Sleep Disorders in Early Psychosis: Incidence, Severity, and Association With Clinical Symptoms

Sarah Reeve; Bryony Sheaves; Daniel Freeman


Schizophr Bull. 2019;45(2):287-295. 

In This Article


We assessed sleep disorders in 60 patients with early nonaffective psychosis. Strikingly, four-fifths of the patients were found to have a comorbid sleep disorder. These were not minor sleep issues—the majority of disorders were rated as severe in their chronicity, frequency, and distress or impairment. In over two-thirds of cases, no treatment for the sleep disorder was reported; in most of these instances the sleep problem had not been discussed with the care team. Even when treatment had been received, it was rarely the recommended treatment for the sleep disorder. The patients with sleep disorders had more severe paranoia, hallucinations, cognitive disorganization, depression, and anxiety. Quality of life was lower in patients with comorbid psychosis and sleep disorder. If the results generalize to the wider population of patients with psychosis, then clinical services need to give a greater priority to sleep than they have to date.

Diagnostic systems such as DSM-5 recommend that sleep problems should be assessed and treated irrespective of other psychiatric comorbidities.[19] This does not appear to be happening in psychosis services. When sleep is assessed in individuals with psychosis, many clinicians report only doing so informally[5]—this may mean that the depth, complexity, and impact of sleep issues are not revealed. The use of brief screening measures such as the Insomnia Severity Index[33] or Disturbing Dreams and Nightmares Severity Index[34] may be of use in facilitating these assessments in clinical practice. Expanding education on sleep disorders within training pathways may also help to address the neglect of these issues.

Provision of treatment for sleep problems is another issue raised by these findings. Where treatment had been given it was often a nonrecommended treatment, eg, antipsychotic medication or sleep hygiene advice for insomnia, neither of which have been shown to be effective.[35,36] The primary recommended treatment for insomnia (CBT) has been shown to be effective in this patient group,[4] and adaptations to the treatment for individuals with psychosis are available,[37] but none of our study participants received this treatment from mental health services. Hypnotic medication was the only recommended sleep intervention received from mental health services. Insomnia treatment guidelines do include hypnotic medication, but they state that it should be reserved for acute phases of insomnia only,[23] due to side effects (including cognitive impairment and increased risk of vehicular accidents) and limited effectiveness once tolerance develops.[38]

Patient factors have previously been cited as a barrier to clinicians referring patients for treatment.[5] However, it has been shown that patients are very much able to engage with treatment for insomnia. This includes outpatients with current delusions and hallucinations[12] and inpatients who were acutely unwell.[39] Many patients highly value the sleep treatment they receive,[4] and our clinical experience is that treating sleep problems results in higher engagement with other clinical interventions. The other issue cited by clinicians is availability of treatment,[5] which does provide a greater barrier to access even when the treatments have been highly manualized.

Insomnia was the most prevalent sleep disorder but nightmare disorder was also found to be common, often severe, yet predominantly not discussed with the care team. A recent student survey found that only 11% of individuals with severe nightmares had discussed them with a health care provider, and those who believed nightmare disorders were untreatable (67.3%) were less likely to report them.[40] It is possible that a lack of awareness regarding the treatability of nightmares might also affect reporting by individuals with psychosis. Although effective treatments for nightmare disorders do exist,[24,41] they have not yet been tested in this clinical group, although such a trial is underway.[42]


The first key issue is the representativeness of the participant group. It is simply unknown how representative the participants are of the wider population of patients with nonaffective psychosis. The study group could, eg, be biased as individuals with sleep problems may have been more motivated to take part. It is worth noting that the insomnia prevalence in this study (50%) is higher than has been reported a larger study (29%, sample n = 623[43]). The current study was advertised as recruiting "good and bad sleepers" to attempt to minimize this bias, and indeed there were a number of individuals with no sleep issues. The demographic data also indicate that our participant group is reasonably representative of those seen within early intervention services in age, gender, and local ethnicity.[44] Given the importance of this issue to patients, and the relative ease of sleep assessment, larger studies should attempt to assess sleep disorders to test if the results are representative of the sleep disorders in the population.

Our screening of several sleep issues is limited due to a lack of polysomnography (PSG). Sleep apnea is a key disorder which we were not able to screen for using the DISP items, and were not able to diagnose due to lack of PSG. Future studies should either use PSG or apnea specific screening measures (eg, STOP-BANG[45]) to further investigate the high prevalence of apnea symptoms reported in this study and others.[46] Polysomnography is also required to provide a confirmed diagnosis for several sleep disorders in this study (eg, PLMS, sleep walking, hypersomnia). Even so, the structured interview and sleep recording used in the current study allowed diagnosis or positive screens to be provided for the majority of sleep disorders, and is more likely than PSG to be available in typical mental health services.

However, it is worth noting that PSG might provide a differential diagnosis in cases where a symptom (eg, excessive sleepiness) might result from multiple disorders. Further work should investigate differential diagnosis of sleep issues in this participant group, where there is clear comorbidity, to provide further clinical guidance. Another issue is that for several sleep disorder diagnoses (eg, hypersomnia, night terrors, sleep walking) the symptoms need to "not be better explained by another sleep disorder, medical or neurological disorder, mental disorder, medication use, or substance use disorder."[22] This raises the question of which (if any) of these diagnoses might be excluded by having a psychotic disorder diagnosis or being in receipt of antipsychotic medication.

The frequency of sleep-related hallucinations is high in our study. In part, this reflects a phenomenological overlap between hallucinations in the context of psychosis and sleep-related hypnagogic or hypnopompic hallucinations. Some participants described the hypnagogic hallucinations as being different in character (eg, more visual) than their daytime hallucinations, or only had sleep-related hallucinations (with no daytime hallucinations)—but in many cases there was no clear divide. Issues in discriminating between these phenomena have been noted in the narcolepsy literature,[47] but further research is clearly needed to investigate the overlap in individuals with psychosis.

Medication use is potentially an important factor in considering sleep in individuals with psychosis. Antipsychotic medication may improve sleep—however daytime sedation is common, and antipsychotics may themselves increase risk of certain sleep issues (eg, sleep apnea[29]). The picture is complicated by antidepressants, which have their own interaction with sleep[48] and are often prescribed alongside or instead of antipsychotics. Our secondary medication analyses indicate that antipsychotic medication dosage did not differ between those with or without an insomnia or nightmare disorder diagnosis, indicating medication is not a significant factor in sleep disorders in our study group, but further research is needed.

Lastly, this study is cross-sectional, therefore we are not able to investigate the direction of effect between sleep disorders and psychotic experiences, mood, or wellbeing. Longitudinal studies to investigate the temporal relationship between sleep disorders and psychotic experiences in clinical populations would therefore be valuable. Nevertheless, there is increasing evidence pointing to sleep disruption as a contributory causal factor in psychosis.[1,17,18] This means that there is a possibility that by improving sleep it may be possible to improve psychosis, representing an exciting new treatment target.[18] However, independent of any relationship with psychotic experiences, our view is that the assessment and treatment of sleep disorders among those with psychosis merits greater clinical attention.