A feasibility study to explore the use of digital treatment of sleep as a first-step intervention to improve adolescent mental health

ABSTRACT

Introduction

Cognitive behavior therapy for insomnia (CBTi), delivered face-to-face or digitally, can improve the mental health of adults. Although insomnia is common in adolescents, the effects of digital CBTi on adolescent mental health have seldom been investigated.

Objectives

The aims of this study were to explore: (i) the acceptability of a digital CBTi intervention, Sleepio, as a first-step intervention for adolescents referred to specialist mental health services (CAMHS), (ii) the impact on sleep and mental health and (iii) subsequent CAMHS interventions.

Method

Sleepio is a computerized CBTi intervention comprised of six sequentially delivered sessions. Digital Sleepio was offered to new referrals to CAMHS with poor sleep and mental health problems. Results. Of the 75 eligible adolescents, 70 (93%; 95% CI: 85% to 98%) accepted Sleepio with 59 starting the programme and consenting to participate in the study. Of these, 37 (63%; 95% CI: 49% to 75%) completed at least half of the programme. There were post-intervention improvements in sleep, mood, and anxiety; the improvement in sleep was greater for those who completed at least half the programme compared to those who did not. Of those who completed all the programme, 55% (15/29) did not need any subsequent specialist CAMHS input. Of the 11 adolescents who accepted but never started Sleepio, none engaged with other CAMHS interventions and were subsequently discharged.

Conclusion

Our study has a number of limitations, in particular the absence of a control group and the loss of follow-up data for programme drop-outs. Nonetheless, these results suggest that digital CBTi may offer a novel and acceptable way of improving the sleep and mental health of adolescents with insomnia. A fully powered randomized controlled trial is required to obtain definitive estimates of the effects of the intervention.

Introduction

Poor sleep during adolescence is common with up to one in four having difficulty falling asleep at night and feeling tired throughout the day (Gariepy et al., 2020; Gradisar et al., 2011). Insomnia, defined as chronic dissatisfaction with sleep quantity and/or quality, is the most prevalent sleep disorder affecting 8–11% of adolescents (American Psychiatric Association, 2013; Dohnt et al., 2012). Adolescent insomnia persists over time and is associated with a variety of negative consequences including poor school performance, increased risk of health conditions, engaging in more risk behaviors, and poor psychosocial health (Roberts et al., 2009; Shochat et al., 2014). Poor sleep is a diagnostic criterion of many mental health problems and the relationship between adolescent insomnia, anxiety, and depression has been documented (Alvaro et al., 2017; Dahl & Harvey, 2007; Johnson et al., 2006).

Given the association between poor sleep and mental health, it is possible that providing treatment for insomnia may also have a positive impact on mental health. If so, the digital treatment of sleep may offer a novel way of addressing some of the present barriers to accessing specialist child and adolescent mental health services (CAMHS). For example, in the UK, high demand has led to long waiting lists resulting in waiting times for mental health interventions growing and problems exacerbating (Crouch et al., 2019). Furthermore, families may experience practical barriers, such as traveling to appointments or missing work/school to attend. These clinic appointments can feel daunting for adolescents due to the emotional exposure involved (King et al., 2006).

The best established and effective treatment for adults with insomnia is cognitive behavior therapy for insomnia (CBTi; National Institute for Health and Care Excellence, 2021). Although the primary goal of CBTi is to improve sleep, a number of studies have also noted secondary improvements in mental health (Belleville et al., 2011; Manber et al., 2011; Taylor & Pruiksma, 2014). The benefits of CBTi on sleep and mental health are retained when delivered in a digital format (Cheng et al., 2019; Christensen et al., 2016; Freeman et al., 2017; Ye et al., 2015). For example, the CBTi programme Sleepio has been shown to improve sleep and mental health when delivered as an unguided online programme in the community (Elison et al., 2017) or as an adjunct within a clinical mental health service (Stott et al., 2021). Similar improvements in mental health and sleep were reported when Sleepio was used as an online programme with brief support within a mental health service (Luik et al., 2017).

Although research using CBTi to improve anxiety and depression is promising in adults, research with adolescents is more limited (Gee et al., 2019). A handful of small and/or pilot studies, have documented positive effects of face-to-face CBTi on both sleep and symptoms of depression (Clarke et al., 2015; Conroy et al., 2019) and anxiety (Rollinson et al., 2021) in adolescents. There is also some limited evidence that adolescents recruited via the community or through mental health services show improvements in sleep, anxiety and depression following digital CBTi, with improvements maintained for up to one year (De Bruin et al., 2015, 2018; Cliffe et al., 2020). If the mental health benefits of digital CBTi for adolescents can be substantiated, the digital treatment of sleep may offer a viable solution to the long waiting lists and delays for treatment experienced by many child mental health services.

The aims of this feasibility study were threefold. Firstly, to determine the acceptability of digital CBTi as a first-step intervention for adolescents referred to specialist mental health services (CAMHS) with accompanying insomnia. Secondly, to explore the effects of digital CBTi on adolescent sleep, anxiety, and depression and finally, to explore the impact of digital CBTi on the need for subsequent mental health interventions.

Method

Ethical review and funding

The project was sponsored by the University of Bath and approved by the North of Scotland Research Ethics Service (ref: 20/NS/0049). The project was funded by NHSX, the information technology unit of NHS England and the Department of Health and Social Care.

Design and setting

We ran a pre-post, mixed methods study involving five specialist child and adolescent mental health service (CAMHS) teams provided by Oxford Health NHS Foundation Trust, UK. CAMHS are specialist NHS-provided services in the United Kingdom for children and young people, generally until school-leaving age (18 years), who are experiencing significant and persistent mental health difficulties. Oxford Health NHS Foundation Trust provides child mental health services to young people living across a large geographical area that includes Bath and North-East Somerset, Buckinghamshire, Oxfordshire, Swindon, and Wiltshire. A national comparison, on a range of indices, found these services to be rated between the lowest and highest performing CAMHS in England (Children’s Commissioner, 2021).

Participants

Eligible adolescents were: (i) accepted by specialist CAMHS, (ii) aged 13–17 years, (iii) presenting with mild to moderate anxiety and/or depression as determined by the clinical team, (iv) presenting with insomnia, and (v) were interested in completing an online sleep intervention. Young people were excluded if there was: (i) active suicidal ideation or planning, (ii) a recent (past 6 months) or ongoing safeguarding investigation or (iii) a significant developmental problem which interfered with the adolescent’s ability to engage with the online programme.

Digital CBTi (Sleepio)

Sleepio is an online, six-session, self-guided CBTi intervention. It includes a behavioral component (sleep restriction, stimulus control, and relaxation), a cognitive component (paradoxical intention, cognitive restructuring, mindfulness, positive imagery and putting the day to rest) and an educational component (psychoeducation and sleep hygiene). The programme is highly interactive, and content is presented by an animated virtual therapist “The Prof”. Participants complete daily sleep diaries throughout the intervention, which are used to provide automated “personalised” help. Sessions and sleep diaries were completed by the young person. Young people could choose to involve their parents, for example, through completing sessions together or joining weekly support calls, but there was no formal role for them or requirement for them to be involved in the programme. A more detailed summary of the programme and content is described elsewhere (Cliffe et al., 2018).

Procedure

Accepted new referrals to CAMHS with significant mental health problems were screened to identify those with possible sleep problems. The presence of insomnia was established through scoring 17 or less on the Sleep Condition Indicator (SCI; Espie et al., 2014). Those that met eligibility criteria and chose to complete the online sleep intervention (Sleepio) were contacted by a Sleepio assistant (NG, EH) to explain the project, identify if they would like to participate in the research evaluation, and to confirm contact details.

The contact details of those interested in participating in the pre-post evaluation were sent to the Research Assistant (AM) at the University of Bath. The researcher contacted each participant, and if under 16 years their parents, to explain the project, seek informed consent and to complete pre- Sleepio assessments. The CAMHS team maintained clinical responsibility for the young people whilst they participated in Sleepio. Any safeguarding concerns or identified risks were referred back to the clinical team for assessment and management.

Participants were sent a code to access Sleepio and weekly telephone calls were arranged to support them through the programme. After completing, or dropping out of digital CBTi, participants were contacted by the Research Assistant and invited to complete the post-assessments. They were offered a £10 Love to Shop Voucher to thank them for their participation.

Young people were then referred back to CAMHS by the Sleepio assistants with the local team assessing whether any additional specialist mental health input was required. The Sleepio assistants reviewed clinical case notes to detail the outcome e.g., whether they were discharged or offered a further assessment or intervention.

Each Sleepio session lasts approximately 20 minutes and were augmented via brief (up to 15 minute), weekly support telephone calls from a trained Sleepio Assistant (NG, EH). The purpose of these calls was to maintain motivation and engagement, signpost content, explain techniques and give suggestions on implementation.

Measures

Self-report standardized measures

The following self-report standardized measures of sleep, anxiety and depression were completed pre- and post- Sleepio.

Insomnia Severity Index (ISI; Morin et al., 2011). This 7-item measure assesses symptoms of insomnia over a 2-week period on a 5-point scale. The items assess sleep onset, sleep maintenance, early morning awakening problems, sleep dissatisfaction, interference of sleep difficulties with daytime functioning, whether sleep problems are noticed by others, and distress caused by sleep difficulties. Higher scores indicate greater sleep disturbance, with a cutoff of 15 or greater identifying 88% of those diagnosed with insomnia (Morin et al., 2011). The ISI has been validated for use with adolescents (Chung et al., 2011; Manzar et al., 2021). Internal consistency in this study was good (Cronbach’s alpha = 0.86).

Sleep Condition Indicator (SCI; Espie et al., 2014). This 8-item measure assesses sleep and its impact on daytime functioning over the previous month on a 4-point scale. It assesses sleep continuity (falling and remaining asleep), satisfaction with sleep (quality and troubled by sleeping), severity (nights per week and problem duration), and consequences of poor sleep (impact on personal functioning and performance). Lower scores indicate greater symptom severity with a cutoff of less than 17 correctly identifying 89% of those with probable insomnia disorder (Espie et al., 2012). The SCI has been used with young adolescents (aged 13–17) and has good internal consistency with this age group (Bandel & Brausch, 2020; Illingworth et al., 2020). Internal consistency in this study was good (Cronbach’s alpha = 0.82).

Revised Child Anxiety and Depression Scale (RCADS; Chorpita et al., 2005). This widely used 47-item questionnaire assesses symptoms of social phobia, separation anxiety, obsessive compulsive disorder, panic disorder, generalized anxiety disorder, and major depressive disorder. Each item is rated on a 4-point Likert scale of frequency ranging from never (0) to always (3). Items are summed to produce subscale and total scores. Higher scores indicate greater symptomatology with age- and gender-related norms used for identifying clinically significant scores (t scores ≥65). Internal consistency in this study was very good (Cronbach’s alpha = 0.95).

Mood and Feelings Questionnaire (MFQ; Angold et al., 1995). This is a 33-item scale, developed and validated for use with adolescents (Angold et al., 1995). Items represents symptoms of depression which are rated as either true (scores 2), sometimes true (scores 1) or not true (scores 0). Higher scores indicate more symptoms of depression with a total score of 27 or more being associated with major depression. The MFQ has high criterion validity and correlates well with other measures of depression (Kent et al., 1997; Wood et al., 1995). Internal consistency in this study was very good (Cronbach’s alpha = 0.91).

Post-use acceptability ratings

Participants who opted for the post-use interview were asked to rate their experience of Sleepio on a 3-point scale of “certainly true” (2) to “not true” (0). They were asked about ease of use, helpfulness, understanding, and whether they would recommend Sleepio to a friend. They rated their overall satisfaction with the programme and the impact on their sleep and mental health on a 10-point scale (1–10), with higher scores indicating a more positive rating.

Sleep efficiency, quality, and onset latency

Participants completed an online sleep diary whilst completing Sleepio. Information was extracted from the diary to assess sleep efficiency (percentage of time in bed asleep) and, sleep quality (rated from very poor (0) to very good (10). Sleep onset latency was assessed by the self-rated item on the Sleep Condition Indicator, “how long does it take you to fall asleep?” There are 5 options ranging from ≥ 61 minutes (scored 0) to 0–15 minutes (scored 4).

Analytic plan

This was a pre-post feasibility study so no formal power analysis was undertaken (Billingham et al., 2013). Guidelines for pilot and feasibility studies suggest a sample of between 30 and 59 participants should be sufficient to provide meaningful data to inform the design of subsequent definitive studies (Billingham et al., 2013; Viechtbauer et al., 2015).

Characteristics were summarized using means and standard deviations (or medians and interquartile ranges) for continuous variables and using numbers and percentages for categorical variables.

Post-intervention scores on continuous measures of sleep, anxiety and mood were compared to pre-intervention scores for the whole sample. The change scores from pre- to post-intervention were compared between participants who did and did not complete at least 3 of the 6 intervention sessions. In order to include in the full analysis some participants that only provided data at pre-intervention, mixed effects (“multilevel”) linear regression models were fitted using the pre- and post-intervention scores as repeated measures, specifying a random effect to allow for the correlation between scores from the same participant. In comparisons between pre- and post-intervention scores for the full sample, study wave (pre- versus post-) was used in the models as a binary predictor to quantify the mean change. In comparisons of mean change scores between those who did and did not complete at least half the programme sessions an interaction term between the binary predictors completer status and study wave was used to quantify the difference. The non-parametric bootstrap method (Efron & Tibshirani, 1993) was used to validate the results, demonstrating that the findings were robust to non-Normality in some of the outcomes.

Further analyses were undertaken to explore whether CBTi completion was associated with gender (boy/girl), age (under/over 16), insomnia severity (total score ISI and SCI), anxiety severity (total score RCADS), depression severity (total score MFQ) and time to start Sleepio (days). The two-sample t-test and the Chi-squared test were used for these analyses.

Analysis was undertaken in STATA version 17.

Results

Participant flow

This study was undertaken from October 2020 until July 2021 during the second and third UK national lockdown of the COVID-19 pandemic and during the disruption that occurred following the move within the NHS from face-to-face to online working. This negatively affected referrals to the project with a total of 86 referrals being received from October 2020- July 2021. Of these, 75 (87%) met inclusion criteria, 70 (93%; 95% CI: 85% to 98%) accepted digital CBTi, although 11 never activated or started the programme. The remaining 59 consented to participate in the study and of these, 37 (63%; 95% CI: 49% to 75%) completed 3 or more sessions with 29 (49%) completing all 6 sessions of the programme. Participants were granted access to Sleepio for 6 months. The mean amount of support provided by the Sleepio assistants was 32.2 (sd = 20.3) minutes per participant.

The mean length of time from the digital CBTi activation code being sent to starting the intervention was 6.2 (sd = 8.3, range 0–43) days. Information provided by Oxford Health Information and Technology department showed a median waiting time for a face-to-face CAMHS appointment for the teams involved in this project of 43.4 days during the course of this project.

For the 11 adolescents who accepted but never started digital CBTi, their referral episode was closed, and they were discharged from CAMHS without any further input. Of these, 6 did not reply to their “opt-in to treatment” letter, 4 were signposted to other services and 1 decided that this was not the right time to seek help.

Of the 29 who completed all 6 sessions, 16 (55.2%) required no further CAMHS input and were discharged from the service. Of the remaining 13 (44.8%), 3 were awaiting further CAMHS assessment and 10 were awaiting or had started other CAMHS face-to-face interventions.

Baseline symptomatology

Baseline data were obtained from 59 participants who were mostly female (34, 58%), of White British ethnicity (52, 88%) and of mean age 15.1 (sd = 1.3, range 13 to 17) years.

Initial levels of symptomatology on standardized measures of sleep, depression and anxiety are summarized in Table 1.

Table 1. Baseline scores and clinical severity of symptoms (N = 59)

Standardized Assessment	Baseline mean (sd)	Clinical Severity
Sleep Condition Indicator (SCI)	7.1 (3.6)	59 (100%) scored <17 suggesting significant insomnia
Insomnia Severity Scale (ISI)	18.2 (4.1)	47 (80%) scored ≥15 suggesting significant insomnia
Mood and feelings Questionnaire (MFQ)	37.4 (12.9)	49 (83%) scored ≥27 suggesting severe depression
Revised Anxiety and Depression Scale (RCADS)	72.3 (24.8)	36 (62%) had t-scores ≥65 suggesting borderline or clinical anxiety/depression

The cohort presented with high levels of mental health psychopathology with 83% presenting in the clinical range for severe depression and 62% in the range for borderline or clinical anxiety.

Post-Sleepio symptomatology

Post-CBTi assessments were completed by 43 participants (73%) with data on sleep efficiency and quality being obtained for an additional two (45). The mean number of days from completing baseline assessments to follow-up was 82.3 (sd = 34.3, range 35 to 210). Table 2 summarizes the comparison of mean outcome scores between pre- and post-intervention for the full sample.

Table 2. Comparison of pre- and post-intervention mean scores

	Pre- score		Post- score		Post – Pre
Standardized assessment	N	mean (SD)	N	mean (SD)	mean diff. (95% CI)	effect size (95% CI)	p value
Sleep Condition Indicator (SCI)	59	7.1 (3.6)	43	15.8 (8.1)	8.7 (6.5 to 11.0)	1.13 (0.84 to 1.42)	<0.001
Insomnia Severity Score (ISI)	59	18.2 (4.1)	43	10.8 (6.0)	−7.4 (−9.0 to −5.8)	−1.36 (−1.64 to −1.07)	<0.001
Mood and Feelings Questionnaire (MFQ)	58	37.4 (12.9)	43	28.1 (15.5)	−8.4 (−12.1 to −4.7)	−0.66 (−0.95 to −0.37)	<0.001
Revised Anxiety and Depression Scale (RCADS)
Total score	58	72.6 (25.1)	43	57.7 (26.1)	−13.5 (−19.3 to −7.6)	−0.67 (−0.96 to −0.38)	<0.001
Social Phobia subscale	58	17.4 (6.6)	43	14.3 (7.9)	−2.9 (−4.4 to −1.5)	−0.60 (−0.89 to −0.30)	<0.001
Panic subscale	58	12.2 (6.1)	43	9.8 (5.5)	−2.3 (−3.5 to −1.1)	−0.55 (−0.84 to −0.26)	<0.001
Depression subscale	58	17.9 (5.7)	43	13.4 (7.2)	−4.0 (−5.6 to −2.4)	−0.72 (−1.01 to −0.43)	<0.001
Separation Anxiety subscale	58	8.2 (4.4)	43	6.2 (4.2)	−1.7 (−2.8 to −0.6)	−0.47 (−0.76 to −0.18)	0.002
GAD subscale	58	9.7 (4.4)	43	7.8 (3.6)	−1.7 (−2.7 to −0.8)	−0.52 (−0.81 to −0.23)	<0.001
OCD subscale	58	7.3 (4.3)	43	6.3 (3.8)	−0.9 (−1.9 to −0.02)	−0.30 (−0.59 to −0.01)	0.04
Sleep Efficiency	49	49.0 (20.4)	45	70.9 (21.0)	21.9 (14.0 to 29.8)	0.80 (0.51 to 1.09)	<0.001
Sleep Quality	49	2.3 (1.7)	45	4.8 (2.6)	2.5 (1.6 to 3.4)	0.76 (0.48 to 1.03)	<0.001
Sleep onset latency	59	0.4 (0.9)	43	2.0 (1.8)	1.5 (1.0 to 2.1)	0.82 (0.53 to 1.11)	<0.001

Effect size is calculated as the mean difference between the pre- and post-intervention scores divided by the standard deviation of the change scores for participants that provided data at both pre- and post-intervention

There were improvements over time on standardized measures of sleep (SCI, ISI), sleep efficiency, sleep quality and sleep onset latency, and on standardized measures and sub-scales for anxiety (RCADS) and depression (MFQ). In terms of clinical severity, 36/43 (84%) did not meet criteria for insomnia on the ISI and 25/43 (38%) on the SCI. In terms of depression, 23/43 (54%) scored below the criteria for severe depression on the MFQ and 24/43 (56%) for clinical anxiety or depression on the RCADS.

Table 3 summarizes the comparison of pre-post change between those who did and did not complete at least half the Sleepio programme sessions.

Table 3. Comparison of change scores between participants who completed at least three of the six Sleepio sessions (“Completers”) and those who did not (“Non-Completers”)

	Completer		Non-Completer		Completer – Non-Completer
Standardized assessment	N	mean (SD)	N	mean (SD)	mean diff. (95% CI)	p value
Sleep Condition Indicator (SCI)	33	10 (7.8)	10	2.8 (4.2)	7.2 (2.5 to 11.9)	0.003
Insomnia Severity Score (ISI)	33	−8.2 (5.5)	10	−4.7 (4.7)	−4.2 (−7.7 to −0.7)	0.02
Mood and Feelings Questionnaire (MFQ)	33	−9.5 (12.0)	10	−1.7 (13.6)	−6.4 (−14.8 to 1.9)	0.13
Revised Anxiety and Depression Scale (RCADS)
Total score	33	−15.4 (18.6)	10	−4.6 (23.4)	−11.3 (−24.7 to 2.2)	0.10
Social Phobia subscale	33	−3.5 (4.8)	10	−0.9 (4.7)	−2.5 (−5.8 to 0.8)	0.13
Panic subscale	33	−2.8 (4.0)	10	−0.6 (4.5)	−2.3 (−5.1 to 0.5)	0.12
Depression subscale	33	−4.4 (5.6)	10	−1.3 (4.8)	−2.7 (−6.4 to 1.0)	0.15
Separation Anxiety subscale	33	−1.9 (2.8)	10	−0.3 (5.6)	−2.0 (−4.5 to 0.5)	0.11
GAD subscale	33	−1.8 (2.9)	10	−1.2 (4.5)	−0.8 (−3.1 to 1.4)	0.46
OCD subscale	33	−1.1 (2.9)	10	−0.3 (4.0)	−0.9 (−3.1 to 1.3)	0.42
Sleep Efficiency	34	25.2 (28.0)	11	12.9 (24.0)	10.8 (−7.5 to 29.1)	0.25
Sleep Quality	34	2.6 (3.1)	11	2.0 (3.8)	0.6 (−1.5 to 2.7)	0.60
Sleep onset latency	33	−2.1 (1.6)	10	0.3 (1.5)	2.5 (1.5 to 3.6)	<0.001

Sample size (N) for means and SDs is based on participants that provided data at both pre- and post-intervention.

Sample size for multilevel analyses (reporting mean difference, 95% CI and p value) is based on participants that provided data on at least one of the pre- or post-intervention assessments. This number was always 37 for the “Completers” and ranged from 12 to 21 for the “Non-Completers” across the outcomes.

There were improvements for people who completed at least three Sleepio session compared to those who did not on the SCI, ISI, and sleep onset latency but not on measures of depression or anxiety.

Variables associated with completing at least half the CBTi programme sessions

There was little evidence of association between completing at least half the CBTi sessions and gender, insomnia, anxiety, or depression severity but there was with age and activation time. Of those young people aged 16 years or over, 88% (22/25) completed 3 or more sessions compared with 41% (15/34) of those under 16 years (χ² = 11.86, df = 1, <.001). Similarly, those who went on to complete three or more sessions had activated Sleepio quicker [mean (SD; range) =4.0 (5.3; 0 to 26) days] than those who did not [mean (SD; range) =10.3 (11.1; 0 to 43) days] (t = 2.88, df = 55, 0.006).

Post-Sleepio acceptability ratings

Post-use acceptability ratings were undertaken by 39 of the 43 participants who completed follow-up assessments. Of these, 32 completed at least three of the programme sessions and 7 did not. Overall, Sleepio was rated very positively. Only 2 (5%) participants endorsed the “not true” option to the statement that Sleepio was easy to use. Similarly, only 6 (15.4%) endorsed the options to indicate that Sleepio was not helpful, 1 (2.6%), that it was not easy to understand with 3 (7.7%) not recommending Sleepio to a friend.

Participants were asked whether they preferred online Sleepio to a face-to-face intervention. Responses were divided with 14 (36%) indicating that they certainly did, whereas 9 (23%) expressed a clear preference for face-to-face interventions. The remainder did not express a definitive view.

Finally, participants were asked to rate changes in sleep, mental health, and overall satisfaction on a 1–10 scale, where higher scores indicate more positive ratings. The mean rating for improvements in sleep [mean (sd; range) =5.8 (2.7; 1 to 10)] and overall satisfaction [mean (sd; range) = 6.7,(2.4; 2 to 10)] with Sleepio were good. The mean rating for improvements in mental health [mean (sd; range) =4.5 (sd = 2.8; 1 to 10)] was lower.

Discussion

The first aim of this feasibility study was to determine the acceptability of digital CBTi as a first-step intervention for adolescents with newly referred mental health problems and insomnia to specialist mental health services (CAMHS). Acceptability as determined by intervention acceptance, programme initiation and completion rates was good. Intervention satisfaction was high with one-third expressing a clear preference for a digital, as opposed to a face-to-face, intervention. There were, nonetheless, approximately half who did not complete all six sessions of the intervention and a quarter who expressed a clear preference for a face-to-face intervention. In the absence of a comparison group, it is not possible to determine how engagement and drop-out rates for digital CBTi compare with those for CAMHS face-to-face interventions during the period of this study. However, drop-out rates in community CAMHS are high, with a systematic review of 30 studies undertaken in community CAMHS settings finding a mean drop-out rate of 50% (De Haan et al., 2013). Our results are comparable and suggest that digital CBTi as a first-step intervention for those with mental health problems and insomnia may be acceptable to new referrals to CAMHS.

Our second aim was to undertake an exploratory analysis of the effect of digital CBTi on sleep and mental health. Post-improvements were found on standardized measures of sleep, anxiety, and depression and ratings of sleep efficiency, sleep quality and sleep onset latency. The effect sizes for sleep outcomes were large with medium effects for anxiety and depression. These findings are consistent with those obtained with adults where online CBTi resulted in improvements in sleep, anxiety and depression (Cheng et al., 2019; Christensen et al., 2016; Freeman et al., 2017; Ye et al., 2015). Although reductions in symptoms of anxiety and depression on standardized measures were large, self-report subjective ratings of mental health improvement were more modest. Whether symptom reductions have not translated into improved functioning or whether adolescents had higher post-intervention expectations is unclear. A clearer and more explicit rationale for why treating sleep might improve mental health and the use of more personalized goal-based outcomes that assess changes in areas of everyday life that are particularly important to the young person may be helpful (Wolpert et al., 2012). Nonetheless, these results are encouraging and suggest that the adoption of an online CBTi pathway within specialist CAMHS for those with insomnia might offer a novel way to improve adolescent mental health.

In terms of our third aim, to assess subsequent CAMHS interventions, over half of those who completed 3 or more sessions of digital CBTi were discharged without any further input. Given the limited capacity of CAMHS and the rising demand for mental health services, the use of digital first-step interventions might offer an efficient and scalable way of increasing access and alleviating some of this pressure on under-resourced services. Mental health therapists could then focus their specialist skills at the later stages of the pathway, on those with more complex and enduring problems or those who have not responded to first-step interventions. Similarly, our finding that all of those who accepted but never activated digital CBTi (n = 11) were subsequently discharged from CAMHS without any further input, may have implications for service providers. In the UK, around 11% of CAMHS appointments are not attended (CAMHS benchmarking report, 2015). Our study is very small and needs replication, but offering digital CBTi as a first-step intervention for those with insomnia and mental health problems might offer a way of testing motivation, engagement, and commitment.

Although Sleepio can be used as an unguided intervention, we offered brief, weekly telephone support. This was provided by an assistant psychologist, not a trained mental health professional, who received half a day of training about sleep and the digital intervention. The aim of support was to maintain engagement and to facilitate implementation of programme content to the adolescent’s specific situation. The mean length of programme support per participant was 32 minutes, comparable to the 38 minutes reported in a previous study (Cliffe et al., 2020). Whilst we chose to use telephone support, other media have been used to deliver support for internet-delivered interventions such as text messages (Ritvo et al., 2021), e-mails (Paxling et al., 2013) or online chat (De Bruin & Meijer, 2017). We did not record unanswered calls and did not assess whether phone calls offer the most acceptable or efficient media for supporting adolescents. Although the use of non-mental health specialists may provide a low cost, viable way of providing this, this would warrant further investigation.

Rates of engagement and programme completion with this guided online intervention were fairly good. 84% (59/70) of participants who elected to use Sleepio activated their account, 63% (37/59) completed 3 or more sessions with 49% (29/59) completing all 6 sessions. This compares favorably with other studies evaluating Sleepio as an unguided intervention. For example, a community study by Freeman et al. (2017) found that 68.9% of adults activated their account although only 17.5% completed the programme. Similarly, Stott et al. (2021) offered unguided Sleepio as an adjunct intervention to adults attending a mental health service and found that 35.7% completed 2 or more sessions. With adolescents, Werner-Seidler et al. (2019) in a community study with volunteers found that whilst 78.7% activated the unguided online Sleep Ninja programme, only 33% completed all sessions. Our results are encouraging and lead us to speculate that some limited guidance may enhance engagement and maintain motivation. This may be particularly important for younger adolescents who may be less motivated or less likely to take ownership of their sleep problems.

If these findings were replicated in a randomized controlled trial, they would have important clinical implications. Firstly, the online treatment of sleep for those with insomnia may offer an alternative and acceptable way of improving adolescent sleep and mental health. Secondly, the use of an automated programme potentially offers a scalable low-cost intervention. Whilst support was offered, this was limited and provided by non-mental health specialists (psychology assistants). Thirdly, access to online CBTi was rapid, thereby reducing the lengthy waiting times for traditional child and adolescent mental health services. Finally, the introduction of online CBTi within the CAMHS referral pathway offers patients greater choice. One third of those participating in this study preferring an online intervention to a face-to-face intervention.

Limitations

Whilst these findings are encouraging our study has a number of limitations. Firstly, this is a small single-arm feasibility study with no randomly allocated comparison group to control for the passage of time on mental health and sleep. However, whilst the observed improvements in outcomes could have occurred naturally, the scale of change we report does suggest that there could be an intervention effect. Indeed, the change on the RCADS we found is of a similar magnitude to that obtained following face-to-face CAMHS interventions within this clinical setting (Gibbons et al., 2021)

Secondly, whilst the majority of eligible participants agreed to participate, follow-up data were only available for 43/59 (73%) of participants. The majority of those we were unable to interview had dropped out of CBTi and so their views may be less favorable than those we report here. Finally, whilst we were able to determine immediate post-CBTi outcomes for those who did and did not complete at least half the programme, we do not have any data on subsequent CAMHS re-referral rates. Whether post-intervention improvements in sleep and mental health are maintained is not known.

Conclusion

This is the first study to report the acceptability and explore the potential benefits and effects of a new first-step digital CBTi pathway for adolescents with insomnia referred to specialist CAMHS. Adolescents participating in CBTi reported large improvements in sleep and mental health with half of those who completed all six sessions not requiring any further CAMHS interventions. A fully powered randomized controlled trial is now required to obtain definitive estimates of the effects of digital CBTi as a first-step intervention on adolescent sleep and mental health.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was funded by the NHSX. Obi Ukoumunne was supported by the National Institute for Health Research (NIHR) Applied Research Collaboration (ARC) South West Peninsula. The views expressed are those of the author(s) and not necessarily those of the NHS 14943, the NIHR or the Department of Health and Social Care.