previous research, we found individuals who spent ≤12 hours outside per week had higher odds of normal sleep compared to those who spent little to no time outside. However, our findings differ from previous work indicating time spent in green spaces improves sleep. That is, we did not find a significant relationship between time outdoors and sleep normality, whereas other studies have found associations between greenness and sleep10,11. Thus, future work needs to consider specific characteristics of time outside (e.g. time outdoors may not involve green spaces or allow for an appreciation of them) and to more clearly specify what time outdoors in nature means, as there is no clear evidence to suggest that time spent in non-green outdoor spaces has the same health benefits as time spent in green outdoor spaces. In addition, this study highlighted associations between less bodily pain and greater odds of normal sleep, and between higher levels of depression and lower likelihood of normal sleep, findings consistent with extant literature19.
Limitations
The study has several limitations. First, the cross-sectional design allows for assessment of association but prohibits assessment of temporality. Thus, future work is needed to examine any possibility of causal relationships. Second, selfreported information may be subject to recall bias. Third, few participants reported high stress, leading to imprecise CIs for this association. Despite the limitations, the study provides insights for future work exploring time outdoors and sleep quality.
CONCLUSIONS
Our findings reinforce previous work on the relationship between experiencing depression or pain and sleep normality. The association of time outdoors with sleep indicates a possible threshold effect, in that spending ≤12 hours outdoors per week increased the odds of reporting sleep normality; however, findings for an overall association were non-significant. Future research should seek to explicate possible relationships between sleep and time in any outdoor setting, in greenness, and in specific outdoor activities.
