European guideline for the diagnosis and treatment of insomnia

Aetiology and pathophysiology

This guideline primarily targets insomnia as an independent disorder, and not as an isolated symptom or a syndrome closely related to, or even directly caused by, other somatic or mental disorders. The type of insomnia addressed here closely resembles the concept of ‘psychophysiological’ insomnia as conceptualized decades ago (Hauri and Fisher, 1986). Given recent developments in the DSM-5 (2013) and ICSD-3 (2014), we will use the terms insomnia and insomnia disorder interchangeably throughout this guideline. Instead of the previously used dichotomy primary versus secondary insomnia, we will follow the concept of co-morbidity.

Several research groups have suggested aetiological and pathophysiological models of insomnia (Espie, 2002; Espie et al., 2006; Harvey, 2002; Levenson et al., 2015; Morin, 1993; Riemann et al., 2012, 2015). Most of these are explicitly or implicitly based on the so-called ‘3P’ model of insomnia by Spielman et al. (1987), which postulates that predisposing, precipitating and perpetuating factors are involved in the aetiology of insomnia. For example, genetic influences (Palagini et al., 2014) or personality characteristics like neuroticism or maladaptive perfectionism are seen as predisposing factors.

Acute stressors, for example, stress at work or interpersonal conflicts, usually trigger acute insomnia. Acute insomnia is very common and often a transient phenomenon, which is relieved after cessation of the stressor (Ellis et al., 2012a; Espie, 2002). Chronic stress exposure may also be seen as an underlying cause for chronic insomnia. In many cases, perpetuating factors have to come into play during the transition from acute to chronic insomnia. Spielman et al. (1987) suggested that maladaptive coping strategies are perpetuating factors, for example, prolonged time in bed or napping in order to catch up on lost sleep. While these behaviours may appear reasonable, they can reduce sleep pressure and may lead to chronic insomnia in the long run. Additionally, Espie et al. (2006) have emphasized the development of a maladaptive focus upon sleep in patients with insomnia, whereby sleep-related attentional biases and direct attempts to control sleep disturb the two-process bioregulation of sleep (Borbély, 1982; Borbély and Achermann, 1999), interfering with the expected default recovery to normal sleep, following episodic stress.

The hyperarousal model of insomnia postulates that increased arousal levels in the cognitive, emotional and physiological domains represent both predisposing ‘and’ perpetuating factors (Perlis et al., 1997; Riemann et al., 2010, 2015). Central to this model are results showing that patients with insomnia have increased power in fast electroencephalographic (EEG) frequencies during non-rapid eye movement sleep. This might also be reflected by an increased cyclic alternating pattern rate (Chouvarda et al., 2012). An increased frequency of microarousals during rapid eye movement (REM) sleep, which contributes to the perception of parts of REM sleep as wakefulness, has also been observed in patients with insomnia, relative to normal sleepers (Feige et al., 2013; Riemann et al., 2012). Neurobiologically, hyperarousal may be driven by a dominance of arousal-generating brain areas relative to sleep-inducing brain areas (Saper et al., 2005).

Cognitive models of insomnia stress the relevance of worry and rumination in the development and maintenance of insomnia (Harvey, 2002). Moreover, Baglioni et al. (2010) have emphasized that patients with insomnia have an increased emotional reactivity, which may also be of aetiological relevance.

Circadian factors are important in a subgroup of individuals, for example in those who undertake shift work or in blind patients, where desynchronization of the sleep–wake pattern and the circadian phase contributes to sleep initiation and sleep maintenance difficulties. This also applies to some cases of sleep-onset insomnia in adolescents/young adults, whereby a circadian phase delay may be the underlying factor, and to elderly patients with early awakening, whereby a phase advance may play a role (Abbott et al., 2016).

Definition of insomnia – diagnostic classification systems

In most European countries, use of the International Classification of Diseases (ICD-10, 1992) is mandatory for physicians and clinical psychologists/psychotherapists in order to get reimbursed through health insurance. For the diagnosis of insomnia, the diagnostic categories ‘Non-organic insomnia’ (F51.0) and ‘Disorders of initiating and maintaining sleep (insomnias)’ (G47.0) are relevant. The definition for non-organic insomnia is presented in Table 1.

The diagnosis of ‘non-organic insomnia’, according to ICD-10, is based solely on the subjective experience of afflicted individuals. No quantitative criteria for sleep-onset latency, sleep duration, or the frequency, or duration, of nocturnal awakenings is required. The term ‘non-organic insomnia’ refers to the fact that this sleep disorder does not have a specific recognizable somatic disorder at its core. However, the use of this term has been discussed critically over the last few years in light of documented neurobiological alterations in patients with insomnia.

DSM-5 (2013) has removed the distinction between primary and secondary insomnia. This distinction was aimed at differentiating ‘pure’ independent insomnia from ‘secondary’ insomnia, i.e. insomnia being related to or even hypothetically being caused by another somatic/mental disorder. Instead, the new umbrella category ‘insomnia disorder’ was introduced, which is also used in the third version of the International Classification of Sleep Disorders (ICSD-3; AASM, 2014). The decision to remove the distinction between primary and secondary insomnia was based on an NIH conference on insomnia in 2005 (National Institutes of Health, 2005), with the lack of evidence that treating the primary disorder would relieve insomnia accordingly, for example in cases of insomnia associated with depression, being the main reason for this change.

The definition of insomnia within the ICSD-3 largely follows that of the DSM-5. Table 2 shows the diagnostic criteria for insomnia according to the ICSD-3. In order to receive the diagnosis, there must be a disturbance of nocturnal sleep (criterion A) and related daytime impairment (criterion B). Furthermore, the sleep disorder has to occur at least 3 nights a week for a period of 3 months to be diagnosed as a clinically relevant disorder. If diagnostic criteria are fulfilled co-morbid with a mental or somatic disorder, both disorders are diagnosed.

As already mentioned, acute insomnia is very common and does not need a specific treatment in all cases (Ellis et al., 2012b). Chronic insomnia, instead, needs to be treated. The definitions for chronicity are, however, varying. ICD-10 requires a minimum duration of 1 month, whereas the ICSD-3 specifies 3 months. The authors of this guideline endorse the use of ICSD-3 for diagnostic purposes, and expect the development of ICD-11 will most likely follow the conceptual innovations of ICSD-3.

Diagnostic procedure

The recommended procedure for the diagnostic management of insomnia disorder, and its co-morbidities, is shown in Table 3.

A medical and psychiatric/psychological anamnesis is mandatory, and has to be tailored to the clinical picture of the patient and his/her symptomatology. With respect to the assessment of medical disorders, it needs to be borne in mind that some somatic causes of insomnia can be specifically treated, for example hyperthyroidism. However, even in the case of a clear somatic cause, many patients with insomnia develop a psychophysiological vicious cycle of insomnia, which includes rumination, worry about the consequences of poor sleep and increased physiological tension. These processes can be successfully treated in these co-morbid cases of insomnia.

Similar considerations should be made for substance use (e.g. alcohol/caffeine), which is important to evaluate in patients with insomnia. In particular, alcohol consumption is a common maladaptive self-treatment strategy in patients with insomnia, and can contribute to sleep-maintenance difficulties. Thus, alcohol consumption should be actively evaluated and considered during treatment planning. Furthermore, many medications can interfere with sleep. Therefore, the use, dosage and timing of medication should also be evaluated.

Mental disorders, especially depression, bipolar disorder or psychosis are also frequently accompanied by sleep-onset or sleep-maintenance difficulties or early morning awakening. A recent meta-analysis (Baglioni et al., 2016) showed that disturbances of sleep continuity (prolonged sleep latency, increased frequency of nocturnal awakenings, prolonged periods of wakefulness after sleep onset) occur transdiagnostically in almost all mental disorders. Patients with chronic insomnia often suffer from a co-morbid mental disorder, which they do not spontaneously report. This may be due to the fact that it is easier for some patients to talk about sleep than to talk about emotional distress. Thus, the presence of mental disorders should also be actively examined. Tiredness/fatigue also occurs in many mental or neurodegenerative disorders. Sleepiness (presumably experienced as a consequence of sleep loss) is usually not a symptom of insomnia per se, but may be due to an accumulation of sleep loss in these patients. As such, tiredness/fatigue and sleepiness should also be assessed.

Table 4 summarizes the major somatic and mental co-morbidities of insomnia.

The diagnostic procedure should also include a clinical interview consisting of a thorough sleep history (to assess sleep hygiene behaviour, sleep habits, sleep environment – including co-sleeping arrangements, work schedules, circadian factors and indications of other sleep disorders, e.g. restless legs syndrome, sleep apnea, circadian sleep–wake disorder, etc.). The consensus sleep diary (Carney et al., 2012), for 7–14 days, is also strongly recommended. Moreover, the Pittsburgh Sleep Quality Index (PSQI) can be used to assess subjective sleep during the previous month (Buysse et al., 1989). The PSQI, however, is not a specific instrument for diagnosing insomnia and should not be used for that purpose. The Insomnia Severity Index (ISI) has been developed to assess the severity of the disorder, and has also been shown to be a reliable and valid instrument to detect patients with insomnia (Bastien et al., 2001). In addition, the Bergen Insomnia Scale (Pallesen et al., 2008) and the Sleep Condition Indicator (Espie et al., 2012) have promising psychometric properties. An overview of the available scales for assessing sleep, and sleep disorders, is provided by Shahid et al. (2012). Furthermore, if indicated, actigraphy or polysomnography should be considered.

A meta-analysis of polysomnographic studies showed that patients with insomnia have a significantly reduced total sleep time, significantly prolonged sleep-onset latencies, and an increased number of nocturnal awakenings and amount of time awake during the night (Baglioni et al., 2014). Furthermore, slow-wave sleep and REM sleep percentages are reduced compared with good sleepers. However, the differences were not very pronounced, for example, total sleep time was reduced by about 25 min. In contrast, subjective total sleep time is reduced by about 2 h in patients with insomnia compared with good sleepers (Feige et al., 2008). This has led to the use of the terms ‘pseudoinsomnia’, ‘sleep state misperception’ or ‘paradoxical insomnia’. Many experts argue that polysomnography is not helpful in the assessment of insomnia because it does not correlate with the subjective perceptions of patients. However, we suggest polysomnography may have an additional diagnostic value ‘because’ it does not correlate with subjective measures and thus may deliver information not inherent in the subjective patient report. In addition, objective measures are mandatory to diagnose potential co-morbid disorders (e.g. PLMD = Periodic Leg Movement Disorder, sleep apnea), which are common. Sleep apnea may have a complex relationship with insomnia, thus being more than a mere co-morbidity (Sweetman et al., 2017). Several studies suggest that the polysomnographically determined microstructure of sleep is altered in insomnia, with increases in fast frequency power and in the number of microarousals. These phenomena are partly independent of the subjective experience of sleep (Riemann et al., 2015), and may become relevant for treatment decisions in the future (see ‘Outlook for the future’). Another recent discovery concerns differences between insomnia with, and without, an objective short sleep duration (Fernandez-Mendoza, 2017; Vgontzas et al., 2013). It is hypothesized that insomnia with polysomnographically documented short sleep duration has primarily biological roots and would thus respond better to biological treatments. If this hypothesis turns out to be true, polysomnography may become even more important in the diagnostic procedure for insomnia.

Epidemiology

Approximately 6% of the adults in industrialized countries suffer from chronic insomnia as a disorder (for overview, see Ohayon, 2002), with a clear-cut preponderance of females compared with males (Zhang and Wing, 2006) and an age-related increase in prevalence rates. More recent data (e.g. from Norway, the UK and Germany) indicate an increase in the prevalence of insomnia, to about 10% of the population, in recent years (Calem et al., 2012; Marschall et al., 2017; Pallesen et al., 2014). Moreover, it appears that the use of hypnotic agents has also increased significantly over a 10-year period (e.g. from 7% to 11% in Norway; Pallesen et al., 2001, 2014). Table 5 shows epidemiological data on the prevalence of insomnia, as a disorder, in 10 European countries (no such data are available for insomnia, on the disorder level, for the other European countries).

Table 5 demonstrates that the prevalence of insomnia varies largely from one European country to the other. This may be, in part, due to differences in methodological quality between studies. At present the prevalence of insomnia, as a disorder, in Europe, seems to vary from a minimum of 5.7% in Germany to a maximum of 19% in France. There is only one comprehensive epidemiological study (Van de Straat and Bracke, 2015) that employed a cross-national approach and studied sleep problems across 16 European countries, but only in older adults. This study did not specifically include questions to derive insomnia diagnoses, just a single item measure of sleep problems. This study showed that the prevalence rate for this type of sleep problem varies from a minimum of 16.6% in Denmark to a maximum of 31.2% in Poland. Our literature search and the study by van de Straat and Bracke suggest an urgent need for Pan-European cross-sectional studies to better understand the size of the problem in Europe, also with respect to co-morbidities.

Studies in general practice or medical specialty settings deliver substantially higher prevalence rates: data from general practice in Germany (Wittchen et al., 2001) indicate that one-fifth of the patients consulting a GP suffer from insomnia; whereas in Norway more than 50% of GP patients have insomnia (Bjorvatn et al., 2017).

In terms of the persistence of insomnia, there is very little information from Europe. However, Morin et al. (2009a) provided data on the natural course of insomnia in Canada, and showed that approximately 70% of the patients show persistent symptoms over the course of 1 year. In this study, 46% of those suffering from insomnia showed persistent symptoms over the course of 3 years.

The prevalence of hypnotic usage, i.e. usage of benzodiazepines (BZ) and benzodiazepine receptor agonists (BZRAs), varies largely from one European country to the other. A UK study reported an increase in hypnotic usage from 0.4% to 0.8% in the general population from 1993 to 2000 – the data remained stable from 2000 to 2007 (Calem et al., 2012). A German study described the prevalence of having taken a hypnotic, at least once, increased from 4.7% to 9.2% from 2009 to 2016 (Marschall et al., 2017). In general, it is not clear how many patients with insomnia in Europe regularly take hypnotics – further research is necessary to determine the exact scale of this issue.

Health risks

Several meta-analyses show that insomnia is a significant risk factor for cardiovascular diseases (Li et al., 2014; Meng et al., 2013; Sofi et al., 2014). Specifically, insomnia is a risk factor for arterial hypertension, myocardial infarction and chronic heart failure (Laugsand et al., 2011, 2014a; Palagini et al., 2013). In addition, Anothaisintawee et al. (2015) showed that insomnia is a risk factor for type 2 diabetes.

Besides insomnia itself, there is evidence suggesting that short sleep duration (sleeping less than 6 h on average) is a risk factor for obesity, type 2 diabetes, hypertension and cardiovascular diseases (Bayon et al., 2014; Buxton and Marcelli, 2010; Cappuccio et al., 2010; Faraut et al., 2012; Patel and Hu, 2008). Consequently, short sleep duration also increases mortality (Liu et al., 2017). However, the association between a short sleep duration and insomnia is not yet fully understood.

Neurological disorders are frequently co-morbid with insomnia (Mayer et al., 2011), and insomnia may play a role in the development of cognitive impairment (Yaffe et al., 2014). In addition, one cross-sectional study suggests a relationship between impaired sleep quality and cortical atrophy in older adults (Sexton et al., 2014). More recent work points to a general involvement of insomnia in the development of neurodegenerative disease, especially dementia (Osorio et al., 2011). Bassetti et al. (2015) stress the bi-directional nature of the relationship between insomnia and brain disorders.

Significant evidence has been gathered with respect to the relationship between insomnia and mental disorders (Riemann and Voderholzer, 2003). In a meta-analysis, Baglioni et al. (2011) showed that people with insomnia have an increased risk for the development of major depressive disorder (odds ratio 2.1), which may also lead to early retirement (Paunio et al., 2015). Similar relationships have been documented for insomnia complaints and suicide ideation, suicide attempts and completed suicides (Malik et al., 2014; Pigeon et al., 2012).

Large epidemiological studies have also demonstrated that insomnia is a risk factor for sick leave, an increased number of accidents in the work place (Laugsand et al., 2014b; Sivertsen et al., 2009a,b) and motor-vehicle accidents (Léger et al., 2014).

Costs of insomnia

The question of the direct and indirect costs of insomnia has been dealt with in several large, well-designed, studies (Daley et al., 2009; Léger and Bayon, 2010; Ozminkowski et al., 2007). Of particular relevance to Europe, the costs of several brain disorders in Europe were compared in 2010 (Gustavsson et al., 2011). This study ranked sleep disorders ninth among all neuropsychiatric disorders with respect to direct and indirect costs. An average total sum (costs) of €790 per year, per patient, was calculated. These overall costs were based on individual costs calculated against the estimated prevalence of insomnia, ranging from 6% to 12%, in the European population (Wittchen et al., 2011). Concerning so-called DALYs (disability-adjusted life-years), a figure of 10.3/10 000 individuals was given for females, and 8.4/10 000 individuals for males – ranking ninth among all neuropsychiatric disorders studied. According to WHO data, insomnia ranked 11th in the list of most important brain disorders with respect to global burden (Collins et al., 2011). Thus, it can be concluded that insomnia represents a high financial burden to European healthcare systems, either through direct costs, i.e. costs for medication or psychotherapeutic treatment, or indirect costs, for example, due to sick leave or early retirement.

Treatment of insomnia

In the presence of co-morbidities, clinical judgement should decide whether the insomnia or the co-morbid condition is treated first, or whether both are treated at the same time. Of note, the grading and recommendations of all the treatment options outlined in this section are collectively summarized in Table 15.