Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Dec 2.
Published in final edited form as: J Sleep Res. 2021 Feb 15;30(5):e13297. doi: 10.1111/jsr.13297

Day-to-day associations between sleep characteristics and affect in community dwelling adults

Patricia M Wong 1,2, Brant P Hasler 3, Thomas W Kamarck 4, Aidan G C Wright 4, Martica Hall 3, Mary A Carskadon 1,2, Lei Gao 5,6, Stephen B Manuck 4
PMCID: PMC8637582  NIHMSID: NIHMS1757695  PMID: 33588521

Summary

Despite the high co-occurrence of sleep and mood disturbances, day-to-day associations between sleep characteristics (sleep duration, continuity, and timing) and dimensions of mood (positive affect and negative affect) remain unclear. The present study aimed to test whether there is a daily, bidirectional association between these sleep characteristics and affective states, while addressing methodological limitations in the extant literature by using actiography and ecological momentary assessment methods. Participants were community dwelling, midlife adults (aged 30–54 years, N = 462, 47% male) drawn from the Adult Health and Behavior Project-Phase 2 study. Participants’ sleep patterns were assessed with actiography over a 7-day monitoring period, and on 4 of those days, participants completed an ecological momentary assessment protocol that included hourly assessments of positive affect and negative affect during their wake intervals. Using hierarchical linear modelling, we tested whether participants’ sleep characteristics on a given night predicted next-day affect and vice versa. We also explored whether nocturnal sleep characteristics would differentially associate with affect at different times of day (morning, afternoon, and evening) while controlling for multiple health behaviours. We found that when participants reported higher positive affect on a given day, they slept later that night (B = 0.22, p = .010). Although we found no other statistically significant associations in our primary analyses (all p > .05), we found several sleep-affect associations specific to time of day (B ranges: 0.01–0.18, all p ≤ .02), which warrants further study. Overall, our findings suggest that healthy adults may be resilient to daily fluctuations in their sleep and mood.

Keywords: affect, daily variability, intra-individual variability, mood, sleep

1 |. INTRODUCTION

Sleep is vital for processing and regulating emotions, and insomnia is associated with increased risk of depression (Hertenstein et al., 2019; Li et al., 2016). However, the temporal relationship between specific sleep features and affective states remains unclear. Experimental studies show that shortened sleep duration, advanced or delayed (i.e. shifted) sleep timing, and poor sleep continuity caused by frequent awakenings lead to more negative and less positive mood the following day (Dinges et al., 1997; Kahn et al., 2014; Taub & Berger, 1976). On the other hand, physiological and cognitive arousal, both of which can stem from anxious and depressed mood, can disrupt sleep (Tang & Harvey, 2004; Zoccola et al., 2009). With the onset of either sleep or mood disturbances, such a bidirectional feedback loop could contribute to the development and maintenance of syndromal sleep and mood symptoms. Understanding the naturally occurring, day-to-day association between sleep characteristics and affective states may shed light on the co-development of sleep and mood problems and thereby inform targeted interventions.

Observational studies of healthy adults have revealed mixed findings about the sleep–affect association with regard to the associations between specific sleep characteristics and affective states (positive affect [PA] and negative affect [NA]), and the directionality of these associations. For instance, some studies found when individuals slept less on a given night, they reported less PA (de Wild-Hartmann et al., 2013; Sonnentag et al., 2008) and/or more NA the follow day (Brissette & Cohen, 2002; Galambos et al., 2009). However, other findings were inconsistent or non-significant (Kalmbach et al., 2014; Totterdell et al., 1994). Studies on self-reported sleep quality and continuity in adults have more consistently found associations with PA and/or NA (Brissette & Cohen, 2002; de Wild-Hartmann et al., 2013; Mccrae et al., 2008; Scott & Judge, 2006; Totterdell et al., 1994), although among studies of adults, only one included actiography-derived sleep continuity and found it unrelated to affect (de Wild-Hartmann et al., 2013). Collectively, the association among sleep duration, continuity, and affect remains unclear.

Methodological limitations in the extant literature may explain the inconsistent and mixed results. First, the majority of naturalistic studies have relied on self-reported sleep assessments. Research shows that subjective accounts of sleep duration and continuity are only modestly correlated with objective, polysomnography measures, and can be confounded by depressed mood and personality factors (Matthews et al., 2018; Silva et al., 2007). An alternative, validated approach to assess sleep patterns is actiography. Actiography is a wrist-worn ambulatory tool that quantifies sleep based on activity counts and reflects the behavioural patterns that are correlated with, but distinct from self-report measures (Girschik et al., 2012; Matthews et al., 2018; Sadeh, 2011).

Most previous studies have also relied on once-a-day, retrospective accounts of mood, which are subject to recall bias and can be influenced by participants’ mood at the time of report (Kihlstrom et al., 2000). An alternative to retrospective report is the use of ecological momentary assessments (EMA), a method in which researchers use repeated sampling strategies to assess a given phenomenon at or close to the moment that it occurs, while participants are in their natural setting (Shiftman et al., 2008; Stone & Shiftman, 1994). By using EMA to assess affect, researchers can maximise ecological validity of their assessments (Shiftman et al., 2008). Only a few existing studies have used EMA and, to our knowledge, no studies of adults have used a combined approach of EMA with actiography.

Fewer studies have considered a bidirectional association and tested whether daily affect influences nocturnal sleep characteristics. This evidence is similarly inconsistent and limited by aforementioned methodological considerations (Brissette & Cohen, 2002; Kalmbach et al., 2014), while others suggest no association (de Wild-Hartmann et al., 2013; Galambos et al., 2009; Totterdell et al., 1994). Finally, with the exception of one (Totterdell et al., 1994), studies have primarily focussed on sleep duration and continuity and have not examined the association between daily sleep timing and affect.

In the present study, we aimed to address the gaps in the literature by examining the day-to-day, bidirectional relationship between sleep characteristics (sleep duration, timing, and continuity) and affect (PA and NA) in a sample of midlife, working adults while using a combination of EMA and actiography assessments. We hypothesised that shorter sleep duration, less sleep continuity, and later sleep timing on a given night would predict lower levels of PA and higher levels of NA the following day. We also hypothesised that the sleep-affect relationship would be bidirectional, such that lower levels of PA and higher levels of NA during the daytime would associate with shorter sleep, less sleep continuity, and delayed sleep timing that same night.

While previous studies have focussed on daily affect levels, it is important to note individuals show a 24-hr rhythm in PA (Clark et al., 1989; Murray et al., 2002) and possibly in NA (Emens et al., 2020; Miller et al., 2015). Given this diurnal pattern in affect and that mood states reflect an interaction between the circadian system and length of time spent awake (Boivin et al., 1997), the influence of sleep on affect and vice versa may differ based on the time at which affect is assessed. In this same community sample of working adults, we previously found participants exhibited a diurnal pattern in both PA and NA (Miller et al., 2015). In the present study, we explored whether participants’ sleep characteristics associated with their next-day affective states differentially by time of day and, conversely, whether affective states experienced at particular times of day associated with sleep characteristics that night.

2 |. METHODS

2.1 |. Participants

Participants were 490 midlife men and women from the Adult Health and Behavior Project-Phase 2 (AHAB-II), a study of psychological, behavioural, and biological risk factors for subclinical cardiovascular disease in healthy individuals. Participants were recruited between March 2008 and October 2011 through mass mailings of recruitment letters to individuals randomly selected from voter registration and other public domain lists. Participant informed consent was obtained in accordance with the guidelines of the University of Pittsburgh Institutional Review Board (IRB).

To be eligible to participate in the AHAB-II, individuals had to be between the ages of 30 and 54 years and working ≥25 hr/week outside the home (this latter restriction due to a substudy focussing on occupational stress). Individuals were excluded from participation if they: (a) had a history of clinically apparent cardiovascular disease, schizophrenia or bipolar disorder, chronic hepatitis, renal failure, neurological disorder, lung disease requiring drug treatment, or Stage 2 hypertension (systolic/diastolic blood pressure ≥160/100 mmHg); (b) consumed alcohol ≥5 portions, 3–4 times/week; (c) used fish oil supplements (because of the requirements for another substudy); (d) were prescribed use of insulin, glucocorticoid, anti-arrhythmic, antihypertensive, lipid-lowering, psychotropic, or prescription weight loss medications; (e) were pregnant; or (f) were shift workers. Participants signed an IRB-approved informed consent agreement when enrolled and received compensation up to $410 (American dollars), depending on extent of participation in study visits and protocol compliance.

2.2 |. Procedure

As part of the larger AHAB-II study, participants completed seven laboratory visits designed to gather a wide range of information including psychosocial, behavioural, biological, neuropsychological, and neuroimaging data. Over the course of these visits, participant demographics, self-reported health behaviours, and baseline depression scores were collected and are included as covariates in the present study (see below).

As shown in Figure 1, participants completed a field (i.e. non-laboratory) monitoring session between Visits 2 and 3. During this time, data regarding subjects’ daily sleep, affect, and social experiences were collected. Actiography data were collected for 7 days (see below) to assess sleep; 4 of these days were EMA monitoring days, and included 3 workdays and 1 non-workday. On EMA monitoring days, participants were instructed to indicate when they awoke using a personal digital assistant (PDA; Palm Z22, software: Satellite Forms). The PDA then prompted participants at hourly intervals, set from time of awakening, to complete a 43-item questionnaire. This questionnaire contained affect and health behaviour items described below. Participants received extensive training and practice using the PDA and received feedback on adherence following a practice day. Additionally, participants were telephoned four times throughout their EMA monitoring period and offered technical support as needed. During this monitoring period, other data including ambulatory blood pressure and saliva samples for measurement of cortisol were collected, but are not analysed for the present study.

FIGURE 1.

FIGURE 1

Open in a new tab

Representative actiography and ecological momentary assessment (EMA) monitoring period. EMA refers to the ecological momentary assessments protocol. Actiography data were collected across each night

2.3 |. Measures

2.3.1 |. Positive and Negative Affect

Participants were administered for the EMA an adapted version of the Positive Affect Negative Affect Schedule-Short Form (PANAS-SF; Thompson, 2007) on an hourly basis throughout their waking phase. In this version, participants rated 13 affect items, each on a 6-point scale. Based on our principal component analysis performed on previous samples, several items with low factor loadings on their respective scales (“ashamed”, “active”, and “alert”) were deleted. Other items were added to the scales: “happy” and “cheerful” from the Profile of Mood States scale (POMS; McNair et al., 1981) to assess high arousal PA states, and “angry”, “lonely”, and “sad” from the PANAS-X for better representation of anger and sadness. The resulting survey is represented in Table 1.

TABLE 1.

Adapted positive affect negative affect scale

Positive affect (PA)
 Inspired
 Determined
 Attentive
 Happy
 Cheerful
Negative affect (NA)
 Upset
 Hostile
 Nervous
 Afraid
 Angry
 Lonely
 Sad
Open in a new tab

In our primary analyses, we focussed on daily affect levels quantified as the average of all momentary reports of PA and NA within a given day. For our exploratory analyses, we separately examined PA and NA levels within three time periods in a given day (wake to 12:00 hours, 12:00–18:00 hours, and 18:00 hours to bedtime).

2.3.2 |. Sleep characteristics

Participants wore the Actiwatch-16 (Bend, OR: Philips Electronics), a wrist accelerometer that samples movement several times per second. Throughout a 7-day period that overlapped with the 4-day EMA monitoring period, participants wore the Actiwatch 24-hr a day and were instructed to keep the watch on even when showering. This period included at least 1 night preceding a free (i.e. non-work) day to capture differences between sleep intervals preceding work and free days. Data were saved in 1-min epochs and scored with Actiware software (v5.59) using automated, standard medium thresholds: Sleep onset was defined as a period lasting at least 10 consecutive min with <40 counts of activity (i.e. movement) per epoch. Wake onset was defined as 10 consecutive min of ≥40 activity counts per epoch.

Sleep duration was quantified as the total time between sleep onset and wake onset. Total sleep time was determined as the total amount of time within a given sleep interval that was scored below the wake threshold described above. Sleep efficiency, an estimate of sleep continuity, was calculated as the percentage of the total rest interval that was scored as total sleep time minus non-sleep time. Finally, sleep timing was calculated as the midpoint between sleep onset and wake onset times.

To test the daily relationships between sleep and affect, we examined the participant’s sleep characteristics on the night preceding and following each EMA monitoring day. To consider individual differences in sleep characteristics, we averaged each of the three sleep characteristics across all available actiography data and included these as covariates in the models.

2.3.3 |. Covariates

Demographics

Participant self-reported age, sex, race/ethnicity, marital status, and work status (full-versus part-time) were included in all models.

Depressive Symptomatology

In order to adjust for possible individual differences in baseline mood, depressive symptomatology was measured using the Center for Epidemiological Studies-Depression (CES-D) scale (Radloff, 1977) and included in all models. This 20-item measure assesses how frequently subjects report experiencing a range of psychological and physical symptoms of depression during the past week. Responses are on a 4-point scale ranging from 0 (rarely or none of the time [<1 day]) to 3 (most or all of the time [5–7 days]). Higher scores indicate more severe depressive symptomatology, with a maximum score of 60. To avoid confounding sleep problems and depression symptoms, the total score minus the one sleep-related item score was used; hence the maximum score was 57.

2.4 |. Health behaviours

The following were included as covariates in all models: smoking status (0, “never or past smoker”; 1, “current smoker”), alcohol intake (average number of alcoholic beverages that participants reported consuming per week), and physical activity (average weekly kilocalories; Paffenbarger et al., 1978). Physical activity was assessed via the Paffenbarger Physical Activity Questionnaire, a widely used instrument for estimating weekly kilocalories expended (Paffenbarger et al., 1978) that queries self-reported activities of daily living (e.g. stairs climbed, blocks walked), and leisure activities requiring physical exertion (e.g. sports, recreational pursuits). We used this questionnaire to reference average weekly levels of physical activity, as experienced over the past year. This instrument has convergent validity with several objective measures of physical activity and fitness (Choo et al., 2010; Nowak et al., 2010).

Each hourly EMA prompt also asked participants to report any cigarette, drug, alcohol, and caffeine use in the past 30 min. The exploratory models that focussed on hourly measures of affect included these hourly-assessed health behaviours as additional covariates.

2.5 |. Statistical Analyses

Statistical analyses were performed with SPSS®, version 26.0 (IBM Corporation) for Windows®/Apple Mac®. Prior to testing, study variables were examined for outliers and to verify assumptions of normality. Table 2 shows the levels of data in the present study. Hierarchical linear modelling (i.e. mixed-effects models) was used to analyse this nested data. Regarding covariates, all analytical models included participant demographics, baseline depressive CES-D score, participants’ average sleep characteristics (duration, continuity, and timing) and health behaviours at the Between-Person, individual difference level. Workday status (workday versus non-workday) was also included as a covariate at the Within-person, Daily level.

TABLE 2.

Study variables organised by model levels

Analyses 1 and 2 Exploratory analyses
Within-person, within-day Level 1
 Positive affect (PA)
 Negative affect (NA)
 Momentary health behaviour
 Time of assessment
Within-person Level 1 Level 2
 PA (daily average)
 NA (daily average)
 Sleep duration
 Sleep continuity
 Sleep timing
 Workday status
Between-Person Level 2 Level 3
 Demographic covariates
 Health behaviour covariates
 Sleep duration (average)
 Sleep continuity (average)
 Sleep timing (average)
Open in a new tab

Momentary Health Behaviours refers to self-reported alcohol, caffeine, drug, and cigarette use at each momentary (hourly) interval. Workday Status indicates whether the date of assessment was a work or non-workday. Demographic covariates refer to: age, sex, race, and baseline depression. Health Behaviour Covariates at the Between-Person level refers to average physical activity, alcohol use, and smoking status.

We first tested baseline models to examine the associations between covariates and affect (PA and NA), and between covariates and each sleep characteristic (duration, continuity, and timing). Next, in Analysis set 1, we used separate two-level models to test whether each sleep characteristic on a given night predicted PA or NA the following day. In Level 1 of these models, the sleep characteristic score was person mean-centred in order to isolate the effect of within-person changes in sleep on mood day-to-day. Level 2 included the participant’s average corresponding sleep characteristic score to model possible associations between individual differences in affect and sleep.

In Analysis set 2, we tested separate two-level models to examine whether PA or NA levels on a given day associated with each sleep characteristics that following night. In Level 1 of these models, the affect term was person mean-centred, and the relevant sleep characteristic from the night prior was included to adjust for possible lag effects. Level 2 included the participant’s average corresponding affect value to model possible associations between individual differences in affect and sleep.

Exploratory analyses involved testing three-level models to examine affect across multiple time points of each day. Separate analyses were performed for PA and NA for three time bins: morning, wake to 12:00 hours, afternoon, 12:00–18:00 hours, and evening, 18:00 hours to sleep time. When testing the effects of sleep on affect, each time interval measure of affect was included in the model as a separate outcome data point. When testing the effects of affect on sleep, the average PA and average NA level for each time bin (e.g. average PA in the morning) was calculated and used as a predictor in separate models.

3 |. RESULTS

3.1 |. Participant characteristics

Of the 490 participants, 22 were missing actiography data. Among these 22 participants, four also had missing electronic diary data. In addition, six participants were missing baseline depression scores (CES-D scale). As a result, a total of 462 participants were included in all primary analyses and included 1,747 total observations in which a participant had both sleep data on a given night and affect data for the next day (for Analysis set 1). Regarding Analysis set 2, 1,684 total observations had both affect data for a given day and sleep data that same night. The distributions of sleep duration and sleep midpoint in both sets of data had moderate kurtosis (kurtosis range: 4.4–7.1). To address this, we excluded 2% of the observations (28 for Analysis set 1 and 35 for Analysis set 2) that included extreme outliers of sleep duration and/or sleep midpoint data (>3 SD from the average of the total sample). After these exclusions, Analysis set 1 included 1,719 observations and Analysis set 2 included 1,649.

Table 3 lists participant characteristics and bivariate correlations of each characteristic with sleep duration, sleep midpoint, and sleep efficiency. Values represent averages across all participants, with correlations referring to between-person correlations. Participants completed 4 days (3 work, 1 non-work) of EMA monitoring. Regarding the actiography monitoring, 51% of participants completed 7 nights, 20.3% 1–6 nights, and 39% 8–11 nights. Intra-class correlations (ICC) analyses showed that within-person variability accounted for 47.0% of the total variance in sleep midpoint, 40.7% in sleep efficiency, and 19.3% in sleep duration.

TABLE 3.

Participant characteristics

Correlations with sleep characteristics
Variable Mean (SD) or % Duration Midpoint Efficiency
Demographics
 Age 42.7 (7.3) −0.01 −0.17** −0.03
 Sex 47.1 Male 0.18** −0.10* −0.17**
 Race 81.8 White −0.08 0.06 −0.12*
 Education, years 17.0 (2.9) 0.11* −0.05 0.10*
 Family income 17.9 >110,000
 Employment status 89.6 Full-time −0.00 0.08 −0.01
 Marital status 62.9 Married −0.07 0.23** −0.18**
Average sleep characteristics
 Chronotype 39.3 (7.1) 0.08 −0.54** 0.01
 Bed time 11:42 hours (1 hr 12 min) −0.46** 0.93** −0.16**
 Wake time 06:31 hours(1 hr 8 min) 0.31** 0.92** 0.01
 Sleep midpoint 03:06 hours (1 hr 4min) −0.09* – −0.09
 Sleep duration 6.8 hr (54 min) – −0.09* 0.22**
 Sleep efficiency, % 83.1 (5.4) 0.22** −0.09 –
 PSQI total 5.1 (2.7) −0.19** 0.06 −0.13**
Depression and average levels of affect
 CES-D Total 8.5 (7.9) 0.00 0.09 −0.03
 Positive affect 4.0 (0.7) −0.01 −0.01 −0.06
 Negative affect 1.9 (0.7) 0.07 0.01 −0.07
Health behaviours
 Physical activity, kilocal/day 2,782.0 (2096.7) 0.00 −0.14** −0.14**
 Alcohol intake, drinks/week 3.1 (4.5) −0.03 0.08 −0.03
 Smoking status 13.5% Smoker −0.06 0.22** −0.12*
Open in a new tab

CES-D, Center for Epidemiological Studies-Depression; PSQI, Pittsburgh Sleep Quality Index.

For categorical variables, point bi-serial correlations were conducted with the described category of each variable serving as the comparison group. All other correlations refer to Pearson bivariate analyses.

*

p < .05

**

p < .01.

In our sample, participants tended to rate high on PA, low on NA, and they showed moderate variability in affect as assessed via EMA at hourly intervals throughout the day: Participants reported an average PA of 4.0 (SD = 0.7; top tertile of sample >4.3, bottom tertile <3.6), and an average NA of 1.9 (SD = 0.7; top tertile >2.2; bottom tertile <1.3). ICC analyses showed that within-person variability accounted for 77.7% of the total variance in PA and 87.2% in NA.

3.2 |. Nocturnal sleep and Next-day affect

Regarding the association with covariates, greater baseline depression was associated with less PA (B = −0.03, p < .001) and more NA (B = 0.03, p < .001), and non-work days were associated with greater PA (B = 0.09, p < .001) and less NA (B = −0.07, p < .001) relative to workdays. Other health behaviours and demographic characteristics were unrelated to PA and NA (all p > .05). All results presented below include baseline depression as a covariate, but results are consistent even after omitting this variable.

Day-to-day variation in sleep midpoint, sleep duration, and sleep timing did not significantly predict next-day PA or NA (all p > .05). There were significant individual differences in the effects of sleep midpoint on affect. Some participants had weaker and some had stronger associations between sleep midpoint and PA (B range: 0.00–0.02, Wald Z = 2.22; p = .026), as well as NA (B range: 0.00–0.01, Wald Z = 3.54; p < .001). While there were statistically significant individual differences in the effect of sleep efficiency on NA, the variance estimates were negligible (estimates = 0.00, Wald Z = 2.34, p = .019).

3.3 |. Daytime affect and subsequent nocturnal sleep

Regarding the association with covariates, older age was associated with earlier sleep midpoint (B = −0.02, p < .001), current smokers tended to have later sleep midpoints compared to non-smokers (B = 0.27, p = .002), and participants tended to sleep later on non-work days than work days (B = −0.19, p = .002). Women slept longer (B = 0.40, p <.001) and had greater sleep efficiency (B = 1.36, p = .001) relative to men, while non-White participants slept less (B = −0.24, p = .004) and had poorer sleep efficiency (B = −1.08, p = .009) than White participants. All three sleep characteristics on a given night were positively correlated with the corresponding sleep characteristic the preceding night (B range: 0.10–0.48, all p < .001).

Day-to-day shifts in PA were associated with sleep midpoint, such that an increase in PA on a given day predicted a later sleep midpoint that night (Table 4; B = 0.22, 95% confidence interval [CI] 0.06–0.39; p = .010). There were no significant effects of PA on sleep duration or sleep efficiency, or any effects of NA on the three sleep characteristics (all p > .05). Participants did not show significant individual differences in the effect of PA or NA on any of the sleep characteristics (all p > .05).

TABLE 4.

Higher levels of positive affect associates with later sleep midpoint

B 95% CI lower limit 95% CI upper limit SE df t p
Fixed effects
 Level 1: Within-Person, daily
  Workday status −0.21 −0.33 −0.09 0.06 1,142.53 −3.34   .001
  Previous midpoint 0.48 0.44 0.52 0.02 764.59 22.15 <.001
  Centred PA 0.22 0.06 0.39 0.09 192.30 2.62   .010
 Level 2: Between-Person
  Intercept 14.94 13.69 16.18 0.64 610.10 23.51 <.001
  Age −0.02 −0.02 −0.01 0.00 219.83 −3.82 <.001
  Sex −0.05 −0.17 0.07 0.06 218.46 −0.81   .418
  Race 0.06 −0.06 0.18 0.06 249.09 0.93   .353
  Alcohol 0.00 −0.01 0.02 0.01 242.20 0.66   .509
  Smoking status 0.28 0.11 0.46 0.09 234.45 3.21   .002
  Physical activity −1.94E–5 −4.75E–5 8.66E–6 1.43E–5 232.41 −1.36   .174
  Baseline depression 0.00 −0.01 0.01 0.00 2,115.94 0.46   .644
  Average PA 0.07 −0.01 0.16 0.04 219.52 1.73   .085
Estimate 95% CI lower limit 95% CI upper limit SE Wald Z p
Random effects
 Residual 1.00 0.91 1.10 0.05 19.80 <.001
 Intercept 0.10 0.05 0.23 0.04 2.43   .015
 Centred PA 0.27 0.09 0.80 0.15 1.80   .072
Open in a new tab

Centred PA, Person-centred positive affect; PA, positive affect.

Baseline Depression refers to the Center for Epidemiological Studies-Depression (CES-D) scale total score minus the sleep-related item. Workday Status indicates if the participant was working on the day of sleep assessment. Previous Midpoint refers to the participant’s sleep midpoint the preceding night.

3.4 |. Associations between sleep and affect across different times of day

When participants slept later than their average sleep time on a given night, they reported greater NA the following morning (before 12:00 hours; B = 0.05, 95% CI 0.02–0.08; p = .004) and afternoon (12:00–18:00 hours; B = 0.04, 95% CI 0.01–0.07; p = .009). When participants had greater sleep efficiency on a given night, they reported higher levels of PA the following morning (before 12:00 hours; B = 0.01, 95% CI 0.00–0.02; p = .001). Finally, when participants slept longer than their average sleep duration, they reported greater NA the next evening (after 18:00 hours; B = 0.02, 95% CI 0.00–0.04; p = .020).

Greater PA in the evenings associated with later sleep midpoint (B = 0.18, 95% CI 0.08–0.29; p = .001). The association between less NA in the evening and longer sleep duration that night just crossed the threshold of significance (B = −0.22, 95% CI −43 to 0.00; p = .049).

4 |. DISCUSSION

The aim of the present study was twofold: first, to examine whether there is a proximal, bi-directional relationship between sleep characteristics and affect; and second, to explore whether there are distinct sleep-affect associations at different times of day. Regarding our primary analyses, we found that a greater increase in a person’s daily PA was related to a later shift in their sleep midpoint. In contrast to our hypotheses, we found that neither sleep midpoint, sleep duration, nor sleep efficiency on a given night predicted next day average affect. We found that PA during the daytime was unrelated to sleep duration and sleep efficiency, and NA was unrelated to all sleep characteristics. While our primary analyses largely revealed non-significant sleep-affect associations, results from our exploratory analyses suggest several associations specific to time-of-day.

Our primary findings suggest that daily fluctuations in behavioural sleep patterns among community dwelling adults generally do not affect daily, overall mood and vice versa. These results add to the existing mixed literature. Although failing to achieve statistical significance, we observed a trend suggesting that later sleep midpoint and longer sleep duration associated with greater average NA the following day. Our time-of-day analyses further showed that when participants slept later than their average sleep time on a given night, they reported significantly greater NA throughout the following morning and afternoon. In addition, when participants slept longer on a given night, they reported greater NA the following evening. We also found greater sleep efficiency associated with higher levels of PA the following morning. Regarding the influence of affect on sleep, we found that the effect of PA on sleep midpoint was specific to evening affect, such that greater PA in the evening related to later sleep.

We originally proposed that previous results were inconsistent in part because studies had limited assessments of sleep and affect. We aimed to address these limitations and by combining both actiography and EMA measures of affect in an adult community sample. Comparison of our participant demographics, study design and methods, and conceptual framework to the larger literature on sleep and mood helps inform interpretation of our findings. In the present sample, participants’ sleep characteristics were consistent with previous reports on healthy adults without sleep disorders (Buysse et al., 2010). While participants were not screened for sleep disorders, their observed sleep efficiency and body mass index (mean [SD] 26.8 [5.2] kg/m2) suggests it is unlikely participants had sleep-related disorders such as obstructive sleep apnea and insomnia. Overall, our present results reflect sleep characteristics of healthy, midlife adults who appear to be free of sleep disorders.

Most previous studies of the sleep–affect relationship used various forms of affect measures and scoring metrics, which precludes direct comparison of previous findings to our present participants’ EMA-reported affect. In our sample, participants tended to rate high on PA and low on NA, with some variability as assessed on hourly intervals throughout the day: participants reported an average PA of 4.0 (average nightly variation = 0.24), and an average NA of 1.9 (average nightly variation = 0.16). Participants were only included in the AHAB-II study if they were free of significant psychopathology and excluded if taking psychotropic medication. These affect characteristics are thus consistent with the demographic of our sample.

Aside from sleep duration, we found a medium to large proportion of variance in our sample’s affect levels and sleep characteristics was attributable to within-person, day-to-day variability. Thus, the present study was equipped to test the daily relationships between sleep and affect, with the exception that some results may reflect limited variability in sleep duration. The largely non-significant sleep–affect findings in our primary analyses suggest that healthy adults without sleep and mood disorders may be largely resilient to co-occurring fluctuations in their sleep or mood.

Our present finding that longer sleep duration associated with more NA the following evening is surprising, both because it is inconsistent with the aforementioned literature linking short sleep to NA, and because the finding was specific to evening affect. It is possible that participants in our present study did not naturally experience fluctuations in their sleep duration large enough to influence their overall (average) mood. Experimental studies have consistently shown that total deprivation or greater sleep restriction alters next day mood, but these studies involved a relatively large degree of sleep loss (e.g. 33%–50% restriction; Dinges et al., 1997; Haack & Mullington, 2005). In contrast, our present participants on average slept 6.8 hr and varied nightly by 58 min, which could approximate a 15% restriction on short sleep nights. Participants may not have naturally experienced the degree of sleep restriction needed to directly influence their mood the next day. Of note, previous studies on sleep restriction across several days have shown increases in negative and decreases in positive mood over time (Dinges et al., 1997; Haack & Mullington, 2005). In the present sample, while sleep data were collected on consecutive days, EMA data were not. This precluded us from examining cumulative effects of sleep characteristics from a succession of previous nights. Future studies are needed to consider such cumulative effects of sleep duration and to replicate our specific time-of-day finding.

Although there is a well-documented relationship between perceived sleep disruptions and affect, the present study aimed to examine whether this finding translates to behaviourally quantified sleep continuity. Our initial finding that actiography-derived sleep efficiency was not associated with average PA or NA the following day is consistent with results from another study of healthy adults (Mccrae et al., 2008). Although non-significant when examining average daily PA, our time-of-day analyses showed that greater sleep efficiency was specifically linked to higher levels of PA the following morning. Individuals tend to experience PA level that is low in the morning and increases throughout the day (Clark et al., 1989; Murray et al., 2002). Our present findings suggest that more consolidated sleep relates to a relative increase in morning PA. As there are documented discrepancies between perceived and objective sleep (Harvey & Tang, 2012), future studies can build on our present findings to examine whether perceived sleep quality and behaviourally quantified sleep continuity differentially associates with affect.

There is a paucity of studies on the effects of sleep timing on mood and our present study examined the associations between naturally occurring shifts in sleep timing and both PA and NA. We found that when participants slept later on a given night, they reported significantly greater NA throughout the following morning and afternoon. Consistent with evidence that 2–4 hr shifts in sleep timing increased NA the next day (Taub & Berger, 1974, 1976), we found that night-to-night shifts of lesser magnitude (average deviation of 39.6 min) similarly related to NA.

We found that greater daytime levels of PA predicted later sleep timing, and this association was specific to evening PA. Our present finding adds to the literature given that the one prior study testing day-to-day effects of PA on sleep timing found no significant effect (Totterdell et al., 1994). We originally hypothesised that greater NA, rather than PA, would predict later sleep timing. This hypothesis was based on existing evidence that depression is associated with later sleep time and a preference for late sleep time (Biss & Hasher, 2012; Hasler et al., 2010; Hidalgo et al., 2009). Our present finding suggests a different interpretation. Higher levels of PA in the evening may be associated with more engagement in positive social interactions and activities, which may involve more cognitive arousal that can delay sleep timing (Tang & Harvey, 2004). Further studies are warranted to examine whether pre-sleep activities and late timing of activities may contribute to the observed association between daytime PA and later sleep time. Aside from the PA-sleep timing relationship, we found no significant effects of either PA or NA on sleep characteristics.

5 |. CONCLUSION

The present study examined whether there is a proximal, bi-directional relationship between sleep characteristics and affect, and to explore whether these associations may differ across time of day. Strengths of the present study include both behavioural and momentary measures of sleep characteristics and affect, respectively, in a large community sample of adults. We found that greater PA predicted a delay in sleep midpoint, but there were no other effects of sleep characteristics on daily average affect or vice versa. There were several significant sleep-affect associations when considering time of day, which warrants further study. Future studies are also needed to consider prolonged effects of sleep disruptions, e.g. short sleep over consecutive nights, on mood that may not be observed in the context of small, daily fluctuations in sleep characteristics.

Funding information

National Institute of Health (NIH), Grant/Award Number: PO1 HL040962

Footnotes

CONFLICT OF INTEREST

None.

DATA AVAILABILITY STATEMENT

Data available on request due to privacy/ethical restrictions.

REFERENCES

  1. Biss RK, & Hasher L (2012). Happy as a lark: Morning-type younger and older adults are higher in positive affect. Emotion, 12(3), 437. 10.1037/a0027071 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Boivin DB, Czeisler CA, Dijk D-J, Duffy JF, Folkard S, Minors DS, Totterdell P, & Waterhouse JM (1997). Complex interaction of the sleep-wake cycle and circadian phase modulates mood in healthy subjects. Archives of General Psychiatry, 54(2), 145–152. 10.1001/archpsyc.1997.01830140055010 [DOI] [PubMed] [Google Scholar]
  3. Brissette I, & Cohen S (2002). The contribution of individual differences in hostility to the associations between daily interpersonal conflict, affect, and sleep. Personality and Social Psychology Bulletin, 28(9), 1265–1274. 10.1177/01461672022812011 [DOI] [Google Scholar]
  4. Buysse DJ, Cheng Y, Germain A, Moul DE, Franzen PL, Fletcher M, & Monk TH (2010). Night-to-night sleep variability in older adults with and without chronic insomnia. Sleep Medicine, 11(1), 56–64. 10.1016/j.sleep.2009.02.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Choo J, Elci OU, Yang K, Turk MW, Styn MA, Sereika SM, Music E, & Burke LE (2010). Longitudinal relationship between physical activity and cardiometabolic factors in overweight and obese adults. European Journal of Applied Physiology, 108(2), 329–336. 10.1007/s00421-009-1203-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Clark LA, Watson D, & Leeka J (1989). Diurnal variation in the positive affects. Motivation and Emotion, 13(3), 205–234. 10.1007/BF00995536 [DOI] [Google Scholar]
  7. de Wild-Hartmann JA, Wichers M, van Bemmel AL, Derom C, Thiery E, Jacobs N, van Os J, & Simons CJ (2013). Day-to-day associations between subjective sleep and affect in regard to future depression in a female population-based sample. The British Journal of Psychiatry, 202(6), 407–412. 10.1192/bjp.bp.112.123794 [DOI] [PubMed] [Google Scholar]
  8. Dinges DF, Pack F, Williams K, Gillen KA, Powell JW, Ott GE, Aptowicz C, & Pack AI (1997). Cumulative sleepiness, mood disturbance and psychomotor vigilance performance decrements during a week of sleep restricted to 4–5 hours per night. Sleep, 20(4), 267–277. [PubMed] [Google Scholar]
  9. Emens JS, Berman AM, Thosar SS, Butler MP, Roberts SA, Clemons NA, Herzig MX, McHill AW, Morimoto M, Bowles NP, & Shea SA (2020). Circadian rhythm in negative affect: Implications for mood disorders. Psychiatry Research, 293, 113337. 10.1016/j.psychres.2020.113337 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Galambos NL, Dalton AL, & Maggs JL (2009). Losing sleep over it: Daily variation in sleep quantity and quality in Canadian students’ first semester of university. Journal of Research on Adolescence, 19(4), 741–761. 10.1111/j.1532-7795.2009.00618.x [DOI] [Google Scholar]
  11. Girschik J, Fritschi L, Heyworth J, & Waters F (2012). Validation of self-reported sleep against actigraphy. Journal of Epidemiology, 22(5), 462–468. 10.2188/jea.JE20120012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Haack M, & Mullington JM (2005). Sustained sleep restriction reduces emotional and physical well-being. Pain, 119(1), 56–64. 10.1016/j.pain.2005.09.011 [DOI] [PubMed] [Google Scholar]
  13. Harvey AG, & Tang NK (2012). (Mis) perception of sleep in insomnia: A puzzle and a resolution. Psychological Bulletin, 138(1), 77. 10.1037/a0025730 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hasler BP, Allen JJ, Sbarra DA, Bootzin RR, & Bernert RA (2010). Morningness–eveningness and depression: Preliminary evidence for the role of the behavioral activation system and positive affect. Psychiatry Research, 176(2), 166–173. 10.1016/j.psychres.2009.06.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hertenstein E, Feige B, Gmeiner T, Kienzler C, Spiegelhalder K, Johann A, Jansson-Fröjmark M, Palagini L, Rücker G, Riemann D, & Baglioni C (2019). Insomnia as a predictor of mental disorders: A systematic review and meta-analysis. Sleep Medicine Reviews, 43, 96–105. 10.1016/j.smrv.2018.10.006 [DOI] [PubMed] [Google Scholar]
  16. Hidalgo MP, Caumo W, Posser M, Coccaro SB, Camozzato AL, & Chaves MLF (2009). Relationship between depressive mood and chronotype in healthy subjects. Psychiatry and Clinical Neurosciences, 63(3), 283–290. 10.1111/j.1440-1819.2009.01965.x [DOI] [PubMed] [Google Scholar]
  17. Kahn M, Fridenson S, Lerer R, Bar-Haim Y, & Sadeh A (2014). Effects of one night of induced night-wakings versus sleep restriction on sustained attention and mood: A pilot study. Sleep Medicine, 15(7), 825–832. 10.1016/j.sleep.2014.03.016 [DOI] [PubMed] [Google Scholar]
  18. Kalmbach DA, Pillai V, Roth T, & Drake CL (2014). The interplay between daily affect and sleep: A 2-week study of young women. Journal of Sleep Research, 23(6), 636–645. 10.1111/jsr.12190 [DOI] [PubMed] [Google Scholar]
  19. Kihlstrom JF, Eich E, Sandbrand D, & Tobias BA (2000). Emotion and memory: Implications for self-report. In Stone AA & Turkkan JS (Eds.), The science of self-report: Implications for research and practice (pp. 81–99). Lawrence Erlbaum. [Google Scholar]
  20. Li L, Wu C, Gan Y, Qu X, & Lu Z (2016). Insomnia and the risk of depression: A meta-analysis of prospective cohort studies. BMC Psychiatry, 16(1), 375. 10.1186/s12888-016-1075-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Matthews KA, Patel SR, Pantesco EJ, Buysse DJ, Kamarck TW, Lee L, & Hall ΜH (2018). Similarities and differences in estimates of sleep duration by polysomnography, actigraphy, diary, and self-reported habitual sleep in a community sample. Sleep Health, 4(1), 96–103. 10.1016/j.sleh.2017.10.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Mccrae CS, McNamara JP, Rowe MA, Dzierzewski JM, Dirk J, Marsiske M, & Craggs JG (2008). Sleep and affect in older adults: Using multilevel modeling to examine daily associations. Journal of Sleep Research, 17(1), 42–53. 10.1111/j.1365-2869.2008.00621.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. McNair P, Lorr M, & Droppleman L (1981). POMS manual. Educational and Industrial Testing Service. [Google Scholar]
  24. Miller MA, Rothenberger SD, Hasler BP, Donofry SD, Wong PM, Manuck SB, Kamarck TW, & Roecklein KA (2015). Chronotype predicts positive affect rhythms measured by ecological momentary assessment. Chronobiology International, 32(3), 376–384. 10.3109/07420528.2014.983602 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Murray G, Allen NB, & Trinder J (2002). Mood and the circadian system: Investigation of a circadian component in positive affect. Chronobiology International, 19(6), 1151–1169. 10.1081/CBI-120015956 [DOI] [PubMed] [Google Scholar]
  26. Nowak Z, Plewa M, Skowron M, Markiewicz A, Kucio C, & Osiadlo G (2010). Paffenbarger Physical Activity Questionnaire as an additional tool in clinical assessment of patients with coronary artery disease treated with angioplasty. Kardiologia Polska, 68(1), 32–39. [PubMed] [Google Scholar]
  27. Paffenbarger RS Jr, Wing AL, & Hyde RT (1978). Physical activity as an index of heart attack risk in college alumni. American Journal of Epidemiology, 108(3), 161–175. [DOI] [PubMed] [Google Scholar]
  28. Radloff LS (1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385–401. 10.1177/014662167700100306 [DOI] [Google Scholar]
  29. Sadeh A (2011). The role and validity of actigraphy in sleep medicine: An update. Sleep Medicine Reviews, 15(4), 259–267. 10.1016/j.smrv.2010.10.001 [DOI] [PubMed] [Google Scholar]
  30. Scott BA, & Judge TA (2006). Insomnia, emotions, and job satisfaction: A multilevel study. Journal of Management, 32(5), 622–645. 10.1177/0149206306289762 [DOI] [Google Scholar]
  31. Shiftman S, Stone AA, & Hufford MR (2008). Ecological momentary assessment. Annual Review of Clinical Psychology, 4, 1–32. 10.1146/annurev.clinpsy.3.022806.091415 [DOI] [PubMed] [Google Scholar]
  32. Silva GE, Goodwin JL, Sherrill DL, Arnold JL, Bootzin RR, Smith T, Walsleben JA, Baldwin CM, & Quan SF (2007). Relationship between reported and measured sleep times: The sleep heart health study (SHHS). Journal of Clinical Sleep Medicine, 3(6), 622. 10.5664/jcsm.26974 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Sonnentag S, Binnewies C, & Mojza EJ (2008). “ Did you have a nice evening?” A day-level study on recovery experiences, sleep, and affect. Journal of Applied Psychology, 93(3), 674. 10.1037/0021-9010.93.3.674 [DOI] [PubMed] [Google Scholar]
  34. Stone AA, & Shiftman S (1994). Ecological momentary assessment (EMA) in behavorial medicine. Annals of Behavioral Medicine, 16 (3), 199–202. [Google Scholar]
  35. Tang NK, & Harvey AG (2004). Effects of cognitive arousal and physiological arousal on sleep perception. Sleep, 27(1), 69–78. 10.1093/sleep/27.1.69 [DOI] [PubMed] [Google Scholar]
  36. Taub JM, & Berger RJ (1974). Acute shifts in the sleep-wakefulness cycle: Effects on performance and mood. Psychosomatic Medicine, 36(2), 164–173. 10.1097/00006842-197403000-00008 [DOI] [PubMed] [Google Scholar]
  37. Taub JM, & Berger RJ (1976). The effects of changing the phase and duration of sleep. Journal of Experimental Psychology: Human Perception and Performance, 2(1), 30. [DOI] [PubMed] [Google Scholar]
  38. Thompson ER (2007). Development and validation of an internationally reliable short-form of the positive and negative affect schedule (PANAS). Journal of cross-cultural Psychology, 38(2), 227–242. 10.1177/0022022106297301 [DOI] [Google Scholar]
  39. Totterdell P, Reynolds S, Parkinson B, & Briner RB (1994). Associations of sleep with everyday mood, minor symptoms and social interaction experience. Sleep, 17(5), 466–475. 10.1093/sleep/17.5.466 [DOI] [PubMed] [Google Scholar]
  40. Zoccola PM, Dickerson SS, & Lam S (2009). Rumination predicts longer sleep onset latency after an acute psychosocial stressor. Psychosomatic Medicine, 71(7), 771–775. 10.1097/PSY.0b013e3181ae58e8 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Data available on request due to privacy/ethical restrictions.

RESOURCES