Internet-delivered cognitive behaviour therapy for adolescents with insomnia comorbid to psychiatric conditions: A non-randomised trial

Abstract

Insomnia is highly prevalent among adolescents with psychiatric conditions and is known to aggravate psychiatric symptoms. Research on cognitive behaviour therapy for adolescents with comorbid insomnia (CBT-I) is still limited. The aim of this study was to investigate feasibility and preliminary effects of internet-delivered CBT for adolescents with insomnia comorbid to a psychiatric condition. Twenty-one patients (13–17 years) with comorbid insomnia were recruited from Child and Adolescent Psychiatry. All patients received 7 weeks of internet-delivered CBT-I with therapist support. Outcomes were assessed at baseline, post-treatment, and at a 4-month follow-up. The proportion of completed assessments was overall acceptable. Participants on average completed 4.48 (sd = 1.97) of the seven treatment modules and therapists on average spent 12.80 minutes (sd = 6.23) per patient and week. Results showed large statistically significant improvements on insomnia severity, sleep efficiency, sleep onset latency and sleep quality. Medium to large improvements were also seen on the psychiatric symptoms of depression, obsessive-compulsive symptoms, interpersonal sensitivity, paranoid ideation and psychoticism. These findings indicate that internet-delivered CBT is feasible and potentially promising for improving sleep and reducing psychiatric symptoms in adolescent psychiatric patients with insomnia and co-morbid psychiatric disorders. A larger randomised trial is warranted to verify these preliminary results.

Keywords

Comorbid insomnia paediatrics cognitive behaviour therapy sleep disorders psychiatry

Introduction

Insomnia is defined as dissatisfaction with sleep quantity or quality with difficulty initiating sleep, difficulty maintaining sleep, or early-morning awakening occurring at least three nights per week, for at least 3 months, despite adequate opportunity for sleep, causing significant distress or daytime impairment (American Psychiatric Association, 2013). The prevalence of insomnia among adolescents is commonly reported to be 9% to 11% in the general population (Chung et al., 2014), but among children and adolescents with psychiatric conditions, prevalence rates are between three and eight times higher (Chorney et al., 2008; Corkum et al., 1998). It has been suggested that insomnia can be a transdiagnostic risk factor bidirectionally related to both the development and course of other psychiatric disorders (Dolsen et al., 2014).

Sleep problems are also related to a higher symptom burden for children and adolescents with psychiatric disorders. In a clinical sample of 553 depressed children and youth, Liu et al. (2007) found that sleep disturbed children were more severely depressed, had a higher number of depressive symptoms, and were more likely to have a comorbid anxiety disorder. In an encompassing review from Harvey (2011) presented a strong case for viewing sleep disturbance as a mechanism in the mood disorders for adolescents and adults, as it is the most common prodrome of mania, is associated with relapse of depression and hypomania/mania, impedes emotional regulation and cognitive performance, and increases the risk of suicidality.

According to European guidelines, cognitive behavioural therapy for insomnia (CBT-I) should be considered as the first treatment choice for chronic insomnia in adults (Riemann et al., 2017). These recommendations are based on findings from a large number of randomised controlled trials, showing moderate to large treatment effects of CBT-I on sleep variables, with results maintained at follow-up (Trauer et al., 2015).

Medium to large effects have also been found for CBT-I on sleep variables for patients with insomnia comorbid with psychiatric and/or medical conditions (Wu et al., 2015). For patients with psychiatric comorbidities the treatment also yielded medium to large effects on psychiatric symptoms, whereas only small effects were seen on medical comorbidities.

CBT-I is a treatment focused on behaviours and cognitions interfering with sleep. The behavioural component suggests that insomnia is caused by classical conditioning, irregular sleep habits, and spending too much time in bed (e.g. Dolsen et al., 2014). This through a process where the sleep environment becomes associated with wakefulness, and counterproductive sleep related behaviours create a homeostatic imbalance in the sleep and circadian systems. The cognitive component proposes that excessive worry about sleep, and selective attention towards sleep-related threat cues, lead to physiological arousal and a tendency to overestimate perceived impairment of sleep and daytime functioning (ibid). Together with inaccurate beliefs about sleep and dysfunctional behaviours performed to compensate for poor sleep, these are the factors contributing to insomnia.

A handful of studies have been conducted on CBT-I for adolescents with primary insomnia (see Åslund et al., 2018 for a review), and insomnia comorbid to physical and psychiatric comorbidities (Åslund et al., 2020; Bootzin & Stevens, 2005; Clarke et al., 2015; Palermo et al., 2017), showing effects on sleep, and on comorbid symptoms such as depression, anxiety and substance use.

In a randomised trial on CBT-I for primary insomnia in adolescents, similar results were found for internet-delivered treatment as for group treatment (de Bruin et al., 2015). This parallels the findings in a large number of randomised trials (n = 103) for a variety of clinical conditions, comparing conventional CBT and internet-delivered CBT (iCBT), where equivalent effects have been found with substantially less therapist time (Hedman et al., 2012). In the discussion de Bruin et al. (2015) notes that all participants from the control group, except one, preferred to start the internet-delivered CBT-I (iCBT-I), once the waiting period was over, indicating that the internet format may be compelling to adolescents with insomnia.

Previous investigations of iCBT-I for adolescents have studied primary insomnia in media-recruited samples, and it is thus unknown whether patients in child and adolescent psychiatry could benefit from this type of treatment. The aim of the present study was therefore to develop and investigate the feasibility and preliminary effects of internet-delivered CBT-I for patients in child and adolescent psychiatric care, with insomnia comorbid to a psychiatric condition.

Methods

Design and procedure

The study was a non-randomised trial with three intakes. Participants were recruited from two tertiary Child and Adolescent Psychiatry clinics. Eligible participants were given written information and consent forms by their regular therapist. Those handing in written consent signed by the adolescent and caregivers were booked for an intake interview containing a structured diagnostic assessment of insomnia and other sleeping disorders (as outlined by Morin & Espie, 2003). The interview was conducted over telephone at the first two intakes, and in-person at intake three. Four psychology master students, and two clinical psychologists conducted the interviews. After the interview, participants were asked to complete the web-based baseline assessment within the next 2 weeks. Also, background variables were collected from the participants’ regular care provider. Based on data from interviews and the regular care provider, criteria for eligibility and sleep diagnoses were evaluated in supervision with a senior clinical psychologist.

Outcome variables were measured at baseline, post treatment and a 4-month follow-up. The primary outcome was also measured weekly. The regular care provider and the participants’ caregivers provided information about changes in the regular treatment that had taken place throughout the trial. For a participant flow chart, see Figure 1. A power calculation was performed beforehand, based on the effect sizes found on an earlier pilot study (de Bruin et al., 2014). Assuming a significance level of 0.05, and a power of 80%, 21 participants needed to be enrolled. The study was conducted in line with the ethical standards of the Helsinki Declaration and its later amendments. It was approved by the regional ethics committee in Uppsala (dnr 2015/326), and registered at clinicaltrials.gov (NCT02612987).

Figure 1.

Participant flowchart.

Participants

Clinical patients were considered eligible if they had difficulties sleeping in addition to their other psychiatric complaints. The inclusion criteria in the study were (a) age between 13 and 17 years, (b) fulfilling the diagnostic criteria of insomnia according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013), (c) having access to a computer connected to the internet, (d) no psychotropic medication or being on a stable dosage for at least 6 weeks before signing up (2 weeks regarding sleeping medicine) and (e) no previous or ongoing CBT for insomnia. The exclusion criteria were: (a) ongoing manic or psychotic episode, (b) being at high risk of triggering manic or psychotic episode and (c) ongoing sleep apnoea. In all 21 participants were enrolled. A large majority of the sample (90%) was female, the average age was 15.48 (sd = 1.29, range 13–17), and a majority (76%) had a duration of insomnia of more than 12 months. The three most common types of main diagnoses represented were ADHD (33%; n = 7), depressive disorders (33%, n = 7) and anxiety disorders (14%; n = 3). Seven participants (33%) also had a second psychiatric diagnosis, and three participants (14%) had two additional diagnoses, apart from the main diagnosis and the sleep disorder. Regarding comorbid sleep disorders, four participants had a circadian rhythm sleep-wake disorder with delayed sleep phase, one had restless legs syndrome and one had nightmare disorder. Other sample characteristics are presented in Table 1.

Table 1.

Sample characteristics.

Characteristic	Total (n = 21)
Characteristic	n	%
Duration insomnia
0 to 3 months	0	0
4 to 12 months	5	23
13 to 24 months	3	14
25 to 36 months	1	5
37 to 48 months	5	24
>49 months	7	33
Duration of contact with psychiatric unit
0 to 12 months	6	29
13 to 24 months	2	10
25 to 36 months	4	19
37 to 48 months	3	14
>49 months	6	29
Ongoing psychiatric care at time of inclusion*
Psychotherapy	6	29
CBT	4	19
DBT	1	5
Not specified	1	5
Support therapy	6	29
Parent counseling	5	24
Collaboration with the patient’s school	1	5
Collaboration with social services	2	10
Weighted blanket	3	14
Ongoing medication prescribed for improved sleep*
No medication	9	43
Melatonin	11	52
Prometazin	2	10
Hydroxizin	1	5
Propiomazin	1	5
Alimemazin	1	5
Other ongoing psychoactive medication*
No medication	7	33
Antidepressants	9	43
Central nervous system stimulants	5	24
Anxiolytics	1	5
Antipsychotics	1	5
Anticonvulsant	1	5

Unrestricted number of response options available.

Intervention

The intervention was an internet-delivered guided self-help, conceptually based on established cognitive behavioural treatment protocols, such as Jernelöv (2008), but with content and language designed for adolescents and the specific conditions surrounding their sleep (such as late night activities on social media, irregular sleeping habits and tendency towards delayed sleep phase). For an overview of the treatment content, see Table 2. Emphasis was put on stimulus control, and restriction of time in bed, that is establishing regular sleeping hours, and limiting the time spent in bed, as these interventions are known to produce significant improvements to sleep (Morin & Espie, 2003). Bed time restriction was applied as follows: (1) mean total sleep time during the past week was calculated, (2) a fixed time for going to bed at night and getting up in the morning was set, based on the mean total sleep time, but starting with a minimum of 7 hours in bed, (3) adjustments of time in bed were done weekly by either (a) leaving it unchanged in case the participant was content with their sleep, and daytime functioning, (b) decreasing time in bed by 15 minutes in case the participant was not content with their sleep, or (c) increasing time in bed by 15 minutes in case the participant was content with sleep, but suffered from severe daytime sleepiness. If the participant continued to compress their sleep throughout treatment, at the end point their time in bed was 6 hours at a minimum.

Table 2.

Content of the iCBT-I intervention.

Module	Content
1	Psychoeducation
2	Bed time restriction
2	Stimulus control
3	Adjustment of bed time restriction schedule
	Continued stimulus control
	Sleep hygiene advice
4	Problem-solving for removing obstacles to change of sleep-related problem behavior
	Adjustment of bed time restriction schedule
	Continued stimulus control
5	Psychoeducation about stress
	Time management
	Deactivation techniques (deep relaxation)
	Adjustment of bed time restriction schedule
	Continued stimulus control
6	Psychoeducation about worry
	Deactivation techniques (applied relaxation)
	Challenging negative automatic thoughts
	Scheduled ‘worry time’
	Adjustment of bed time restriction schedule
	Continued stimulus control
7	Review of treatment content
7	Relapse prevention

The treatment consisted of seven modules, one per treatment week. Participants could only proceed to the next module if the homework assigned to their ongoing module had been completed. The treatment was provided on a secure treatment platform accessible from computer or smartphone, and consisted of text, pictures, illustrations, sound files and films. The therapists logged in to the platform 2 days per week. Treatment was delivered by two psychologists involved in developing the treatment protocol, and a final year psychology master student with theoretical and practical training in CBT. The student received clinical supervision weekly. Caregivers had no contact with the therapist during treatment, but were initially provided with a pdf-file, containing the summary of the different modules, and advise on how to best support the adolescent during treatment. Communication between participant and therapist could take place either as messages or as comments in the text. Messages and comments from therapists mainly consisted of clarifications regarding treatment content, positive remarks regarding the work the participant had done, comments encouraging the participant to continue working, and specifications on how the general advice were applicable to the participant.

If the participant had not logged on to treatment platform for ten days they were sent a mobile phone text message, if they had not logged in for 20 days they were contacted by telephone, and finally, if the participant did not reply to the phone call, their parents were contacted by telephone. Time for texting and calling participants is included in the specified therapist time.

Measures

Background variables

Background variables were collected from the participant’s regular care provider, and during the structured intake interview. The background variables included the participants’ current psychiatric diagnoses (given at the latest appointment with the medically responsible psychiatrist), duration of the current psychiatric health care contact, and ongoing psychiatric interventions at inclusion, including ongoing medication, presence and duration of insomnia, and other sleep disorders (restless legs syndrome, narcolepsy, obstructive sleep apnoea hypopnea, delayed sleep phase, non-rapid eye movement sleep arousal disorders, nightmare disorder).

Feasibility variables

Feasibility was evaluated in terms of completion of planned assessments, practicality (i.e. amount of therapist contact required), and treatment compliance. Completion of planned assessments were considered acceptable if the proportion of missing data was ⩽20%, while data was missing at random or completely at random (Higgins et al., 2019). Based on earlier trials of iCBT (Hedman et al., 2012), <20 minutes of therapist contact per participant and week was considered practical. Treatment compliance was measured as the number of modules completed and the number of messages and comments sent to the therapist. Participants were considered to have completed the treatment if they completed the first four modules, in agreement with the minimum amount of CBT-I sessions found to be effective for adults (van Straten et al., 2018).

Outcome

Primary and secondary outcomes were collected by self-report administered online.

Primary Outcome

Insomnia severity was measured using the Insomnia Severity Index (ISI; Morin, 1993). The ISI is a seven-item questionnaire, with scores on each item ranging from 0 to 4, with higher scores implying more severe insomnia. The ISI has shown high reliability regarding internal consistency and item response analyses and a high concurrent validity in relation to results on sleep diaries, polysomnographic variables and clinicians’ ratings (Morin et al., 2011). The ISI has also shown sensitivity to treatment-induced changes in perceived degree of sleep problems (ibid). Treatment response was defined as >8 points change on the ISI compared to baseline, and treatment remission as an absolute ISI score of less than 8, as suggested by Morin et al. (2011).

Secondary outcomes

Sleep variables

A sleep diary was filled in by the participants daily during the course of 1 week at baseline, post-treatment and at follow-up. The diary contained registrations of bedtime, time of falling asleep, length of night-time awakenings, time of waking up and time of getting out of bed. These registrations were used in order to calculate average sleep efficiency (SE), sleep onset latency (SOL), wake after sleep onset (WASO) and total sleep time (TST). The diary also contained a daily registration of perceived sleep quality, on a scale between 1 (very poor) and 5 (very good).

Depression

Montgomery-Åsberg Depression Rating Scale – Self-report (MADRS-S; Montgomery & Åsberg, 1979) is a nine-item questionnaire with scores on each item ranging from 0 to 7, where higher scores indicate more depressive symptoms. MADRS-S has shown high reliability, and a high concurrent validity in relation to clinicians’ ratings and other validated instruments measuring depression (Svanborg & Åsberg, 2001).

Psychiatric symptom severity

Symptom Check List 90-items (SCL-90; Derogatis et al., 1974) is a 90-item questionnaire with scores on each item ranging from 0 to 4, where higher scores indicate more psychiatric symptoms. The scale is subdivided in nine different symptom dimensions: somatisation, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation and psychoticism. The psychometric properties of the scale have been evaluated repeatedly; with high reliability, and in a large Swedish validation study the subscales could discriminate between clinical and normal samples, and between meaningful subgroups in the clinical sample (Fridell et al., 2002).

Data analysis

Statistical analyses were performed in SPSS, version 23.0 (IBM Corp., 2015). Little’s MCAR test was conducted to establish whether or not data was missing completely at random. Frequency distributions of variables and their residuals were plotted in histograms and Q-Q plots. Normality, skewness and kurtosis were also formally tested. Outliers were detected on basis of the outlier-labelling rule (Hoaglin et al., 1986).

To analyse outcome on continuous variables linear mixed effects models fitted with full information maximum likelihood estimation were performed. Linear mixed models generate unbiased estimates of missing data and thus constitute an intention-to-treat analysis. The analytic assumptions of this approach are that the residuals are normally distributed, and that data is missing at random. The primary outcome was measured at several points during treatment. Therefore we first analysed changes during treatment, and secondly changes from post- to follow-up. For the remaining outcomes we included all three measurement points (pre, post, follow-up) in the same analysis. In all analyses fixed effects error terms were held equal across time and were not correlated. Regarding random effects, these were included in the model when significant. In case where both random intercept and random slope contributed to the model, two different covariance structures were tested for the random effects: one where the covariance was constrained to be zero and one where the covariance was estimated, as suggested by Hesser (2015). Model fit was tested systematically with the log-likelihood ratio test for nested models.

Effect sizes are reported as Cohen’s d within-group effect size with 95% confidence intervals, calculated from the linear slope, as outlined by Feingold (2009).

Two kinds of sensitivity analyses were conducted. Robust repeated measures ANOVAs were conducted on data with non-normal distributions. A second sensitivity analysis was conducted by including treatment changes in the regular psychiatric treatment during the period from baseline to follow-up as a covariate in the linear mixed effects models.

Results

Completion of planned assessments and analytic assumptions

At post treatment and follow-up, data was missing completely for two participants (9.52%). The sleep diary was missing for five participants at follow-up (23.81%). Data were found to be missing completely at random (Little’s MCAR: χ² = 301.91, df = 2168, p = 1.00). One significant outlier was found on MADRS-S at post treatment, and WASO at follow-up, by using the outlier-labelling rule (Hoaglin et al., 1986). The outliers were adjusted through winsorizing, that is replacing the value with the second highest/lowest value at the same assessment point (Duan, 1997). Significant skewness of residuals was detected for ISI at follow-up, MADRS-S at follow-up, and SOL at post-treatment. This was handled through sensitivity analyses with robust repeated measures ANOVAs, as recommended by Field and Wilcox (2017). In the sensitivity analyses missing data was not replaced. Data on WASO was found to have a negative binomial distribution and thus a negative binomial regression was used for calculating the results on this variable.

Practicality and treatment compliance

The therapists on average spent 12.80 minutes (sd = 6.23), sent 1.71 (sd = 0.40) messages, and posted 0.68 (sd = 0.97) comments, per participant and week. Regarding treatment compliance the mean number of completed modules was 4.48 (sd = 1.97) out of 7. In all 67% (14/21) of participants completed at least four modules and were thus categorised as completers. On average participants sent 2.38 (sd = 2.97) messages to their therapist, and wrote 1.86 (sd = 3.51) comments.

Evaluation of treatment outcome

Means and standard deviations for all outcome measures at pre-, post- and follow-up are presented in Table 3, together with results from linear mixed effects models.

Table 3.

Means and standard deviations for observed data on outcome variables, and results from linear mixed-effects regression analyses, with within-group effect sizes.

	m (SD)		Effect size (95% CI)	Linear mixed effect model (95% CI)
	m (SD)		Cohen’s d	Estimate	p
Insomnia severity^a
Baseline	15.43 (4.87)
Post-treatment	7.21 (4.04)	Pre-post	−1.88 (−2.55, −1.21)	−1.08 (−1.47, −0.70)	<.001
Follow-up	7.53 (5.73)	Post-fu		0.03 (−0.20, 0.26)^a	.771
		Pre-fu	−1.80 (−2.42, −1.17)
Sleep efficiency %
Baseline	66.47 (14.30)
Post-treatment	81.29 (9.54)
Follow-up	78.81 (11.64)	Pre-fu	1.01 (0.30, 1.73)	0.60 (0.20, 1.00)	.006
Sleep onset latency hour:minute
Baseline	2:27 (1:23)
Post-treatment	1:00 (0:43)
Follow-up	1:16 (0:57)	Pre-fu	−1.04 (−1.67, −0.40)	−3.44 (−5.55, −1.33)	.002
Wake after sleep onset hour:minute
Baseline	0:26 (0:31)
Post-treatment	0:14 (0:16)
Follow-up	0:12 (0:15)	Pre-fu	−0.45 (0.03, −0.84)	n/a
Total sleep time hour:minute
Baseline	6:58 (1:57)
Post-treatment	7:38 (1:10)
Follow-up	7:16 (1:22)	Pre-fu	0.24 (−0.37, 0.85)	1.08 (−1.69, 3.85)	.433
Sleep quality
Baseline	2.72 (0.65)
Post-treatment	3.34 (0.71)
Follow-up	3.38 (1.09)	Pre-fu	1.06 (0.05, 2.09)	0.033 (0.01, 0.06)	.046
Depression^b
Baseline	23.86 (10.83)
Post-treatment	19.26 (10.72)
Follow-up	17.95 (12.48)	Pre-fu	−0.55 (−0.93, −0.16)	−0.29 (−0.49, −0.09)	.006
Somatisation^c
Baseline	0.95 (0.80)
Post-treatment	0.88 (0.80)
Follow-up	0.94 (1.02)	Pre-fu	0.04 (−0.53, 0.61)	0.002 (−0.021, 0.024)	.877
Obsessive-compulsive^c
Baseline	1.90 (0.80)
Post-treatment	1.62 (0.91)
Follow-up	1.17 (0.78)	Pre-fu	−0.87 (−1.18, −0.47)	−0.037 (−0.05, −0.02)	<.001
Interpersonal sensitivity^c
Baseline	1.52 (0.95)
Post-treatment	1.19 (1.07)
Follow-up	0.90 (0.87)	Pre-fu	−0.68 (−1.06, −0.21)	−0.032 (−0.05, −0.01)	.002
Depression^c
Baseline	1.52 (0.82)
Post-treatment	1.21 (0.92)
Follow-up	1.14 (1.02)	Pre-fu	−0.42 (−0.88, 0.01)	−0.019 (−0.04, 0.00)	.057
Anxiety^c
Baseline	1.46 (0.82)
Post-treatment	1.26 (0.98)
Follow-up	1.18 (0.98)	Pre-fu	−0.28 (−0.64, 0.06)	−0.013 (−0.03, 0.003)	.120
Hostility^c
Baseline	1.06 (0.63)
Post-treatment	1.00 (1.03)
Follow-up	0.78 (0.80)	Pre-fu	−0.30 (−0.75, 0.25)	−0.012 (−0.03, 0.01)	.202
Phobic anxiety^c
Baseline	1.12 (0.98)
Post-treatment	0.95 (1.02)
Follow-up	0.89 (0.90)	Pre-fu	−0.23 (−0.63, 0.21)	−0.011 (−0.03, 0.01)	.196
Paranoid ideation^c
Baseline	1.31 (0.95)
Post-treatment	1.06 (1.03)
Follow-up	0.75 (0.85)	Pre-fu	−0.62 (−1.06, −0.21)	−0.029 (−0.05, −0.01)	.001
Psychoticism^c
Baseline	0.79 (0.57)
Post-treatment	0.72 (0.94)
Follow-up	0.47 (0.45)	Pre-fu	−0.53 (−0.93, −0.01)	−0.017 (−0.03, −0.00)	.034

Note. n/a = not applicable.

Measured on Insomnia Severity Index.

Measured on Montgomery-Åsberg Depression Rating Scale – self-report.

Measured on Symptom Check List 90 items.

Primary Outcome

The mixed effect model analyses showed a large negative effect of time on insomnia severity from pre to post (d = 1.88), and no significant effect of time from post to follow-up, indicating an improvement during treatment that was maintained at follow-up.

At post-treatment, ten participants (47.6%) had a clinically significant response to treatment (decrease of ⩾8 points on ISI) as compared to the pre-treatment assessment, and eleven (52%) were in remission (ISI < 8). At follow-up, eleven participants (52%) had a clinically significant response to treatment, and fourteen (67%) were in remission.

Secondary outcome

The linear mixed effect model analysis for secondary outcomes from pre- to follow-up showed large improvements on sleep efficiency (d = 1.01), sleep onset latency (d = 1.04), sleep quality (d = 1.06) and obsessive-compulsive symptoms (d = 0.87). Moderate improvements were found on depressive symptoms (d = 0.55, on MADRS-S), interpersonal sensitivity (d = 0.68), paranoid ideation (d = 0.62) and psychoticism (d = 0.53). No statistically significant changes were seen on total sleep time, somatisation, depression (on SCL-90), anxiety, hostility, or phobic anxiety.

The negative binomial regression with LSD-corrected pairwise comparisons showed no statistically significant improvement in Wake After Sleep Onset (χ²(2,21) = 5.26, p = .072).

Sensitivity analyses

Robust repeated measures ANOVAs were conducted on ISI, MADRS-S, and SOL, as they were found to be significantly skewed at one or more measurement points. The results of the robust ANOVAs replicated the findings in the linear mixed models.

A second sensitivity analysis was conducted by including treatment changes as a covariate in the linear mixed models analyses. In all, seven (33%) of the participants made changes in their in regular psychiatric treatment during the period from baseline to follow-up. Two participants started a medication to induce sleep, two participants started a Cognitive Behavioural Treatment focusing on their main diagnoses, one received a heavier weighted blanket than the one previously prescribed, one participant quit taking antidepressants and one participant quit taking antidepressants and increased the dosage of ADHD-medication. There were no statistically significant interaction effects when entering treatment changes as a covariate, indicating that changes in the patient’s regular treatment did not influence the outcome.

Discussion

This study investigated feasibility and preliminary effects of internet-delivered CBT-I for adolescent psychiatric patients with insomnia comorbid to a psychiatric condition. With respect to feasibility we found acceptable levels of treatment compliance, and overall proportion of completed assessments. However, the proportion of sleep diaries completed at follow-up was somewhat less than optimal. Amount of therapist contact required was practical (i.e. would be realistically achievable in usual practice). Preliminary outcome analyses showed large statistically significant improvements from pre- treatment to follow-up assessments on insomnia severity, sleep efficiency, sleep onset latency and sleep quality. Medium to large improvements were also seen on the psychiatric symptoms of depression, obsessive-compulsive symptoms, interpersonal sensitivity, paranoid ideation and psychoticism. No statistically significant changes were found on total sleep time, wake after sleep onset, somatisation, anxiety, hostility or phobic anxiety.

The average therapist time spent on treatment per participant and week was around 13 minutes, which is approximately one-third of the time spent per session in standard CBT. This indicates that iCBT-I could potentially save a substantial amount of therapist time when treating adolescents with insomnia comorbid with psychiatric conditions. In comparison to conventional CBT, internet-delivered CBT is accessible at any time and location and requires less therapist time. Thus the adolescent does not have to skip school to undergo the treatment, can participate irrespective of how far away from the clinic they live, and the therapist can treat more patients in a given timeframe. Also, the internet-delivered treatment is a form of self-help, thus the therapy work is mainly the adolescents’ own doing, a fact that potentially clarifies the common mix-up among patients between ‘attending sessions’ and ‘doing therapy’. Furthermore, since most of the treatment is highly standardised, it is likely to be less sensitive to therapist drift, a previously underestimated threat to the implementation of evidence-based therapy (Waller & Turner, 2016). However, a web-based format including no parent intervention, while compelling, may introduce a decreased possibility to engage parent support. This aspect of the treatment delivery should be evaluated.

Participants completed on average 4.5 modules, that is around two-thirds of the treatment. Module five and six addressed stress and worry: aspects not relevant to all participants. It is possible that completion rates would have increased if this content had been presented only to a selected subgroup of participants, a procedure called tailored interventions. Tailored internet delivered CBT has been developed as a means to adjust standardised protocols to different patient profiles, by offering different treatment content to different patients, according to their individual needs and preferences. In a randomised controlled trial by Johansson et al. (2012), tailored iCBT showed a larger recovery rate and reduction of depressive symptoms, as compared to standardised iCBT, among depressed patients with a high symptom burden.

Results on sleep outcomes were in line with those found of iCBT-I in samples of adolescents with no psychiatric comorbidity (de Bruin et al., 2014, 2015). The fact that 52% had a clinically significant response to treatment, and 67% were in remission is encouraging given the fact that 57% of the sample had a duration of insomnia of >36 months and that sleep disturbances in adolescence is a problem that can persist into adulthood (Dregand & Armstrong, 2010). Moreover, 61% of the sample had already at baseline received a treatment for their insomnia with an insufficient effect, of which 12 participants had a medication prescribed for their sleep, and three participants had a weighted blanket. Thus, the preliminary effect on sleep, in most cases was accomplished in a group of participants with a high level of chronicity of their insomnia, and with a complexity of presentation both with regard to psychiatric diagnoses and sleep difficulties.

The preliminary results found on psychiatric symptoms contribute to a growing understanding of how improved sleep can be beneficial with regards to lessened psychiatric symptom burden in adolescents (e.g. Åslund et al., 2020; Clarke et al., 2015; Palermo et al., 2017).

This non-randomised trial is limited by the small sample size, the uncontrolled design, the absence of an objective sleep measure (such as actigraphy) and the relatively high degree of missing values for the sleep diary at follow up (24%). Further, the study contained no measure of acceptability, although a convenience sample of 11 of the participants answered an informal interview about their experiences of treatment, after their 4-month follow-up. Another limitation is the fact that the sample consisted to 90% of female participants, so we cannot tell if males would experience similar outcomes

In conclusion, the findings indicate that iCBT-I is feasible and potentially promising for improving sleep and reducing psychiatric symptoms in adolescents with psychiatric disorders. A larger randomised trial is warranted to verify these preliminary results, with a secondary aim to identify predictors of treatment outcome. Before conducting such a trial, results from the informal evaluation of the treatment should be used as a basis for further refinements of treatment content and outline.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Financial support for the study was provided from Uppsala-Örebro Regional Research Council and through the regional agreement on medical training and clinical research between Uppsala County Council and Uppsala University.

ORCID iDs

Vendela Zetterqvist

Lie Åslund

Author biographies

Vendela Zetterqvist is an associate professor in clinical psychology and a clinical researcher in the field of behavioural medicine. She is a licensed clinical psychologist working at the child- and adolescent psychiatric services. Together with Linda Hasbar and Susanna Jernelöv, she has developed the clinical program evaluated in this trial.

Charlotte Lundén is a licensed clinical psychologist and psychotherapist undergoing doctoral studies. Her research focus is on insomnia and comorbid psychiatric disorders in adolescents. Her clinical practice is at the psychiatric services.

Anders Herrmann is a licensed clinical psychologist with experience of working at the child- and adolescent psychiatric services conducting treatment of insomnia disorder.

Linda Hasbar is a licensed clinical psychologist with experience of working at the child- and adolescent psychiatric services.

Najah Khalifa, MD, PhD, is a senior specialist in paediatric neurology and habilitation. Her main research focus is on tics and Tourettes syndrome. She has her clinical practice at the child- and adolescent psychiatric services.

Mats Lekander is a professor of psychoneuroimmunology at Stockholm University, and of health psychology at Karolinska Institutet. His research interests span from consequences of immune-to-brain communication to development and implementation of psychological treatment for somatic and mental disorders.

Lie Åslund is a licensed clinical psychologist undergoing doctoral studies. Her research focus is on insomnia and comorbid psychiatric disorders in adolescents. She has experience of working in the child- and adolescent psychiatric services.

Susanna Jernelöv, PhD, is a licensed clinical psychologist who has conducted several research studies on psychological treatments for sleep problems, including long-term follow-ups. She is currently heading a project on cognitive behavioural treatment of insomnia (CBT-i) and other sleep difficulties in psychiatric populations, and a project on mechanisms in CBT-i.

References

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). American Psychiatric Association.

Åslund

Arnberg

Kanstrup

Lekander

(2018). Cognitive and behavioral interventions to improve sleep in school-age children and adolescents: A systematic review and meta-analysis. Journal of Clinical Sleep Medicine, 14(11), 1937–1947.

Åslund

Lekander

Wicksell

R. K.

Henje

Jernelöv

(2020). Cognitive-behavioral therapy for insomnia in adolescents with comorbid psychiatric disorders: A clinical pilot study. Clinical Child Psychology and Psychiatry, 25(4), 958–971. https://doi.org/10.1177/1359104520929376

Bootzin

Stevens

(2005). Adolescents, substance abuse, and the treatment of insomnia and daytime sleepiness. Clinical Psychology Review, 25(5), 629–644.

Chorney

Detweiler

Morris

Kuhn

(2008). The interplay of sleep disturbance, anxiety, and depression in children. Journal of Pediatric Psychology, 33(4), 339–348.

Chung

Kan

K. K.

Yeung

(2014). Insomnia in adolescents: Prevalence, help-seeking behaviors, and types of interventions. Child and Adolescent Mental Health, 19(1), 57–63.

Clarke

McGlinchey

Hein

Gullion

Dickerson

Leo

Harvey

(2015). Cognitive-behavioral treatment of insomnia and depression in adolescents: A pilot randomized trial. Behaviour Research and Therapy, 69, 111–118.

Corkum

Tannock

Moldofsky

(1998). Sleep disturbances in children with attention-deficit/hyperactivity disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 37(6), 637–646.

de Bruin

Bögels

Oort

Meijer

. (2015). Efficacy of cognitive behavioral therapy for insomnia in adolescents: A randomized controlled trial with internet therapy, group therapy and a waiting list condition. Sleep, 38(12), 1913–1926.

10.

de Bruin

Oort

Bögels

Meijer

. (2014). Efficacy of internet and group administered cognitive behavioral therapy for insomnia in adolescents: A pilot study. Behavioral Sleep Medicine, 12(3), 235–254.

11.

Derogatis

L. R.

Lipman

R. S.

Rickels

Uhlenhuth

E. H.

Covi

(1974). The Hopkins Symptom Checklist (HSCL): A measure of primary symptom dimensions. In Pichot

(Ed.), Psychological measurements in psychopharmacology. Karger.

12.

Dolsen

M. R.

Asarnow

L. D.

Harvey

A. G.

(2014). Insomnia as a transdiagnostic process in psychiatric disorders. Current Psychiatry Reports, 16(9), 1–7.

13.

Dregand

Armstrong

(2010). Adolescence sleep disturbances as predictors of adulthood sleep disturbances—A cohort study. Journal of Adolescent Health, 46(5), 482–487.

14.

Duan

(1997). The robustness of trimming and Winsorizing when the population distribution is skewed (Unpublished doctoral dissertation). Tulane University.

15.

Feingold

(2009). Effect sizes for growth-modeling analysis for controlled clinical trials in the same metric as for classical analysis. Psychological Methods, 14(1), 43.

16.

Field

A. P.

Wilcox

R. R.

(2017). Robust statistical methods: A primer for clinical psychology and experimental psychopathology researchers. Behaviour Research and Therapy, 98, 19–38.

17.

Fridell

Cesarec

Johansson

Malling-Andersen

(2002). Svensk Normering, Standardisering och Validering av Sympstomskalan SCL90 (Symptoms Checklist 90) [Swedish standards and validations of the SCL90 symptom scale] (SiS-rapport 4/2002). Statens Institutionsstyrelse.

18.

Harvey

(2011). Sleep and circadian functioning: Critical mechanisms in the mood disorders. Annual Review of Clinical Psychology, 7, 297–319.

19.

Hedman

Ljotsson

Lindefors

(2012). Cognitive behavior therapy via the Internet: A systematic review of applications, clinical efficacy and cost–effectiveness. Expert Review of Pharmacoeconomics & Outcome Research, 12(6), 745–746.

20.

Hesser

(2015). Modeling individual differences in randomized experiments using growth models: Recommendations for design, statistical analysis and reporting of results of internet interventions. Internet Interventions: The Application of Information Technology in Mental and Behavioural, 2(2), 110–120.

21.

Higgins

J. P. T.

Savović

Page

Elbers

Sterne

J. A. C.

(2019). Assessing risk of bias in a randomized trial. In Higgins

J. P. T.

Thomas

Chandler

Cumpston

Page

M. J.

Welch

V. A.

(Eds.), Cochrane handbook for systematic reviews of interventions (2nd ed., pp. 205–228). Wiley-Blackwell.

22.

Hoaglin

D. C.

Iglewicz

Tukey

J. W.

(1986). Performance of some resistant rules for outlier labeling. Journal of the American Statistical Association, 81(396), 991–999.

23.

IBM Corp. (2015). IBM SPSS statistics for windows (Version 23.0). IBM Corp.

24.

Jernelöv

(2008). Sov Gott: Råd och tekniker från KBT. Wahlström & Widstrand.

25.

Johansson

Sjöberg

Sjögren

Johnsson

Carlbring

Andersson

Rousseau

Andersson

(2012). Tailored vs. standardized internet-based cognitive behavior therapy for depression and comorbid symptoms: A randomized controlled trial. PLoS One, 7(5):e36905.

26.

Liu

Buysse

D. J.

Gentzler

A. L.

Kiss

Mayer

Kapornai

Vetró

Á.

Kovacs

(2007). Insomnia and hypersomnia associated with depressive phenomenology and comorbidity in childhood depression. Sleep, 30(1), 83–90.

27.

Montgomery

Asberg

(1979). A new depression scale designed to be sensitive to change. The British Journal of Psychiatry, 134(4), 382–389.

28.

Morin

C. M.

(1993). Treatment manuals for practitioners. In Morin

C. M.

(Ed.), Insomnia: Psychological assessment and management. Guilford Press.

29.

Morin

C. M.

Belleville

Bélanger

Ivers

(2011). The Insomnia Severity Index: Psychometric indicators to detect insomnia cases and evaluate treatment response. Sleep, 34(5), 601–608.

30.

Morin

C. M.

Espie

C. A.

(2003). Insomnia: A clinical guide to assessment and treatment. Kluwer Academic/Plenum Publishers.

31.

Palermo

T. M.

Beals-Erickson

Bromberg

Law

Chen

(2017). A single arm pilot trial of brief cognitive behavioral therapy for insomnia in adolescents with physical and psychiatric comorbidities. Journal of Clinical Sleep Medicine, 13(3), 401–410.

32.

Riemann

Baglioni

Bassetti

Bjorvatn

Dolenc Groselj

Ellis

J. G.

Espie

C. A.

Garcia-Borreguero

Gjerstad

Gonçalves

Spiegelhalder

(2017). European guideline for the diagnosis and treatment of insomnia. Journal of Sleep Research, 26(6), 675–700.

33.

Svanborg

Åsberg

(2001). A comparison between the Beck Depression Inventory (BDI) and the self-rating version of the Montgomery Asberg Depression Rating Scale (MADRS). Journal of Affective Disorders, 64(2–3), 203–216.

34.

Trauer

J. M.

Qian

M. Y.

Doyle

J. S.

Rajaratnam

S. M. W.

Cunnington

(2015). Cognitive behavioral therapy for chronic insomnia: A systematic review and meta-analysis. Annals of Internal Medicine, 163(3), 191.

35.

van Straten

van der Zweerde

Kleiboer

Cuijpers

Morin

C. M.

Lancee

. (2018). Cognitive and behavioral therapies in the treatment of insomnia: A meta-analysis. Sleep Medicine Reviews, 38, 3–16.

36.

Waller

Turner

(2016). Therapist drift redux: Why well-meaning clinicians fail to deliver evidence-based therapy, and how to get back on track. Behaviour Research and Therapy, 77, 129–137.

37.

J. Q.

Appleman

E. R.

Salazar

R. D.

Ong

J. C

. (2015). Cognitive behavioral therapy for insomnia comorbid with psychiatric and medical conditions: A meta-analysis. JAMA Internal Medicine, 175(9), 1461–1472.