Insomnia

The epidemiology of insomnia is complicated by several factors, including the distinction of symptom from disorder. First, because epidemiologic studies of sleep disorders rely on self-reports, biases may occur because of inexact wording of questions, length or form of interview, subjects’ avoidance or exaggeration of symptoms, measurement of confounding comorbidities, concurrent medication use, and memory difficulties (Moul et al. 2004). In addition, deciding when a complaint is clinically significant is a challenge for epidemiological and clinical studies. Finally, the validity of epidemiological estimates of specific insomnia disorders may be limited by the low degree of interrater reliability of clinical diagnoses (Buysse et al. 1994). In-depth reviews of insomnia epidemiology are available elsewhere (Ohayon 2002), but major findings are reviewed below.

The one-year prevalence of insomnia symptoms in the United States and other Western nations is approximately 30%–40% in the general population and up to 66% in primary care and psychiatric settings. The prevalence of primary insomnia as a specific disorder is in the range of 5%–10% of the general population. Aside from estimating the prevalence of nighttime symptoms or syndromal insomnias, investigators continue to expand the study of daytime symptoms in primary insomnia, among them poor concentration, fatigue, dysphoria, sleep dissatisfaction, and mental inefficiency (Moul et al. 2002).

Risk factors for insomnia can be categorized into predisposing, initiating, and maintaining factors (Spielman et al. 1987). Established vulnerability factors include advancing age, female sex, being divorced or separated, unemployment, and comorbid medical and psychiatric illness. Genetic factors, although inadequately studied, represent another possible risk. Factors that commonly initiate or maintain insomnia include psychosocial stresses such as moves, relationship difficulties, occupational and financial problems, and caregiving responsibilities (Kappler and Hohagen 2003). An additional maintaining factor that forms the basis for many therapeutic interventions is the adoption of counterproductive sleep habits (described later in this chapter). It remains difficult to estimate the exact roles of different risk factors in the initiation and maintenance of insomnia in specific populations.

The natural history of insomnia, determined from population-based studies, has not been well described. However, follow-up studies of clinical patients and population samples indicate that insomnia often persists for years and may become more severe with time (Mendelson 1995; Vollrath et al. 1989), although remission has been described in one study of older adults (Foley et al. 1999). The longitudinal epidemiology of insomnia is complicated by normal aging, which leads to lighter and more fragmented sleep. Studies of older adults again suggest that insomnia is strongly related to medical and psychiatric conditions and that it decreases when these problems are adequately treated (Katz and McHorney 1998).

The consequences of insomnia can be substantial. Several studies have identified insomnia as a risk factor for the later development of depressive, anxiety, and substance use disorders (Breslau et al. 1996; Chang et al. 1997; reviewed in Riemann and Voderholzer 2003). In patients with major depressive disorder (MDD), insomnia is associated with worse treatment outcomes (Buysse et al. 1999), suicidal ideation (Agargun et al. 1997), symptom persistence (Moos and Cronkite 1999), and recurrence of MDD (Perlis et al. 1997a; Reynolds et al. 1997). Insomnia is also associated with persistence and relapse in alcohol dependence (Brower et al. 2001). Insomnia is associated with increased medical care utilization, more absenteeism from work (Simon and Von Korff 1997), and increased direct and indirect medical costs (Walsh and Engelhardt 1999). Insomnia may be associated with increased motor vehicle and other accidents (Powell et al. 2002), as well as increased incidence of falls in the elderly (Brassington et al. 2000).Various studies have also substantiated reduced quality of life in individuals with insomnia (Hatoum et al. 1998; Léger et al. 2001) that is similar in magnitude to that seen with chronic conditions such as congestive heart failure and MDD (Katz and McHorney 2002). Quality of life is associated with the severity of the insomnia complaint, and in individuals with medical disorders, quality of life has an effect independent of the comorbid condition (Karlsen et al. 1999; McCall et al. 2000). Quality of life problems among individuals with insomnia include role impairments across a broad set of domains such as job performance, social life, and family life (Léger et al. 2002).

Insomnia is often considered to be a disorder of increased arousal. Arousal may refer to many different phenomena, but a working definition of arousal is the individual’s state of central nervous system (CNS) activity and reactivity, ranging from sleep at one end of the spectrum to wakefulness with excitement or panic at the other (Coull 1998). Arousal is related to the function of wake-promoting structures in the ascending reticular activating system, hypothalamus, and basal forebrain, which interact with sleep-promoting brain centers in the anterior hypothalamus and thalamus (see Chapter 1, “Introduction”). Hyperarousal is a state characterized by a high level of alertness that may be present tonically or in response to specific situations, such as the sleep environment.

Psychophysiological and metabolic evidence for hyperarousal in insomnia patients includes findings of increased body temperature and galvanic skin response near sleep onset and increased heart rate and decreased heart period variability during sleep (reviewed in Bonnet and Arand 1997a). Whole-body metabolic rate measured by volume of oxygen consumption per unit of time (VO₂) is higher during sleep in patients with insomnia compared with healthy sleepers (Bonnet and Arand 1995, 1997b), again supporting the general concept of hyperarousal.

Electrophysiological evidence for hyperarousal comes from studies showing elevated high-frequency electroencephalographic (EEG) activity (beta activity) during non–rapid eye movement (NREM) sleep in insomnia patients (Krystal et al. 2002; Merica et al. 1998; Perlis et al. 2001). Beta activity is normally associated with mental activity during wakefulness. Reduced homeostatic sleep drive in insomnia patients, indicated by reduced delta EEG activity, has been found in some studies as well (Merica and Gaillard 1992).

Neuroendocrine evidence for hyperarousal includes increased cortisol and adrenocorticotropic hormone levels before and during sleep, particularly during the first half of sleep, in insomnia patients (Rodenbeck et al. 2002; Vgontzas et al. 2001). Decreased melatonin levels have been found less consistently (Attenburrow et al. 1996; Hajak et al. 1995).

Finally, hyperarousal is suggested by functional neuroanatomic studies of arousal, as indicated by patterns of regional brain metabolic activity during NREM sleep during single-photon emission computed tomography (SPECT) and positron emission tomography (PET) scans. In the first PET study of primary insomnia, patients had increased global glucose metabolic rates during both wakefulness and sleep compared with healthy control subjects, and the usual sleep-related decline in metabolism in brainstem arousal centers was attenuated (Nofzinger et al. 2004) (Figure 1, part A). During wakefulness, primary insomnia patients have reduced dorsolateral prefrontal cortical activity. This constellation of findings suggests hyperarousal during NREM sleep and frontal hypoarousal during wakefulness, which may correspond to patients’ sleep- and wake-related complaints (Figure 1, part B). In patients with insomnia in the context of major depression, EEG beta activity was positively associated with metabolic activity in the orbitofrontal cortex and with complaints of poor sleep quality (Nofzinger et al. 2000), further substantiating the hyperarousal hypothesis. On the other hand, a study using [^99mTc]-HMPAO SPECT in primary insomnia patients (M.T. Smith et al. 2002a) found hypoperfusion across eight preselected regions of interest (including frontal medial, occipital, and parietal cortex) during NREM sleep, with the most prominent effects in the basal ganglia. Thus, imaging studies have begun to show functional neuroanatomic changes during NREM sleep associated with primary and secondary insomnia, although the precise nature of these changes awaits further confirmation.

All cognitive-behavioral models of insomnia outline the relationships between sleep-related thoughts, behaviors, and perceived arousal, their antecedents, and their consequences. However, the core dysfunction underlying insomnia differs across models. In the behavioral model of Spielman et al. (1987) (Figure 2), individual predisposing factors (e.g., heightened physiological or cognitive arousal) interact with external precipitating factors (e.g., life stressors) to produce insomnia. Perpetuating factors (e.g., maladaptive coping strategies such as spending more time in bed) maintain and reinforce insomnia even after the original precipitants recede.

In Morin’s (1993) model, cognitive hyper-arousal, indicated by sleep-focused, ruminative thoughts, particularly around bedtime, is the central component. Cognitive arousal increases physiological arousal. With repetition, this pairing facilitates conditioning between temporal (e.g., bedtime routines) and environmental (e.g., bed, bedroom) cues and sleeplessness. Resulting sleep disturbance and ruminations ultimately lead to daytime consequences such as mood disturbances and fatigue. Over time, these experiences can alter one’s beliefs regarding the ability to sleep and the consequences of sleep difficulties and can lead to the development of maladaptive strategies aimed at maximizing sleep (e.g., spending more time in bed) that further reinforce sleep disturbances and cognitive hyperarousal.

Perlis and colleagues (1997b) proposed a neurocognitive model of insomnia to explain discrepancies between objective polysomnographic findings and subjective reports of poor sleep quality and sleep misperception. Specifically, this model emphasizes brain cortical arousal as a central component and suggests that both physiological and cognitive arousal arise from increased cortical arousal at or around the sleep onset period, as measured by beta EEG activity. The authors posited that this arousal can be conditioned and that it further contributes to the insomnia experience.

Another cognitive model, by Harvey (2002), emphasizes the role of attention biases in the maintenance of insomnia. The model proposes that cognitive strategies (rather than behavioral strategies) employed by people with insomnia are maladaptive and maintain sleep disturbances. Excessive negative cognitive activity leads to increased physiological hyperarousal, which in turn leads to selective attention to and monitoring of autonomic symptoms and environmental cues indicative of sleeplessness. Such selective attention distorts perceptions regarding sleep deficits and adverse daytime effects. Individuals develop maladaptive safety behaviors, such as thought suppression and emotional inhibition, to avoid sleep loss. However, these behaviors further reinforce the negative cognitive activity, physiological arousal, and erroneous beliefs regarding sleep and daytime deficits seen in insomnia.

The assessment of patients with insomnia rests on a detailed clinical history. As with any clinical disorder, the history for insomnia should focus on specific symptoms, chronology, exacerbating and alleviating factors, and response to previous treatments. However, some aspects of the insomnia history differ from those of other disorders and deserve special emphasis.

Although the clinical history is the key to making an insomnia diagnosis, several other tools may aid this process.

Sleep-wake diaries covering 1 or 2 weeks, in which patients record their actual sleep hours and sleep experiences, can be invaluable. Diaries are useful for establishing patterns of sleep, as well as for indicating the day-to-day variability in sleep hours and sleep problems. An example of a sleep diary for insomnia patients is shown in Figure 1–4 in Chapter 1.

Actigraphy is an objective means of assessing rest-activity patterns by use of a motion-sensitive device worn on the nondominant wrist. Commercially available software provides descriptive statistics and graphical displays of rest-activity patterns. Validation studies have shown a strong correlation between actigraphy patterns and sleep as monitored by polysomnography (PSG), although actigraphy tends to overestimate the actual amount of sleep (Sadeh and Acebo 2002). Like the sleep diary, actigraphy can be useful for examining temporal patterns, variability, and responses to treatment.

Polysomnography, or a sleep study, is the gold standard for quantifying sleep and sleep disturbances. PSG is not routinely recommended for the evaluation of chronic insomnia (Sateia et al. 2000) because in most cases, PSG simply confirms the patient’s subjective report without indicating a cause for awakenings. However, PSG may be useful in specific situations involving patients with symptoms of sleep apnea (see Chapter 3), periodic limb movements (Chapter 5), or parasomnias (Chapter 6). Patients who have atypical complaints or a history of poor response to usually efficacious treatments may be candidates for polysomnography.

Cognitive-behavioral interventions for insomnia are described in this section and are listed in Table 3.

Stimulus control therapy is based on principles of operant learning (Bootzin and Nicassio 1978) and aims to reinforce associations between sleepiness, sleep, and the sleep environment. The patient is instructed to use the bed and bedroom only for sleep and sex, and to go to the bedroom only when feeling sleepy. If awake for long periods of time in bed (e.g., 20 minutes or longer), the individual is instructed to get out of bed and leave the bedroom until feeling sleepy again.

Sleep restriction therapy involves restricting the time spent in bed by setting strict bedtime and rising schedules that closely match the number of hours of actual sleep reported by the patient (Spielman et al. 1987). The aim is to increase sleep efficiency, or the ratio of total time spent asleep to total time spent in bed. Initially, sleep restriction is often associated with slight to moderate sleep deprivation, which increases sleepiness and enhances the ability to fall asleep and to maintain sleep. Sleep restriction may lead to increased daytime sleepiness, and patients should be cautioned about operating machinery or performing duties that require high levels of alertness.

Relaxation techniques aim to reduce muscular tension and cognitive arousal, which are incompatible with sleep (Jacobson 1974; Woolfolk and McNulty 1983). Several specific relaxation techniques have been evaluated for insomnia. Autogenic training involves focusing on physiological sensations, such as heat, to induce relaxation via modulation of the autonomic nervous system. Biofeedback relies on a similar principle, but the process of reducing autonomic arousal is aided by the use of visual or auditory feedback to monitor biological signals such as heart rate or skin temperature to indicate arousal levels to the patient, who can then voluntarily reduce physiological arousal through active concentration. Acupuncture is distinct from conventional relaxation and other behavioral techniques, but preliminary evidence suggests that it too may be efficacious for insomnia (see Sok et al. 2003 for meta-analysis).

Cognitive interventions involve the identification, challenge, and replacement of initial erroneous beliefs or fears regarding loss of sleep (Gross and Borkovec 1982). Cognitive distortions increase arousal and tension, further preventing sleep. Challenging the erroneous beliefs and fears can be done through education and discussion within a psychotherapeutic context. Replacing the irrational beliefs and fears requires behavioral changes to test the new beliefs and cognitive strategies to face resurgence of distorted thoughts about sleep. Cognitive therapy is most often combined with more behaviorally based interventions (Morin 1993). Paradoxical intention is another example of a cognitive technique used for insomnia. The rationale is that actively trying to maintain wakefulness will reduce performance anxiety related to falling asleep and thus will facilitate sleep onset. Data indicate that paradoxical intention may provide some benefit, but responses are highly variable (Morin et al. 1999).

Sleep hygiene refers to a set instructions aimed at reinforcing sleep-promoting behaviors and reducing the frequency of behaviors that may interfere with sleep (Hauri 1991). Exercising, relaxing evening routines, avoiding stimulants and naps, and limiting alcohol intake are examples of behaviors that may enhance sleep quality. Although sleep hygiene alone has limited efficacy to reduce insomnia (Lacks and Morin 1992), it is often combined with other behavioral interventions.

Behavioral interventions for insomnia significantly reduce sleep onset latency, reduce wake time after sleep onset, and improve total sleep time. Two meta-analyses support the efficacy of behavioral or cognitive-behavioral treatments of chronic insomnia (Morin et al. 1994; Murtagh and Greenwood 1995). Overall, a majority (70%–80%) of insomnia patients benefit from behavioral interventions, and improvements are maintained or enhanced at follow-up. Stimulus control, sleep restriction, and relaxation show the most robust effects among specific behavioral interventions. A third meta-analysis (Montgomery and Dennis 2003) investigated the efficacy of behavioral interventions in older adults. Overall, cognitive-behavioral therapy was effective at reducing sleep maintenance insomnia.

One additional meta-analysis (M.T. Smith et al. 2002b) compared the efficacy of cognitive-behavioral treatments and pharmacologic agents for insomnia. When compared with hypnotic medications, stimulus control and sleep restriction therapies were associated with similar improvements in wake time after sleep onset, number of awakenings, and sleep quality ratings. Behavioral interventions were associated with greater reductions in sleep onset latency, but hypnotics were associated with greater increases in total sleep time.

New behavioral treatment modalities for insomnia are currently being investigated to disseminate these treatments and reduce patient burden and costs. These new developments include inhome visits (Espie et al. 2001), telephone interventions, Internet-based interventions (Strom et al. 2004), and self-help material (Mimeault and Morin 1999). Initial results suggest that behavioral interventions can be effectively delivered via these nontraditional methods. In elderly patients, however, one study suggest that in-person, group behavioral interventions may be more effective than home-based relaxation treatment to reduce insomnia complaints (Rybarczyk et al. 2002).

Cognitive-behavioral interventions for insomnia are also effective at improving sleep quality in patients with comorbid medical and psychiatric illnesses (Lichstein et al. 2000) and specific conditions such as cancer (D’Ambrosio et al. 1999; Simeit et al. 2004) and chronic pain (Currie et al. 2000; Espie et al. 2001; Morin et al. 1989). Multifaceted psychological insomnia interventions have been shown to improve sleep quality in caregivers of dementia patients (McCurry and Ancoli-Israel 2003). A growing number of studies show that cognitive-behavioral treatments delivered in the primary and psychiatric care settings are also efficacious (Edinger and Sampson 2003; Espie et al. 2001).

Most of the behavioral treatments described above have been evaluated in clinical trials using approximately six 1-hour sessions delivered by trained therapists. However, these treatments share a few basic principles that can be used in a variety of clinical settings as a basic behavioral treatment package. These specific interventions are reasonable straightforward and rely on principles of the “two-process” model of sleep regulation (see Chapter 1, “Introduction”) and simple conditioning. These recommendations are summarized in Table 4.

Trazodone is a sedating antidepressant drug that is a weak but a specific inhibitor of the serotonin reuptake transporter, with minimal affinity for nor-epinephrine or dopamine reuptake. Trazodone also inhibits serotonin 5-HT_1A, 5-HT_1C, and 5-HT₂ receptors. Although it has no anticholinergic activity, it does have moderate antihistaminergic activity and is a weak antagonist of α₂ receptors and a somewhat more potent agonist of α₁ receptors (Golden et al. 2004). Trazodone is rapidly absorbed, with peak concentrations in 1–2 hours, and it has a half-life of approximately 5–9 hours. One of the metabolic products of trazodone is m-chlorophenylpiperazine (m-CPP), which itself possesses serotonergic activity and may be responsible for some of its side effects.

The sleep effects of trazodone have been described in small studies including control subjects and depressed patients, as well as one large study using only subjective outcomes in primary insomnia patients (reviewed in James and Mendelson 2004).

Sleep effects of trazodone and other nonbenzodiazepine medications used for insomnia are summarized in Table 6. Trazodone generally improves subjective sleep quality in healthy subjects and those with depression (Parrino et al. 1994; Saletu-Zyhlarz et al. 2002), but not consistently (van Bemmel et al. 1992). Trazodone has inconsistent effects in terms of improving PSG-defined sleep latency, wakefulness, total sleep time, and sleep efficiency (Saletu-Zyhlarz et al. 2002; van Bemmel et al. 1992). Unlike other sedating antidepressants, trazodone has little effect on REM sleep, with studies showing no significant change or a small decrease (Saletu-Zyhlarz et al. 2002; van Bemmel et al. 1992). Furthermore, trazodone is different from BzRA by virtue of increasing Stage 3/4 NREM sleep (Parrino et al. 1994; Saletu-Zyhlarz et al. 2002). In one large study comparing the effects of trazodone and zolpidem in patients with primary insomnia, trazodone improved subjective sleep latency, awakenings, wake time during sleep, and sleep time over 2 weeks of treatment, but these effects were significant only during the first week (Walsh et al. 1998).

Side effects of trazodone include sedation, orthostatic hypotension, lightheadedness, weakness, and the infrequent but serious risk of priapism.

Sedating tricyclic antidepressants (TCAs) such as doxepin, trimipramine, and amitriptyline have been used in low doses to treat insomnia. Doxepin and amitriptyline inhibit both serotonin and nor-epinephrine reuptake transporters, but trimipramine has few such effects. Sedating TCAs also antagonize peripheral α₁- and α₂-adrenergic receptors. Finally, they are potent cholinergic and histamine antagonists. TCAs are rapidly absorbed, with maximum concentrations in 2–6 hours. Their half-lives range from 15 to 30 hours (Nelson 2004).

Sedating tertiary TCAs have polysomnographic effects including reduced sleep latency, reduced wakefulness during sleep, and improved sleep efficiency (Feuillade et al. 1992; Roth et al. 1982; Shipley et al. 1985). Doxepin and amitriptyline also decrease REM sleep amount and increase eye movement activity during REM sleep (Roth et al. 1982; Shipley et al. 1985). Trimipramine, on the other hand, has very few effects on REM sleep. Sedating TCAs have little effect on Stage 3/4 sleep, although some reports show a small increase.

Small studies examining the effects of TCAs in primary insomnia have shown positive effects on both subjective measures and polysomnographic measures such as sleep deficiency (Hajak et al. 2001; Hohagen et al. 1994b).

Mirtazapine is a strong serotonin 5-HT₂ receptor antagonist, with strong antihistamine and α_1-antagonist properties as well. It also antagonizes α₂ noradrenergic receptors (Flores and Schatzberg 2004). Mirtazapine is rapidly absorbed but has a half-life of approximately 20–40 hours (Flores and Schatzberg 2004). Few data are available regarding its effects on sleep. In one study of healthy adults, mirtazapine decreased sleep latency, awakening, and light NREM sleep, and it increased Stage 3/4 sleep (Ruigt et al. 1990). A small study of depressed patients showed similar effects but no change in REM or Stage 3/4 sleep (Winokur et al. 2000). Curiously, mirtazapine has stronger sedative effects at lower doses than at higher doses—presumably because higher doses are associated with more noradrenergic effect.

Antihistamines are reversible antagonists of CNS histamine H₁ receptors. Histamine, localized in the tuberomammillary nuclei, promotes wakefulness. Sedating antihistamines penetrate the blood-brain barrier readily, whereas nonsedating antihistamines do not. In addition, sedating antihistamines have anticholinergic and serotonergic effects. Diphenhydramine is well absorbed and has a half-life of approximately 4–8 hours. Despite the widespread use of antihistamines, their effects on sleep are not well documented. Clinical trials using doses of 12.5–50 mg have shown subjective improvements in sleep latency, nocturnal awakenings, sleep duration, and sleep quality (Kudo and Kurihara 1990; Meuleman et al. 1987; Rickels et al. 1983). Polysomnographic studies have focused on daytime sedation rather than nighttime sleep effects. In this context, antihistamines produce more rapid sleep onset than placebo, although tolerance to these effects is often seen within a few days (Richardson et al. 2002). Significant side effects of antihistamines include impaired psychomotor performance, cognitive impairment, decreased appetite, and constipation. In older adults, the anticholinergic effects of these medications can be associated with urinary retention.

Melatonin is a hormone normally produced by the pineal gland and secreted exclusively at night. Melatonin production is suppressed by bright light and regulated by the circadian timing system. The most likely effect of melatonin on sleep-wake regulation is its interaction with specific receptors in the suprachiasmatic nucleus of the hypothalamus, the site of the circadian pacemaker. Exogenous melatonin is rapidly absorbed, with peak levels occurring in 20–30 minutes and a very short elimination half-life of 40–60 minutes (DeMuro et al. 2000). Sustained-release preparations can extend the duration of melatonin’s effects to several hours.

Melatonin may affect sleep as both a chronobiotic and a hypnotic. As a chronobiotic, melatonin in appropriately timed administration can shift circadian rhythms to an earlier or later phase in a manner that is exactly opposite to that of bright light (see Chapter 7, “Circadian Rhythm Sleep Disorders”). As a hypnotic, melatonin has inconsistent effects. Studies of subjective effects in healthy young adults given melatonin during the daytime, when activity of the suprachiasmatic nucleus is greatest, support its hypnotic efficacy (Cajochen et al. 2003). When given at night, melatonin can decrease subjective sleep latency, although other effects are less consistent. The efficacy of melatonin demonstrated by polysomnographic measures is less well documented. Some studies have shown increased sleep efficiency, but there is no clear indication of an effect on sleep duration (Olde Rikkert and Rigaud 2001; Sack et al. 1997). Melatonin is generally well tolerated, sedation being its only substantial side effect.

Valerian preparations are derived from roots of the plant genus Valeriana. The exact amount of various constituents in commercially available preparations is not well described. Likewise, the exact mechanism of valerian preparations is unknown. Some reports suggest GABA-like activity, as indicated by its sedative, anxiolytic, myorelaxant, and possible anticonvulsant effects. Some components of valerian extracts may inhibit GABA metabolism, and interactions with serotonin and adenosine receptors have also been discussed (Houghton 1999; Krystal and Ressler 2001). The pharmacokinetic properties of valerian are not well described because of its multiple constituents.

Sleep studies of valerian show subjective sedative effects, as well as decreased sleep latency and improved sleep quality. Polysomnographic studies have shown a reasonably consistent improvement in sleep latency, but other potential effects, including improved sleep efficiency, increased Stage 3/4 sleep, and reduced Stage 1 sleep, are inconsistent (Balderer and Borbély 1985; Donath et al. 2000). Side effects associated with valerian preparations are typically mild, including sedation, headache, and weakness.

Gabapentin was initially developed and marketed as an anticonvulsant drug, although it is widely used for treatment of conditions such as chronic pain and insomnia. Its exact mechanism of action is not well understood, although it may promote formation of GABA in the CNS or antagonize N-methyl-d-aspartate receptors (Rose and Kam 2002; Stahl 2000). More recent evidence indicates that gabapentin and a related drug, pregabalin, are ligands for the α₂δ subunit of voltage-sensitive calcium channels (VSCC). Binding of α₂δ ligands to activated VSCC reduces calcium influx and reduces neurotransmiter release (Stahl 2004). Gabapentin has relatively low bioavailability and an elimination half-life of 5–9 hours. Although gabapentin is subjectively sedating, few studies have been conducted on its effects on sleep PSG. A study of epilepsy patients showed decreased awakenings and Stage 1 sleep, as well as increased REM sleep and slow-wave sleep (Foldvary-Schaefer et al. 2002). A study in restless legs syndrome showed similar findings, with increased sleep continuity and increased Stage 3/4 sleep (Garcia-Borreguero et al. 2002). No studies have formally evaluated gabapentin in insomnia.

Tiagabine is an anticonvulsant drug with a well-defined mechanism of action. It inhibits the GABA transporter GAT-1, reducing GABA reuptake into presynaptic neurons. Tiagabine is rapidly absorbed and metabolized by the cytochrome P450 enzyme CYP3A4, with a half-life of approximately 8 hours. Efficacy studies have not been conducted in insomnia patients. One study in healthy adults demonstrated increased slow-wave sleep and improved sleep efficiency (Mathias et al. 2001). Tiagabine has been associated with new-onset seizures in a small number of patients. The exact frequency of this adverse effect is not known, but it can occur at low doses in patients with no seizure history.

Sedating antipsychotics have been used increasingly in the treatment of insomnia, particularly among patients with bipolar and psychotic disorders. The two most commonly used drugs are olanzapine and quetiapine. These drugs have a variety of receptor effects, including antagonist effects at 5-HT₂ receptors, dopamine antagonism, and anti-histamine effects. Olanzapine is slowly absorbed and has a half-life of 20–54 hours. Quetiapine has a more rapid onset of action, with a peak concentration in 1.5 hours and a half-life of only 6 hours (Baldessarini and Tarazi 2001; Stahl 2000).

A limited number of polysomnographic studies suggest the hypnotic efficacy of olanzapine, which decreases wakefulness, Stage 1 sleep, and REM sleep but increases Stage 3/4 NREM sleep and subjective sleep quality (Lindberg et al. 2002; Sharpley et al. 2000). No published studies are yet available regarding quetiapine as a hypnotic. Consideration should be given to the potentially serious neurological side effects of these medications, as well as their potential effects of weight gain and impaired glucose metabolism.

Despite the wide array of agents used for the treatment of insomnia, there are no published guidelines regarding the selection of specific drugs, the optimal duration of treatment, the choice of drugs for specific types of insomnia, or strategies to use in cases of partial or complete nonresponsiveness. In most cases of primary insomnia, a short-acting benzodiazepine receptor agonist is the drug of first choice. Specific patients may require a benzodiazepine receptor agonist with a longer or shorter duration of action, depending on their pattern of symptoms, sensitivity to medication, and rate of metabolism. For older adults with primary insomnia, smaller doses should be used initially. In cases of partial or no response, substitution or addition of a drug from a different class, usually a sedating antidepressant, is a reasonable choice. Third-line agents would include tiagabine and gabapentin. For patients with concurrent mood or anxiety disorders, BzRA or sedating antidepressants are again appropriate, but only when a primary therapy for the underlying condition, such as psychotherapy or a selective serotonin reuptake inhibitor, is also being used. For patients with known substance use disorders and significant respiratory disease, non-BzRA such as sedating antidepressants should be considered as first-line agents.

In the absence of specific guidelines regarding duration of use, initial treatment should be aimed at short-term use of 2–4 weeks. Gradual taper and discontinuation should then be attempted, by decreasing both the dose per administration and the number of doses per week. If longer-term treatment is needed, intermittent use may be preferred to avoid potential tolerance and adverse effects. In some chronic insomnia patients, long-term use of BzRA or other hypnotic drugs is appropriate. Regular monitoring to establish continued efficacy and tolerability of side effects is essential.

Behavioral treatments should be considered and utilized whenever possible, even in patients receiving pharmacotherapy. Long-term efficacy is better established for behavioral treatments than for pharmacologic treatments, and it seems prudent to avoid medications and their potential side effects when such efficacious other treatments are available.