CLINICAL REVIEW
Sleep changes in the disorder of insomnia: A meta-analysis of polysomnographic studies

https://doi.org/10.1016/j.smrv.2013.04.001Get rights and content

Summary

Insomnia is a highly prevalent health problem worldwide. Primary insomnia (PI), i.e., insomnia not due to another disorder or substance use, represents a model to elucidate the pathophysiology of sleep. However, prior research in patients with PI has failed to demonstrate consistent abnormalities in the state-of-the-art assessment of sleep (polysomnography). The aim of this meta-analysis was to clarify whether there are identifiable polysomnographic sleep changes that correspond to the subjective complaints of patients with PI. Medline and PsycInfo databases were searched from 1994 to 2012. Effects were calculated as standardized mean differences. Studies were pooled with the random-effects meta-analytic model. Twenty-three studies met inclusion criteria. In total, 582 patients with PI and 485 good sleeper controls (GSC) were evaluated. The results showed that patients with PI present a disruption of sleep continuity and a significant reduction of slow wave sleep (SWS) and rapid eye movement (REM) sleep compared to GSC. The observed changes in sleep architecture, i.e., reductions in SWS and REM sleep, hitherto did not count among the typical polysomnographic findings in patients with PI. An advanced knowledge of the polysomnographic changes in PI may add to foster the understanding of the pathophysiology of sleep and its bi-directional relationships with somatic and mental disorders.

Introduction

Insomnia is defined as complaint of difficulties initiating/maintaining sleep or non-restorative sleep accompanied by decreased daytime functioning persisting for at least four weeks (DSM-IV1; DSM-IV-TR2). The current version of the DSM (DSM-IV-TR)2 identifies two main categories of insomnia: 1) primary insomnia in which the condition of insomnia is not attributable to other medical or psychiatric disorders; and 2) secondary insomnia in which another medical or psychiatric disorder has to be identified as the primary cause. This dual interpretation has been strongly criticized and will be eliminated in the next edition of the DSM, the DSM-V.3 The new definition will refer to “insomnia disorder” as the main diagnostic category for insomnia with respect to which clinically comorbid conditions (both medical and psychiatric) can be coded.

The identification of insomnia as an independent diagnostic entity has stimulated research about two main questions: what are the psychophysiological characteristics underlying the disorder of insomnia; and which are the psychophysiological characteristics explaining its high comorbidity with other clinical conditions Because the DSM-V has not yet been published, researchers have until now referred to the DSM-IV condition of primary insomnia – i.e., insomnia in the absence of other clinical conditions or substance abuse – as a useful research model to understand the pathophysiology of disrupted sleep. The condition of primary insomnia in the absence of other clinical disorders or substance abuse is consistent with the definition of insomnia disorder with no specified clinically comorbid conditions which will be adopted in the DSM-V. Accordingly, this condition has been the focus of the present study and will be indicated with the abbreviation “PI”.

One of the main controversial results in the study of PI is whether it is associated with changes in sleep or not, based on the findings from many experimental studies which failed to demonstrate consistent electrophysiological correlates of the subjective complaints. However, these controversial findings may be related to different methods or definitions. For this reason, the aim of this meta-analysis was to analyze all studies in which the electrophysiological correlates of sleep were recorded with polysomnography (PSG) in patients with PI as compared to healthy good sleeper controls (GSC).

In the following paragraphs of the introduction, a brief overview of insomnia in terms of its prevalence and costs will be presented and the history of the change in the conceptualization from symptom to disorder will be discussed. Then, the assessment of the symptoms will be described, and finally, the focus will be specifically on studies evaluating electrophysiological sleep recordings in PI.

Chronic insomnia afflicts about 10% of the population in western industrialized countries4 with women being more frequently affected than men5 and an increasing prevalence with older age.4 More than 70% of subjects experiencing insomnia today will still suffer from it the next year as the disorder tends to take a chronic course.6 PI occurs in about 2–4% of the adult population.4 Recent research has emphasized the economical and societal costs of insomnia. For example, in the US, the costs of insomnia due to low work performance and absenteeism have been estimated to exceed 60 billion $ per year.7 Additionally, the condition of insomnia is associated with increased healthcare utilization and indirect costs, for example linked with a higher rate of accidents (e.g.,8).

The classic psychiatric view has conceptualized insomnia as a symptom of an underlying mental disorder, with affective disorders being the most important ones. This conceptualization has been substantially challenged based both on clinical observations and experimental findings showing that, at least with respect to major depression, insomnia often precedes it and it does not remit after successful treatment of depression (e.g.,9). In addition, a recent meta-analysis found insomnia to be an independent clinical predictor of major depression.10 Symptoms of insomnia are a pervasive problem not only for patients with depression, but also for patients with bipolar disorders, anxiety disorders, alcoholism, eating disorder, and schizophrenia.11, 12, 13 Recent theories have proposed that insomnia may be involved in the causation and the maintenance of psychiatric disorders by representing a transdiagnostic mechansim.14 Consistently, sleep-related processes play an important role for healthy emotional and cognitive functioning.15 Moreover, recent evidence has shown that insomnia, and specifically PI, is also associated with emotional alterations (for a review see16). However, PI is not the same as sleep deprivation, as it involves adaptation processes and, as mentioned above, it has not been ascertained that it is associated with alterations of sleep physiology. The understanding of if and which physiological sleep changes are associated with PI should help in clarifying its association with emotional processes and psychopathology.

The sleep complaint in patients with insomnia can be evaluated through technical assessment or by subjective self-monitoring. The gold standard of technical sleep assessment is polysomnography (PSG) including electrophysiological recordings of brain activity (EEG), muscle activity (EMG), and eye movements (EOG). The recording is scored and categorized into different stages – wake; sleep stage 1; sleep stage 2; slow wave sleep; rapid eye movement (REM) sleep – and typically PSG-derived sleep parameters are grouped into sleep continuity variables (e.g., sleep onset latency (SOL), number of awakenings), sleep architecture variables (e.g., duration of each sleep stage in minutes and %) and REM sleep variables (i.e., REM latency, REM density).

Subjective self-monitoring of sleep is generally achieved by sleep diaries, asking subjects to provide an estimate of sleep-related variables (SOL, WASO, TST, etc.) every morning. Additionally, subjects are asked to document their daytime behaviors (e.g., medication or alcohol intake) in the evening.17

As insomnia is defined as complaint of poor sleep, its assessment is based primarily on subjective description of the symptoms, and not necessarily on pathological polysomnographic recordings. The international classification of sleep disorders (ICSD-2),18 for example, identifies a type of insomnia, called paradoxical insomnia, characterized by a profound subjective complaint of poor sleep in the absence of polysomnographic alterations (total sleep time ≥ 6.5 h and a sleep efficiency ≥ 85%). In general, PSG recordings, allowing conventional sleep staging referring to standard criteria,19, 20 are not required for the clinical assessment of insomnia (e.g., consensus reports21, 22).

While potential PSG alterations associated with insomnia play a secondary role in clinical settings, they are of high relevance in order to describe the pathophysiology of the disorder. As mentioned above, research has focused on PI as a useful research model. Current pathophysiological models agree that the main etiological factor in the onset and the maintenance of PI is cognitive, emotional and physiological hyperarousal which is present during night and daytime (e.g.,23; for review, please refer to24, 25, *26, *27). If people with PI are characterized by increased psychophysiological arousal, it is reasonable to assume that this should be associated with alterations of sleep continuity and architecture, such as lighter and less deep sleep. However, PSG impairment of sleep in PI is still a matter of controversy. Several studies applying PSG in patients with PI failed to find significant differences to GSC28, 29 or found only a modest impairment in the patient group.30, 31, 32, 33 Accordingly, many authors have criticized the use of PSG data to evaluate the clinical impairment of patients with PI. It can be speculated that contradictory results found in studies using PSG depend on different methods or definitions of insomnia. In addition, some authors have suggested the use of alternative EEG analyses, as for example, power spectral analysis (PSA) which is a method for quantifying the EEG's constituting frequency components including those that reflect the general arousal level of the brain.34, 35, 36

To our knowledge, the only previous quantitative summary of the literature on PSG data in patients with insomnia was provided by Benca et al.37 In this meta-analysis, the authors analyzed all studies measuring sleep polysomnographically in psychiatric disorders, including insomnia.37 This was an extensive study; however, as it was published 20 years ago, it had some limitations which might have influenced the results. First, the sleep variables were taken from either one night or from the average of all recorded nights. As a consequence, no attempt was made to eliminate the possible confounding influence of the “first night effect”, which is the tendency of individuals to sleep worse during the first night of PSG,38 or the “reverse first night effect”, which may be encountered in patients with PI who have been reported to sleep better in the first laboratory night.39 Second, no attempt was undertaken to avoid a possible overlap between subjects published repeatedly in different publications. Third, the definition of sleep variables was not taken into consideration as a source of bias. Fourth, considering the meta-analytic calculations, all studies were entered into the analysis with equal weight. Today, state-of-the-art software is available which automatically weights studies for sample size. Additionally, effect sizes and mean differences can be calculated in order to estimate the clinical importance of the findings. Last, with specific relation to the analyses of insomnia studies, no DSM-IV1 definition of primary insomnia was available in 1992. Consequently, insomnia was defined either based on PSG abnormalities or on self-reported sleep patterns with no consideration of the daytime symptoms. This variability could of course, have had a strong impact on the results. Moreover, the authors did not focus specifically on PI, which may imply that the results could have been biased by comorbidity with other disorders.

In the present study, we aimed to quantitatively summarizing all the data published so far in which conventional PSG was measured in patients with PI and in GSC. The objective was to describe the sleep profile of patients with PI and to identify future avenues for research.

Section snippets

Method

The meta-analysis was performed according to MOOSE40 and PRISMA41 guidelines. The first and second authors independently conducted the literature search and screened the titles and abstracts of potentially eligible studies. The first and the third author examined the full texts and collaborated to extract the data for the analyses.

Results

Fig. 1 illustrates the search flow. Inserting the keywords, a total of 1227 abstracts were identified (768 abstracts in PubMed, 459 in PsycInfo). The screening of titles and abstracts led to 92 studies for which the full text was evaluated. The examination of the reference lists of the full texts identified 17 additional papers. The evaluation of the full texts led to a final sample of 23 studies. The 86 studies excluded and the reasons for the exclusion are listed in Table S1 (supplemental

Polysomnographic profile of people with PI

The current meta-analysis indicates that PI seems to be characterized by impairments in both sleep continuity and sleep architecture. More specifically, compared with GSC, patients with PI sleep about 23 min less; take on average 6 min longer to fall asleep; and present about six awakenings more per night. These alterations in sleep continuity result in a reduced SEI that, in patients with PI, was below the generally considered threshold for “good sleep” of 85%. Surprisingly patients with PI,

Conclusions

Consistent with the hyperarousal model,*23, *26, *27 our meta-analysis showed that patients with PI present a disruption of sleep continuity associated with a moderate, albeit significant reduction of deep sleep and REM sleep. The length of SWS has been shown to correlate positively with cognitive processes, i.e., the consolidation of individual declarative memories (e.g.,15). Patients with PI have been shown to display deficits in sleep-related memory consolidation.77 A loss of slow wave sleep

Acknowledgements

The authors wish to thank Prof. C. Bastien and Prof. M. Perlis for their kind replies and for sharing with us additional information about their studies.

References (83)

  • K. Spiegelhalder et al.

    Increased EEG sigma and beta power during NREM sleep in primary insomnia

    Biol Psychol

    (2012)
  • N. Srinivasan

    Cognitive neuroscience of creativity: EEG based approaches

    Methods

    (2007)
  • R. Ferri et al.

    Periodic leg movements during sleep and periodic limb movement disorder in patients presenting with unexplained insomnia

    Clin Neurophysiol

    (2009)
  • M. Irwin et al.

    Nocturnal catecholamines and immune function in insomniacs, depressed patients, and control subjects

    Brain Behav Immun

    (2003)
  • F. Jurysta et al.

    The impact of chronic primary insomnia on the heart rate–EEG variability link

    Clin Neurophysiol

    (2009)
  • L. Parrino et al.

    Paradoxical insomnia: the role of CAP and arousals in sleep misperception

    Sleep Med

    (2009)
  • R.J. Salin-Pascual et al.

    Caffeine challenge in insomniac patients after total sleep deprivation

    Sleep Med

    (2006)
  • A.N. Vgontzas et al.

    Chronic insomnia is associated with a shift of interleukin-6 and tumor necrosis factor secretion from nighttime to daytime

    Metabolism

    (2002)
  • M.C. Lopes et al.

    Non-REM sleep instability in patients with major depressive disorder: subjective improvement and improvement of non-REM sleep instability with treatment (Agomelatine)

    Sleep Med

    (2007)
  • S. Leistedt et al.

    Characterization of the sleep EEG in acutely depressed men using detrended fluctuation analysis

    Clin Neurophysiol

    (2007)
  • V. Pillai et al.

    A meta-analysis of electroencephalographic sleep in depression: evidence for genetic biomarkers

    Biol Psychiat

    (2011)
  • C. Baglioni et al.

    Psychophysiological reactivity to sleep-related emotional stimuli in primary insomnia

    Behav Res Ther

    (2010)
  • C.F. Reynolds et al.

    The DSM-V sleep–wake disorders nosology: an update and an invitation to the sleep community

    Sleep

    (2010)
  • B. Zhang et al.

    Sex differences in insomnia: a meta-analysis

    Sleep

    (2006)
  • C.M. Morin et al.

    The natural history of insomnia: a population-based 3-year longitudinal study

    Arch Intern Med

    (2009)
  • R.C. Kessler et al.

    Insomnia and the performance of US workers: results from the America insomnia survey

    Sleep

    (2011)
  • R.J. Ozminkowski et al.

    The direct and indirect costs of untreated insomnia in adults in the United States

    Sleep

    (2007)
  • R. Manber et al.

    Insomnia and depression: a multifaceted interplay

    Curr Psychiatry Rep

    (2009)
  • J.M. Monti et al.

    Sleep disturbance in schizophrenia

    Int Rev Psychiatr

    (2005)
  • C.E. Carney et al.

    The consensus sleep diary: standardizing prospective sleep self-monitoring

    Sleep

    (2012)
  • AASM (American Academy of Sleep Medicine Work)
  • A. Rechtschaffen et al.

    A manual of standardized terminology, scoring system for sleep stages of human subjects

    (1968)
  • C. Iber et al.
  • S.J. Wilson et al.

    British Association for Psychopharmacology consensus statement on evidence-based treatment of insomnia, parasomnias and circadian rhythm disorders

    J Psychopharmacol

    (2010)
  • C.A. Kushida et al.

    Practice parameters for the indications for polysomnography and related procedures: an update for 2005

    Sleep

    (2005)
  • M.H. Bonnet et al.

    24-Hour metabolic rate in insomniacs and matched normal sleepers

    Sleep

    (1995)
  • M.L. Perlis et al.

    Beta EEG activity and insomnia

    Sleep Med Rev

    (2001)
  • E.A. Nofzinger et al.

    Functional neuroimaging evidence for hyperarousal in insomnia

    Am J Psychiat

    (2004)
  • K. Spiegelhalder et al.

    Heart rate and heart rate variability in subjectively reported insomnia

    J Sleep Res

    (2011)
  • Cited by (0)

    The most important references are denoted by an asterisk.

    View full text