In adults, bipolar disorder occurs less commonly than unipolar depression but, when present, comprises an extremely impairing condition, with a strong likelihood of recurrence of mood episodes during a person's life and a high potential for disruption to educational and vocational functioning as well as interpersonal relationships (e.g., Coryell et al., 1993). With a mean age of onset in the early 20s, it is often life threatening, with approximately one in five individuals completing suicide and with rates of attempts far higher (Isometsä, 1993). The lifetime prevalence for bipolar I and II is estimated to be 3.9%, based on the National Comorbidity Survey Replication (Kessler et al., 2005), although more liberal definitions of hypomania identify many more patients with bipolar spectrum disorder (Angst, Gamma, & Endrass, 2003). It is crucial that both pharmacologic and adjunctive psychosocial interventions can produce major clinical improvements for individuals with bipolar disorder (Craighead, Miklowitz, Frank, & Vajk, 2002; Miklowitz & Johnson, 2006), although treatment adherence remains a key issue.
There is growing consensus that bipolar spectrum disorders do occur in children and adolescents, that they impose serious impairments for such youth, and that they represent a significant public health concern (Geller & DelBello, 2003). Although the typical age of onset for the pediatric bipolar phenotype is not yet well established, one carefully diagnosed sample reported an average onset of mania at 6.8 years old (Tillman et al., 2003). Estimating the prevalence of this phenotype across childhood and adolescence is difficult given the current state of the literature. Existing studies providing such estimates tend to differ across a number of important domains including diagnostic criteria, symptom measures, and sampling methods (i.e., clinical vs. community). With such caveats in mind, one large community sample reported that by high school, rates are approximately equal to those found in adulthood (Lewinsohn, Klein, & Seeley, 1995).
To clarify the diagnostic ambiguity surrounding bipolar in childhood and adolescence, investigators have begun to work diligently toward documenting its phenomenology. From these studies, it appears that childhood-onset bipolar disorder is often characterized by very rapid cycling (Geller et al., 1998), with coexisting irritability and elation (Geller et al., 2003), as well as significant mood lability and aggression (Faedda, Baldessarini, Glovinsky, & Austin, 2004; Geller et al., 2002). In fact, some investigators believe that many child and adolescent cases are marked less by “classic” cycles of mania and depression and more by rather chronic displays of irritable, explosive, and “mixed” mood states (Biederman et al., 2004). In addition, it appears that the childhood phenotype may include aspects of sleep disturbance consistent with those exhibited by adults with bipolar disorder (Geller et al., 2002). Although evidence has started to accrue to suggest a possible causal role for sleep and/or circadian rhythm dysfunction in the multifactorial causation of bipolar disorder in adults—and particularly to the maintenance of this condition—the applicability of these important findings to childhood-onset bipolar is not yet known.
The aim of this paper is to discuss the potential importance of the sleep–wake cycle in children and adolescents with bipolar disorder. We present a review of the evidence suggestive of a possible causal role for sleep and/or circadian rhythm dysfunction in adults. We do this in two sections: (a) empirical results that together converge on the hypothesis that sleep and/or circadian rhythms are one possible variable of considerable influence leading to relapse in bipolar disorder, and (b) theoretical perspectives on why sleep and/or circadian rhythms constitute such important mechanisms. We then move on to review the small empirical literature on sleep in children and adolescents who have a diagnosis of bipolar disorder. An agenda for future research in this area is next offered, including a discussion of the possible adaptation of existing evidence-based interventions for sleep disturbance. To initiate discussion, we provide an overview of human sleep, with an emphasis on the large changes in sleep patterns across development.
A Primer on Human Sleep
Human sleep can be divided into nonrapid eye movement (NREM) sleep, which can be subdivided into four stages (Stages 1, 2, 3, and 4), through which sleep progressively deepens, followed by rapid eye movement (REM) sleep. That is, in adults, sleep follows an organized pattern starting with Stage 1 NREM sleep, deepening down to Stage 4 NREM sleep, and then moving into REM, with each NREM–REM cycle spanning 70–120 min (Hirshkowitz, Moore, & Minhoto, 1997; Shneerson, 2000).
The sleep–wake cycle is regulated by an interaction between two processes. The first is the circadian rhythm, which is driven by small structures in the hypothalamus called suprachiasmatic nuclei (SCN). The circadian rhythm can be estimated by measuring fluctuations in substances like melatonin and cortisol and in core body temperature, which are all regulated by the SCN across the 24-hr day. The second is the homeostatic process, which is determined by prior sleep and wakefulness. Specifically, the longer a person stays awake, the more homeostatic pressure exists to sleep. This pressure then resolves with sleep. It is the interaction of the circadian rhythm and the homeostatic process that determines the temporal distribution and duration of sleep and wakefulness. During the day they act together to help maintain wakefulness, and during the night they act together to promote sleep (Borbely, 1982; Borbely & Achermann, 2005).
Measurement of sleep
In this section we present the most commonly used methods. In recognition of the complex issues involved in assessing sleep (Smith, Nowakowski, Soeffing, Orff, & Perlis, 2003), a combination of methods is typically employed. Together, these measures (a) each provide a unique viewpoint, (b) enable sampling prospectively and retrospectively, and (c) provide both objective and subjective estimates, which are often discrepant from each other and may reflect different processes (Argyropoulos et al., 2003; Armitage, Trivedi, Hoffmann, & Rush, 1997).
Objective estimates
The gold standard measure of sleep is polysomnography (PSG). PSG is used to classify sleep into the aforementioned stages. It involves placing surface electrodes on the scalp and face to measure electrical brain activity (electroencephalogram), eye movement (electrooculogram), and muscle tone (electromyogram). The data obtained are used to classify each epoch of data by sleep stage and in terms of sleep cycles (NREM and REM). Disadvantages associated with PSG include its expense, discomfort for participants, and labor-intensive nature.
Actigraphy is an alternative means of providing an objective estimate of sleep. Actigraphs are small devices worn on the wrist, within which are located a sensor, a processor, and memory storage. The sensor samples physical motion; the processor translates it into numerical digital data, summarizing the frequency of motions into epochs of specified time duration and storing the summary in memory. These data are then downloaded to a computer and analyzed to generate various sleep parameters. Because the body becomes more quiescent during sleep, actigraphy can be used to differentiate between periods of wakefulness and periods of sleep. In fact, the correlation between actigraphy- and polysomnography-defined estimates of total sleep time is quite strong, ranging from 0.88 to 0.97 in adult nonpatients (Cole, Kripke, Gruen, Mullaney, & Gillin, 1992; Jean-Louis et al., 1997; Sadeh, Alster, Urbach, & Lavie, 1989). The accuracy of actigraphy in identifying sleep and wakefulness is 82% in adult nonpatients (Blood, Sack, Percy, & Pen, 1997), with accuracy referring to the proportion of polysomnography epochs (sleep and wake) correctly identified by actigraphy. Although most of the validation studies have been conducted on adults, previous research has established the utility of actigraphy in measuring the sleep of infants and children (Sadeh, Hauri, Kripke, & Lavie, 1995; Sadeh, Lavie, Scher, Tirosh, & Epstein, 1991) and in adolescents (Axelson et al., 2003).
The key advantages of actigraphy are that it allows long-term monitoring and that it is minimally intrusive. Deriving the sleep estimate based on movement information means, however, that actigraphy is a relatively “blunt” instrument, with particular difficulties in detecting the precise onset of sleep. Indeed, accuracy is reduced for patients with depression or insomnia who can often lie still for long periods (e.g., Kushida et al., 2001). In addition, it does not allow sleep to be staged (i.e., cannot be used to determine Stages 1–4 or REM).
Subjective estimates
Subjective reports of sleep quantity and quality are important measures. Sleep diaries, completed immediately on waking, are highly correlated with objective measures of sleep (Wilson, Watson, & Currie, 1998). Moreover, individuals' perceptions of their own sleep (i.e., how they believe or feel they have slept) have been shown to affect their daytime functioning (Neitzert Semler & Harvey, 2005).
In addition, several questionnaire measures have been validated to index the presence of sleep disturbance (e.g., Pittsburgh Sleep Quality Index; Buysse, Reynolds, Monk, Berman, & Kupfer, 1989), insomnia (e.g., Insomnia Severity Index; Bastien, Vallieres, & Morin, 2001), and daytime sleepiness (e.g., Stanford Sleepiness Scale; Hoddes, Zarcone, Smythe, Phillips, & Dement, 1973). However, it should be noted that none of these questionnaires have been specifically developed for, or validated on, children or adolescents. Indeed, there are only a handful of questionnaires designed and validated as measures of the sleep of children and adolescents. One such measure is the parent-report Child Sleep Habits Questionnaire (Owens, Spirito, & McGuinn, 2000), which screens for commonly reported sleep problems in children aged 4 to 12 years. A complementary self-report questionnaire for children over 7 years old, known as the Sleep Self-Report, has also been developed (Owens, Maxim, Nobile, McGuinn, & Msall, 2000), although psychometrics are not yet available. A second measure is the Sleep Habits Questionnaire, a 20-item measure that provides an overview of a child's sleep habits, sleepiness, and fatigue. It is completed by the child and the parent (Gruber, Sadeh, & Raviv, 1997), and has good psychometric properties (Sadeh, Raviv, & Gruber, 2000).
Sleep diaries provide prospective estimates of sleep (or, at least, concurrent estimates),1
A caveat to the claim that the sleep diary is prospective is that it is completed immediately on waking. As such, it involves a retrospective judgment to be made over the past 6–10 hr.
Functions of sleep
The function of sleep continues to be debated (Dement, 1999), with NREM sleep thought to be important for conservation of energy and restoration, given that cell division in some tissues is most rapid and protein synthesis is increased during this phase of sleep (Shneerson, 2000). Hypotheses about the function of REM include that it is involved in learning (Karni, Tanne, Rubenstien, Askenasy, & Sagi, 1994; Smith, Huang, & Manber, 2005) and memory consolidation (Dewan, 1968; Karni et al., 1994; Smith & Lapp, 1991), and may also be a mechanism for unlearning irrelevant information (Crick & Mitchison, 1983).
An additional likely function of REM, and perhaps the one most relevant to the topic of this paper, is that it is thought to play a role in emotional processing and mood/emotion regulation (Cartwright, Luten, Young, Mercer, & Bears, 1998; Cartwright, Young, Mercer, & Bears, 1998; Perlis & Nielsen, 1993). Perlis and Nielsen (1993) summarized three lines of empirical support for the hypothesis that REM plays a key role in emotional processing and mood/emotion regulation. First, abnormal REM sleep can be observed across various groups of patients diagnosed with a psychiatric disorder including schizophrenia, major depression, borderline personality disorder, and eating disorder. Of course, causal relations are impossible to discern from such associations, but it is tempting to conclude that disruptions in REM sleep are associated in more than a casual fashion with important psychiatric conditions.
Second, presleep mood and stress appear to influence the content of dreaming, the emotion within a dream, and latency to REM as well as REM density (e.g., Breger, Hunter, & Lane, 1971; Cartwright, Bernick, Borowitz, & Kling, 1969; De Koninck & Brunette, 1991). In other words, stress, mood and emotion appear to be linked directly to important sleep parameters.
Third, REM sleep and dreaming have been shown to influence daytime mood. Evidence for this linkage can be derived from REM deprivation studies, which produce anxiety, irritability, and reduced adaptability the following day (see Perlis & Nielsen, 1993, for a review of these studies). Prospective studies, such as those conducted by Cartwright and colleagues, provide additional evidence for this third point. To give one example, Cartwright, Young, et al. (1998) reported the findings from an investigation of 61 people who had recently been separated from a marital partner. Each participant was assessed for depression at baseline and 1-year follow-up, and their sleep was monitored with PSG for three nights at each time point as well. At baseline, 39 of the participants were classified as suffering from depression. One year later, 72% of these people could be classified correctly as remitted or not remitted based on the presence of negative/unpleasant dreams at baseline. The direction of this relationship was that the more negative/unpleasant dreams were at baseline, particularly in the first half of the night, the less likely the person suffered from depression 1 year later. Presumably, unpleasant dreams reflect effective processing and resolution of the emotional issues surrounding the separation. Additional research is required both to replicate such a provocative finding and understand underlying mechanisms.
The emerging neuroscience literature has demonstrated the presence of amygdala activity during REM sleep (e.g., Maquet et al., 1996). As the amygdaloid structures have been implicated in the acquisition of emotional memories, this observation is also consistent with the possibility that REM sleep has a role in emotion-related regulation.
Moving away from the focus on REM sleep, there is a further line of evidence that lends support to the hypothesis that sleep has an important emotion regulatory function. That is, one of the strongest adverse effects of sleep deprivation is increased negative mood (Dinges et al., 1997; Pilcher & Huffcutt, 1996; Van Dongen, Maislin, Mullington, & Dinges, 2003). A recent study by Zohar, Tzischinsky, Epsten, and Lavie (2005) suggests that context, particularly extent of progress toward/away from valued goals, may have an influence on the direction of the mood effect. These researchers examined the relationship between sleep loss and emotion reactivity in 78 medical residents who were monitored for 5 to 7 days every 6 months over a 2-year period. The results indicated that sleep loss not only intensified negative emotions following a goal-thwarting event, but also following a goal-enhancing event, diminished positive emotions.
Summary
This section has aimed to provide a brief overview of the key concepts in the sleep literature to which we will refer to throughout this paper. We described the architecture of sleep (Stages 1–4 and REM), the measurement of sleep (covering subjective and objective estimates), and the possible functions of sleep (emphasizing the effects on mood). We now move on to consider the important differences in the sleep of children and adolescents, relative to adults.
A Developmental Perspective on Sleep
Whereas the previous section focused almost entirely on adult sleep, here we emphasize that sleep patterns change enormously across the life cycle. In newborns, average total sleep time is approximately 16 to 18 hr, organized into 3- to 4-hr sleep periods across the 24-hr cycle. The average amount of sleep obtained by a 5-year-old is 11.1 hr and in a 9-year-old is 10.2 hr (Hoban, 2004). In adolescence, nighttime sleep reduces from an average of 9 hr at age 13 to 7.9 hr at 16 (Hoban, 2004). In young adults, average nighttime sleep varies between 7 and 9 hr and in the middle adult years between 6 and 8 hr. Nocturnal sleep is further reduced in older adults, but there is an increase in daytime napping (Shneerson, 2000).
Accompanying these normative changes in the amount of sleep are alterations in sleep architecture over the course of development. Newborn infants are thought to start sleep with REM and then move into NREM, with each REM–NREM cycle lasting about 50 min (Carskadon & Dement, 2005). In newborns, REM and NREM phases are called “active” and “quiet” sleep, respectively, because of the difficulty in differentiating sleep stages at this age. Whereas at birth approximately 50% of sleep is spent in active sleep, once a child is 2 years of age this percentage is reduced to 20–30% of total sleep time. Between the ages of 6 and 11, the amount of Stage 3 and 4 sleep is reduced and Stage 2 sleep is increased (Hoban, 2004). Across the adolescent years, the “adult” sleep cycle length becomes established, with Stages 3 and 4 further decreasing in length, accompanied by increases in Stage 2 sleep (Carskadon & Dement, 2005).
A key feature of sleep preceding and during the adolescent years is the delay in circadian phase and corresponding delay in sleep onset, often shifting past midnight to the early morning hours (Carskadon, 2002; Tate, Richardson, & Carskadon, 2002). This effect may be attributable to a number of influences, which include a tendency toward increasing autonomy in deciding what time to go to bed, which coincides with both a natural biological delay in the circadian cycle plus irregularity in the sleep schedule associated with psychosocial stress and social activities (Carskadon, 2002; Hoban, 2004). This tendency toward a delayed circadian phase, combined with early school start times (Carskadon, Wolfson, Acebo, Tzischinsky, & Seifer, 1998) and paid work responsibilities (Rosa, 2002), means that adolescents often do not obtain sufficient sleep. In addition, a recent survey of 1,602 adolescents (11–17 years old) conducted by the National Sleep Foundation (2006) found that 90% of respondents had electronic music devices in their bedroom, 57% had a television, 42% had a cell phone, and 21% had internet access. Of course, such devices hold great potential appeal for activating social interaction once parents have retired to sleep (National Sleep Foundation, 2006). This state of affairs is of critical concern, as there is thought to be an increased biological need for sleep associated with pubertal development (Dornbusch, 2002), along with the important role for sleep in mood and emotion regulation (e.g., Perlis & Nielsen, 1993), learning, and memory consolidation (e.g., Karni et al., 1994).
Summary
The changes in sleep across the life cycle are striking. There are clear changes in total sleep time and sleep architecture across development, and at each stage of development new challenges are faced. In particular, around the time of adolescence, there are multiple pressures on sleep from (a) the natural delay in the circadian phase, (b) reduced parental influence on time of getting into bed, and (c) the greater participation in social events.
Why Are Sleep and Circadian Rhythms Important in Bipolar Disorder?
Sleep disturbance versus circadian rhythm
Before launching into this section, we emphasize that sleep disturbance and circadian rhythm disturbance are not identical processes. Although there is an interaction between them, the amount and patterning of sleep depends on many factors (e.g., life stress) of which the circadian rhythm is only one (Reid & Zee, 2005). As will become evident in this section, both sleep disturbance and circadian rhythm disturbance have been implicated as key processes in the multifactorial causal pathways related to bipolar disorder.
Empirical evidence
There are at least three lines of evidence that converge to highlight the key importance of sleep disturbance in bipolar disorder.
First, sleep disturbance is a core symptom of bipolar disorder. During mania there is often a reduced need for sleep, and during depression people may suffer insomnia or hypersomnia (American Psychiatric Association, 2000; Hudson et al., 1992). A recent study suggests that sleep disturbance is common even when patients with bipolar disorder are between episodes (Harvey, Schmidt, Scarna, Semler, & Goodwin, 2005). However, relative to the interepisode phase, sleep disturbance escalates just before an episode and worsens further during an episode (American Psychiatric Association, 2000; Jackson, Cavanagh, & Scott, 2003; Wehr, Sack, & Rosenthal, 1987). In short, sleep appears to be a sensitive trait marker and state marker for affective illness.
Second, a recent meta-analysis concluded that sleep disturbance is the most common prodrome of mania and the sixth most common prodrome of depression (Jackson et al., 2003). A prodrome is a symptom or sign that precedes an episode (Jackson et al., 2003). As such, sleep difficulties can serve as an important clinical signal of incipient episodes.
Third, experimentally induced sleep deprivation is associated with the onset of hypomania or mania in a proportion of patients (Colombo, Benedetti, Barbini, Campori, & Smeraldi, 1999; Kasper & Wehr, 1992; Leibenluft, Albert, Rosenthal, & Wehr, 1996; Wehr et al., 1987), and sleep loss is highly correlated with daily manic symptoms (Barbini, Bertelli, Colombo, & Smeraldi, 1996). Although the rate of switch reported across these studies is variable (Colombo et al., 1999; Wu & Bunney, 1990), in a recent review Boivin (2000) declared “the dramatic effect of sleep deprivation and sleep restoration on mood swings” to be an important clinical feature of bipolar disorder (p. 452).
We recognize that each of these lines of evidence varies in the extent to which it provides convincing evidence of a causal relationship (Bradford Hill, 1965), with the third point (experimental evidence) being the most suggestive of a causal relationship between sleep and relapse in bipolar disorder. Nonetheless, when considered together, these lines of evidence converge on the conclusion that sleep disturbance and/or problems with the circadian rhythm are correlates of mood disorders, particularly bipolar disorder, and raise the hypothesis that sleep disturbance may, in fact, serve as one important aspect of the multifactorial causal chain leading to relapse in bipolar disorder (Goodwin & Jamison, 1990; Wehr et al., 1987). Whether sleep disturbance is causal with respect to initial episodes—that is, to the onset of the condition—is less well established.
Mechanisms
In this section we review several of the existing theoretical viewpoints regarding the mechanisms by which sleep disturbance is associated with relapse in bipolar disorder (see Rao, 2003, for a review of others). The distinction we drew earlier in this section between the sleep–wake cycle and the circadian cycle as overlapping but separable processes should be kept in mind while reading this section, as should the evidence for the potential role of REM sleep in emotional processing and mood/emotion regulation. As will become evident, the first three theories emphasize the role of the circadian cycle and the fourth recognizes a role for the sleep–wake cycle.
Abnormality in circadian functioning: An emphasis on the timing of sleep
The theory that has perhaps attracted the most attention in the context of bipolar disorder was proposed by Wehr, Wirz-Justice, and colleagues (Wehr, 1990, 1991; Wehr et al., 1987; Wehr & Wirz-Justice, 1981). Their hypothesis is that the timing of the sleep–wake schedule is critical. They draw on evidence that the risk of depression is greatest when sleep coincides with a circadian phase that is most vulnerable to sleep's depressant effects. Specifically, the data reveal that advancing the timing of the sleep schedule 6 hr earlier improves depression (e.g., Wehr, Wirz-Justice, Goodwin, Duncan, & Gillin, 1979). In contrast, delaying the timing of sleep induces depressive symptoms (e.g., Surridge-David, MacLean, Coulter, & Knowles, 1987).
One mechanism that may underlie these sleep timing effects centers on REM sleep. A robust finding is that after falling asleep, the latency to the first REM episode is shorter in patients with depression, relative to individuals without depression (see Boivin, 2000, for review). Hence, REM sleep occurs earlier in depressed patients. Based on this finding, it has been reasoned that by advancing the timing of sleep, the time to the first REM episode will be delayed, and this delay should improve depression. In contrast, by delaying the timing of sleep, the time to the first REM episode will be advanced (i.e., REM latency will be even shorter) and this process should intensify depression (Rao, 2003).
Abnormality in circadian functioning: An emphasis on the amount of sleep
Wu and Bunney (1990) reviewed two key findings: relapse into depression often follows a full night of sleep in the majority of patients who have shown clinical improvement following sleep deprivation, and reducing the amount of time slept is associated with the onset of hypomania or mania in a proportion of patients (Colombo et al., 1999; Kasper & Wehr, 1992; Leibenluft et al., 1996; Wehr et al., 1987). Based on these findings it has been proposed that a depressogenic substance may be released during sleep, which is metabolized or stored during the day, or a euphorogenic substance may be released during wake periods, which is metabolized or stored during sleep (see Rao, 2003, for review). To the best of our knowledge, no such substance has, as of yet, been identified.
Social zeitgeber theory
Several theorists have proposed that people with bipolar disorder have a genetic vulnerability to circadian and sleep–wake cycle disruption (Ehlers, Frank, & Kupfer, 1988; Frank, Swartz, & Kupfer, 2000; Goodwin & Jamison, 1990). Social rhythms are the routines that are associated with personal relationships, social demands, or tasks that entrain biological rhythms. In other words, social rhythms are conceptualized as operating as “timekeepers” or “zeitgebers.” Instability in social rhythms are proposed to disrupt the circadian rhythm and the sleep–wake cycle, culminating in an episode of mania or depression. The evidence pertaining to the importance of social rhythms is steadily accumulating. Disregarding the severity of the life event, life events characterized by significant disruption to social rhythms uniquely predict the onset of manic episodes, but not depression, among individuals with bipolar disorder. However, when the severity of the stressful life events is considered, severe events are related to the onset of both manic and depressive episodes (Malkoff-Schwartz et al., 1998, 2000). This theoretical perspective has resulted in the development of interpersonal and social rhythm therapy (IPSRT), an adjunctive intervention tested on adults with bipolar disorder, as will be discussed below.
Emotion regulation
The evidence reviewed earlier suggesting that sleep has an important role in emotion regulation (e.g., Perlis & Nielsen, 1993) provides a further mechanism by which sleep disturbance may contribute to bipolar disorder. Emotion regulation generally refers to the multiple processes that initiate, sustain, intensify, and modulate the experience and expression of emotions, often in the service of a particular goal (Eisenberg et al., 1997; Thompson, 1994). To the best of our knowledge, a hypothesis linking sleep to emotion dysregulation has not been tested in the context of bipolar disorder, although we note evidence that bipolar disorder is fundamentally a disorder of emotion regulation (Goodwin & Jamison, 1990; Hyman, 2000); that sleep disturbance is characteristic of patients with bipolar disorder, even when they are euthymic (Harvey et al., 2005); and that sleep loss is highly correlated with daily manic symptoms (Barbini et al., 1996). We suggest, therefore, that it is plausible that sleep disturbance promotes emotion dysregulation, which would be a negative occurrence for persons at risk for bipolar disorder.
Sleep and/or Circadian Rhythms in Children and Adolescents With Bipolar Disorder
Thus far only a handful of empirical studies have been conducted assessing the sleep of children and adolescents with bipolar disorder and, to the best of our knowledge, none of the theoretical frameworks relating to the mechanisms just discussed have been adapted to take account of the developmental changes in the sleep experienced by children and adolescents. Therefore, we begin by reviewing the empirical evidence that has accrued on sleep in children and adolescents with bipolar disorder and then proceed to raise a series of considerations for adapting existing theoretical frameworks and/or developing new ones specifically for children and adolescents with this condition.
Empirical evidence
In this section we review the relatively new literature on sleep and/or circadian functioning in children and adolescents who have a diagnosis of bipolar disorder. All of the studies conducted to date are descriptive in nature. Although providing few clues as to causal mechanisms, descriptive research fills an important gap in knowledge by documenting the relevance of sleep disturbance in youth diagnosed with bipolar disorder. We divide the first part of this section according to the methodology employed for measuring sleep.
Sleep estimated via self-report
As part of a large, longitudinal investigation of the prepubertal and early adolescent bipolar phenotype, Geller and colleagues (2003) compared the clinical features of children and adolescents, between the ages of 7 and 16 years old, with bipolar disorder (n = 93) or attention-deficit/hyperactivity disorder (ADHD; n = 81) without any diagnosis (n = 94). Bipolar disorder was diagnosed carefully using both a mother and child report on a psychometrically validated semistructured interview, the Washington University (St. Louis) Kiddie Schedule for Affective Disorders and Schizophrenia (Geller et al., 2001). This instrument covers each of the DSM-IV criteria for bipolar I or II, including the presence of at least one “cardinal” symptom (elation or grandiosity). Whereas little detailed information was reported for sleep problems, the children with a diagnosis of bipolar disorder slept less compared to the ADHD and control participants. Thirty-nine percent of the children diagnosed with bipolar disorder were reported to have difficulty in sleeping, compared to 6.2% for children diagnosed with ADHD, and 1.1% for the healthy controls.
Another recent study analyzed the clinical records of a sample of 82 children diagnosed with bipolar disorder, who were between ages 3 and 17 (M = 10.6 years; Faedda et al., 2004). The diagnoses were made using parent questionnaires combined with clinician assessment. The participants met modified DSM-IV criteria for bipolar disorder, with the episode duration criterion for both depressive and manic episodes omitted to permit the inclusion of rapid-cycling participants (who may be overrepresented in children and adolescents with bipolar disorder). Comorbidities were common in this sample, with nearly 60% of the sample meeting criteria for ADHD. Rates of sleep disturbance were elevated as well, with 42.6% of participants reported to display insomnia or a parasomnia. A parasomnia is a sleep disorder that is characterized by arousals during sleep (e.g., sleep waking, sleep talking, restless legs syndrome). Unfortunately, this study did not include a control group, and the measure of sleep relied solely on parental retrospective recall.
Sleep estimated via actigraphy
One recent study employed actigraphy to measure the sleep and circadian rhythms of a sample of 3- to 17-year-olds (M = 9.6) with bipolar disorder (Faedda, Teicher, Glovinsky, McGreenery, & Goldfarb, in press). Diagnostic status was determined via a questionnaire given to parents, employing DSM-IV criteria. Comorbidity rates were high; 95.8% of participants had at least one comorbid psychiatric disorder, with anxiety disorders presenting in 52% of cases. Other comorbidities present included oppositional defiant disorder and ADHD. Each participant wore an actigraph over 3–5 school days. The results consistently demonstrated the presence of a disrupted sleep pattern, with difficulties of both falling and staying asleep. Activity levels were also abnormal, with most of the participants demonstrating consistent diurnal and nocturnal hyperactivity. It is important that this study demonstrated the feasibility of actigraphy techniques with child bipolar samples. Because a comparison group was not included, it is difficult to gauge the significance of sleep and circadian abnormalities relative to other conditions or to normally developing youth. Moreover, although less than half of the sample was medicated during data collection, the data for medicated versus unmedicated participants were combined for analysis, potentially obscuring important effects of bipolar disorder per se.
Sleep estimated via polysomnography
To the best of our knowledge, only one study has employed polysomnography to analyze the sleep patterns of children with bipolar spectrum disorder. Mehl et al. (2006) administered the Child Behavior Checklist (CBCL; Achenbach, 1991) to screen a community sample of 438 children. Thirteen (3%) of these children exhibited a pediatric “bipolar disorder profile.” They were carefully matched with healthy children. During overnight polysomnography, bipolar profile children exhibited impaired sleep efficiency2
Sleep efficiency is the total sleep time divided by the total time in bed, multiplied by 100.
Unipolar depression
Although our literature search revealed only one study that had employed PSG to measure sleep among youth with bipolar disorder, a handful of studies have been conducted on youth with depression (Dahl et al., 1990; Emslie, Rush, Weinberg, Rintelmann, & Roffwarg, 1994; Puig-Antich et al., 1982; Young, Knowles, MacLean, Boag, & McConville, 1982). Only one of these studies (Emslie et al., 1994) found a significant difference between the sleep of depressed children and healthy controls, with depressed children exhibiting increased sleep latency and decreased latency to the first REM episode. Studies on adolescents are more plentiful (see review by Rao, 2003), with parallel findings: that is, adolescents with depression take longer to fall asleep and exhibit a shorter time interval to the first REM episode.
One of the adolescent depression studies is particularly interesting. Dahl et al. (1990) prospectively followed adolescents with unipolar depression 7 years later, when they were in their early 20s. A clinical interview was administered to determine their disease course (Rao et al., 2002). Five of the 26 patients who were diagnosed with unipolar depression at the first assessment had switched to bipolar disorder by the time of the follow-up. Based on these new categorizations, the original polysomnography data were reanalyzed. The results indicated that depressed participants who followed a unipolar course showed reduced REM latency, higher REM density, and increased overall REM sleep, whereas those who moved to a bipolar course demonstrated increased Stage 1 sleep and decreased Stage 4 sleep. Although these results do not necessarily pertain specifically to the child or adolescent bipolar phenotype, they do provide preliminary evidence of adolescent sleep abnormalities that distinguish bipolar from unipolar trajectories.
Children of parents with bipolar disorder
A small body of literature has begun to examine the sleep patterns of children with affectively ill parents, a group that may be at significant risk for developing bipolar disorder. To the best of our knowledge, two of these studies have included measures of sleep.
Stoleru, Nottelmann, Belmont, and Ronsaville (1997), using a subjective measure of sleep, were able to detect significant sleep disturbance in the children of affectively ill mothers. This study followed a cohort of children of mothers with a diagnosis of bipolar disorder, major depression, and healthy controls, assessing them three times over a 4-year span using the CBCL. Both groups of children from affectively ill mothers demonstrated more severe and persistent sleep problems (i.e., present at more than one assessment point), relative to their healthy offspring peers.
A recent study by Jones et al. (2006) used both objective and subjective sleep estimates to compare adolescent offspring of parents diagnosed with bipolar disorder with children of healthy parents. The participants (ages 13–19 years) wore actigraphs for 7 days and nights, and also completed questionnaire assessments of sleep quality. Fourteen of the 19 bipolar offspring participants had current or lifetime mood diagnoses, whereas only two children of healthy parents reported past or current depression. Somewhat surprisingly, the actigraph results indicated that bipolar offspring exhibited shorter sleep onset latency, slightly greater sleep duration, and less variable sleep fragmentation than their peers from healthy parents. In contrast, their self-ratings revealed higher rates of sleep disturbance than their peers. The authors suggested that early dysregulation of sleep (in the form of objective sleep appearing to be adequate on objective estimates but not being perceived to be adequate) may be an early marker for bipolar disorder. However, given the relatively small sample size, the authors urged caution in interpreting the findings.
Although this domain of research is at a very early stage of development, studying sleep problems in the offspring of parents with bipolar disorder could potentially be fruitful. This group is at risk for the development of bipolar disorder, as well as myriad other forms of psychopathology (Chang, Steiner, & Ketter, 2000). It may be the case that examining sleep and circadian functioning in these children will reveal important information about the relevance of such processes to the pathogenesis of this disorder.
Theoretical considerations
As we have already stated, existing theoretical frameworks specifying potential mechanisms underlying the linkage between sleep and bipolar disorder are based almost exclusively on research with adults. Hence, the challenge ahead will be to derive theories that are sensitive to the considerable developmental changes taking place, in sleep and emotion/mood regulation ability, across childhood and adolescence. In this next section we offer a number of considerations that we see as important in such an endeavor.
The preliminary evidence just reviewed documents that, similar to adults with bipolar disorder, children and adolescents with this condition also display significant sleep disturbance. We therefore proceed to raise several considerations for adapting existing theoretical frameworks and/or developing new ones specifically for children and adolescents with bipolar disorder.
At the outset, a point that is relevant to both children and adolescents relates to the accruing evidence that sleep has an important role in learning (Maquet, 2001; Stickgold, Hobson, Fosse, & Fosse, 2001). It is also associated with daytime sleepiness and mood regulation difficulties (Carskadon, Acebo, & Jenni, 2004). This evidence is important because even if sleep disturbance turns out not to be causal to mood and emotional disturbances in children and adolescents with bipolar disorder, it remains of concern because of potential contributions to impairment at school (Carskadon et al., 1998; Dahl, 1996a, 1996b) and to peer relationship problems. Treatment implications related to both core bipolar symptomatology and to associated sleep problems are therefore paramount.
It is crucial that the causal relationships between sleep problems and the well-being of children and adolescents appear to be complex and bidirectional (Carskadon et al., 1998; Dahl, 1996a, 1996b; Sadeh, 2005; Sadeh et al., 2000), meaning that not only may sleep problems contribute to psychopathology, but also that behavioral and emotional problems may contribute to disrupted sleep. This bidirectionality is particularly well articulated in the work of Dahl and colleagues.
Dahl (2004) describes adolescence as “a unique developmental period” and a time of great “vulnerability and opportunity.” Of course, the majority of adolescents progress through these years, often helped by “social scaffolding” from parents and other adults. However, Dahl highlights that rates of disability and death rise 200% during adolescence and raises the hypothesis that these alarming statistics may be related to a tendency toward (a) erratic and emotionally influenced behavior and (b) affect regulation difficulties, defined as the processes involved in modulating one's feelings “in adaptive ways in order to achieve goals, to act with the norms of social rules and expectations and in ways that support rather than interfere with decision-making” (p. 19). Dahl (2004) also distinguishes between “hot cognition” and “cool cognition” and suggests that under conditions of cool cognition adolescents have the potential for excellent cognitive and reasoning capacities to make wise judgements. However, conditions of hot cognition are much more challenging; the origin of this challenge may relate to a cascade of maturational changes in the neural and hormonal systems underlying emotion and motivation, in combination with psychosocial challenges caused by the waning of parental influence and the encountering of other new challenges, such as increasing pressure in academic work, first romantic relationships, and substantial paid work responsibilities.
Dahl (2002) has referred to the critical task for adolescents as the integration of cognitive and emotional processes such that “an optimal link between the relevant cognitive processes and the appropriate emotional processes” can be achieved (p. 277). It is this level of cognitive–emotional functioning that is thought to be particularly vulnerable to disruption from the effects of sleep deprivation. Adolescents are prone, in fact, to get caught in an escalating vicious cycle whereby sleep disturbance leads to difficulty in regulating emotion, and the consequences of emotion regulation difficulties include increased arousal, which can produce further sleep disturbance (Dahl, 2002). We speculate that these predictions may be highly relevant to youth with bipolar disorder, for whom high levels of daytime emotionality would be expected to establish a marked and severe vicious bidirectional cycle between daytime emotional dysregulation and nighttime sleep disturbance.
Agenda for Future Research
The likely importance of sleep and/or circadian rhythm functioning among children and adolescents with bipolar disorder is indicated by the preliminary evidence, from descriptive studies, indicating the presence of sleep disturbance and from the developmental literature indicating that (a) sleep patterns change markedly over the course of development, and (b) sleep is crucial for optimal daytime functioning. In the hope of cultivating research in this domain, we offer a range of avenues that we see as particularly important targets for future research.
Descriptive research
Many more basic descriptive studies are needed to clearly document the sleep patterns of youth with bipolar disorder. These investigations should include multiple measures of sleep that encompass subjective indicators (e.g., sleep diaries), objective markers (e.g., actigraphy and/or PSG), and estimates of sleep based on retrospective reports as well as prospective reports. Each provides a unique insight and all are important for a comprehensive assessment.
Sleep studies conducted across changes in mood state are needed to document the differential associations between sleep during manic, depressed, mixed, and euthymic states. Such studies, which require prospective data collection, will provide an opportunity to check the extent to which the robust finding in the adult literature, that sleep is an important prodrome of episode onset (Jackson et al., 2003), replicates among children and adolescents with a diagnosis of bipolar disorder or those at risk for bipolar disorder. But we recognize once again that youth diagnosed with bipolar disorder may well display less clear episode distinction and instead a more chronic pattern of disturbance (or very rapid cycling; Biederman et al., 2004). As a result, it may be difficult to delineate prodromes.
Indeed, the reliability and validity of the diagnosis of bipolar disorder prior to adolescence is in question (Carlson, 2005). That is, it is uncertain whether child-onset bipolar disorder conforms to the classic pattern of alternating periods of mania and depression or whether it constitutes a more chronic pattern of affect dysregulation featuring either extremely rapid cycles and/or high levels of mixed agitation, anger, and dysphoria. Thus, how sleep may contribute to such forms of intermingled positive and negative affect is not yet certain.
The presence of sleep inertia among children and adolescents with bipolar disorder has been reported clinically. Sleep inertia is a normal transitional phase between sleep and wake, typically lasting from 5 to 30 min just after waking. Arousal is low and cognitive and performance decrements are observed during this period. Even after sufficient sleep, sleep inertia is experienced to a moderate degree on waking, with individual differences across the population. Yet, high levels of sleep inertia may be indicative of insufficient sleep; they may also reflect the stage of sleep the person was in immediately prior to waking, with Stages 3 and 4 sleep just before waking resulting in the most sleep inertia (Tassi & Muzet, 2000). Given the clinical reports, we suggest that including an assessment for the presence of high levels of sleep inertia is a potentially fruitful domain in future studies of children and adolescents with bipolar disorder.
Comorbidity
A striking result from studies of child and adolescent bipolar disorder is the high rates of comorbidity reported (e.g., Faedda et al., in press). This finding is consistent with reports of adults with bipolar disorder (Kessler, Rubinow, Holmes, Abelson, & Zhao, 1997). Hence, in future research it will be important to carefully assess and report the presence of threshold and subthreshold comorbid conditions in order address whether sleep problems are related specifically to bipolar disorder versus other conditions that coexist with this serious mood disorder. Without careful ascertainment of the role of comorbidity, research in this domain will be plagued by the murkiness of nonspecificity.
Medication effects on sleep: Implications for research
The contribution of psychoactive medications for bipolar disorder is likely to be an important variable. As will be discussed in more detail below, most medications commonly prescribed for bipolar disorder are sedating and/or alter sleep architecture. Because a medication-free group of youth with bipolar disorder would be likely to be unrepresentative, and because medication-free periods are not easily attained (given both medication half-life issues and potential safety concerns), research participants cannot be excluded on the basis of medications prescribed for bipolar disorder in most instances. Strategies to overcome these difficulties include keeping a careful record of medication use throughout the duration of the study and documenting any changes over the course of the study. In addition, in collaboration with the participant's physician and if not compromising the safety and optimal functioning of the participant, it may be possible to keep medication constant for the duration of the study. In rare cases, medication free periods could be considered, but these are not likely to be routine in research on bipolar disorder. Clearly, investigators should use caution when interpreting sleep data from medicated samples. Remaining cognizant of medication effects on sleep patterns and architecture, combined with careful study design, should still enable investigators to draw substantial conclusions about sleep irregularities in child and adolescent bipolar disorder, independent of medication effects.
Sleep versus circadian influences
It will also be important for future research to parse the independent contribution of sleep versus circadian influences. Such research will be important for testing the theoretical frameworks presented earlier. Basic sleep research provides a methodology by which separation of function is possible (Czeisler, Buxton, & Khalsa, 2005). In the forced desynchrony protocol, the participant resides in a strictly controlled environment. Exposure to light, the length of the day, the amount of activity in which the person engages, and the intake of food and drink are closely controlled. Of importance, the participant does not have access to external time cues. The key feature is that a non-24-hr sleep–wake schedule is imposed. That is, a schedule that is shorter or longer than the circadian rhythm is introduced by controlling exposure to light and access to sleep. During the procedure continuous measures of the circadian rhythm are taken, typically including rectal temperature, melatonin, and cortisol. With the addition of measures of mood (e.g., Koorengevel, Beersma, den Boer, & van den Hoofdakker, 2003), it may be that this paradigm could illuminate the interaction of the sleep–wake cycle and the circadian rhythm in bipolar disorder. Although there are practical and ethical challenges to consider before attempting such protocols with children and adolescents with bipolar disorder, the potential yield is great.
Causal consequences of sleep disturbance
Prospective and experimental studies are greatly needed to determine the extent to which sleep disturbance is causally contributing to mood episodes in children and adolescents with bipolar disorder. For example, various designs have been used in the adult bipolar literature to identify predictors of relapse (e.g., Johnson & Miller, 1997). These could be readily adapted for use with children and adolescents. Although care is needed when completing experimental studies of sleep deprivation with vulnerable individuals, it may be that partial sleep deprivation designs are achievable, even in children and adolescents. Such research will provide a test of the applicability to children and adolescents of the theories reviewed earlier, which were developed for adults with bipolar disorder; and, of course, if a causal relationship is present, it will provide an opportunity for developing pharmacological and psychological preventative interventions targeting sleep disturbance that could be of substantial clinical benefit for individuals with bipolar disorder.
Treatment
Taking together the findings that children and adolescents with bipolar disorder exhibit sleep disturbance, and that sleep has important functions, including the crucial domains of emotion regulation and learning, the development of treatment approaches to reduce sleep disturbance is another important avenue for future research. The section that follows is divided into subsections that address different aspects of the treatment of sleep disturbance in bipolar disorder: (a) the impact of pharmacological treatments for bipolar disorder on sleep, (b) the evidence relating to the treatment of sleep disturbance pharmacologically, (c) the extent to which adjunctive psychological interventions for bipolar disorder already address sleep disturbance, and (d) the evidence relating to the treatment of sleep disturbance using psychological interventions. The reason for including a discussion of interventions for (not related to bipolar) sleep disturbance is that several of these advances in knowledge have not yet been incorporated into treatments for bipolar disorder. Hence, they may provide a potential basis for developing new interventions.
Pharmacological treatments for bipolar disorder: Impact on sleep
The most commonly used pharmacological interventions for bipolar disorder are antidepressants, lithium, the anticonvulsants (e.g., carbamazepine and lamotrigine), and the novel antipsychotics (e.g., olanzapine; Grunze, 2005; McIntyre, 2002). Most of these medications have an impact on sleep. Typically they are either sedating (Ghaemi, Manwani, Katzow, Ko, & Goodwin, 2001; Makela, Miller, & Cutlip, 2003; McIntyre, 2002; Reischies, Hartikainen, & Berghofer, 2002; Sorge, Pollmaecher, & Lancel, 2004) and/or alter sleep architecture (Bazil, 2003; Gann, Riemann, Hohagen, Muller, & Berger, 1994; Geurkink, Sheth, Gidal, & Hermann, 2000; Kupfer, Wyatt, Greenspan, Scott, & Snyder, 1970; Sharpley, Vassallo, & Cowen, 2000; Sharpley, Walsh, & Cowen, 1996). Effects on sleep architecture include that lithium carbonate (Kupfer et al., 1970), carbamazepine (Geurkink et al., 2000), and olanzapine (Sharpley et al., 2000) are all associated with a reduction in REM and an increase in Stages 3 and 4 sleep. In contrast, both valproic acid and lamotrigine appear to have minimal effect on sleep architecture (Geurkink et al., 2000). Note that because some antidepressants used to reduce sleep disturbance may trigger mania or accelerate episode cycles, they tend to be avoided or used with caution (e.g., Grunze, 2005; Saiz-Ruiz, Cebollada, & Ibañez, 1994).
Knowledge about the delivery of these pharmacological treatments to children and adolescents with bipolar disorder is at a very early stage of development, relying heavily on the adult literature just discussed (Kowatch & DelBello, 2005; Ryan, 2003). Unfortunately, to the best of the authors' knowledge, studies of pharmacotherapy in youth with bipolar disorder have not as yet included sleep disturbance as an outcome measure. Hence, we do not know the extent to which the sleep-related effects described above, found in adults, will replicate in younger individuals.
To date, for obvious reasons, there has been a focus on developing pharmacological interventions to manage mood dysregulation in bipolar disorder. If sleep disturbance turns out to be a causal variable, perhaps an additional focus of research should be to develop compounds that effectively reduce sleep disturbance at the same time. Having said that, as we review in the section that follows, finding a highly effective medication for insomnia that is low on side effects and that has enduring effects (even when medication is terminated) has proved nigh on impossible, to date. However, we note that psychological approaches to sleep disturbance (to be reviewed in a later section) have shown considerable promise.
Pharmacological treatments for sleep disturbance
Several classes of medications are used to manage insomnia including the benzodiazepines (e.g., flurazepam, temazepam, triazolam, estazolam, quazepam) and the nonbenzodiazepine hypnotics (e.g., zolpidem, zopiclone, zaleplon), which act on benzodiazepine receptor subtypes or on the related GABAA complex. These medications are more effective than placebo for improving sleep (Holbrook, Crowther, Lotter, Cheng, & King, 2000; Kupfer & Reynolds, 1997; Nowell et al., 1997; Roth & Roehrs, 1991). Their effects on sleep stages vary with the specific class of medication. Although they all increase Stages 1 and 2 sleep, the benzodiazepines tend to reduce slow-wave (Stages 3–4) sleep and, to a lesser extent, REM sleep. These latter changes, however, are less pronounced with zolpidem and zopiclone (Hoehns & Perry, 1993; Wadworth & McTavish, 1993). The main limitations of hypnotic medications are their residual effects (e.g., sedation the next day) and the associated risks of tolerance and dependence (Greenblatt, 1991). Zolpidem and zopiclone have the advantage of producing fewer withdrawal symptoms (i.e., rebound insomnia) upon discontinuation (Hoehns & Perry, 1993; Wadworth & McTavish, 1993). Another disadvantage is that there is no evidence that the effects of these medications endure upon cessation of the intervention (Morin & Espie, 2003).
To the best of our knowledge, there are no clinical trials of these medications for individuals with bipolar disorder, but the initial case reports have been far from encouraging. Weilburg, Sachs, and Falk (1987) reported that triazolam induced brief episodes of mania in an eldery woman, and Sattar, Ramaswamy, Bhatia, and Petty (2003) reported a case where an interaction between valproic acid and zolpidem induced somnambulism (sleep walking) in a 47-year-old man with bipolar disorder. Mandated are careful, experimental trials.
Adjunctive psychological treatment for bipolar disorder
In the adult literature on bipolar disorder, several adjunctive interventions have been developed that include one or more components designed to assist individuals with their sleep. IPSRT includes instruction in behavioral techniques designed to regularize routines and reduce social rhythm disruption (e.g., Frank et al., 2000). Cognitive behavior therapy administered individually (e.g., Lam et al., 2003) or in groups (e.g., Patelis-Siotis et al., 2001) provides education about the importance of a regular social routine and sleep–wake cycle via monitoring of mood and sleep and the use of activity scheduling. It also involves working with the patient to enhance his or her ability to recognise prodromes, which often include sleep disturbance (Jackson et al., 2003), and to develop effective coping strategies for such prodromes. Moreover, family interventions (Miklowitz & Goldstein, 1997), group psychoeducation, and individual psychoeducation (see Miklowitz & Johnson, 2006, for review) also include content that encourages sleep–wake stabilization.
Although these approaches have been shown to reduce relapse in adults with bipolar disorder (Craighead et al., 2002; Frank et al., 2005; Lam, Hayward, Watkins, Wright, & Sham, 2005; Miklowitz et al., 2000), to the best of our knowledge they have not yet been tested in youth. Nonetheless, particularly relevant to the present paper is the finding that stabilization of social rhythms (which include the sleep–wake cycle) during the acute treatment phase in adults receiving IPSRT predicted longer delays prior to relapse (Frank et al., 2005).
Of interest, these approaches have yet to draw from the literature reporting on the effectiveness of psychological treatments for sleep difficulty among adults with insomnia (Morin et al., 1999a). As these interventions may provide a basis for developing new interventions for sleep in patients with bipolar disorder, we now describe them briefly.
Adjunctive psychological treatment for sleep disturbance
An effective intervention for adults with chronic insomnia has been developed. It is known as cognitive behavior therapy for insomnia (CBT-I). CBT-I is a multicomponent treatment that typically involves several of the following components: stimulus control, sleep restriction, sleep hygiene, paradoxical intention, relaxation therapy, and cognitive restructuring for unhelpful beliefs about sleep (for review, see Morin & Espie, 2003; Perlis, Smith, Jungquist, & Posner, 2005). In a review of the evidence for each component, the Standards of Practice Committee of the American Academy of Sleep Medicine concluded that for the treatment of insomnia there is sufficient evidence to justify the use of stimulus control, progressive muscle relaxation, paradoxical intention and biofeedback, and a moderate amount of evidence associated with sleep restriction and the multicomponent behavioral therapy package (Chesson et al., 1999; Morin et al., 1999b). The components are administered over 4 to 10 weekly sessions in either individual (e.g., Edinger, Wohlgemuth, Radtke, Marsh, & Quillian, 2001) or group (e.g., Espie, Inglis, Tessier, & Harvey, 2001) format. This intervention is superior to wait-list and credible alternative treatments. It can be effectively disseminated into the community and the effects endure to 24 months after the conclusion of the intervention (for review, see Harvey & Tang, 2003).
Although the area is underresearched, there is a small evidence base reporting on the effectiveness of some of the interventions just described with children and adolescents who suffer from sleep disturbance (Owens, France, & Wiggs, 1999; Sadeh, 2005). The relevant strategies fall into four broad categories.
Bedtime routine. A presleep routine that helps children and adolescents to reduce their arousal, regulate their mood, and prepare for sleep is recommended, as is maintaining consistent bed and wake times. The latter has the dual advantage of ensuring the child is ready for bed at bedtime and promoting stability in the circadian rhythm (Hoban, 2004).
Bedtime resistance. This difficulty can be a source of sleeplessness and stress for the whole family. This intervention is based on the assumption that difficulty going to sleep is, at least in part, maintained by parental attention. It aims to reduce bedtime behaviors that are inconsistent with getting to sleep by reducing parental attention to them. Owens et al. (1999) describe several modifications to what is essentially an extinction procedure, the goal of which is to increase acceptance by parents, who are often concerned about harmful effects of withdrawing attention from their child. The procedures include the following. First, there is “parental presence” to reduce separation anxiety in the child during the first week of the program. During the next week the parent sleeps in the child's bedroom but in a separate bed, yet without interacting with the child during the night. For the third week, the parent moves to a separate room. Second, there is “graduated extinction and fading,” in which the parent reduces their response slowly over time either in the form of increasing the latency to respond or decreasing the length of the response. Finally, parents can provide “minimal brief checks,” at regular intervals of 5 to 20 min, while crying persists.
In older children, this intervention needs to be altered. It involves facilitating enforcement of bedtime by the parent and maintenance of regular bed and wake times with age appropriate incentives.
Insomnia involves difficulty settling to sleep at the beginning of the night or awakening during the night and finding it difficult to return to sleep. Insomnia in children and adolescents is addressed with similar treatments as for adults, particularly stimulus control (Morin & Espie, 2003; Morin et al., 1999b). Stimulus control involves asking the insomnia sufferer to go to bed only when he or she is tired, to limit activities in bed to sleep, to get up at the same time every morning, and when sleep onset does not occur within 10 to 20 min, to get up and go to another room. The underlying rationale of this treatment is that insomnia is the result of maladaptive conditioning between the environment (bed/bedroom) and sleep incompatible behaviors (e.g., worry/frustration at not being able to sleep). The intervention, developed by Bootzin (1972), aims to reverse this association by limiting the sleep incompatible behaviors that may be engaged in. Because stimulus control often results in short-term sleep deprivation, it would need to be carefully adapted and may even be contraindicated among individuals with bipolar disorder, given the evidence reviewed earlier that sleep deprivation can trigger relapse in a proportion of patients.
Parents may ask how to decide between stimulus control instructions (only go to bed when sleepy) versus the bedtime resistance instructions (go to bed at a set time). In our opinion, many children and adolescents with bipolar disorder may need to do the latter (get into bed even though not yet sleepy) in order to begin the process of downregulating sufficiently to fall off to sleep.
Other insomnia treatments that may be well suit individuals with bipolar disorder include cognitive restructuring and behavioral experiments (Harvey, 2005). The latter involve arranging experiences that give the client an opportunity to test out their unhelpful thoughts, beliefs, behaviors, and processes (Bennett-Levy et al., 2004). These interventions are suggested on the basis of recent evidence indicating that adults with bipolar disorder are anxious about their sleep (because they have noticed that disturbed sleep can be a prodrome of an episode) and often hold unhelpful beliefs about sleep (Harvey et al., 2005), both of which may increase vulnerability to sleep disturbance and both of which are treatable.
As already discussed, a key feature of sleep leading up to and during the adolescent years is the delay in circadian phase, such that sleep onset does not occur until the early hours of the morning. Although the literature for treating delayed phase is small, the interventions that have been tested include exposure to bright light, although the optimal timing and dose of the exposure remains to be established for children and adolescents (Hoban, 2004). More simply, such procedures as gradual advances in bedtime and wakeup times (say, by 15 min/day), avoiding daytime napping, and maintaining consistency on the weekend are recommended; or when the bedtime is very late, it may be easier to institute delays in bedtime. The reason that moving in this direction may be favorable is that the circadian cycle naturally runs over 24 hr. Hence, delays that capitalize on the natural tendency for the circadian system to run over 24 hr may be more easily achieved than advances in bedtime (see Dahl, 1992, for further details and a case example).
To summarize, although it seems unlikely that the treatments described above are targeting the core of sleep disturbance in children and adolescents with bipolar disorder, they may provide a starting point for new treatment development efforts. We caution that one cannot reason backwards from treatments to etiology, which is akin to positing that because aspirin helps flu, flu must be caused by a deficit in aspirin (Salkovskis, 2002). However, if evidence that interventions for sleep improve mood in children and adolescents with bipolar disorder does emerge, it would add strength to evidence obtained from stronger tests of causality (experiments and prospective studies) in support of the hypothesis that sleep is an important influence in the multifactorial causal chain leading to relapse.
Conclusion
The aim of this paper has been to highlight the potential importance of the sleep–wake cycle in children and adolescents with bipolar disorder. We have reviewed evidence that, in adults, sleep and/or circadian rhythms are one possible variable of considerable influence in the complex causal chain that leads to relapse in bipolar disorder. Given the considerable sleep disturbance evident among children and adolescents with bipolar disorder, along with the key role that sleep appears to play in emotion regulation and learning, we conclude that sleep among children and adolescents with bipolar disorder is a critical domain for future research. One goal of gaining this knowledge is to maximally inform intervention. It seems easy to imagine that a high-quality psychological intervention for sleep disturbance will be an important component of a multimodal intervention that combines intensive psychosocial intervention and well-delivered pharmacological intervention for children and adolescents with bipolar disorder. The psychological intervention approach to sleep disturbance is emphasized here, given that in the adult literature, currently available sleep-inducing medications are associated with significant side effects, and follow-up data indicate that the sleep improvements obtained are quickly lost after discontinuation of the medication. In contrast, the effects of psychological interventions are well sustained (Morin, Colecchi, Stone, Sood, & Brink, 1999).
