Sleep quality predicts positive and negative affect but not vice versa. An electronic diary study in depressed and healthy individuals

Highlights

• Changes in affect are not associated with subsequent changes in sleep quality.

• Changes in sleep quality are associated with subsequent changes in affect.

• Fatigue mediates the association between sleep quality and subsequent affect.

• Depression status is not a moderator of the association between sleep quality and affect.

Abstract

Background

The exact nature of the complex relationship between sleep and affect has remained unclear. This study investigated the temporal order of change in sleep and affect in participants with and without depression.

Methods

27 depressed patients and 27 pair-matched healthy controls assessed their sleep in the morning and their affect 3 times a day for 30 consecutive days in their natural environment. Daily sleep quality and average positive affect (PA) and negative affect (NA) were used to examine whether changes in sleep quality preceded or followed changes in PA and NA, and whether this was different for patients and healthy controls. Second, presumptive mediating factors were investigated. We hypothesized that fatigue mediated the effect of changes in sleep quality on subsequent PA/NA, and that rumination mediated the effect of changes in PA/NA on subsequent sleep quality.

Results

Multilevel models showed that changes in sleep quality predicted changes in PA (B=0.08, p<0.001) and NA (B=−0.06, p<0.001), but not the other way around (PA: B=0.03, p=0.70, NA: B=−0.05, p=0.60). Fatigue was found to be a significant mediator of the relationship between sleep quality and PA (Indirect Effect=0.03, p<0.001), and between sleep quality and NA (Indirect Effect=−0.02, p=0.01). Rumination was not investigated because of non-significant associations between PA/NA and sleep quality. The associations were not different for patients and controls.

Limitations

The analyses were restricted to self-reported sleep quality, and conclusions about causality could not be drawn.

Conclusions

Improvements in sleep quality predicted improvements in affect the following day, partly mediated by fatigue. Treatment of sleep symptoms would benefit affect in clinical care and beyond.

Introduction

It is well known that sleep and affect are related (Gershon et al., 2012, Pilcher and Huffcutt, 1996 Sonnentag et al., 2008). A disturbance of positive and negative affect is one of the core symptoms of a major depressive disorder (MDD) (American Psychiatric Association, 2013, Peeters et al., 2006). The disturbance of sleep is also one of the symptoms of a diagnosis of MDD; approximately 70% of depressed patients report changes in sleep (Soehner and Harvey, 2012, Soehner et al., 2014). The relationship between sleep and affect in depression is complex. Depression can be related to difficulty initiating or maintaining sleep (=insomnia), but sometimes also to sleeping too much (=hypersomnia), or both (Soehner et al., 2014). Similarly, interventions are targeted at increasing sleep time, but also at the opposite (i.e., sleep deprivation), even in patients who have insomnia. Sleep deprivation seems to have a positive effect on affect in some patients with depression, although the effect does not last long (Dopierala and Rybakowski, 2015, Hemmeter et al., 2010). In sum, although there is consensus in the literature that sleep and affect are associated with each other, the exact nature of the association remains unclear (Cousins et al., 2011, Harvey, 2008, Kahn et al., 2013, Sonnentag et al., 2008, Tavernier and Willoughby, 2014).

Along with sleep disturbances comes increased fatigue, which can be part of an MDD diagnosis as well (Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 2013)). Decreased subjective sleep quality (Fung et al., 2014) and decreased subjective sleep duration (Sonnentag et al., 2008) have been found to be highly correlated with same-time fatigue. Feelings of fatigue have been found to be associated with same-time positive and negative affect (Kahn et al, 2013). Based on these findings it may be argued that fatigue mediates the temporal association between sleep and subsequent affect. However, in these previous studies there was no temporal separation between mediator and outcome, and no multilevel approach was used. The association in the opposite direction, from affect to subsequent sleep, may be mediated by rumination. In a recent review (Kahn et al., 2013) it was shown that rumination had a significant impact on sleep quality. Next to that, rumination has been found to mediate the relationship between negative affect and sleep quality in healthy adults (Slavish and Graham-Engeland, 2015). In the latter study there were two months between assessment of predictor versus mediator and outcome, and besides that there was again no temporal separation between mediator and outcome.

Sleep and affect can both change on a daily basis (Fung et al., 2014, Peeters et al., 2006). These day-to-day changes can be captured by the use of multiple repeated assessments (i.e., daily diary study), and such a design enables the examination of the temporal order of these changes. This temporal order of change in sleep and affect has been investigated before in healthy adults and children (Cousins et al., 2011, Galambos et al., 2009, Sonnentag et al., 2008), and in two of these studies a bidirectional effect was found between sleep and affect. In patients with depression, the one-way association between sleep quality and subsequent affect was investigated in only one study, in a design of 6 consecutive days (Peeters et al., 2006). Neither sleep duration nor sleep quality were associated with subsequent affect in this study. Thus, a depressive disorder might moderate the association between sleep and affect: the difference in results between the abovementioned studies may be explained by depression status. Possibly, depression weakens the temporal association between sleep and affect because of reduced emotional reactivity in depressed patients (Bylsma et al., 2008). This notion needs more research, however, because neither of the abovementioned studies examined both healthy and depressed persons within one and the same study. Moreover, in the study by Peeters et al. (2006) the relationship between sleep and affect was investigated only unidirectional, and only six time points were used for analyses. The examination of the bidirectional association between sleep and affect in depressed patients and healthy controls with multiple repeated assessments could reveal whether depression status indeed moderates the association between sleep and subsequent affect, and whether it moderates the association between affect and subsequent sleep. The design with multiple repeated within-day assessments is the major strength of our study compared with earlier studies on this topic. This approach enabled to investigate how sleep and affect are temporally associated with each other within the day, and to study how the proposed mediators fatigue and rumination differentially contribute to these within-day associations. In contrast to previous studies on this topic, we were able to better separate the proposed mediators from the outcome measures in time. An additional benefit of a multiple repeated assessments design is the possibility to take heterogeneity among participants into account (Hamaker, 2012). The first aim of the present study was to examine the temporal order of change in affect and sleep, with depression status as a potential moderator. In case of significant associations between sleep and affect, we investigated mediating factors of the significant association. We hypothesized that affect during the day predicts sleep quality the subsequent night, and that rumination in the evening mediates this association. Further, we hypothesized that sleep quality the previous night predicts affect during the day, and that fatigue in the morning mediates this association.