Objective: Although stress can elicit profound and lasting effects on sleep, the pathways whereby stress affects sleep are not well understood. In this study, we used autoregressive spectral analysis of the electrocardiogram (EKG) interbeat interval sequence to characterize stress-related changes in heart rate variability during sleep in 59 healthy men and women.
Methods: Participants (N = 59) were randomly assigned to a control or stress condition, in which a standard speech task paradigm was used to elicit acute stress in the immediate presleep period. EKG was collected throughout the night. The high frequency component (0.15–0.4 Hz Eq) was used to index parasympathetic modulation, and the ratio of low to high frequency power (0.04–0.15 Hz Eq/0.15–0.4 Hz Eq) of heart rate variability was used to index sympathovagal balance.
Results: Acute psychophysiological stress was associated with decreased levels of parasympathetic modulation during nonrapid eye movement (NREM) and rapid eye movement sleep and increased levels of sympathovagal balance during NREM sleep. Parasympathetic modulation increased across successive NREM cycles in the control group; these increases were blunted in the stress group and remained essentially unchanged across successive NREM periods. Higher levels of sympathovagal balance during NREM sleep were associated with poorer sleep maintenance and lower delta activity.
Conclusions: Changes in heart rate variability associated with acute stress may represent one pathway to disturbed sleep. Stress-related changes in heart rate variability during sleep may also be important in association with chronic stressors, which are associated with significant morbidity and increased risk for mortality.
AR = autoregressive;, BMI = body mass index;, EEG = electroencephalogram;, EKG = electrocardiogram;, EMG = electromyogram;, EOG = electro-oculogram;, FFT = Fast Fourier transform;, GSI = Global Severity Index;, HF = high-frequency;, HRV = heart rate variability;, IBI = interbeat interval;, LH = low-frequency;, NREM = nonrapid eye movement;, PSG = polysomnography;, PSQI = Pittsburgh Sleep Quality Index;, PSWQ = Penn State Worry Questionnaire;, REM = rapid eye movement;, SWS = slow-wave sleep;, VAS = visual analogue scale.
Stress can elicit profound and lasting effects on sleep (1–3). It is intuitively appealing to hypothesize that worry and intrusive thoughts delay the onset of sleep, and studies have supported this hypothesis (4–6). Similarly, stress-related intrusive thoughts and negative affect may color one’s subjective perceptions about the quality of one’s sleep (2,7). However, the pathways whereby stress reaches into the night to produce frequent awakenings from sleep, to lighten nonrapid eye movement (NREM) sleep, or to affect quantitative and qualitative components of rapid eye movement (REM) sleep are less clear. Emerging evidence suggests that stress also affects more subtle aspects of sleep such as high-frequency power in the beta-range (8).
Sustained autonomic nervous system arousal may represent one pathway whereby stress affects sleep. Spectral analysis of heart rate variability provides a noninvasive technique for indirectly measuring sympathetic and parasympathetic modulation during sleep. Fast Fourier transform (FFT) and autoregressive techniques have identified specific heart rate variability components that are indirect indices of sympathetic and parasympathetic modulation (9,10). Both FFT and autroregressive techniques transform heart rate signals from the time domain (beat-to-beat intervals) to the frequency domain (power in bands that provide indices of sympathetic and parasympathetic modulation). The high-frequency component (HF, 0.15–0.4 Hz) represents the respiratory cycle and is mediated by the parasympathetic nervous system. Because the low-frequency component (LF, 0.04–0.15 Hz) reflects inputs from both branches of the autonomic nervous system, sympathetic modulation is generally imputed from the ratio of the LH to HF components (11). Studies of acute sympathetic withdrawal during ganglionic blockade and muscle sympathetic nerve activity as measured by microneurography indicate that the LF:HF ratio is an acceptable measure of sympathetic modulation (12,13).
These frequency domain techniques reveal reliable patterns of autonomic modulation across different sleep/wake states. Previous studies have compared levels of sympathetic and parasympathetic modulation during 3-minute to 5-minute epochs of discrete stages of sleep and wakefulness (eg, wake, NREM sleep, REM sleep). In general, NREM sleep is characterized by parasympathetic predominance, whereas REM sleep and wakefulness show increased levels of sympathetic modulation (14–16). These relationships are seen during discrete epochs of sleep at the beginning, middle, and end of the night (16). Gradations in autonomic modulation are also seen within NREM sleep (16,17). Slow-wave sleep (SWS) is associated with increased power in the HF (parasympathetic) component, and concomitant increases in sympathovagal modulation are seen during stages 1 and 2 compared with SWS. Levels of sympathovagal balance during stages 1, REM, and wakefulness are often indistinguishable.
Time series analysis of heart rate variability (HRV) has revealed a close temporal coupling between autonomic modulation and electroencephalogram (EEG)-assessed sleep that suggests common control mechanisms. For example, Bonnet and Arand (16) demonstrated that shifts in heart rate variability precede arousals from stage 2 sleep and the onset/offset of REM sleep by 10 to 20 beats. Analyses of interbeat autocorrelation coefficients also reveal increases in sympathovagal modulation 1 to 2 minutes before shifts to lighter stages of sleep, as defined by EEG activity (15). Time series analysis of SWS has revealed that oscillations in delta wave activity mirror changes in sympathovagal modulation (18). During individual NREM sleep cycles, increases in delta activity are associated with decreases in sympathetic modulation. Conversely, decreasing levels of delta activity during NREM and REM sleep are accompanied by increasing levels of sympathovagal balance.
Although previous research has established a link between autonomic modulation and sleep, this technique has not been applied to acute stress paradigms. We sought to evaluate the impact of stress on heart rate variability during sleep. A standard speech task was used as the stressor, given evidence that experimental speech tasks are reliably associated with increases in subjective stress and cardiovascular, endocrine, and immune responses (19,20). Stress was defined by experimental group (speech task vs. control condition), and heart rate variability was calculated using autoregressive spectral analysis of the electrocardiogram (EKG) interbeat interval sequence, which was recorded throughout sleep. It was hypothesized that stress would be associated with a decrease in parasympathetic modulation and an increase in sympathovagal balance throughout NREM and REM sleep. It was also hypothesized that stress-related changes in HRV would be associated with disrupted sleep.