Sleep and circadian disruption in bipolar disorders: From psychopathology to digital phenotyping in clinical practice

Abstract

Sleep and biological rhythms are integral to mood regulation across the lifespan, particularly in bipolar disorder (BD), where alterations in sleep phase, structure, and duration occur in all mood states. These disruptions are linked to poorer quality of life, heightened suicide risk, impaired cognitive function, and increased relapse rates. This review highlights the pathophysiology of sleep disturbances in BD and aims to consolidate understanding and clinical applications of these phenomena. It also summarizes the evolution of sleep and biological rhythms assessment methods, including ecological momentary assessment (EMA) and digital phenotyping. It underscores the importance of recognizing circadian rhythm involvement in mood regulation, suggesting potential therapeutic targets. Future research directions include elucidating circadian clock gene mechanisms, understanding environmental impacts on circadian rhythms, and investigating the bidirectional relationship between sleep disturbances and mood regulation in BD. Standardizing assessment methods and addressing privacy concerns related to EMA technology and digital phenotyping are essential for advancing research. Collaborative efforts are crucial for enhancing clinical applicability and understanding the broader implications of biological rhythms in BD diagnosis and treatment. Overall, recognizing the significance of sleep and biological rhythms in BD offers promise for improved outcomes through targeted interventions and a deeper understanding of the disorder's underlying mechanisms.

Sleep and biological rhythms play a critical role in the regulation of mood across the lifespan and in bipolar disorder (BD), changes in sleep phase, structure, or duration are reported across all mood states. ¹ These changes in sleep patterns have been associated with a lower quality of life, greater risk of suicide attempts, worse clinical and cognitive functioning, and higher relapse rates of mood episodes. ² The maintenance of regular sleep–wake cycles and other biological rhythms is crucial for preserving affective stability in BD, ¹ emphasizing the importance of investigating biological rhythms in this disorder.

Circadian disruption is a term that became popular in the translational investigation of BD, and it refers to situations when the natural organization of physiology and behavior over a 24‐h period is significantly disturbed. ³ This disruption can lead to desynchronized rhythms, potentially resulting in adverse health effects. When such disturbances persist beyond a certain homeostatic threshold and become chronic, this process is referred to as chronodisruption. ³ In individuals with bipolar disorder (BD), the prevalence of chronodisruption is estimated to be higher than the general population, with studies showing ranges from 10% to 80% depending on definitions of key concepts of chronodisruption, methodologies, and management of potential confounding factors. ⁴ Rhythm irregularities are present in all BD spectra, even in euthymic states, and have been associated with disorder onset, severity, and prognosis. ⁵

Recent review articles address important aspects of neurobiological and behavioral mechanisms of circadian rhythm disruption ⁶ and measurement of circadian function ⁷ in BD, as well as provide fundamental insights on chronotherapeutics. ⁸ This review aims to summarize the most recent findings regarding the assessment of sleep and biological rhythms, emphasizing up‐to‐date evidence that show direct clinical implications in terms of diagnosis, treatment, and improving the quality of life for individuals with BD. It describes the pathophysiology of sleep disturbances and their phenomenology in the course of BD, and describes the evolution of ecological momentary assessment and digital phenotyping in the study of mood disorders, with a focus on BD.

Biological Rhythms, Sleep and Mood Regulation: Pathophysiological Insights

Biological rhythms refer to the innate, cyclical patterns of physiological and behavioral processes that occur within an organism. These rhythms are regulated by external cues, primarily the 24‐h light–dark cycle, which generate circadian rhythms. ⁹ , ¹⁰ From molecular to multi‐organ physiological systems, these rhythms play a profound role in shaping an individual's behavior, cognition, and mood. ¹¹ , ¹²

The disruption of biological rhythmicity, whether due to shift work, jet lag, nighttime light exposure, or irregular sleep patterns, may lead to marked alterations in physiology and behavior. ¹³ , ¹⁴ , ¹⁵ , ¹⁶ Notably, research has increasingly connected these factors with the pathophysiology and course of mood disorders, including BD. ¹ Given the characteristic cycling between manic and depressive episodes that occurs in BD, it is not surprising that researchers have examined connections between BD and disturbances in biological rhythms for many years. In the subsequent sections, we delve into various concepts that explore the intricate interplay between biological rhythmicity and mood regulation, supported by evidence from both animal models and clinical studies (Table 1).

Table 1.

Summary of main findings of the sleep and circadian rhythm correlates of bipolar disorder

Clock and clock‐related genes
Gene or protein	Description
Clock (mutation)	Manic‐like or “mixed state” behavior in mice
CLOCK (polymorphism)	Sleep dysregulation, evening chronotype, higher risk for BD in humans
Per1 and Per2 (mutation)	Manic‐like behavior in mice
PER1 (expression phase advance)	Mania phase compared to depression in humans
Per3 (deficiency)	Depressive‐like behavior in mice
PER3 (haplotypes and SNP)	Diagnosis of BD in humans
Bmal1 (haplotypes and SNP)	Diagnosis of BD in humans
TIMELESS (SNP)	Diagnosis of BD in humans
PER4/4 (allele variant)	Later onset of BD and extreme evening chronotype and delayed sleep phase disorder in humans
PER5/5 (allele variant)	Early onset of BD and extreme morning chronotype in humans
FBXL3 (mutation)	Manic‐like behavior in mice
GSK3β (protein overexpression)	Manic‐like behavior in mice
	Age of onset, presence of core BD symptoms, and response to lithium and valproic acid therapy in humans
Rev‐erbα (knockout)	Depressive‐like behavior in mice
Cry1 (knockout)	“Mixed state” behavior in mice

Sleep disturbances
Sleep variable	Description
Sleep deficits in evening chronotypes	Evening chronotypes are a trait of vulnerability due to social demands; BD is generally associated with eveningness.
	Higher prevalence of circadian rhythm sleep–wake disorders in samples of BD individuals compared to the general population
	Eveningness confers greater functional impairment in individuals with BD
Sleep deprivation	Inducing manic‐like behavior in mice
	Correlation with surge of manic and hypomanic symptoms in humans
	increased risk of suicide
Insomnia	Reported by 40–80% of individuals with bipolar depression (in different samples)
Hypersomnia	Reported by 30–78% of individuals with bipolar depression (in different samples)
Decreased need for sleep	Subjective perception of 69–99% of patients in manic or hypomanic episodes
Fragmented sleep	Increased WASO, lower total sleep time, and higher sleep latency in all phases of BD, including euthymia.
	Measured by subjective reports, actigraphy and PSG
Sleep efficiency	Time sleeping relative to the total time in bed; reduced in individuals with BD
	Measured by subjective reports, actigraphy and PSG
Sleep quality	Perception of poor sleep quality is a risk factor for mood episode recurrence
REM sleep (PSG records)	Reduction in REM latency, increase in REM sleep percentage and REM density in manic episodes
	Earlier REM sleep latency in depressive episodes
Sleep apnea	Higher prevalence of obstructive sleep apnea in individuals with BD compared to the general population

Circadian rhythm of activity measured by actigraphy
Activity parameter	Description
Daily amplitude	Individuals with BD show dampened degree of changes in magnitude from the highest to the lowest values; studies report lower relative amplitude of activity.
Daily mean	Lower daily average of daily activity in individuals with BD
Onset	Later activity onset in depressive states
Acrophase	The moment of highest activity happens later in individuals with BD, associated with the higher occurrence of eveningness
Low evening activity	Characteristic of depressive states
Fragmentation	Higher intradaily variability in BD individuals; indicates more life‐time manic and hypomanic episodes

Other circadian rhythms
Rhythm	Description
Shortening of light exposure (seasonality)	Onset of depressive episodes in individuals with BD
Melatonin	Inconclusive results: delayed release in euthymia and increased secretion in mania in humans
Cortisol	Inconclusive results: diminished cortisol reactivity and diurnal slope
Body temperature	Inconclusive results: phase advance and dampened amplitude in humans

Digital phenotyping ‐ preliminary reports
Profile	Description
Screen time	Longer screen time associated with severity of bipolar depression
Phone calls	Shorter phone calls when euthymic
	Fewer outgoing calls and less answered incoming calls associated with severity of bipolar depression
Geolocation	Shorter distance traveled associated with severity of bipolar depression
	Longer distance traveled associated with severity of manic symptoms
Messaging	More frequent outgoing text messages, shorter incoming text messages associated with severity of manic symptoms
Smartphone‐based self‐assessments of mood and cognition	Prospective associations between activity and mood in the following days
	Instability of mood associated with lower daily functioning in individuals with BD

BD, Bipolar Disorder; SNP, Single Nucleotide Polymorphism.

Circadian Clock Genes in Mood Regulation and Bipolar Disorder Pathophysiology

Numerous studies have pointed towards the significant involvement of circadian clock genes in the regulation of anxiety, mood, and reward. ¹⁷ The mechanisms by which circadian clock genes contribute to the pathophysiology of BD involve a myriad of homeostatic mechanisms related to mood regulation, including monoamine signaling, immune function, HPA axis regulation, metabolism, oxidative stress regulation, and neurogenesis (Fig. 1B). ²⁰ , ²¹

Fig. 1.

The intersections of circadian disruption and sleep disturbances in BD. Panel A presents the mood‐related structures and the light inputs to the brain. Contours highlight brain structures that receive direct light signals from the environment. The remaining structures are regulated by the circadian system via the SCN. Panel B shows a theoretical model of the intersections of circadian rhythms, sleep and mood states in BD, adapted from Watling et al. ¹⁸ and De Prisco et al. ¹⁹ Of note, eveningness is cited as a trait marker of BD in situations of circadian disruption. AN, arcuate nucleus; BNST, bed nucleus of the stria terminalis; DMH, dorsomedial hypothalamus; DR, dorsal raphe; LC, locus coeruleus; LHb, lateral habenula; meA, medial amygdala; PVN, paraventricular nucleus; SCN, suprachiasmatic nucleus; VTA, ventral tegmental area.

Manic‐like behavior, portrayed by hyperactivity, decreased sleep, and increased sensitivity to reward was observed in mice with induced mutations for the genes Clock, ²² Per1 and Per2, ²³ and F‐box protein 3 (FBXL3). ²⁴ , ²⁵ These results were also seen when there is an overexpression of the GSK3β, a regulatory kinase, whose signaling pathways, via phosphorylation, form feedback loops with the molecular circadian clock to modulate circadian amplitude and period. Notably, Per3 deficient ²⁶ and Rev‐erbα knockout ²⁷ mice show depressive‐like behaviors. Another relevant finding from animal models was that Cry1 knockout mice show a “mixed state”, leading mice to be less anxious and hyperactive, yet with an increased depression‐related behavior. ²³ It is interesting to note that induced Clock mutation selectively in the ventral tegmental area (VTA) also brings up a “mixed state”. ²⁸ When a functional CLOCK protein is restored in the VTA, most of the manic‐like behaviors are normalized. In mice, lithium use minimizes most of the mania‐like behaviors and restores normal levels of dopamine in the VTA.

Several clinical studies have shown that clock genes may be related to the pathophysiology and clinical presentation of BD. ² Clock gene variants may increase BD susceptibility, influencing age at onset, and sleep/wake patterns. ²⁹ , ³⁰ , ³¹ However, these gene effects are probabilistic and non‐specific, most likely a result of interaction with environmental factors. The CLOCK gene is one of the most evaluated genes concerning BD in clinical populations and its alterations are potentially associated with the sleep dysregulation in BD. For example, polymorphism of the CLOCK gene has been generally associated with higher “eveningness” in subjects carrying at least one copy of the C allele. ³² As described in the sections below, eveningness is an important individual trait in BD. In their study following individuals for over 5 years, Benedetti and colleagues ³³ found a higher chance of BD recurrence in homozygotes for the C variant, and the same associations were corroborated by two additional studies. ³⁴ , ³⁵ These findings suggest a genetic predisposition and increased vulnerability to negative repercussions in certain individuals. Polymorphisms in GSK‐3β, resulting in overexpression of this protein, might be associated with the age of onset, some core BD symptoms, and response to lithium and valproic acid therapy. ³⁶ Also, the long allele variant of PER5/5 has been linked to early onset of BD and extreme morning chronotypes, whereas the shorter allele PER4/4 is associated with later onset of BD extreme eveningness and delayed sleep phase syndrome. ³⁷ , ³⁸ Additionally, there might be different expressions of these genetic profiles depending on the phase of bipolar illness. For example, one study describes that the phase of expression of PER1 gene is advanced in individuals in mania compared with those in depression. ³⁹ Finally, some studies show the associations of BD psychopathology with haplotypes in Bmal1 and PER3 ³¹ and single nucleotide polymorphisms in Bmal1, PER3, and TIMELESS genes. ⁴⁰ Some of these studies found that similar clock genes associated with BD were also associated with other mood and psychiatric disorders. ²⁹ , ⁴⁰

Circadian Disruption As a Potential Underlying Factor in Bipolar Disorder

Zeitgeber, a German word meaning “time giver”, defines the external cues that align internal rhythms to the environment. Light is the main zeitgeber for extrinsic coordination of the timing of endogenous clocks in mammals. ⁴¹ Hence, photoentrainment is at the base of circadian rhythms regulation, modulating signaling between the brain and other peripheral tissues via a well‐known pathway starting with retinohypothalamic tract projections to the hypothalamic suprachiasmatic nucleus (SCN). ¹³ , ⁴² One canonical pathway connects SCN projections to the pineal gland, enabling the production and release of melatonin in the absence of light. Neural projections of the SCN and the neuro‐endocrine signaling of melatonin orchestrate circadian patterns, working to maintain the homeostasis of physiological processes so that they occur at the most opportune moment of the circadian cycle. ⁴³ , ⁴⁴

Among the neural projections of the SCN within the central nervous system, various regions are involved with mood and behavior (Fig. 1A): the hypothalamic paraventricular nucleus (key to the regulation of the hypothalamo‐pituitary–adrenal axis ⁴⁵ ), the dorsomedial hypothalamus, ⁴⁶ the locus coeruleus, ⁴⁷ the dorsal raphe, ⁴⁸ the arcuate nucleus, ⁴⁹ the bed nucleus of the stria terminalis, ⁵⁰ and the lateral habenula. ⁵¹ While such projections may be derived from retinal stimuli to the SCN – and thus influence mood via regulation of biological rhythms – recent research points to a direct pathway connecting light stimuli in the retina to some of these brain regions independently of the SCN. ¹² , ⁵² The medial amygdala and the lateral habenula receive direct retinal input and project to the ventral tegmental area, raphe, and lateral hypothalamus, all of which regulate mood and cognition. ⁵² , ⁵³ , ⁵⁴ , ⁵⁵

With the understanding of the main mechanisms by which circadian rhythms are regulated, it is possible to define models for the study of chronodisruption (Box 1). Changes in expected light–dark transitions may impact physiology and behavior by temporarily disrupting the rhythmic expression of clock and clock‐related genes within the SCN. ⁵⁶ Night shift work, jet lag, or exposure to artificial light at night are good examples of situations of the challenges one's body faces when re‐synchronizing the circadian system to abrupt changes of light exposure. In both human and animal research, this process may lead to marked cognitive deficits and affective changes. ¹³ , ¹⁴ A striking result derived from a long cohort of 26 BD individuals (and a total of 31 mood episodes) describes an average of about 7 h of advanced phase of clock gene expression in manic episodes and a 4 to 5 h delay in depressive phases. ⁵⁷ These phase‐shifts returned to baseline along with the recovery of the mood episode. Such results indicate an important phase‐shift in the circadian clock during mood episodes, possibly suggesting a misalignment of circadian rhythms.

Box 1. Environmental Impact and Conditions that Lead to Circadian Disruption.

1. Shift Work: People who work irregular or night shifts often experience circadian disruption because their work schedules conflict with their natural sleep–wake patterns. This can lead to sleep disturbances, fatigue, and health issues.

2. Jet Lag: When individuals travel across multiple time zones, their circadian rhythms may not immediately adjust to the new local time, leading to jet lag. This can result in sleep problems, fatigue, and other symptoms until the body's internal clock synchronizes with the new time zone.

3. Social Jet Lag: This occurs when people maintain different sleep schedules on workdays versus weekends. It can lead to misalignment of circadian rhythms.

4. Artificial Light Exposure: Exposure to artificial light in the evening, especially the blue light emitted by electronic devices, can interfere with the body's ability to produce melatonin, a hormone that helps regulate sleep. This can disrupt the sleep–wake cycle.

5. Irregular Sleep Patterns: Inconsistent sleep patterns, such as irregular bedtimes and wake times, can disrupt circadian rhythms and make it challenging to maintain a consistent sleep schedule.

6. Social Isolation: Lack of social interaction and regular daily routines, as seen in certain forms of social isolation or confinement, can disrupt circadian rhythms.

7. Aging: As people age, their circadian rhythms can change, leading to altered sleep patterns and difficulties in maintaining a regular sleep–wake cycle.

8. Medical Conditions: Certain medical conditions, such as circadian rhythm sleep disorders, can result in chronic circadian disruption, affecting an individual's ability to maintain a typical sleep–wake cycle.

9. Medical Treatments and Medications: Certain medications, such as those for psychiatric disorders, can disrupt circadian rhythms.

The length of the light period (i.e. the photoperiod) has also shown profound impact in animal and human behavior and mood. Short photoperiods, similar to the shorter daylight hours in the winter, lead to depressive behavior and hippocampal learning deficits in rodents. ⁵⁸ , ⁵⁹ Additionally, mice with reduced dopamine transporter expression have shown seasonal‐induced switching between behavioral states that mimic BD. ⁶⁰ , ⁶¹ In humans with BD, the beginning of autumn, a period of progressive shortening of daily photoperiod, marks a shift towards the depressive phase. ⁶² , ⁶³ Besides external light stimuli, the SCN also shows the ability to entrain biological rhythms using exogenous administration of melatonin, dopaminergic, or glutamatergic agonists. ⁶⁴

Beside the light–dark signal, social cues (e.g., meal timing, bedtimes, work schedules, and timing of exercise) are implicated in the entrainment of circadian rhythms. ⁶⁵ , ⁶⁶ , ⁶⁷ , ⁶⁸ These non‐photic zeitgebers define the periods of light exposure by establishing social routines and may influence circadian oscillators in peripheral tissues. The misalignment of social cues with one's inherent preference for timing of sleep and activity (i.e. chronotype) may also contribute to chronodisruption. Eveningness is often associated with greater emotional dysregulation, affective instability, and sleep complaints, ⁶⁹ , ⁷⁰ with worse outcomes in major psychiatric diagnoses. Individuals with BD, even in euthymic states, have a higher prevalence of eveningness chronotype than healthy individuals. ⁷¹ Additionally, previous studies demonstrate that the evening chronotype in BD individuals may determine a clinical subphenotype of the disorder. This cluster would be composed of relatively younger individuals who had an earlier age of onset of BD, more mood episodes, higher rates of rapid cycling, more past suicide attempts, comorbid anxiety, and substance use disorders. ⁷² , ⁷³

In most studies, evening preference was not associated with polarity type, or mood state in BD, suggesting that this characteristic may be a trait marker that increases vulnerability to circadian disruption (Fig. 1B). ⁷⁴ That is, eveningness is not pathological in itself, but rather a state of vulnerability because of social demands. This characteristic is associated with a reduced sleep quality and quantity during weekdays, suggesting that individuals experience a sleep deficit during weekdays due to social and work/study commitments. ⁷⁵

Bipolar Disorder Express Distinct Rhythms in Physiology

Several studies show that symptoms of BD might be accompanied by changes in the rhythms of physiological variables. ⁷⁶ For example, a few studies evidenced an increase in melatonin secretion in mania episodes and a decrease in depressive episodes. ³⁹ , ⁷⁷ , ⁷⁸ Samples of euthymic individuals with BD also present a delayed release of melatonin compared to controls. ⁷⁸ It has been suggested that BD individuals could express a hypersensitivity of the pineal gland to regulation by ambient light. ⁷⁹ These results, however, are not consistent throughout the literature. ⁸⁰ , ⁸¹

Cortisol production and release is also influenced by the circadian system, given that its levels are usually higher in the morning and depends on melatonin production. ⁸² However, in individuals with BD, results are inconclusive. One study showed that the 24 h cycle of cortisol is higher, irrespective of the illness phase, ⁸³ but another evidenced morning cortisol higher in non‐manic states. ⁸⁴ In addition, individuals who had experienced numerous previous episodes exhibited elevated overall cortisol levels, diminished cortisol reactivity in response to negative events, and shallower diurnal cortisol slopes compared to individuals with fewer episodes. ⁸⁵

Body temperature in bipolar depression shows a paradoxical increase at night accompanied by a decrease in the morning, resulting in an anticipation of daily peak and a reduced amplitude of this rhythm. ⁸⁶ Interestingly, appropriate treatment of mood symptoms is able to normalize these alterations. Improvement of mood symptoms was also accompanied by normalization of the circadian rhythms of prolactin, TSH, growth hormone, and some urinary metabolites. ⁸⁶ , ⁸⁷ However, a recent study was not able to reproduce the findings on temperature rhythm by exploring wrist temperature with actigraphy. ⁸⁸

Sleep Disturbances and Impaired Mood Regulation

Individuals with BD struggle with mood symptoms and emotion dysregulation that are commonly linked with sleep disturbances (Fig. 1B), being sleep deprivation the most prevalent type. Sleep deprivation was found to induce manic episodes in animal models of BD, characterized by increased motor activity, vocalization, and aggressive behaviors. ²⁴ , ⁸⁹ , ⁹⁰ In humans, chronic sleep loss has been implicated in the surge of manic symptoms in case series and cohorts of small sample size. ⁹¹ , ⁹² , ⁹³ , ⁹⁴ However, one study with over 200 individuals with BD under chronic sleep deprivation shows that about 5% of these individuals would switch from depression into mania, and 6% into hypomania. ⁹⁵

Sleep deprivation produces a gradual decline in extracellular serotonin levels, both in the hippocampus and in the frontal cortex. ⁹⁶ On the other hand, other monoamines (i.e., dopamine, noradrenaline, and adrenaline) tend to increase during sleep deprivation in mice, sometimes remaining high even during subsequent recovery. ⁹⁷ In addition, it has been observed that acute sleep deprivation leads to increased neurotrophins in healthy humans. ⁹⁸ Acute or chronic sleep deprivation in the form of sleep debt can also impact cognitive functioning. Special focus has been given to the impaired performance on psychomotor vigilance tasks that require vigilant attention, which increase the risk for accidents like work injuries or car crashes. ⁹⁹ These are often accompanied by sleepiness, fatigue, and feelings of cognitive impairment.

Notable effects to individuals with BD include negative mood states and decreased levels of positive mood states, as well as decreased adaptive emotion regulation ¹⁰⁰ , ¹⁰¹ possibly leading to greater mood instability over time. ¹⁰² Studies also consistently report heightened emotional response and affective lability, as well as an increased tendency towards negativity bias as a consequence of chronic sleep loss. ¹⁸ So far, the literature on the topic suggests that the effects of sleep deprivation are greater for outcomes related to mood than those related to cognition or executive function. ¹⁰³ Studies using magnetic resonance imaging have found that sleep‐deprived individuals exhibit heightened amygdala activity and decreased activity of the medial–prefrontal cortex, suggesting a disruption of the emotional modulation circuitry. ¹⁰⁴ , ¹⁰⁵ Therefore, imaging studies also support the hypothesis that chronic sleep loss would lead to emotional dysregulation as a result of diminished cognitive functioning by affecting frontal brain regions and diminishing control over executive functions. ¹⁰⁶ , ¹⁰⁷ Watling and colleagues (2017) discuss how emotional dysregulation and sleep loss may form a vicious cycle as individuals deprived of sleep might find it difficult to control their emotions, thus impairing their ability to get sufficient sleep at night. ¹⁸ We expand on this discussion by pointing out that (1) circadian disruption and other forms of sleep disturbance (not limited to sleep deprivation) might lead to similar effects on emotion regulation, and (2) that this cycle, over time, might lead to a vulnerability of future mood episodes in BD ¹⁸ , ¹⁹ (Fig. 1B).

Sleep Disturbances in Bipolar Disorder: Clinical Insights

Sleep disorders are the most commonly studied manifestation of circadian rhythm dysfunction in BD ¹⁰⁸ and they occur independently of mood states. ⁷⁴ Prevalence studies including individuals with BD in any mood state report at least one type of sleep disturbance in over 60% of the sample, with insomnia being the most frequent, present in 40–50% of all cases, followed by hypersomnia in around 30% of cases. ⁴ , ¹⁰⁹ , ¹¹⁰ When considering individuals in a current mood episode, this prevalence is even higher.

Although insomnia and reduced necessity for sleep are the sleep features most commonly associated with BD, several other sleep disturbances are present in the illness course. The following sections are dedicated to discussing sleep in all phases of BD, major sleep syndromes associated with mood episodes, as well as sleep disturbances that are less commonly explored in individuals with BD (Table 1). We draw special attention to the study of circadian rhythm disturbances, including the circadian rhythm sleep–wake disorders (CRSD). Of note, sleep disturbances and circadian rhythm disturbances might overlap as several behavioral manifestations of sleep described here might be a result from an interaction of both homeostatic and circadian sleep regulation. For instance, symptoms of insomnia, short sleep duration, higher levels of nocturnal exposure to light and motor activity, lower sleep efficiency, are higher variability in sleep episodes might all be a result of individuals with eveningness or even delayed sleep phase type that are unable to regulate sleep–wake schedules according to their chronotype.

Sleep Patterns in Bipolar Disorder Phases

Given its importance, alterations in sleep patterns are core aspects of the diagnostic criteria for BD in the DSM‐5 and the ICD‐11. In the classification of manic and hypomanic episodes, one criterion refers to “decreased need for sleep” and in depressive episodes, “insomnia or hypersomnia” or, generically, “changes in sleep”. ¹¹¹ During episodes of mania, 69–99% of individuals show decreased need for sleep, ¹¹⁰ which implicates this symptom as a critical marker of the manic phase. In depression, most individuals show insomnia, and more than 78% might experience hypersomnia. ¹¹⁰

Insomnia, hypersomnia, or reduced necessity for sleep are often explored in clinical settings as part of the diagnostic criteria. However, considering that sleep has an intricate relationship with mood, several other aspects of sleep show clinical relevance and deserve the attention of mental health practitioners. For example, a recent consensus of the National Sleep Foundation based on up‐to‐date relevant studies argues in favor of the importance of sleep regularity independently of the evaluation of sleep duration. In this context, there is a gap in the current literature of BD, whereby most studies do not report the consistency of sleep onset and offset timing. ¹¹² Fragmented sleep, characterized by an increased wake after sleep onset, lower total sleep time, and higher sleep latency are significantly more pronounced in all mood phases. ¹¹³ , ¹¹⁴

In addition to sleep behavioral patterns, alterations in sleep architecture are also associated with clinical manifestations of sleep disturbances, though findings vary greatly for individuals with BD. A study assessed manic individuals using polysomnography at the beginning of the symptoms and after 3 weeks of treatment and observed an increase in total sleep time, sleep efficiency, and sleep continuity. ¹¹⁵ This was accompanied by a reduction in REM latency and an increase in REM sleep percentage and REM density, a measure theorized to be reflective of a decreased need for sleep. ¹¹⁵ Similar findings of shorter REM latency and greater REM density were seen in a cross‐sectional study, ¹¹⁴ alongside a reduction in delta sleep (characteristic of slow‐wave sleep). Depressive states are marked by a change in the REM sleep timing, characterized by earlier REM sleep latency. ¹¹⁴

Sleep in the prognosis and observed clinical course of bipolar disorder

There is evidence that sleep disturbance and poor sleep quality in adolescents and young adults increase the risk of future BD development. ¹¹⁶ Sleep disturbances may appear 1 year before the onset of BD, or even earlier, frequently during childhood or adolescence. ¹¹⁷ Among such disturbances, an increased REM density prior to BD onset was observed in one study. ¹¹⁴

Poor sleep quality is also a risk factor for mood episode recurrence, independent of residual mood symptoms. Because of this association, sleep disturbance may be considered a prodromal symptom for recurrence of mood episodes. ¹¹⁸ , ¹¹⁹ Sleep disturbance is the most common prodrome of mania and total sleep time can be a predictor of future manic episodes. ¹¹⁵ , ¹¹⁷ Alterations of total sleep time found in BD individuals are trait features that indicate more severe clinical conditions, worse quality of life and greater impairment in functioning. ¹²⁰ For instance, abnormal sleep duration (<6 or ≥9 h), and higher variability in total sleep time in recovered BD individuals has been associated with earlier depressive relapse. ¹¹⁹ , ¹²¹ Additionally, sleep loss confers a poor prognosis and an increased risk of suicide in BD individuals with a suicide attempt history. ¹²² It is also known that the improvement of sleep in the early stages of treatment of mood episodes indicate a faster recovery and a reduced time until discharge from psychiatric units. ¹²³

Sleep disturbances are commonly observed even outside of acute mood episodes BD. Euthymic individuals with BD often exhibit a clinically significant sleep disturbance, ¹⁰⁸ which is frequently the last symptom to resolve nearing the end of an affective episode and, for many individuals, sleep disturbance may not fully remit, remaining a persistent interepisode symptom. ⁴ , ¹²¹ Euthymic BD individuals show greater sleep fragmentation, as well as higher variability of sleep–wake cycles and sleep efficiency. ¹²⁴ Studies also found that euthymic BD individuals have increased movement during sleep. ¹²⁵ , ¹²⁶ , ¹²⁷ Increased sleep fragmentation with higher movement after sleep onset was associated with a greater rate of relapse in a prospective study. ¹²¹ The persistent biological rhythm disruption in the interepisode period had been associated with functional impairment. ¹²⁸ These results emphasize the importance of addressing sleep issues across all phases of BD for improved individual outcomes and prediction of relapse.

Sleep Disorders Comorbid with Bipolar Disorder

Individuals with excessive daytime sleepiness and significant clinical comorbidities, especially obesity and dyslipidemia, would benefit from investigating the possibility of having obstructive sleep apnea and hypopnea syndrome (OSAHS). There are reports that individuals with BD show an increased prevalence of this condition, as high as 47.5%, when compared to the general population. ¹²⁹ One study describes OSAHS to be associated not only to cardiovascular diseases in this population, but also more refractory and severe BD. ¹³⁰ There might also be the case that these individuals suffer from side effects of commonly used antipsychotics and mood stabilizers that cause daytime sedation. ⁴

Sleep disorders related to circadian rhythm (CRSD) are common and often underdiagnosed in BD. ¹³¹ These disorders are characterized by sleep disturbances due to a primary alteration of the circadian system or misalignment between biological circadian rhythms and social cues. The latter mechanism is the most common in BD, often leading to excessive sleepiness and/or significant insomnia. Since most clinicians are unaware of these disorders, and the typical manifestations are similar (if not identical) to those of an insomnia disorder, it is evident why there is frequent confusion in the differential diagnosis.

DSM‐5 presents five specifiers of CRSD with identified characteristics related to epidemiology, course, and prognosis. ¹³² Among them, the “delayed sleep phase type”, characterized by a delayed timing of the main sleep period compared to the desired bedtime and wake‐up time, is highly prevalent in euthymic BD individuals. ¹³³ When individuals with delayed sleep phase type can adapt their own schedules, they exhibit normal sleep quality and duration for their age. ¹³⁴ The problem lies primarily when the demands of social routines (e.g., work, school, familial demands) are not aligned with one's circadian biology. As individuals with BD have a greater tendency towards eveningness, a significant percentage develops CRSD of the delayed sleep phase type, a phenotype that confers these individuals worse sleep outcomes, as well as greater functional impairment. ¹³¹ , ¹³⁵ A longitudinal study by Takaesu and colleagues showed that individuals with BD and delayed sleep phase type CRSD had a significantly greater risk of relapse in a new mood episode. ¹³⁵ Furthermore, it has been debatable whether or not a delayed‐phase phenotype might be secondary to BD itself. ¹³¹ The evidence thus far points towards circadian dysfunction as one of the main ways through which sleep disturbances arise in individuals with BD.

Objective Assessment of Sleep and Activity Rhythm Using Passive Sensing

Ecological Momentary Assessment (EMA) is a data collection tool aimed at assessing individuals in their daily environment without the constraints of laboratory‐based assessments. Data collection with EMA decreases recall bias and increases ecological validity as data is collected in real time and in the context of an individual's environment. EMA encapsulates multiple methods that aim to assess health states in short time intervals, which can be either passive or active. Passive EMA methods use devices such as actimeters and smartphones to continuously collect data without the active involvement of the individual. Active EMA describes data reported by the individual and may be collected in the form of event‐based sampling (only specific events) or experience sampling (random or scheduled times). This wide range of different methods of assessment allows EMA the potential to be a valuable and dynamic approach to understanding the rhythms of behavior and symptomatology.

Actimetry and The Origins of Naturalistic Sleep Tracking

While self‐reported data and polysomnography have been valuable tools in understanding sleep, they have their inherent drawbacks, such as recall bias and the artificial sleep environment of a lab, respectively. This has led to the growing popularity of actimetry and passive data sensing technologies in the field of sleep research. Actimetry (or actigraphy) is a method of continuous monitoring of motor activity using an accelerometer worn on the wrist or hip. ¹³⁶ This technology emerged in the early 1970s, providing data that enables comprehensive analysis of time series data for activity and rest, which serve as a proxy to define sleep measures. ¹³⁷ Some available devices also monitor exposure to light and wrist temperature.

Literature of actimetry in euthymic individuals with BD is summarized in a meta‐analysis by Geoffroy and colleagues. ¹³⁸ Pooled results indicate that individuals with BD show a greater sleep latency, sleep duration, and wake after sleep onset, as well as lower sleep efficiency. Another more recent meta‐analysis by De Crescenzo and colleagues describe the same results for pooled analyses. ¹³⁹ The results of longer sleep duration in BD is corroborated by other reports, ¹⁴⁰ , ¹⁴¹ which might indicate either a trait of BD or, at least in part, a side effect of common medications used to treat the disorder. ¹⁴² On the other hand, this also indicates a discrepancy in objective assessment and subjective perception; that is, an observed longer objective sleep duration with daytime sleepiness or needing less sleep.

Some studies show that these parameters have satisfactory correlations with those assessed by polysomnography. ¹⁴³ , ¹⁴⁴ However, other studies show that actigraphs record shorter sleep latency, advanced sleep onset time, increased number and duration of nighttime awakenings, increased nighttime sleep duration, and increased number and duration of naps compared to subjective sleep reports. ¹⁴⁵ Therefore, considering these low predictive values and overestimation of sleep, some researchers currently disqualify actimetry as a precise sleep–wake indicator. This discussion grows as we encounter several actigraphs in the market that do not disclose the algorithms used to calculate sleep variables based on motor activity/acceleration. Nevertheless, actimetry can still be useful for measuring the circadian period and has face validity as a measure of rest‐activity, ¹⁴⁶ a fundamental variable for characterizing biological rhythms. Hence, recent research has focused on identifying circadian activity disturbance through objective sleep assessment.

Shifting Gears – Rest‐activity Rhythms Beyond Sleep

The rest‐activity rhythm refers to the profile of an individual's active and non‐active periods throughout the day. When analyzing rest‐activity rhythms, the magnitude (i.e., the numerical quantity, including highest, lowest, and average values) is only one possible interpretation of this actigraphic parameter (Fig. 2). A growing body of research describes the amplitude of activity, which represents the degree of changes in magnitude from the highest to the lowest values, as a possible clinical marker of mood disorders. A few studies have described a lower daily average of activity as a marker of BD. ¹³⁹ , ¹⁴¹ , ¹⁴⁷ On the same note, motor retardation differentiates bipolar from unipolar major depressive disorder. ¹⁴⁸

Fig. 2.

Typical variations in rest‐activity rhythms assessed via continuous monitoring. Bars represent levels of motor activity at a given minute of the day for four consecutive days.

In addition to magnitude, one may analyze the phase of the rest‐activity rhythm to determine timing of peaks and troughs. An interesting prospective study found that the rest‐activity rhythm during depressive periods differs from euthymia or mania/hypomania because of an overall lower activity level, later activity onset, midday elevation of activity, and low evening activity. ¹⁴⁹ Further studies show that individuals with BD in the euthymic state show higher activity levels during their least active 5 h and lower circadian relative amplitude of activity than healthy individuals. ¹²⁵ , ¹²⁶ Individuals with BD also show earlier acrophases of activity and lower overall daily activity levels. ¹²⁷ Song and colleagues show a prospective pattern of associations of circadian phase disturbances and mood states in individuals with BD. ¹⁵⁰

Furthermore, rest‐activity can be reported in terms of regularity and stability over time. The two metrics that have been extensively used in research are intradaily variability (IV) and interdaily stability (IS). ¹⁵¹ IV is a measure of circadian rhythm disturbance through activity and is calculated as a ratio of bin‐to‐bin (e.g., minute‐to‐minute) variability to overall activity. IS is calculated as a ratio of the variance of average activity to the mean 24‐h profile and represents the degree of stability of the activity rhythm across days. Several studies have created analytical methods for assessing regularity of activity rhythms in BD. One study observed a higher IV in BD individuals, indicating more fragmented activity rhythms and less predictable daily activity profile. ¹²⁵ However, a larger study of 23,000+ individuals showed that both IV and relative amplitude differentiate people with diagnosis of BD from controls. ¹⁴⁰ Another report calculated individual activity variability, separating groups in terms of stability of rest‐activity cycles and evidenced that the group with greater rest‐activity instability showed more variability in mood symptoms, higher reports of sleep disorder, and later and more variable wake‐up‐times and bedtimes. ¹⁵²

A few other studies chose to combine assessments of activity regularity with magnitude and phase in order to identify BD phenotypes of activity profiles. One study separated 145 adults with BD in three groups based on profiles of acrophase, IV, IS, and relative amplitude. ¹⁵³ The results indicate that a group of individuals with a later phase and more irregular rhythms showed more lifetime manic‐hypomanic and depression symptoms as compared to the group with an earlier and robust profile.

Together, these results reinforce a few hypotheses described in the sections above. Firstly, they indicate that a later profile of activity (possibly indicating later chronotypes) are a risk phenotype for mood episodes in BD. Secondly, they show that irregularity of activity rhythms (and likely abnormalities in the sleep–wake cycle) relate to a higher risk of mood episodes. Finally, they confirm the significant decrease in daily activity and amplitude seen especially in depressive states.

New tools of Ecological Momentary Assessment and Digital Phenotyping

This section delves into the application of EMA using smartphones and active sampling in the context of BD, shedding light on how innovative methodologies enable researchers and clinicians to capture real‐time data on sleep patterns, circadian rhythms, and mood fluctuations.

The rise of technology has introduced novel methods of assessment that led to the development of the term digital phenotype, which describes one's digital patterns that may predict the diagnosis and/or prognosis of a disorder. ¹⁵⁴ In individuals with BD, parameters from passive and active sensors have been associated with mood symptoms and shown good predictive capacity for mood states (Table 1 ). Using geolocation data from smartphones of 29 individuals with BD and 18 controls for 12 weeks, severity of depressive symptoms was associated with longer screen time, more received and less answered incoming calls, fewer outgoing calls, and less physical activity as assessed by location of cell towers. On the other hand, severity of manic symptoms was associated with more outgoing text messages, shorter incoming text messages, fewer phone calls, and more physical activity. ¹⁵⁵ Another study assessing 23 individuals with BD for 10 days showed they had shorter call durations when euthymic than when in a depressed or mixed state. ¹⁵⁶ A similar finding was reported in an independent sample of 203 individuals with BD and 109 controls assessed for 12 weeks, where the control group had a shorter duration of calls and less missed calls. The BD group also had greater sleep time variability and less sleep between midnight and 6:00am. ¹⁵⁷ Nighttime bedroom light exposure was associated with a hypomanic state in 184 individuals with BD assessed for 7 days. ¹⁵⁸

Using continuously improving technology of smartphones and other electronic devices also allows researchers to capture real‐time perception of mood and calculate longitudinal mood instability. The following work is part of a larger project to develop a self‐monitoring system for individuals with BD ¹⁵⁹ and all samples were assessed daily for 9 months. Smartphone‐based self‐assessments of mood, cognition, activity, and sleep on 84 individuals with BD were compared to gold‐standard assessments of depression (Hamilton Depression Rating Scale), for which the model had moderate effect size, and mania (Young Mania Rating Scale), for which the model had a very weak effect size. ¹⁶⁰ A related study assessing the prevalence of symptoms in 117 individuals with BD reported a significantly positive association between mood and activity. ¹⁶¹ Mood instability was associated with greater perceived stress and worse quality of life and functioning in a sample of 84 individuals with BD. ¹⁶² In comparison to individuals with BD‐I, those with BD‐II had lower intensity of manic symptoms but greater mood instability during depressive episodes. ¹⁶³ In an independent study of 139 individuals assessed for 7 days, greater variability of negative affect was associated with bipolar spectrum disorders and predicted the development of these disorders after a 3‐year follow‐up. ¹⁶⁴

A great volume of EMA work in BD has been targeted towards the predictive capacity of EMA assessments to distinguish between disorders and affective states. Cho et al., ¹⁶⁵ collected data using smartphones, actimeters, and daily self‐reported mood states over the course of 2 years to form a predictive model of mood in a mixed sample of 55 individuals with major depressive disorder, BD‐I, and BD‐II. The accuracy of the cumulative model to predict mood state for the next 3 days were 65%, 64%, and 65% of each group, respectively. For all individuals, this model was able to predict if they were in a depressive (87% accuracy), manic (94% accuracy), or hypomanic episode (91% accuracy), as well as when euthymic (85% accuracy). There are numerous studies that analyze the predictive capacity of EMA assessments ‐ their methodology, sample, length, and frequency of data collection are summarized in comprehensive systematic reviews. ¹⁶⁶

Research to date supports the usefulness of long‐term tracking for informing both the individual and their clinicians on the trajectory of BD, making the inclusion of EMA in clinical practice much more probable. A conceptual framework has also been suggested to incorporate passive sensing to aid in detecting early signs of BD. ¹⁶⁷ Through EMA, researchers will be able to uncover how temporal dynamics differ between individuals with disorders in the BD spectrum and ultimately develop more precise and personalized interventions.

Challenges and Recommendations for Clinical Practice

The study of circadian function and sleep plays is crucial to understand various aspects of BD, like the evolution and fluctuation of sleep and mood, which is critical for predicting mood episodes. However, the challenges in measuring circadian function present significant obstacles to achieving diagnostic and prognostic validity of recent EMA technologies. These techniques provide valuable insights into circadian rhythms, but their utility is contingent on the reliability and reproducibility across various populations. Establishing standardized protocols for data collection, processing, and analysis is imperative to enhance the clinical utility of these methods. With technological advancements, however, there are limitations to consider and continuously revise, which include challenges related to technology (e.g., battery drainage, sensor precision), methodology (e.g., adherence to protocol, sample sizes and biases), and privacy (e.g., data anonymization and storage). ¹⁶⁸ These are challenges worth addressing, however, as recent research by Tonon et al., ¹⁶⁹ Pilz et al., ¹⁷⁰ and Slyepchenko et al. ¹⁷¹ highlight the importance of robust methodologies for handling data in studies utilizing actimetry and passive data sensing.

To address these challenges, recommendations should focus on standardizing assessment methods, improving data management techniques, and addressing privacy concerns associated with EMA technology. Collaborative efforts across the research community will be essential in advancing the clinical use of passive data sensing in BD. The use of technology in healthcare is inevitable, and therefore, further understanding how to incorporate new technology in the management of BD will contribute to a more refined understanding of this complex disorder.

Conclusion

Bipolar disorder is characterized by a spectrum of mood states, and alterations in biological rhythms, including sleep phase, structure, and duration are consistently reported throughout all illness states. These disruptions in biological rhythms and sleep are associated with various clinical outcomes, including lower quality of life, increased risk of suicide attempts, impaired cognitive functioning, and higher relapse rates of mood episodes.

Understanding the connections between circadian rhythms, sleep disturbances, and BD pathophysiology can inform clinical states, treatment, and interventions aimed at improving the quality of life for BD individuals. Specifically, the identification of circadian clock genes' involvement in mood regulation provides potential targets for therapeutic interventions. Additionally, the observation of phase‐shifts in circadian rhythms during mood episodes suggests that interventions aimed at realigning these rhythms could be beneficial in managing BD. Moreover, recognizing the impact of chronotype on mood fluctuations underscores the importance of chronotherapeutic treatment approaches that account for these variations.

To further advance our understanding of biological rhythms, sleep, and their implications in BD, future research should focus on several key areas. First, investigating the precise mechanisms by which circadian clock genes influence mood regulation will provide valuable insights into potential therapeutic targets. Second, examining the impact of light–dark transitions, photoperiod variations, and non‐photic zeitgebers on circadian rhythms in BD is essential to guide strategies for minimizing circadian disruptions and improving overall stability in BD individuals. Third, the bidirectional impacts of sleep disturbances and mood regulation in BD requires further investigation. Finally, to overcome the challenges in measuring circadian function, it is imperative to work on standardizing assessment methods, refining data management techniques, and addressing privacy concerns associated with EMA technology. Collaborative efforts within the research community are crucial to establish a unified approach that enhances the clinical applicability of passive data sensing in bipolar disorder studies. By addressing these challenges, researchers can contribute to a more comprehensive and reliable understanding of circadian function and its implications for mental health. Recognizing the significance of biological rhythms and sleep patterns in the context of BD offers hope for more effective diagnosis and treatment, ultimately leading to better outcomes for individuals affected by this challenging disorder.

Disclosure statement

No competing interest.

Author contributions

ACT contributed to the conception and design of the study, drafting the manuscript and the figs. AN contributed to the search for relevant literature, drafting the manuscript and the figs. MM contributed to the search for relevant literature, and drafting the manuscript. FAG contributed to drafting the manuscript, and revising the final text. MPH contributed to the conception and design of the study, drafting the manuscript, and revising the final text. BNF contributed to the conception and design of the study, drafting the manuscript, and revising the final text.