From the Editor

If students sleep better, are they less likely to have mental health problems like paranoia?

In this week’s Reading, we look at a new study from The Lancet Psychiatry. In this single-blind, randomized controlled trial, Oxford professor Daniel Freeman et al. consider students from 26 universities with insomnia, assigning them CBT (offered over the internet) or the usual care.

Spoiler alert: the students with CBT did better.

Sleep: good for babies, teddy bears, and students

In this Reading, we review that paper and consider the broader implications.



Sleep and Mental Health

The effects of improving sleep on mental health (OASIS): a randomised controlled trial with mediation analysis

Daniel Freeman, Bryony Sheaves, Guy M Goodwin, Ly-Mee Yu, Alecia Nickless, Paul J Harrison, Richard Emsley, Annemarie I Luik, et al.

The Lancet Psychiatry, October 2017  (Open Access)


Sleep problems are a common occurrence in patients with mental health disorders. The traditional view is that disrupted sleep is a symptom, consequence, or non-specific epiphenomenon of the disorders; the clinical result is that the treatment of sleep problems is given a low priority. An alternative perspective is that disturbed sleep is a contributory causal factor in the occurrence of many mental health disorders. An escalating cycle then emerges between the distress of the mental health symptoms, effect on daytime functioning, and struggles in gaining restorative sleep. From this alternative perspective, the treatment of sleep problems attains a higher clinical importance. We are particularly interested in the putative causal association between disturbed sleep and psychotic experiences. The interventionist–causal model approach to establishment of a causal association is to manipulate the hypothesised mechanistic variable and observe the effect on the outcome of interest; if a causal association exists then the outcome variable should alter. The effects of the manipulation can then be substantiated further by use of mediation analysis. In the present study, we aimed to improve sleep in individuals with insomnia to determine the effect on psychotic experiences. This approach therefore informs both theoretical understanding and clinical practice.

The most common form of sleep disruption is insomnia, comprising sustained difficulties in initiating or staying asleep, or both, which cause problems during the day. The association of insomnia with psychotic experiences in the general population has been established. There are multiple, independent, psychotic experiences. Each psychotic experience exists on a spectrum of severity in the general population with differing heritability and differing strength of association with insomnia. Paranoia and hallucinations have the strongest links with insomnia. However, the effect of altering the amount of sleep disruption—eg, by targeted sleep treatment—on these psychotic experiences remains to be established. Clinical guidelines recommend the use of cognitive behavioural therapy (CBT) as the first-line treatment for insomnia. Digital forms of CBT for insomnia that require no therapist to be present have been shown to be efficacious as well. In patients with current delusions and hallucinations, results of our pilot randomised controlled trial have shown that insomnia can be substantially reduced with CBT, but the trial was underpowered to establish with sufficient precision the consequences for psychotic experiences. Therefore, we undertook a clinical trial that was large enough to definitively test the causal association between insomnia and self-reported psychotic experiences.

Daniel Freeman

So begins a paper by Freeman et al.

Here’s what they did:

  • In a single-blind, randomized trial, students from 26 universities across the UK were eligible.
  • Inclusion criteria: they were attending university; they had a positive score for insomnia; they were 18 or older. Exclusion criteria: none.
  • The intervention used Sleepio, a CBT program for insomnia, delivered through the internet, consisting of six sessions of roughly 20 minutes; there were also tools (sleep diaries and relaxation audios) accessible through the web. Sleepio has educational, cognitive, and behavioural components. Behavioural techniques, for example, included sleep restriction, stimulus control, and relaxation.
  • Assessments were done at weeks 0 (baseline), 3, 10 (end of therapy), and 22. The primary outcomes: insomnia, paranoia, and hallucinatory experiences. Various scales were used (SCI-8 for insomnia, GPTS for paranoia, and a Specific Psychotic Experiences Questionnaire for hallucinations).
  • Statistical analyses were done, with a particular focus on symptoms of paranoia. Secondary outcomes considered nightmares and other symptoms.

Here’s what they found:

  • Between March 5, 2015, and Feb 17, 2016, 3,755 participants were randomized to receive digital CBT for insomnia (n=1,891) or the usual practice (n=1,864).
  • Demographically: participants were “predominately female, studying for their first university degree, and two-thirds were of white British ethnicity…. Around a fifth of the participants were in contact with mental health services.”
  • “Regarding the primary measures, the sleep treatment was associated with significant reductions, at all timepoints, in insomnia, paranoia, and hallucinations compared with the control group (all p<0·0001).”
  • “The reduction in insomnia after treatment was large, while the reduction in psychotic experiences was small… After treatment, 454 (62%) of 733 individuals in the treatment group and 326 (29%) of 1142 individuals in the control group scored outside the clinical cutoff for insomnia used for trial entry.” To speak more technically: “Compared with usual practice, the sleep intervention at 10 weeks reduced insomnia (adjusted difference 4·78, 95% CI 4·29 to 5·26, Cohen’s d=1·11; p<0·0001), paranoia (–2·22, –2·98 to –1·45, Cohen’s d=0·19; p<0·0001), and hallucinations (–1·58, –1·98 to –1·18, Cohen’s d=0·24; p<0·0001).”
  • “The sleep treatment also led to improvements in depression, and improvements in anxiety, prodromal symptoms, nightmares, psychological wellbeing, and functioning, and all these improvements were maintained over time…”
  • “Treatment uptake was relatively low. In the intervention group, 1302 participants (69%) logged on for at least one treatment session, 953 (50%) accessed at least two sessions, 672 (36%) accessed at least three sessions, 497 (26%) accessed at least four sessions, 390 (21%) accessed at least five sessions, and 331 (18%) accessed six sessions…”

They go on to note:

Students randomly assigned to the sleep intervention showed small, sustained reductions in paranoia and hallucinations, suggesting that disrupted sleep has a contributory causal role in the occurrence of these psychotic experiences in a specific population of young adults… Insomnia might not be the largest cause of psychotic experiences but it is not an epiphenomenon. Hence, this study adds to our understanding of the causes of psychotic experiences and might indicate a promising route into the early treatment of some psychotic problems.

A few thoughts:

  1. This is a good paper.
  1. This is an interesting paper on several levels.

First, it’s a good example of prevention in mental health care. There are limitations here – much of the data is gathered from self-reported surveys and the results aren’t necessarily generalizable to an adult population. Still, the authors sought to reduce symptoms of paranoia and they did (as well as symptoms of mood, anxiety, nightmares, etc.).

Second, they opt for a digital solution. Instead of training dozens of therapists, they use an approach that is far less labour intensive. Economically, that makes sense and allowed the authors to put together “the largest randomised controlled trial of a psychological intervention for a mental health problem…” The paper runs with a Comment from the Finnish Institute of Occupational Health’s Tea Lallukka and Børge Sivertsen. They make a similar argument:

A consensus now exists that psychological and behavioural interventions, such as cognitive behavioural therapy (CBT) for insomnia, should be the first-line treatment option, and are preferred over pharmacological treatments in the management of chronic insomnia. Still, use is limited due to high cost, low availability, and unequal access. Particularly since 2014, internet-based CBT for insomnia has become available and has been successful in reducing insomnia symptoms, and evidence is emerging that CBT for insomnia might improve mental health problems…

Tea Lallukka

You can find the piece here:

(Note: it’s open access.)

  1. As is the case with some digital interventions, attrition rates can be problematic. (Lallukka and Sivertsen note that this study’s attrition rate wasn’t larger than smaller studies with traditional CBT).
  1. What to make of the author’s view that paranoia is connected with sleep? They did complex analyses and concluded: “improvements in sleep accounted for almost 60% of the change in paranoia after treatment…” We are left wondering, of course, if there are long-term implications – what if the study period had been 220 weeks, not 22 weeks? Still, the authors have a robust result.
  1. Happy #DigitalHealthWeek.


Reading of the Week. Every week I pick articles and papers from the world of Psychiatry.