1. Introduction

Traumatic brain injury (TBI) is a major cause of death among civilians and military personnel, as a consequence of motor vehicle accidents and explosive devices. Every year in the US, around 1.4 million civilians sustain TBI, of whom 124 000 develop long-term disability.[1] Presently, there are over 3.3 million Americans living with long-term disability related to TBI.[2] According to a recent systematic review, an estimate of the aggregate annual incidence of fatal TBI in the EU was approximately 235 per 100 000,[3] although substantial variation was seen among the member nations of the EU. A prospective study of hospitalized patients in the New South Wales state of Australia found that approximately 4 per 100 000 deaths were caused by CNS-related injuries associated with high-energy trauma, while 6.4 deaths per 100 000 were caused by CNS injuries from low-energy trauma.[4] A report from the Japanese trauma network indicated that the incidence of TBI was 27 per 100 000,[5] while more recent studies from Turkey and Malaysia report a much higher incidence of TBI in their populations.[6,7]

Recent studies have revealed a high prevalence of sleep disorders in the TBI population including narcolepsy, post-traumatic hypersomnia (PTH), periodic limb movements in sleep (PLMS) and obstructive sleep apnoea (OSA).[8–15] These disorders can have serious consequences on the overall well-being of the patient. Subjective complaints of ‘sleeping problems’ are common, and have been reported in 47% of 639 patients presenting to a minor head injury clinic.[16] Sleep, and especially rapid eye movement (REM) sleep, is believed to be involved with learning and consolidation of memories,[17–19] even with the development of insight.[20] Sleep loss/disruption may have a deleterious effect on hippocampal neurons involved in memory formation.[21,22] Cognitive deficits may result from sleep disturbances[23] and these may worsen the impairment in TBI.[8]

Sleep disorders that can affect daytime function such as insomnia, narcolepsy, shift-work sleep disorder and OSA are significant risk factors for motor vehicle accidents and can increase the risk of TBI.[24–27] Thus, there appears to be a complex relationship between disorders of sleep and TBI. Part of the confusion stems from the high risk of accidents among those with sleep disorders, especially sleep apnoea.[28–30] There is often a question about the existence of a sleep disorder prior to TBI, which may have been the underlying cause of the accident in the first place. Despite this complex relationship, we cannot discount the increased prevalence of sleep disorders, including sleep-disordered breathing (SDB), in patients with TBI who were asymptomatic prior to injury.

This review presents the epidemiology, pathophysiology and management of various sleep disorders that have been described in association with TBI.

2. Epidemiology of Sleep Disorders in Traumatic Brain Injury

2.1 Hypersomnia

Subjective excessive daytime sleepiness (EDS) can be estimated by the Epworth Sleepiness Scale (ESS). The ESS is a self-administered eight-item questionnaire to assess subjective daytime sleepiness in which the subject rates from 0 to 3 the likelihood of dozing in specific situations.[31,32] The objective measure of daytime sleepiness used is the Multiple Sleep Latency Test (MSLT).[33] This a series of four to five naps at 2-hour intervals of 20 minutes duration after a normal night of sleep documented by nocturnal polysomnography. Subjects sleep with EEG leads in place and sleep stages are scored in 30-second epochs. The mean sleep latency over these naps is the MSLT score. EDS is defined as an MSLT score of <10 minutes and pathological sleepiness as an MSLT score of <5 minutes.

EDS, defined as a score of >10 on the ESS, was found in 10.9% of a cohort of young and middle-aged healthy subjects.[34] Other estimates of questionnaire-based daytime sleepiness range from 5% to 15% in the general population.[35] Multiple studies have resulted in a general conclusion that EDS is common after TBI.[9,10,36] In a prospective multicentre study of 87 TBI patients,[9] EDS was found in 25%, while in an earlier case series of 71 patients with TBI,[36] 47% of subjects had EDS and 18.3% of those patients had pathological sleepiness as measured by the MSLT score. In a more recent retrospective study,[10] where overnight polysomnographic data were obtained in 54 patients with TBI, EDS was reported by 52% of all patients as documented by an ESS score of >11. Fifty-three percent (n = 15) of the 28 subjects with an ESS score of >11 who underwent MSLT had a sleep-onset latency of <5 minutes, suggestive of severe hypersomnia.

2.1.1 Hypersomnia Related to CNS Pathology

Narcolepsy

Case reports of ‘post-traumatic narcolepsy’ were published as early as 1941,[37] but it was impossible at the time to distinguish other sleep disorders such as sleep apnoea and narcolepsy from what is today known as PTH.[38] It is possible that narcolepsy could have preceded the head trauma or even contributed to it. On the other hand, a brain injury that is not of a traumatic nature such as irradiation of pituitary tumour has also been reported to cause narcolepsy with cataplexy.[39] Thus, it is conceivable that TBI can trigger events to unmask the disease in subjects who are at risk for developing it. In a retrospective study by Verma et al.,[10] the authors reported that 32% (9 of 28) of their subjects with TBI had severe hypersomnia (MSLT score of ≤5 minutes) and two or more episodes of sleep-onset REM periods (SOREMPs) compatible with a clinical diagnosis of narcolepsy. Only two patients in this group were positive for HLA-DRB1-15 and DQB1*0602 antigens, while this is found in 85% of those with idiopathic narcolepsy and cataplexy.[40] In a prospective multicentre study,[9] we found that 6% of TBI patients with a mean sleep latency of <10 minutes on the MSLT met the diagnostic criteria for narcolepsy, which is significantly higher than the sporadic form of the disease in the general population (0.056%).[41]

Post-Traumatic Hypersomnia

PTH is a disorder of excessive sleepiness that occurs as a result of a traumatic event involving the CNS.[42] The diagnosis of PTH is made when the following criteria are met: (i) sleepiness begins only after TBI; (ii) other causes of sleepiness are excluded by history and nocturnal polysomnography; and (iii) the MSLT score is <10 minutes with less than two SOREMPs.[43] It is important to exclude all other causes of hypersomnia in patients with TBI, including the sedative effects of medications such as antiepileptic drugs frequently used in this population.[15] In a retrospective study of 184 patients presenting to an outpatient sleep clinic with EDS after head and neck trauma, 49% had PTH.[44] A study of brain-injured patients in an inpatient rehabilitation facility found that 30% of these subjects had PTH.[36] In the only prospective multicentre study that has evaluated unselected TBI patients with nocturnal polysomnography and the MSLT, the prevalence of PTH was 11%.[9]

2.1.2 Hypersomnia Related to Sleep-Disordered Breathing

The syndrome of SDB encompasses a group of breathing disorders in sleep of varying severity such as OSA, upper airway resistance syndrome and central sleep apnoea.[45,46]

The most clinically important and the most widely studied SDB in patients with TBI is OSA. This is caused by narrowing or complete occlusion of the upper airway accompanied by continued attempts to breathe against this increased resistance. This results in intermittent hypoxemia, sleep disruption and EDS or difficulty sleeping.

Multiple studies with varying study designs have estimated that 23–70% of patients with TBI have SDB,[9,10,44,47,48] which is significantly higher than the expected prevalence in the population.[49] However, the true prevalence is likely closer to 23% as estimated by a prospective multicentre cohort study.[9] Cognitive deficits most commonly involving vigilance, attention, arousal, memory and executive functions have been associated with OSA.[23] It is also well established that cognitive impairment is associated with TBI and increases with increasing severity of injury.[50] Patients with TBI and OSA have a greater impairment of neurocognitive functions, especially of memory and sustained attention, as compared to TBI patients without SDB.[8]

2.2 Insomnia

Insomnia is defined as a repeated difficulty with sleep initiation, duration, consolidation or quality that occurs despite adequate time and opportunity for sleep, and results in some form of daytime impairment.[51] A complaint of sleep disruption over the previous year was reported in 30% of the general population, while 10% of these individuals had accompanying symptoms of daytime impairment.[52,53]

Even though EDS is a very important chronic symptom after TBI, hospitalized patients with a recent episode of TBI are more likely to complain of difficulty in initiating or maintaining sleep that can persist after hospital discharge.[54] Even minor head injury can be associated with decreased sleep quality and an increase in sleep interruptions as compared to the pre-injury state.[55,56] It appears that at least 30–50% of patients with TBI in the setting of outpatient rehabilitation centres have difficulty sleeping, of whom 64% complain of early morning awakenings and around 45% have problems initiating sleep.[11,57,58] When compared to subjects who have had orthopaedic injuries[59] or other non-neurological injuries,[60] patients with TBI are more likely to have difficulty initiating and maintaining sleep.

It appears that insomnia after TBI is associated with milder forms of brain injury in the presence of pain or depression.[60,61] Assault, as a causal mechanism of head injury, can in itself influence the effect of TBI on sleep. In a questionnaire study of patients with mild head injury where personal assault was the second most common cause of TBI (44%), 46% of patients complained of sleep disturbance that persisted at 6 months.[16] Despite the disproportionately high dropout rate among patients with TBI secondary to assault, this underscores the possibility that the additional psychological trauma of being personally attacked may contribute to insomnia as distinct from a head injury sustained in a road traffic accident.[11]

Even though, patients with TBI report poor-quality sleep, polysomnographic findings from a study[12] of ten patients with TBI and age- and sex-matched controls showed comparable sleep-onset latency, arousal index and sleep efficiency between the two groups. However, patients with TBI had significantly more awakenings, concordant with their self-reported symptoms noted on an earlier study.[11] These findings are consistent with those of an earlier study of 14 patients with mild to severe TBI compared to healthy controls,[62] which documented a disparity between subjective and objective measures of insomnia in these patients. The majority of the TBI patients (71.4%) in this sample had subjective insomnia by questionnaires. Of the five subjects reporting good sleep quality, three had objective (polysomnographic) evidence of insomnia as defined by sleep-onset latency or a wake after sleep onset of over 30 minutes. Of the four patients with objective evidence of good-quality sleep, two met criteria for insomnia by questionnaires. Similar to hypersomnia, subjective complaints of insomnia may not be reliable to affirm the diagnosis of insomnia in this population.

2.3 Circadian Rhythm Disorders

The literature is lacking in prospective studies specifically designed to describe the epidemiology of circadian rhythm disturbances after TBI. However, there have been published reports that suggest the existence of post-traumatic delayed-sleep phase syndrome (DSPS).[63–65] A case report of a patient with a non-24-hour sleep-wake cycle (hypernyctohemeral) syndrome as a late complication of TBI has been reported.[66] Usually, a minority of patients presenting with insomnia in the general population have a circadian rhythm disorder,[67] but in a recent study by Ayalon and colleagues,[68] 36% of patients with minor TBI, who were referred for evaluation of insomnia, were diagnosed with a circadian rhythm disorder. Of these individuals, 52% had DSPS and the rest had an irregular sleep-wake pattern (ISWP). Three patients in the ISWP group had a decrease in the amplitude of the oral temperature cycle as compared to the DSPS group. In another study by Steele and colleagues,[69] ten patients with DSPS post-acute TBI had similar habitual sleep times (assessed by Morningness-Eveningness Questionnaire) and onset of melatonin secretion as compared to their controls. Conflicting results from the two studies could have resulted from lack of statistical power from a small sample size and/or the inherent differences between the pathophysiological processes associated with nature and timing of the injury. Future studies are necessary to better describe the nature of circadian rhythm disorder associated with TBI.

2.4 Periodic Limb Movements in Sleep

PLMS is a movement disorder of sleep characterized by slow and rhythmic movements of predominantly lower limbs (although upper limb movements may also occur) during non-REM sleep. Approximately 80% of patients with restless legs syndrome (RLS) also have PLMS,[70] but RLS is a disorder of wakefulness characterized by a subjective feeling of discomfort accompanied by a need to move the extremities. This occurs more during the evening and night and may cause sleep-onset insomnia. Periodic limb movements (PLMs), on the other hand, occur during sleep and may be a cause for sleep-maintenance insomnia. PLMs are rhythmic anterior tibialis contractions of 0.5–5 seconds duration in groups of at least four contractions interrupted by 5–90 seconds of quiescence.[71] Currently, more than 15 PLMs per hour (PLM index, PLMI) is considered abnormal in adults, while an index of >5 PLMs per hour is considered abnormal in children. All of the current studies on TBI have used the older criteria of PLMI ≥5 per hour as the minimal criteria for the diagnosis of PLMS. The proportion of institutionalized TBI patients with PLMS was reported to be as high as 25.4%,[36] while in a more recent multicentre study of both institutionalized and ambulatory TBI patients the prevalence of PLMS was 7%.[9]

2.5 Other Movement Disorders and Parasomnias

There is scant literature describing the prevalence of parasomnias associated with TBI, but the one most commonly reported is REM sleep behaviour disorder (RBD). Clinical RBD is characterized by dream enactment and the presence of REM sleep without normally occurring atonia. Subclinical RBD is the descriptive term for the polysomnographic detection of excessive electromyogram tone during REM sleep and/or movement activity without overt dream enactment.[72] In a recent study, parasomnias were the presenting complaint in 25% of patients with TBI, of which clinical or subclinical RBD was the most common, having been found in 7 of 54 (13%) patients with sleep complaints.[10]

Sleep-related bruxism is also known as nocturnal tooth grinding.[72] Since bruxism is fairly common in the general population, it is not surprising that this should also be found in patients with TBI, but there are no reliable studies that have measured the prevalence of bruxism in this population.

Jactatio nocturna or head banging is a sleep-related rhythmic movement disorder rarely seen in adults. There is one report of the onset of jactatio nocturna after TBI in an adult patient who also had global encephalopathy related to the injury.[73]

2.6 Dreams

There have been conflicting reports with regard to potentially reduced REM sleep in patients with minor TBI.[74,75] Even though early studies including case series have reported a decrease in dreaming among patients post-TBI,[76,77] a more recent study based on patient questionnaires[78] reported a change in the quality rather than the quantity of dreams. In 51 subjects with TBI, frequent dreaming with threatening content increased by 23.5%, while dreams with sexual content decreased by 9.8% after TBI.

3. Pathophysiology

While there have been no studies to date that have described a localizing neuroanatomical injury as a predictor of hypersomnia in patients with TBI, there is reason to believe that hypothalamic injury with decreased levels of wake-promoting neurotransmitters such as hypocretin (orexin-A) and histamine may be involved in the pathophysiology of EDS associated with TBI. Early studies have shown that hypothalamic damage appears to be common in TBI patients.[79,80] Low CSF levels of both hypocretin and histamine are associated with hypersomnia.[81] Measured levels of CSF hypocretin were low (<320 pg/mL) in 95% of 44 patients with acute TBI.[82] The lowest levels of CSF hypocretin were found in patients who were comatose. When studied 6 months post-injury, 4 of the 14 patients with EDS had low hypocretin levels (289 ± 64 pg/mL), while those without EDS had normal levels (444 ± 113 pg/mL; p = 0.05).[14]

The role of CSF histamine as a diagnostic predictor in PTH is less clear. A recent study reported low CSF histamine levels in idiopathic hypersomnia but not in patients with OSA.[81] Low CSF histamine levels have also been reported in patients with narcolepsy with or without cataplexy and with or without low CSF hypocretin levels.[83] Since histamine in the CNS is produced largely by the tuberomammillary nucleus,[84,85] this suggests an area to be further explored in future investigations in patients with TBI. The relationship between narcolepsy and TBI is still controversial, since this is generally regarded as a disease with a genetic basis and presumed to be present prior to injury. Case reports of post-traumatic narcolepsy have been published as early as 1941.[37] However, these reports do not have the necessary components of polysomnography to make a confident diagnosis of narcolepsy. On the other hand, more recent investigators have questioned whether narcolepsy after TBI is ‘unmasked’ in subjects who have pre-existing risk factors.[86]

The relationship between SDB and TBI is complex. Pre-existing untreated SDB associated with hypersomnia can predispose individuals to motor vehicle accidents, which can lead to TBI. However, injury to upper respiratory muscles along with TBI can cause post-traumatic OSA.[47] It is also important to consider factors such as weight gain, supine sleep and medication effects such as reduced muscle tone and respiratory depression that can predispose TBI patients to SDB.

Insomnia may develop in TBI patients as a result of ‘psychological trauma’ resulting from personal assaults and accidents resulting in TBI. Psychophysiological insomnia may arise in the context of the recovery period after TBI and persist long after discharge from the acute care hospital.

The development of circadian rhythm disorders after TBI can be explained by one of two mechanisms. The first would be injury to the suprachiasmatic nucleus or other thalamic structures. While this is possible, there is no evidence to support that mechanism in reports to date. The second possibility is that sleep/wake disturbances arise as a result of prolonged hospitalization, loss of Zeitgebers and use of sedative-hypnotics and analgesics following TBI. This can be seen in other hospitalized patients, especially in intensive care units and more so after mechanical ventilation. However, prolonged hospitalization as a possible mechanism for circadian rhythm disorders in all patients with TBI is debatable. A study of 42 patients with mild TBI evaluated for insomnia in an outpatient clinic revealed that, while 36% of those subjects had a circadian rhythm disorder,[68] prolonged hospitalization did not play a significant role in the pathophysiology of these cases.

4. Management

4.1 Diagnosis

Based upon the high prevalence of sleep disorders in TBI such as SDB and PTH,[9] and the fact that SDB associated with TBI contributes to additional impairment,[8] we recommend that patients with TBI should undergo nocturnal polysomnography followed by the MSLT. The MSLT may be cancelled if significant SDB is found on nocturnal polysomnography, but is necessary for the diagnosis of narcolepsy and PTH. Since the correlation between objective measurement of sleepiness by the MSLT and subjective sleepiness score by the ESS is poor in TBI subjects, patient selection for nocturnal polysomnography based on the ESS can be unreliable.[9,36]

The diagnosis of insomnia is usually weighted by the historical information obtained from patients, family members or caregivers.[52] However, in patients with TBI, due to the poor predictive value of subjective measures and a high prevalence of SDB, PLMS and circadian rhythm disorders[9] that can also present with insomnia, further testing using nocturnal polysomnography is necessary to detect these co-morbid sleep disorders. It should be noted, however, that polysomnographic characterization of sleep architecture has not been systematically evaluated in this population.

Actigraphy and/or sleep logs are crucial in the diagnosis and management of circadian rhythm disorders associated with TBI.

4.2 Treatment

SDB in this population is effectively treated with continuous positive airway pressure therapy, which is titrated in the sleep laboratory with polysomnography until all apnoea, hypopneas and respiratory effort-related arousals are eliminated.[87] It should be noted that despite elimination of SDB with continuous positive airway pressure, reversal of hypersomnia and neurocognitive deficits may not follow in this population.[87] This could be due to persisting PTH after treatment of co-morbid SDB.

The data on the efficacy of modafinil in the treatment of PTH and narcolepsy after TBI are equivocal at a dose of 200 mg/day.[87] This is due to the variable response seen in a small sample size. It is possible that higher doses of modafinil or other medications such as armodafinil (150–250 mg/day), methylphenidate (5–60 mg/day) or dextroamphetamine (5–60 mg/day) may be of benefit. Modafinil in doses up to 400 mg/day failed to show improvement in fatigue and subjective sleepiness in a larger group of patients with TBI, but these patients did not undergo polysomnography and exclusion of SDB.[88]

Cognitive behavioural therapy appears to be of benefit in TBI patients with insomnia[89] and has also been shown to improve the emotional well-being of these patients.[90] Treatment of post-TBI insomnia with hypnotics has not been systematically studied, even though as many as 20% of TBI patients may receive such treatment.[91] Currently used medications include the melatonin receptor agonist ramelteon and the nonbenzodiazepine hypnotics that act on the Ω-1 benzodiazepine receptor subtype located in the GABAA receptor complex: zolpidem, zaleplon, zopiclone and eszopiclone. These are all indicated for sleep-onset insomnia, but zaleplon has an onset of action of 14–30 minutes and a short, 1-hour half-life that enables its use in the middle of the night as needed.[92] Eszoplicone has a long half-life of 6 hours and is approved for long-term use, as well as sleep-maintenance insomnia. Ramelteon is also approved for long-term use in chronic insomnia, and has no potential to worsen SDB. Zolpidem has an onset of action of 30 minutes and a half-life of 1.5–4.5 hours, with an extended-release formulation available. It has been associated with cognitive impairment,[93] and also with parasomnias such as sleep walking.[94] Zopiclone has a half-life of 3–6 hours and is not available in the US. Benzodiazepine hypnotics are not recommended for routine use after TBI.[95]

Even though the administration of methylphenidate did not have adverse effects, it was not shown to have significant benefit in an attempt to improve the sleep/wake cycle of patients in an inpatient brain injury unit.[96] Bright-light therapy (10 000 lux) and melatonin (0.1–0.5 mg/night), useful in the treatment of circadian rhythm disorders, have not been systematically studied in TBI patients. At this time, there is no literature to support the use of ramelteon in circadian rhythm disorders.

Bruxism can be treated with a tooth guard when feasible to prevent permanent damage to dentition. Successful treatment of bruxism with botulinum toxin-A in a patient with TBI has been reported.[97]

Pramipexole at a dose of 0.375 mg/day was effective in the treatment of PLMS after TBI in one prospective study, but only 4 of the 57 TBI patients studied had PLMS.[87] In non-TBI patients, PLMS is treated with pramipexole at 0.125–0.5 mg/day or ropinirole at 0.25–4 mg/day.

Imipramine was used successfully to resolve jactatio nocturna in one patient with TBI.[73]

5. Conclusion

It is clear that all TBI patients are at considerable risk for various sleep disorders, the most common being OSA, PTH, narcolepsy, circadian rhythm disorders and insomnia. Because of the unreliability of subjective sleep symptoms in these patients, objective testing with nocturnal polysomnography and the MSLT are necessary to diagnose co-morbid sleep disorders. Future studies must systematically evaluate new and existing treatment modalities for these disorders in TBI patients with meaningful outcome measures and long-term follow-up.