Partner-assisted cognitive behavioural therapy for insomnia versus cognitive behavioural therapy for insomnia: a randomised controlled trial

Insomnia is a highly prevalent and impairing sleep disorder, affecting 10–15% of adults worldwide [1‐5]. In Australia, 34.5% of the adult population report regular difficulty falling asleep or staying asleep, hallmark symptoms of insomnia, and many develop a diagnosable insomnia disorder [6]. Insomnia is associated with substantial personal and economic costs. Individuals with insomnia report lower quality of life [7], are at greater risk of accidents [8‐11], and have high rates of psychiatric comorbidities [1]. For example, in Australia from 2009 to 2010, half of the clients seen for depression in General Practitioners’ offices also presented with insomnia [12]. In addition to depression, individuals with insomnia are more likely to develop a range of mental health issues, including suicidality, substance abuse, and post-traumatic stress disorder (PTSD), as well as physical health problems such as hypertension, diabetes, and cardiac events [6]. Insomnia carries substantial economic burden ($10.9 billion in Australia in 2010) due to high healthcare costs, reduced work productivity, and increased accidents both at work and on the road [6].

There is a strong evidence base for the non-pharmacological treatment of insomnia [13, 14] with behavioural intervention [15]. Cognitive behavioural therapy for insomnia (CBTI) has been identified as the gold standard intervention for treating insomnia. CBTI is a multi-component intervention based on Speilman’s behavioural model of insomnia [16]. CBTI is efficacious in the short-term; individuals maintain therapeutic gains at long-term follow up, and research indicates its long-term superiority to pharmacotherapy in several randomised controlled trials [17, 18]. Treatment effects are seen in primary and comorbid insomnia [19]. Furthermore, CBTI produces greater subjective and objective sleep improvements when compared with no treatment, progressive relaxation therapy, and pharmacologic and non-pharmacologic placebo interventions [20]. CBTI also has a number of practical advantages: it is relatively brief (i.e. seven to eight sessions on average), can be delivered effectively in multiple formats (e.g. individual or group [20]), and is generally preferred by clients over medications [18].

Dismantling studies have shown that the two components carrying the most variance in treatment outcomes in CBTI are sleep restriction therapy (SRT) and stimulus control (SC) [21]. SRT requires the individual with insomnia to stay up much later and/or get out of bed much earlier than they typically do; this builds sleep pressure, in essence by sleep depriving clients. SC requires getting out of bed when the individual is unable sleep, something many individuals find very difficult in the middle of the night. Both strategies are challenging and require significant effort to maintain adherence. Therefore, while CBTI is efficacious, as with any behavioural intervention, it only produces the desired outcome when the individual adheres to the treatment recommendations and completes the treatment. Although the challenges of sleep restriction and stimulus control are typically overcome in tightly controlled randomised controlled trials (RCTs) due to the added support provided by the study and the typically homogenous sample, they pose a challenge for mainstream clinicians without those added resources and enrolment restrictions.

The two places where the potential difficulties associated with the behavioural changes demanded by CBTI would be expected to diminish outcomes in more real-world settings are: 1) adherence to the treatment protocol; and 2) attrition (i.e. treatment drop-out).

A recent review shows that adherence to CBTI in the clinical settings ranges from 32–52% [21]. The lowest rates were in clinical effectiveness studies and studies with comorbid populations [21]. Not surprisingly, even modest improvements in CBTI adherence can improve outcomes [22]. Findings from a study of 696 clients with insomnia found that those who completed treatment with high adherence improved significantly more than completers with low adherence [22]. Additionally, adherence is the strongest predictor of long-term (e.g. 1 year) outcomes [22]. Attrition also reduces the benefits of treatment, especially in clinical settings where dropout rates are reported to be as high as 39% [23, 24]. A study on the national dissemination of CBTI in the US Veteran Affairs Healthcare System reported that 33% of clients dropped out [24]. Together, these studies suggest a sizable proportion of clients are likely not to be fully benefiting from CBTI. Efforts to broadly implement CBTI would benefit from ways to improve treatment adherence and completion rates. There currently are no evidence-based recommendations for augmenting either treatment adherence or completion in CBTI.

One attractive option for increasing adherence and completion in CBTI is via the involvement of bed partners due to 60% of Australian adults sharing a bed. A growing body of literature supports the effectiveness of couple-based interventions for treating individual psychiatric disorders including mood, obsessive compulsive disorder, PTSD, eating disorders, and substance-related disorders [25‐28]. Within the sleep field, qualitative studies have suggested including the bed partner in continuous positive airway pressure treatment for sleep apnoea [29, 30]. In “partner-assisted” interventions, partners play a supportive role by acting as surrogate therapists, helping clients make behavioural changes outside of the therapy setting [31]. Treatment focuses on improving the client’s symptoms without directly targeting the couple’s relationship [32]. Partners in this role are taught to help provide an environment favourable to behaviour change and to reinforce positive client changes [32]. This is important, because well-intentioned partners can negatively impact clients’ psychiatric conditions and treatment gains [32]. For example, caring partners can inadvertently accommodate clients’ maladaptive behaviours in an effort to demonstrate support, avoid conflict, or decrease distress for clients or themselves. Such accommodation can often reinforce maladaptive behaviours and work against treatment goals [32]. Therefore, the need to educate partners may well be relevant in insomnia.

To our knowledge, the only partner-assisted intervention for insomnia to be tested was performed by the investigator team during the treatment development phase prior to the current RCT. Thus, this study aims to determine whether enlisting partner support in adhering to difficult behaviour change, and teaching partners to reduce ineffective, albeit well intentioned, accommodating behaviours, could serve to optimise important environmental and behavioural conditions in the treatment of insomnia.

The literature on couples and sleep is relatively sparse, especially in the context of insomnia. However, recent reviews and the few observational studies in this domain highlight several important points. Importantly, sleep is often a social experience since the majority of adults have bed partners [33]. Consistent with this idea, couples prefer to sleep in the same bed and report worse sleep when they sleep alone, despite objective evidence for negative consequences on sleep from co-sleeping [34]. Moreover, couples influence each other’s sleep. For example, Lee et al. found sleep duration covaried within couples on a daily basis, and sleep quality covaried within couples when averaged across several days [35]. Sleep is also related to relationship satisfaction. For example, Troxel et al. found sleep disturbance, including insomnia, and relationship satisfaction were associated [33]. Similarly, those with sleep-onset insomnia were more likely to report relationship problems as a result of their own or their partner’s sleep issues [36]. Another study found higher marital satisfaction was linked to higher sleep-wake concordance in couples [37], and a large study (n = 405 couples) found a spouse’s sleep problems predicted higher levels of marital unhappiness [38]. Finally, couple-level variables, including bed partner behaviours, can perpetuate insomnia [34]. For example, bed partners can encourage a client to sleep in or take a nap after a bad night of sleep, or to read, watch TV, etc. in bed until they become sleepy, or they can increase a client’s attentional preoccupation with sleep by asking about their sleep in the morning. While these behaviours are generally well-intended, they are all antithetical to treatment. Therefore, evidence suggests that partners affect each other’s sleep and relationship satisfaction, and thus incorporating a partner into CBTI has the potential to significantly impact outcomes. Surprisingly, though, no published studies report development or testing of a CBTI intervention incorporating the partner.

To examine the impact of adding the partner to treatment on adherence and completion.

To examine the impact of adherence on insomnia symptoms in the short- and long-term (6 months post-treatment completion). This will test if adherence is directly related to outcomes and if the combined effects of the partner and CBTI are superior to either alone.

Study participants (n = 120 completers) are clients and their partners recruited from the Monash Healthy Sleep Clinic (Melbourne, Australia) and the community via advertisement, e.g. radio, newspaper, flyer drops, and social media. Our estimated recruitment numbers shown in Fig. 1 are based on a 4:1 referral-to-screening ratio, and our initial screening visit goal assumes a 20% screen failure rate. Our randomisation numbers assume 20% attrition. These rates are similar to those we have experienced in other similar insomnia trials.

Participant eligibility criteria are outlined in Table 1. We considered carefully whether to exclude participants taking sleep medications but decided against this based on the recommendations of Morin et al. [42]. We are, however, tracking medications throughout the study.

When participants first express interest in the study, they complete a brief pre-screening questionnaire either online or via telephone. This questionnaire is administered to both clients and partners and asks questions pertaining to sleep, mental health, time travel, shift work, and relationship status.

At the initial visit, and following written informed consent, a 2- to 3-h study eligibility screening and pre-treatment assessment is conducted. This assessment involves: (a) Duke Structured Interview for Sleep Disorders (DUKE; [39]) to assess insomnia and other potential sleep disorders (client only); (b) Structured Clinical Interview for DSM-5 (SCID-5; [40]) to document any other psychiatric disorders (client only); (c) medical and sleep history; and (d) a series of self-report measures as shown in Table 2. Participants also undergo an overnight polysomnography to rule out excluded sleep disorders.

Following pre-treatment assessments, couples are randomised to one of three interventions: PA-CBTI, i-CBTI, or PA-SMT. All three interventions consist of seven weekly, 1-h sessions. Participants can discontinue the treatment or the study at any stage. If one of the members of the couples wishes to withdraw, they must withdraw as a couple. Post-treatment and 6-month follow-up data will be collected for those who discontinue the treatment unless they withdraw from the study.

Manuals are used in all treatment sessions with the psychologist to ensure treatment fidelity, and specific training is provided to all clinicians for all manuals. To ensure fidelity to the delivery of treatment, all sessions are audio recorded. Treatment fidelity is emphasised in several ways: (a) therapists receive intensive training on all interventions; (b) therapists receive weekly group supervision, including the supervisor listening to selections of that week’s audio recordings; and (c) audio recordings of every session from the first two cases of each treatment and 10% of all subsequent sessions are evaluated for treatment fidelity (based on standardised check lists of session content).

A limitation of the current study is that neither clients nor clinicians are blinded to the treatment condition, a problem inherent to any intervention of psychotherapy. However, clients are blinded to the study hypotheses and staff who score polysomnography, actigraphy, and questionnaires are blind to treatment condition.

In addition, because we want to be reflective of a clinic-based population, and for the sake of feasibility and to reduce participant burden, the ‘baseline’ sleep diary and actigraphy data are collected between weeks 1–2 of treatment. Hence, this may not be considered as a true baseline. However, between weeks 1–2 the active intervention has not yet commenced as the first session consists of education only. Similarly, our ‘post-treatment’ sleep diary and actigraphy data are collected between weeks 6–7 of treatment. However, the final treatment session is a summary session only and no additional treatment strategies are introduced. All other post-treatment data (e.g. questionnaires) are collected after completion of treatment session 7.

We do not conduct a diagnostic assessment in the partners. Thus, it is possible some of them meet the criteria for a sleep disorder. While we considered conducting such screens, conceptually we do not believe the presence of a sleep disorder should have an impact on the partner’s ability to perform the role defined for them within the intervention. We do, however, have the STOP-Bang from screening, as well as Insomnia Severity Index, sleep diary, and actigraphy at each assessment time point. Thus, we can assess the risk of obstructive sleep apnoea and track symptoms of insomnia in the partners. If we have a couple where both partners report insomnia upon initial contact with the research team, we ask the couple to decide who will enrol as the “client” and who will enrol as the “partner.” While the partner is not required to adopt treatment recommendations themselves, they are welcome to change their behaviour in accordance with treatment if they so desire. This is true whether or not the partner reports sleep problems. Not having formal diagnostic information on the partners limits our ability to assess effectiveness of the interventions in couples where both individuals experience a diagnosed sleep disorder.

Finally, our lack of an individual SMT treatment condition could be considered a limitation of this trial. However, the decision not to include this group was made due to feasibility. An individual SMT group would have required a significantly larger sample allocated to a condition for which we do not expect to have favourable outcomes.

Strengths of the current project include the range of outcome variables, which permits high-quality analysis. Our focus on optimising external validity by maximising generalisability to clinical settings and implementing relatively few exclusion criteria is a significant strength, given clinic-based studies and clients with comorbidities show reduced adherence and completion relative to more tightly controlled RCTs. In addition, our control condition, sleep management therapy (PA-SMT), is a valuable control given that it is considered ‘treatment as usual’ in the community. Results could lead to the dissemination of a new CBTI intervention that helps more people adhere to and complete treatment, ultimately benefiting a greater number of people who seek treatment for insomnia. PA-SMT allows us to assess for the unique contribution of the partner independent of the active treatment. Another strength of this trial is our multimodal assessment of sleep; that is, both prospective (e.g. sleep diaries) and retrospective (e.g. ISI) data are collected. Furthermore, we control for clinician time and effort in that clinicians dedicate the same amount of time and effort to each of the three treatment conditions. This trial is also not limited to the investigation of how the interventions impact on sleep since data on broader outcomes will be analysed, including relationship functioning and mental health. Finally, this trial will allow a robust analysis of the impact of insomnia and its treatment on the bed partner, something not examined to date (Additional file 1).