Effects of Improving Sleep on Mental Health

A Critical Appraisal of: The Effects of Improving Sleep on Mental Health (OASIS): A Randomised Controlled Trial with Mediation Analysis

Lay Abstract

Mental health is becoming an increasingly discussed and researched subject and for good reason. Poor mental health is common and disabling, affecting 16.7 million people in the UK at any one time and accounting for 15 per cent of all the disability due to disease. It is estimated to cost at least £77 billion annually in England alone and severe forms of mental illness are associated with social exclusion and deprivation [1]. Therefore, it is vital that more research is carried out on the diagnosis and treatment of mental health disorders.

Psychosis is a mental health problem that causes people to perceive or interpret things differently from those around them. Paranoia is a persistent, irrational feeling that people are ‘out to get you’ or of constantly being watched or listened to, whilst hallucinations are where a person hears, sees and, in some cases, feels, smells or tastes things that aren’t there. As many as 16 million adults in the UK suffer from sleepless nights [26] and whilst previously it was assumed that insomnia – a sleep disorder characterised by difficulty falling and/or staying asleep – was a symptom of mental illnesses, recent research has attempted to prove that there is a causal relationship between insomnia and psychotic experiences. This study, by means of recruiting a large student population who suffer from insomnia and treating half of them using a type of behavioural therapy, seeks to further establish this causal relationship and demonstrate that treating the insomnia will in turn treat the psychotic experiences. The results show that the treatment leads to a large reduction in insomnia and a small reduction in psychotic experiences.

Background Information and Rationale

The paper I have chosen, “The Effects of Improving Sleep on Mental Health (OASIS): A Randomised Controlled Trial with Mediation Analysis” [2] aims to find a link between insomnia and different mental health conditions. Specifically, the researchers are looking at the link between insomnia and psychotic disorders: namely paranoia and hallucinations. The researchers point out that while the traditional view is that poor sleep is a symptom of mental health conditions, there is an alternative view that disrupted sleep is actually a contributory causal factor in a wide variety of mental health illnesses [3]. The lead researcher on this paper, Professor Daniel Freeman, a British psychologist, paranoia expert, professor of clinical psychology and research professor, has conducted previous studies in which he sought to establish the link between insomnia and psychotic experiences [4][5]. In one of these studies, the research team conducted a systematic review of 66 research papers and found that there was strong evidence supporting the co-occurrence of sleep dysfunction and psychotic experiences, especially paranoia [4]. However, the conclusion from this study was that ‘studies are needed that: determine the types of sleep dysfunction linked to individual psychotic experiences; establish a causal connection between sleep and psychotic experiences’ [4].

In future studies, Professor Freeman and his associates managed to establish a causal connection between insomnia and paranoia [6][7]. Therefore, the purpose of this study was to find out whether improving sleep would reduce the occurrence of mental health illnesses, primarily psychotic experiences. The researchers explained that an extensive search on PubMed revealed that there were only 2 randomised controlled trials examining the effects of improving sleep on reducing psychotic experiences, the larger of which was a study with 50 patients who had schizophrenia or related disorders and was actually conducted by the research team themselves [8]. Therefore, it was clear that a large study was required since the previous trials were underpowered to accurately convey a link between improving sleep and reducing psychotic experiences. The research consistently mentioned that they believe this is the largest randomised controlled trial of its kind and felt it necessary in informing both theoretical and clinical practice [2].

Approaches to the Question and Results

Methods

The study was a single-blind, randomised controlled trial of digital CBT (cognitive behavioural therapy) versus treatment as usual; meaning that if selected individuals were already receiving treatment for insomnia, then they should carry on using it as usual. Cognitive behavioural therapy is a form of psychotherapy that treats problems and improves morale by modifying dysfunctional emotions, behaviours, and thoughts. Unlike traditional Freudian psychoanalysis, which looks at traumatic experiences in childhood to identify the root causes of conflict, CBT focuses on solutions, encouraging patients to challenge distorted ways of thinking and change destructive patterns of behaviour [9]. Following a recent study by the American College of Physicians, clinical guidelines for the treatment of insomnia recommend CBT as the first line of treatment and it is for this reason that the researchers have chosen to use CBT in this study [10]. Furthermore, the researchers identified a number of studies indicating that digital CBT was similarly effective in treating insomnia [11] [12] [13]. The reason that the researchers opted for digital CBT was that since it does not require the participants to physically come in and see a therapist, they would be able to enrol a much larger group into the study and be more likely to have a powerful enough trial to prove their hypothesis.

The primary outcome of the trial was to find out whether CBT for insomnia, compared with a usual practice control group, reduced insomnia and reduced paranoia and hallucinations by the end of treatment, and whether the changes in insomnia mediated the changes in psychotic experiences [2]. This is known as mediation analysis as the researchers are carrying out a study treating insomnia but seeing whether this change mediates a reduction in paranoia and hallucinations. However, in addition to these primary outcomes, the secondary outcomes were to investigate whether digital CBT for insomnia, compared with usual practice, reduced depression, anxiety, nightmares, and mania; improved psychological wellbeing; and led to the occurrence of fewer mental health disorders [2].

The study involved 3755 participants, 1891 of which were allocated to receive digital CBT and 1864 of which were allocated to usual practice. There were no exclusion criteria for the study but there were inclusion criteria. Participants were eligible if they were attending university; had a positive screen for insomnia, as indicated by a score of 16 or lower on the Sleep Condition Indicator (SCI); and were 18 years or older [1]. The principal method of recruitment was sending an email within universities that contained a link to the web-based screening. When an email was not possible, recruitment was via advertisement on websites and displaying posters. A total of 26 universities in the UK took part in the study, which the researchers claim ensured a range of geographical regions in order to get a varied population sample. Whilst the ethnicity of both groups was predominantly white, the percentages of different ethnicities was very similar in both groups. Further to this, age, sex, mean total UCAS points and course level were also very similar in both groups. So, even though the researchers used the technique of simple randomisation using an automated online system, the groups ended up well balanced, avoiding the largest problem of using simple randomisation which is groups which are not balanced in important co-varieties and so there is no allocation bias [14]. However, the researchers themselves mention that a limitation to the recruitment method was that students were self-selecting in responding to the invitation to take part in the study and no doubt this will have impacted the representativeness of the sample. Further to this there is another limitation in recruitment due to the fact that there were no exclusion criteria. This means that the possibilities of confounders were not ruled out as there are many things that may affect sleep such as whether the participant is a recreational drug user or drinks a lot of alcohol; particularly considering this was a student population [19].

The digital CBT programme used by the researchers was called Sleepio and was chosen as a couple of studies showed it to be effective in reducing insomnia [12] [15]. It is provided in six sessions, lasting an average of 20 minutes each and the full programme is accessible via any web browser and via smartphone. The treatment includes behavioural, cognitive, and educational components, which are all discussed at further length in the paper and are all presented by a virtual therapist on-screen [2]. The participants had their progress assessed at weeks 3 (only primary outcomes), 10 (end of treatment) and 22. Participants received an email prompt to complete the assessments online and if they did not complete the assessment, then they received up to two email reminders two days apart in an effort to get them to complete the assessment. As explained previously, the main reason that the researchers opted to use digital CBT was so that they could have a very large number of participants, without compromising the quality of the CBT provided. However, I believe that the decision to use digital CBT was one of the biggest flaws of the study. The reason that I believe this is that the number of participants who completed all six sessions of the CBT was just 18% and only 69% of participants even logged on at all [2]. The fact that there is no physical doctor/therapist to actually go in and see means that there is less motivation and less of a reason for the participants to actually participate. They may have just signed up out of interest and never really committed to completing the study. It is a lot of trust on the researchers’ part to leave it in the hands of nearly 2000 people as to whether they log on and actively use the treatment. Furthermore, this trust extends to the fact that the participants may not have been using the digital CBT correctly/effectively and whether or not they were putting effort into accurately updating their sleep diaries and answering questions truthfully about their progress. So, whilst the intention of using digital CBT was a good one, in the end I believe it had a detrimental effect on the analysis given the issues highlighted above.

Results

This study was carried out on an intention to treat analysis. This means that the results include all the participants, regardless of their adherence to the treatment. The benefit of this is that it avoids over-optimistic estimates of the efficacy of an intervention that would occur should those who do not comply to the treatment are included in the results. This reflects the fact that non-adherence and protocol deviations are likely to occur in clinical practice [16]. The researchers state that based on the standard deviations observed from a previous study for the Green et al. Paranoid Thoughts Scale [17] for which the standard deviation was 10·4, a total sample size of 2614 participants (i.e. 1307 per group) would provide 90% power to detect a small effect size in paranoia, with a standardised mean difference of 0.15, while in turn accounting for a high amount of expected attrition of 40%. They also mention that in a study amendment the sample size was increased due to a larger than expected dropout. It is not clear whether they restarted the study or if the new participants were done at a separate time.

The researchers used a linear mixed effects regression model to measure continuous outcomes and a logistic mixed effects model to measure binary outcomes i.e. outcomes where there are only two possible outcomes. Linear regression is used to study the linear relationship between a dependent variable and one or more independent variables [27] and so is ideal for explaining the relationship between sleep and the primary outcomes. A mixed effects model is a model containing both fixed and random effects. According to a study from 2004, mixed models are the recommended model for clinical trials where data is collected at repeated timepoints i.e. weeks 3, 10 and 22. Furthermore, this method has the advantage of dealing with missing values which was very ideal for this trial with researchers expecting attrition of 40%.

The results show that for the primary measures, the sleep treatment resulted in significant reductions, at all three measurements (weeks 3,10,22), in insomnia, paranoia, and hallucinations compared with the control group. This can be seen in table 2 of the results with all the p values being p<0.0001, meaning that they are statistically significant at the 95% confidence interval. However, whilst the reduction in insomnia was significant for the treatment group, the reduction in paranoia and hallucinations was much smaller; although still present. After treatment, 62% individuals in the treatment group and 29% of 1142 individuals in the control group scored outside the clinical cut off for insomnia [18] used for trial entry. The large improvement in insomnia is can be seen in table 4.  In terms of the secondary outcomes, tables 4 and show that the sleep treatment also led to improvements in depression, anxiety, prodromal symptoms, nightmares, psychological wellbeing, and functioning. Furthermore, table 5 shows that the participants who received the sleep treatment were less likely to meet criteria over the course of the trial for a depressive episode, anxiety disorder, or ultra-high risk of psychosis. Finally, the sleep treatment led to a small, sustained increase in symptoms of mania, although the reason for this is not known.

In terms of the mediation analysis, the researchers looked at the extent of mediation of the week 3 and week 10 insomnia scores on the week 10 paranoia and hallucination outcomes. The results provided in table 3 showed that change in sleep over 3 weeks accounted for 30% of the treatment effect on paranoia at 10 weeks, whilst change in sleep over 10 weeks accounted for 58% of the treatment effect on paranoia. When looking at hallucinations, change in sleep over 3 weeks accounted for 21% of the treatment effect on hallucinations at 10 weeks, whilst change in sleep over 10 weeks accounted for 39% of the traatment effect on hallucinations. Therefore, it can be seen clearly that changes in sleep mediate significant reductions in psychotic experiences.

Figure 1 – Table 2 showing the primary outcome results from the study [2]

The data was presented very clearly throughout a number of different tables, making it easy to draw conclusions from the presentation. This can be seen below with Table 2 taken from the paper.

Likely Impact of Research Outcome and Future Work

This paper was published in the Lancet journal in September 2017 and has already been cited 55 times, showing the impact that it is having on the medical world. These include studies examining the treatment of insomnia using CBT as an adjunctive to antipsychotics in treating schizophrenia [20], the link between prenatal stress and the development of psychopathology [21] and the contribution of sleep to neurodegeneration in mood disorders [22]. Therefore, it is clear that the work of Professor Freeman and his associates is opening up the research done in looking at the relationship between sleep and mental illnesses. Whilst the results are significant in what they show, the question of whether there is an immediate impact of the results lies in whether the data is transferrable to a wider population. The researchers themselves admit that the answer to this question is unknown even in terms of the wider student population [2]. However, they state that since they used a modified treatment developed for adults and achieved similar results in a different study with adults [23], they are confident that the results will be transferrable. The main impact of this research however is demonstrating that there is a need for reconsideration in the clinical services of the priority that quality of sleep is given in practice.

In terms of future work, the next logical step would be to carry out this study on very different groups of people. The study can be repeated in adult populations of different age groups and ethnicities, perhaps by conducting research abroad. Also, as this study was predominantly female, it would be logical to ensure the study had a large proportion of males, especially given the fact that a large proportion of those who dropped out of the study were male [2]. Furthermore, Professor Freeman and his associates have pointed out several avenues of future research. Firstly, there is a need to determine the mechanisms linking insomnia to psychotic experiences which will shed further light on the causes of psychosis and potentially enable treatment improvement. Also, studies can be done to look at the effects of improving sleep for patients attending clinical services with ultra-high risk of psychosis, or established clinical psychotic experiences, or at the early stages of relapse. Indeed, since this study, Professor Freeman has already undertaken 2 further studies, one looking at stabilising sleep for patients admitted at acute crisis to a psychiatric hospital [24] and one looking at treating sleep problems in young people at ultra-high risk of psychosis [25].

Conclusion

So, to conclude. I believe that this paper had good intentions and built well on previous work carried out by Professor Freeman and his associates. However, I think that the study was undermined by flaws in its design and execution that take away from what would otherwise be a very good paper – specifically in terms of the recruitment method and its implications and in terms of the very high drop-out which may have skewed the results. However, it is laudable that the researchers acknowledge the many limitations of this study and will no doubt bear them in mind when carrying out future research. Nonetheless, the statistical analysis reveals that the researchers achieved their aim of demonstrating a causal relationship between insomnia and psychotic experiences and more importantly showing, via a mediation analysis, that improving sleep leads to a reduction in psychotic experiences.

References

1. Medical Research Council. [Internet]. Mrc.ukri.org. 2010 [cited 13 November 2018]. Available from: https://mrc.ukri.org/documents/pdf/mrc-mental-health-research-report-2010/
2. Freeman D, Sheaves B, Goodwin G, Yu L, Nickless A, Harrison P et al. The effects of improving sleep on mental health (OASIS): a randomised controlled trial with mediation analysis. The Lancet Psychiatry. 2017;4(10):749-758.
3. Harvey A, Murray G, Chandler R, Soehner A. Sleep disturbance as transdiagnostic: Consideration of neurobiological mechanisms. Clinical Psychology Review. 2011;31(2):225-235.
4. Reeve S, Sheaves B, Freeman D. The role of sleep dysfunction in the occurrence of delusions and hallucinations: A systematic review. Clinical Psychology Review. 2015;42:96-115.
5. Freeman D, Pugh K, Vorontsova N, Southgate L. Insomnia and paranoia. Schizophrenia Research. 2009;108(1-3):280-284.
6. Sheaves B, Porcheret K, Tsanas A, Espie C, Foster R, Freeman D et al. Insomnia, Nightmares, and Chronotype as Markers of Risk for Severe Mental Illness: Results from a Student Population. Sleep. 2016;39(1):173-181.
7. Freeman D, Brugha T, Meltzer H, Jenkins R, Stahl D, Bebbington P. Persecutory ideation and insomnia: Findings from the second British National Survey Of Psychiatric Morbidity. Journal of Psychiatric Research. 2010;44(15):1021-1026.
8. Freeman D, Waite F, Startup H, Myers E, Lister R, McInerney J et al. Efficacy of cognitive behavioural therapy for sleep improvement in patients with persistent delusions and hallucinations (BEST): a prospective, assessor-blind, randomised controlled pilot trial. The Lancet Psychiatry. 2015;2(11):975-983.
9. Unknown. Cognitive Behavioral Therapy | Psychology Today UK [Internet]. Psychology Today. 2018 [cited 13 November 2018]. Available from: https://www.psychologytoday.com/gb/basics/cognitive-behavioral-therapy
10. Qaseem A, Kansagara D, Forciea M, Cooke M, Denberg T. Management of Chronic Insomnia Disorder in Adults: A Clinical Practice Guideline From the American College of Physicians. Annals of Internal Medicine. 2016;165(2):125.
11. Ritterband L, Thorndike F, Ingersoll K, Lord H, Gonder-Frederick L, Frederick C et al. Effect of a Web-Based Cognitive Behavior Therapy for Insomnia Intervention With 1-Year Follow-up. JAMA Psychiatry. 2017;74(1):68.
12. Espie C, Kyle S, Williams C, Ong J, Douglas N, Hames P et al. A Randomized, Placebo-Controlled Trial of Online Cognitive Behavioral Therapy for Chronic Insomnia Disorder Delivered via an Automated Media-Rich Web Application. Sleep. 2012;35(6):769-781.
13. Christensen H, Batterham P, Gosling J, Ritterband L, Griffiths K, Thorndike F et al. Effectiveness of an online insomnia program (SHUTi) for prevention of depressive episodes (the GoodNight Study): a randomised controlled trial. The Lancet Psychiatry. 2016;3(4):333-341.
14. www-users.york.ac.uk/~djt6/NIcourse/randomisation.ppt
15.  Luik A, Bostock S, Chisnall L, Kyle S, Lidbetter N, Baldwin N et al. Treating Depression and Anxiety with Digital Cognitive Behavioural Therapy for Insomnia: A Real World NHS Evaluation Using Standardized Outcome Measures. Behavioural and Cognitive Psychotherapy. 2016;45(01):91-96.
16. Heritier SR, Gebski VJ, Keech AC. Inclusion of patients in clinical trial analysis: The intention-to-treat principle. Med J Aust. 2003;179:438–40.
17. Freeman D, Lister R, Evans N. The use of intuitive and analytic reasoning styles by patients with persecutory delusions. J Behav Ther Exp Psychiatry 2014; 45:454–58.
18. Espie CA, Kyle SD, Hames P, Gardani M, Fleming L, Cape J.

The sleep condition indicator: a clinical screening tool to evaluate

insomnia disorder. BMJ Open 2014; 4: e004183.

19. Shibley H, malcolm R, veatch L. Adolescents with Insomnia and Substance Abuse: Consequences and Comorbidities. Journal of Psychiatric Practice. 2008;14(3):146-153.
20.                      Izuhara M, Matsuda H, Saito A, Hayashida M, Miura S, Oh-Nishi A et al. Cognitive Behavioral Therapy for Insomnia as Adjunctive Therapy to Antipsychotics in Schizophrenia: A Case Report. Frontiers in Psychiatry. 2018;9.
21. De Weerth C. Prenatal stress and the development of psychopathology: Lifestyle behaviors as a fundamental part of the puzzle. Development and Psychopathology. 2018;30(03):1129-1144.
22.                      Hardeland R. The Underrated Circadian System and Its Contribution to Neurodegeneration in Mood Disorders. JOURNAL OF MENTAL HEALTH AND CLINICAL PSYCHOLOGY. 2018;2(3):17-22.
23. Zachariae R, Lyby M, Ritterband L, O’Toole M. Efficacy of

internet-delivered cognitive-behavioral therapy for insomnia— a systematic review and meta-analysis of randomized controlled

trials. Sleep Med Rev 2016; 30: 1–10.

24.                      Sheaves B, Freeman D, Isham L, McInerney J, Nickless A, Yu L et al. Stabilising sleep for patients admitted at acute crisis to a psychiatric hospital (OWLS): an assessor-blind pilot randomised controlled trial. Psychological Medicine. 2017;48(10):1694-1704.
25. Bradley J, Freeman D, Chadwick E, Harvey A, Mullins B, Johns L et al. Treating Sleep Problems in Young People at Ultra-High Risk of Psychosis: A Feasibility Case Series. Behavioural and Cognitive Psychotherapy. 2017;46(03):276-291.
26.                      Sleepless cities revealed as one in three adults suffer from insomnia [Internet]. Aviva.com. 2018 [cited 20 November 2018]. Available from: https://www.aviva.com/newsroom/news-releases/2017/10/Sleepless-cities-revealed-as-one-in-three-adults-suffer-from-insomnia/
27.  Schneider A, Hommel G, Blettner M. Linear Regression Analysis. Deutsches Aerzteblatt Online. 2010.
28.                      Gueorguieva R, Krystal J. Move over ANOVA: progress in analysing repeated-measures data and its reflection in papers published in the Archives of General Psychiatry. Arch Gen Psychiatry 2004; 61: 310–17.