Research report
The sleep of remitted bipolar outpatients: a controlled naturalistic study using actigraphy

https://doi.org/10.1016/S0165-0327(03)00055-7Get rights and content

Abstract

Background: Several sleep laboratory studies suggest sleep abnormalities in bipolar disorder. However, this is the first study to compare remitted bipolar subjects with controls on actigraphic and subjective sleep parameters in a naturalistic setting over 5 nights. Methods: Nineteen subjects with Bipolar I Disorder and 19 age- and gender-matched healthy controls were included. Objective sleep parameters were estimated using wrist actigraphs. Subject-rated sleep diaries and mood ratings were also completed. Sleep data were averaged for each subject across nights, and raw score standard deviations were calculated as a measure of within-subject variability. Results: Multivariate analyses of variance found significant group differences for both actigraphic (F(4,33)=3.80, P=0.012) and subjective measures (F(4,31)=3.18, P=0.027). Univariate analyses identified reliable differences in sleep onset latency (subjective), sleep duration (subjective), and variability of sleep duration and night wake time (actigraphic). Binary backward stepwise logistic regression demonstrated that a combination of three sleep measures correctly predicted disorder status in 84% of cases. Limitations: Failure to match on sociodemographic and employment status is a limitation that may provide an alternative explanation for some findings. Furthermore, in the bipolar group 18 of 19 subjects were in receipt of psychotropic medication, compared to none of the healthy control group. Also, no information was recorded about family history of mental disorders in the control group. Conclusions: The study suggests that the sleep of remitted bipolar outpatients measured in naturalistic settings is characteristically different from controls: bipolar subjects sleep longer, report longer onset latencies, and display greater variability across nights.

Introduction

Bipolar disorders (BP) are common, severe and persistent disorders, which affect 1.3–1.7% of the population (Kessler et al., 1996). Hyposomnia and hyperactivity are two defining features of mania, while reduced activity and hypersomnia characterise bipolar depression (Leibenluft et al., 1995). The pervasiveness of sleep disturbances in BP suggests that they are not merely symptomatic, but may be central in the aetiology of symptoms.

There is evidence that changes in sleep and activity may act as sensitive markers for relapse in BP, with reduced sleep and increased activity being associated with the onset of a manic episode (Barbini et al., 1996; Klein et al., 1991, Klein et al., 1992; Leibenluft et al., 1996; Nowlin-Finch et al., 1994). There is also compelling evidence that sleep abnormalities are trait markers of mood disorder, which persist even when subjects are not in an episode. The evidence is more clear in the case of unipolar (UP) depression, where it has been shown that reduced REM latency is a stable sleep characteristic in those with a history of UP depression, even when remitted (Rush et al., 1986; Giles et al., 1990, Giles et al., 1993).

In one of the few studies exploring sleep characteristics of remitted BP subjects, Sitaram et al. (1982) reported an increased density and percentage of REM sleep in remitted BP patients (n=14) compared to a control group. The study also found that the patients were more sensitive to the effects of arecoline, an acetylcholine agonist that can reduce REM latency. Knowles et al. (1986) undertook electroencephalographic assessment of remitted BP patients on five consecutive nights. In many respects, the sleep of the 10 remitted patients was similar to that of 10 controls. However, the BP patients had a higher percentage of time in stage 1 sleep, and more disturbed sleep than controls.

The above data are consistent with the view that BP patients may have characteristic sleep disturbances. However, a major limitation of this research is its reliance on polysomnography (PSG). For logistic reasons polysomnographic studies are usually confined to one or two nights. In addition, the artificial nature of the sleep laboratory may disrupt natural rhythms. For example, the fact that subjects are woken at predetermined times may lead to underestimation of sleep duration in hypersomnic subjects (e.g., Wehr et al., 1985). A potential solution to these problems is offered by actigraphy, which employs a small motion-sensitive device, worn like a wristwatch, that can sample physical activity levels continuously for prolonged periods (Sadeh et al., 1995). A range of sleep-related variables is calculated from raw movement data (e.g., sleep onset latency, sleep duration, amount of wake time during the night, and ‘sleep efficiency’). Although actigraphy is less precise than polysomnography, and does not discriminate sleep stages, comparative studies have reported 80–90% agreement between PSG and actigraphic recordings with respect to determination of sleep and wakefulness (Sadeh et al., 1989). The practical advantage of actigraphy is that recordings of daytime and nighttime activity can be made with minimal disruption to a subject’s normal life. Actigraphic measurement has been utilised successfully in group comparisons where stability of the sleep–wake cycle is of particular interest (e.g., Gruber et al., 2000). This capacity for naturalistic measurement makes actigraphy the obvious choice to study underlying circadian abnormalities in BP subjects.

The present study compared a group of euthymic BP I subjects with age and gender matched controls on a number of actigraphically estimated, and subjectively estimated sleep variables over a 5-day period. The study also explored the relationship between sleep variables, and mood ratings to determine whether objective and subjective sleep parameters and/or daily mood ratings can be used to classify subjects as cases or controls.

Section snippets

Methods

With ethical approval, a two-group design was employed that compared subjects with a confirmed psychiatric diagnosis of BP I with healthy age- and gender-matched controls on a number of sleep parameters. Cases and controls all gave informed consent to participate in the study. No participant was involved in any other research study.

Description of the groups

Thirty-two BP subjects were identified for potential inclusion. Eleven chose not to take part and, of the 21 who consented to participate, two subjects dropped out within the first day of the study. The BP group comprised eight males and 11 females; their ages ranged from 26 to 68 years (mean=47.3, S.D.=10.61). Time since diagnosis ranged from 1 to 35 years (median 15 years) and time since last episode ranged from 1 to 96 months (median 12 months).

Eighteen BP subjects were receiving

Discussion

This study suggests that the more prolonged, naturalistic measurement offered by actigraphic techniques is a useful method for studying possible sleep–wake cycle abnormalities in this disorder. It is noteworthy that the differences in sleep parameters between BP outpatients and controls identified would almost certainly have been masked by the constraints of laboratory-based methods of sleep measurement. Using actigraphy appears to be a sensitive and convenient way of identifying these

References (29)

  • C Bowden

    Strategies to reduce misdiagnosis of bipolar depression

    Psychiatr. Serv.

    (2001)
  • C Ehlers et al.

    Social zeitgebers and biological rhythms: a unified approach to understanding the etiology of depression

    Arch. Gen. Psychiatry

    (1988)
  • C.A Espie
  • M.B First et al.
  • Cited by (149)

    • The circadian systems genes and their importance of human health

      2023, Advances in Protein Chemistry and Structural Biology
    • Seasonal effects on bipolar disorder: A closer look

      2020, Neuroscience and Biobehavioral Reviews
    View all citing articles on Scopus
    View full text