Cognitive behavioural therapy for insomnia in patients with rheumatoid arthritis: protocol for the randomised, single-blinded, parallel-group Sleep-RA trial

Rheumatoid arthritis (RA) is a chronic autoimmune inflammatory disease with a prevalence worldwide of 0.4–1.3%. Women are affected three times more frequently than men, and onset is most often at 50–60 years of age [1, 2]. Symptoms include joint swelling, pain and stiffness, and often significant disability. Disease activity is most commonly monitored with Disease Activity Score-28 C-reactive protein (DAS28-CRP) based on the number of swollen joints, the number of tender joints, serum CRP and patient global assessment [3, 4]. Improvements in early diagnosis and medical treatment with anti-rheumatic drugs have led to a reduced incidence of swollen and tender joints, pain and joint destruction [5]. Nevertheless, patients still report a wide range of symptoms with a high prevalence; among the most frequently reported are sleep disturbances, fatigue, depressive symptoms, pain, reduced functional status and health-related quality of life (HRQoL) [6–8]. In evaluating their medical treatment with antirheumatic drugs, patients indicate that sleep is one of the most important parameters. This highlights the major impact that sleep has on physical and mental well-being [9, 10].

Sleep disturbances are common in patients with RA, with a prevalence of 60–80% compared to 10–30% in the general population [11–13]. Furthermore, sleep quality can be associated with disease activity; for example, patients with moderate and high disease activity reported fewer hours asleep than patients with low disease activity or those in remission [14–17]. Sleep disturbances in patients with RA are characterised by significantly worse sleep efficiency (SE), sleep quality, sleep latency, number of awakenings, and time awake after sleep onset than healthy control groups [14, 18–20].

According to the International Classification of Sleep Disorders third edition (ICSD-3) and the Diagnostic and Statistical Manual of Mental Disorders the diagnostic criteria for insomnia are: sleep initiation or maintenance problems; inadequate opportunity and circumstances to sleep; and daytime consequences [21, 22]. The ICSD-3 duration criterion for chronic insomnia disorder is 3 months and a frequency criterion of at least three times per week [21]. Like RA, insomnia is associated with fatigue, depressive symptoms, pain, functional status and cardiovascular diseases [23–25].

The association between disease activity and sleep is acknowledged in patients with inflammatory arthritis, but the exact underlying mechanisms of insomnia in general have not been identified. Insomnia is often considered to be linked to hyperarousal, which decreases the likelihood of sleep; however, the heterogeneity in cause, symptoms, course, comorbidities and consequences complicates the establishment of a universal aetiological or pathophysiological model of insomnia [26, 27]. According to existing evidence on the pathophysiology of insomnia, insomnia is most likely to develop in people who have increased genetic risk and who experience abnormalities in neurobiological processes—vulnerabilities that may lead to hyperarousal and to psychological and behavioural processes that increase the risk for developing insomnia [27–29].

According to the conceptual “3P model” of insomnia by Spielman et al., three elements contribute to the development and maintenance of chronic insomnia: predisposing factors, precipitating events, and perpetuating mechanisms [30]. The predisposing factors include the fact that some individuals may have an increased risk of insomnia by having certain genetic predisposing factors of personal vulnerability that might contribute to a highly sensitive or malfunctioning biological sleep system or a hyperactive arousal system that interferes with sleep [31]. The time of diagnosis can be a precipitating event as being diagnosed with a chronic disease like RA can cause a cognitive and emotional reaction. These reactions can manifest as worry or depressed mood, which may also exploit the vulnerability of a highly reactive sleep system [32]. For patients already diagnosed with RA, several conditions related to their arthritis can represent precipitating events that contribute to the development of insomnia—for example, fatigue, pain, reduced functional status, flares and disease activity [33–35].

When insomnia becomes chronic, sleep is characterised by inappropriate habits and dysfunctional thoughts related to sleep, i.e. the perpetuating mechanisms. Insomnia is often linked to maladaptive strategies, such as avoidance behaviour during waking hours. Avoidance behaviour may include cancelling planned activities either because of feeling too tired or due to fear that the activities will interfere with sleep. Also, increased sleep effort is common, e.g. spending excessive time in bed and developing rigid sleep-related rituals. In response to insomnia, some people also develop sleep-interfering cognitions, such as overestimating and worrying about the negative consequences of insomnia and approaching bed time with fear of failure [30, 36].

The predominant treatment for insomnia is hypnotic drugs [37]. However, hypnotic drugs cause many side effects, including increased risk of falls resulting in fracture, cognitive impairment, impaired motor coordination, sedation, confusion, motor vehicle accidents, tolerance and dependence [38, 39], and are associated with increased mortality and risk of dementia [40–42]. In the search for non-pharmacological treatments for insomnia without severe side effects and with longer-lasting effects, clinical trials have tested the effects of exercise, thought control, imagery training, acupuncture, paradoxical intention, mindfulness and cognitive behavioural therapy for insomnia (CBT-i) [43–45]. Currently, non-pharmacological treatment for insomnia in patients with RA has focused only on physical activity and exercise training [46, 47].

However, hypnotic drugs do not address the precipitating events or perpetuating mechanisms that maintain chronic insomnia, and precipitating events or perpetuating mechanisms are not the primary focus or purpose of physical activity or exercise training. Furthermore, some patients can be disabled to a degree that inhibits them engaging in physical activity or exercise training. CBT-i is recommended as the first-line treatment for insomnia as it has demonstrated more effects than hypnotic drugs and has demonstrated a long-term effect [48–51]. In clinical settings, CBT-i is most often carried out by psychologists; however, treatment with CBT-i in groups led by nurses has been tested and the results showed significant positive effects [52, 53].

According to Morin and his model “The vicious circle of persistent insomnia”, the effect of CBT-i is caused by its ability to modify thought patterns and behaviours that reinforce poor sleep [54]. Thus, it is highly relevant to investigate the efficacy of CBT-i in patients with RA and, to our knowledge, this is the first clinical trial to investigate CBT-i in this population.

Participants will be included if they: are diagnosed with RA according to the American College of Rheumatology/European League Against Rheumatism 2010 criteria [55]; have low to moderate disease activity, defined as DAS28-CRP ≤5.1 [4] and are 18 years of age or older; have both insomnia, defined as Insomnia Severity Index ≥11 [56] and chronic insomnia, according to the ICSD-3 duration criterion [29, 57]; have unchanged treatment with anti-rheumatic drugs for the preceding 3 months and no indication for changes of treatment with antirheumatic drugs during the intervention; and are able to understand and read Danish.

Patients will be excluded if they: have had glucocorticoid administered by intramuscular, intra-articular or intravenous methods within the preceding 4 weeks; are on a continuous current daily intake of oral glucocorticoids >5 mg/day; have had a change in oral glucocorticoid dose within the preceding 3 months; have had prior or current professional non-pharmacological treatment for insomnia; work at night during the intervention, are pregnant or nursing a baby; have a current alcohol or drug-use disorder, according to their medical journal; have patient-related circumstances (physical or mental) that impede their ability to give informed consent and adhere to the programme.

Participants are prohibited from participating in any professional non-pharmacological treatment for insomnia other than Sleep-RA during the trial period.

The primary outcome is SE at week 7 assessed by polysomnography (PSG) [68]. SE percentage = (TST/total time in bed (lights out to lights on)) × 100. In healthy persons, SE is 85–90%.

PSG is a comprehensive study of the biophysiological changes that occur in the human body during sleep. In Sleep-RA, 17 electrodes will be attached to each participant, recording data from electroencephalogram, eye movements electro-oculogram, electromyogram muscle activity or skeletal muscle activation, heart rhythm, respiratory airflow, respiratory effort and peripheral pulse oximetry [69]. All PSG studies will be conducted in the participants’ own homes. Data from PSG will be analysed in epochs of 30 s and managed at the Danish Center for Sleep Medicine (DCSM), Rigshospitalet, Glostrup.

We will examine other secondary outcomes after treatment at week 7 and at follow-up at week 26 if CBT-i has a significant positive effect on the primary outcome of SE after treatment at week 7.

Additional information includes sociodemographic characteristics (education, employment, civil status, children living at home), lifestyle (smoking, alcohol, drugs, caffeine, physical activity), RA-related information (duration of RA, rheumatoid factor of the immunoglobulin M class, anti-citrullinated protein antibody, CRP), sleep-related information (duration of insomnia, sleep stages, number of awakenings, apnoea-hypopnoea index, leg movements), comorbidity (diabetes, hypertension, heart attack, other heart disease, stroke, chronic obstructive pulmonary disease, cancer, osteoarthritis, osteoporosis, asthma, depression) and current medicine (antirheumatic drugs, glucocorticoids, antidepressant drugs, hypnotic drugs).

Randomisation will be conducted after baseline assessment by computer-generated random numbers with blocks of ten participants.

During randomisation, participants will be stratified by disease activity to equalize the risk of change in treatment with antirheumatic drugs during the trial period and the effect this might have on sleep. Stratification will be set to disease activity according to DAS28-CRP ≤3.2 (low disease activity) and >3.2 (moderate disease activity) measured at baseline [97].

Allocation concealment will be ensured as it will not be possible to conduct the computer-generated randomisation until the baseline assessment is completed for each participant.

An external collaborator (Zitelab Aps, http://www.zitelab.dk/↗) will generate the allocation sequence. This collaborator is not involved in outcome assessments or the CBT-i intervention. KML will enrol participants and, immediately after randomisation, inform them to which group they have been allocated.

Outcome assessments are conducted by project nurses specialising in rheumatology, medical laboratory technologists or neurophysiology assistants, all of whom will be blinded to participant allocation.

Application of PSG equipment and instructions on how to use the actigraph and subsequent analysis of data will be conducted by neurophysiology assistants at the DCSM. Patient-reported outcome measures will be collected via questionnaires on tablets handed out by a project nurse. To assess the outcome of disease activity, a project nurse will examine the joints of participants to identify the number of swollen and tender joints related to DAS28. Blood samples will be taken and analysed by a medical laboratory technologist.

KML will conduct the CBT-i intervention and is therefore not blinded to participant allocation.

The trial design is open-label with only the outcome assessors being blinded so unblinding will not occur.

Data will be collected from PSG, actigraphy, physical examination, blood samples, a medical journal and questionnaires as presented in Table 2 and as follows.

The data collected from PSG cover SE, SOL, WASO, TST, sleep stages, number of awakenings, apnoea-hypopnoea index and leg movements. The data collected from actigraphy cover SE, SOL, WASO and TST for 7 continuous nights. From the medical journal we will collect data on duration of RA, anti-rheumatic drugs and glucocorticoids. Questionnaires will provide data on patient-reported SE, SOL, WASO, TST, insomnia, sleep quality, fatigue, impact of RA, depressive symptoms, pain, functional status, HRQoL, patient global assessment, physician global assessment, sociodemographic information, lifestyle, RA-related information, comorbidity, duration of insomnia, antidepressant drugs and hypnotic drugs. From the physical examination we will obtain data on swollen joints and tender joints counts (of 28 examined joints), and we will collect data on CRP from serum samples.

Dates for assessments will be planned by KML together with each participant, and all participants will receive a text reminder on their telephone or an email the day before a scheduled assessment. If participants withdraw consent for follow-up assessment of one outcome, they will be given the option to continue with assessments for other outcomes.

All participants will receive an identification code and all data will be de-identified.

A list of identifiable participant information associated with each identification code will be stored electronically separately from the research data for 5 years. Patient-reported information will be completed electronically on a tablet, which will be used only for participants in this trial and is linked to the DANBIO Reuma electronic case report form.

All information collected during the trial will be kept confidential in accordance with Danish Data Protection Agency rules. All data will be processed confidentially according to the Act on Processing of Personal Data.

There will be no biological specimens collected for genetic or molecular analysis in this trial.

All main analyses will be conducted based on the intention-to-treat (ITT) population. We plan to conduct both an analysis of the full analysis set (patients with outcome data available at baseline) and a per-protocol analysis so that any differences between them can be explicitly discussed and interpreted. The ITT principle states the effect of a treatment policy (i.e. the planned treatment regimen) rather than the actual treatment given. The consequence of conducting ITT analyses is that participants allocated to a treatment group should be followed up, assessed and analysed as members of that group, irrespective of their compliance to the planned course of treatment. Robustness is a concept that refers to the sensitivity of the overall conclusions to various limitations of the data, assumptions and analytic approaches to data analysis. Robustness implies that the treatment effect and primary conclusions of the trial are not substantially affected when analyses are carried out based on alternative assumptions or analytical approaches.

We will control the α level across only the most important hypotheses by use of a hierarchical testing strategy to preserve a type 1 error rate of 5% for the primary outcome of change in SE assessed after treatment at week 7, followed by SE assessed at follow-up after 26 weeks. All the patient-reported outcome measures and other secondary outcome measures will, by principle, only be clinically relevant after 26 weeks. The sleep-related key secondary outcomes (SOL, WASO, TST, insomnia and sleep quality) will only be considered statistically significant with a P value <0.01 (0.05/5). The RA-related key secondary outcomes (fatigue, impact of RA and depressive symptoms) will only be considered statistically significant with a P value <0.017 (0.05/3) as described in Fig. 2.

In the Sleep-RA trial with repeated measures, participants will be randomly assigned to treatment groups, and outcome observations are made at two time points on each patient. We anticipate that measures on the same patient at different times are correlated and that measures taken close together in time will be more highly correlated than measures taken further apart in time; observations on different patients will be assumed to be independent.

Data will be analysed using the PROC MIXED procedure of the statistical program SAS System, with baseline level as a covariable, using a multilevel repeated-measures random-effects model, with participants as the random effect factor and based on a restricted maximum likelihood estimate. For the primary outcome measure, the after-treatment value will be the response variable, and the baseline values of treatment group (two levels), stratum (i.e. two levels according to the randomisation) and time (two levels) will be the covariates. Assessment of these baseline values (main effects) will be of interest, along with the interaction between treatment group and time. This statistical model holds all between-group comparisons at both assessment points and allows for evaluation of the average effect over the period from baseline to follow-up at 26 weeks.

The SAS statistical package (v.9.4; SAS institute Inc., Cary, NC, USA) and R 3.0.1 (http://www.R-project.org↗, the R Foundation for Statistical Computing) will be used for the statistical models.

We plan to include 60 patients, and the trial period for each participant is 26 weeks. This is a non-pharmacological 6-week intervention with no expected adverse or harmful events, and the trial is therefore not subject to independent safety monitoring and periodical review, e.g. interim analyses.

We have no intention to conduct subgroup and adjusted analyses.

We will apply the analysis framework suggested by White et al. for ITT analysis that depends on making plausible assumptions about the missing data and including all participants in sensitivity analyses [98]. We will: Attempt to follow-up all randomised participants, even if they withdraw from allocated treatment.Perform a main analysis of all observed data that are valid under a plausible assumption about the missing data (missing at random). Our primary analysis population will be all participants with available data statistically modelled using repeated-measures linear mixed models (see below). These models will be valid if data are missing at random (any systematic difference between the missing values and the observed values can be explained by differences in observed data).Perform sensitivity analyses to explore the effect of departures from the assumption made in the main analysis (data as observed; valid assuming that data are missing completely at random).Account for all randomised participants, at least in the sensitivity analyses. We will analyse all variables with missing data (due to withdrawals) being replaced by imputation of the baseline level, i.e. interpreted as assuming that those who dropped out returned to their baseline level [99]. These models will potentially be valid even if data are ‘missing not at random’ (even after the observed data are considered, systematic differences remain between the missing values and the observed values).

In addition, data will be analysed according to the per-protocol population, where data from participants who participated in a minimum of four out of six sessions of CBT-i are included. The participants from the control group will be considered per-protocol individuals if data are available from all three assessments (baseline, after treatment at week 7, and follow-up at week 26).

The datasets analysed during this trial are available from the corresponding author on reasonable request.

This trial proceeds from COPECARE (Copenhagen Center for Arthritis Research), a multidisciplinary centre which is internationally renowned and provides day-to-day support for the trial. COPECARE represents the steering group in collaboration with the University Hospitals Centre for Nursing and Care Research, Parker Institute and DCSM.

The steering committee consists of health professionals with solid research and management competences. The committee represents knowledge at a high level from different professions: nurses, rheumatologists, psychologist, neurophysiologist and statistician. This ensures scientific quality, multidisciplinary width and possible implementation in clinical practice depending on the results of the trial. The managers of the rheumatology department participate in the steering committee and ensure clinical anchoring and implementation.

Once a year an external assessor from the University of Copenhagen is invited to a meeting with the project management group to discuss progress in the trial and methods to improve quality and stringency.

The daily project manager (KML) is affiliated to COPECARE. She identifies all potential participants from the recruiting sites through DANBIO (as explained in the “Recruitment” section). She provides oral information and received written consent.

The intervention in this trial is a non-pharmacological intervention with low or no risk of adverse or harmful events. Therefore, there is no data monitoring committee. The trial is registered on www.clinicaltrials.gov↗ and approved by the Regional Ethics Committee and the Danish Data Protection Agency.

Participants will be monitored throughout the period of the intervention to detect any unintended events.

Should a PSG show signs of a sleep disorder requiring further examination or treatment the participant will be contacted by DCSM after the final assessment is completed.

The project management group meet twice a year and conduct an audit of the trial.

When a participant has completed self-reported questionnaires on the tablet used in this trial, the project nurse ensures that data are complete and imported to the electronic case report form. When objective sleep measures of PSG and actigraphy are completed, all data are assessed for completeness by a neurophysiology assistant. If any data are missing or of low quality, the participant is invited to a new assessment.

Important protocol amendments which may impact the content or conduct of the trial will be decided in the steering group. Important protocol amendments will be sent for approval to the ethics committee, notified to the participating sites and addressed in the written and oral information. Furthermore, the protocol registered on www.clinicaltrials.gov↗ will be updated prior to implementation.