Predicting sleep quality from stress and prior sleep – A study of day-to-day covariation across six weeks

Abstract

Background/objectives

The connection between stress and sleep is well established in cross-sectional questionnaire studies and in a few prospective studies. Here, the intention was to study the link between stress and sleep on a day-to-day basis across 42 days.

Methods

Fifty participants kept a sleep/wake diary across 42 days and responded to daily questions on sleep and stress. The results were analyzed with a mixed model approach using stress during the prior day to predict morning ratings of sleep quality.

Results

The results showed that bedtime stress and worries were the main predictors of sleep quality, but that, also, late awakening, short prior sleep, high quality of prior sleep, and good health the prior day predicted higher sleep quality.

Conclusions

Stress during the day predicts subsequent sleep quality on a day-to-day basis across 42 days. The observed range of variation in stress/worries was modest, which is why it is suggested that the present data underestimates the impact of stress on subsequent sleep quality.

Introduction

Stress is seen as the primary cause of persistent psychophysiological (or primary) insomnia [1]. In the International classification of sleep disorders (ICSD2) [2] “adjustment insomnia” represents the effects of acute stress, whereas “primary insomnia” represents the long-term consequences of mainly adjustment insomnia.

Most research on stress and sleep has been experimental in nature and has not dealt with stress in daily life [3], except for some cross-sectional epidemiological studies [4], [5], [6], [7], [8], [9]. All indicate a clear negative correlation between stress and sleep quality.

While informative, cross-sectional studies have problems with causal links. The first prospective study of work stress and sleep seems to be the one by Ribet et al. [6] who studied more than 21,000 subjects in France using a sleep disturbance index. Shift work, a long working week, exposure to vibrations, and “having to hurry” appeared to be the main risk factors, controlling for age and gender. Another prospective self-report study across three points in time showed a relation between work demands and sleep quality [10]. Similarly, Linton et al. related “negative psychosocial work conditions” to insomnia, with an Odds Ratio across five years of 2.15 [11]. Jansson and Linton [12] controlled for initial insomnia in a similar design and found that high work demands at the outset predicted impaired sleep one year later. Using another approach, Mezick et al. showed that prior stressful life events were related to a later increase in daily variability of sleep [13]. Hall et al. [14] demonstrated that financial strain was related to poorer subjective sleep quality and to reduced sleep efficiency.

An alternative to prospective epidemiological studies across a period of years is to focus on the day-to-day variation in stress in daily life and its corresponding day-to-day variation in sleep quality. Does a stressful day in daily life lead to a reduced sleep quality that night, and would such a relation hold across a number of days in daily life? In one study in which a low and high stress day were selected out of five measures the results showed reports of reduced sleep quality as well as reduced sleep efficiency and increased latency to Stage 3 sleep after a high stress day [15].

While occasional days with heavy stress might affect sleep, one may wonder whether the, presumably, modest variation in stress in daily life during a longer period of time led to disturbed sleep. If stress would show a coupling to subsequent sleep on a day-to-day basis, this would add credibility to earlier large but coarse-grained studies on this issue. To the best of our knowledge only one study of the relation between daily stress and subsequent sleep has been carried out. It was based on longitudinal, intraindividual correlations between stress and sleep ratings mainly in insomnia patients [1]. It showed a rather weak, but significant, correlation indicating that increased stress was followed by reduced sleep quality. Today, a more timely approach would be to use multilevel modeling, which would permit both between and within-groups analyses and estimates of the shape of the relation [16]. Such an approach has been used for investigating the daily variation of sleep on the daily variation of mood across two weeks, finding that reduced sleep affects mood the subsequent day [17].

The aim of the present study was to investigate how the daily variation in stress during the day/evening would be related to the daily variation in reported sleep quality the next morning over a period of 42 days in healthy individuals. The hypothesis was that day/evening stress would result in reports of poor sleep the next morning. The design also included other factors that might affect sleep and confound the relation between stress and sleep, for example alcohol intake in the evening, time of turning off the lights to sleep (lights out), self-rated health (SRH), sleep quality the night before the day of rated stress, a day off, etc. In addition, background factors such as age, gender, depression, and anxiety were used as possible modifiers.