Effects of talk therapy and bright light therapy on insomnia in young people who stay up late: study plan for a controlled trial

Insomnia is the most common sleep complaint in youths, with a prevalence rate of 7.8–23.8% for insomnia disorder [1–6] and up to 40% for insomnia symptoms [2]. Youth insomnia tends to run a chronic course and persist into adulthood [7, 8]. Fernandez-Mendoza and colleagues followed a group of children (n = 502, median age of 9 at baseline) for 15 years and found that children with insomnia symptoms continued to have sleep problems in adolescence (~ 54%; median age, 16) and young adulthood (~ 62%; median age, 24), respectively [8]. If left untreated, insomnia may be associated with long-term adverse outcomes, such as an increased risk for physical and mental health problems [9], incidences of substance use [6], impaired daytime functioning, and increased school absenteeism [10].

Another significant change during adolescence is a gradual shift in circadian preference towards eveningness. This increased evening preference is intrinsically associated with pubertal development [11, 12], during which peak eveningness is reached at the end of adolescence [13]. Eveningness in youths is often associated with a constellation of adverse outcomes, including an increased risk for physical [14], mental [15–17], and sleep problems [18]. It is also frequently implicated in insomnia symptoms in adolescents, with those having an eveningness tendency having up to a 2.5-fold risk of insomnia problems. Existing research suggested that eveningness is a unique contributing factor to insomnia in adolescents, where a misaligned circadian rhythm potentially contributes to the aetiology of insomnia and implicates the phenomenology of insomnia, particularly with sleep-onset insomnia [19, 20]. Moreover, increasing research has demonstrated the potential synergistic effects of eveningness and insomnia on daytime sleepiness and mental health outcomes, such as depressive and anxiety symptoms. Previous studies showed that adolescents with both eveningness and insomnia were 1.3–4.9 times more likely (compared to insomnia alone) and 2.8–6.9 times more likely (compared to eveningness alone) to develop mental health problems [17, 21]. While both insomnia and eveningness are linked to a wide range of adverse outcomes in youths, the research on the treatment for this vulnerable group has been limited to date. The intricate relationship between insomnia and eveningness and their associated consequences underscore the need for concurrently addressing sleep and circadian issues in youths.

Although cognitive behavioural therapy for insomnia (CBT-I) is recommended as the first-line treatment for insomnia in adults with substantial empirical support [22, 23], research on the efficacy of CBT-I in adolescents has been limited. Åslund et al. (2018) conducted a meta-analysis on the effect of CBT-I in school-age children and adolescents and found only six RCTs (528 pooled samples, mean age 14.6 years) which showed a small-to-large magnitude of improvements in sleep parameters, specifically in sleep onset latency (SOL) [24]. Another meta-analysis by Blake et al. (2017) on adolescent cognitive behavioural sleep interventions included 9 trials (357 pooled samples, mean age 15.0 years) and found that cognitive and behavioural interventions improved not only various subjectively and objectively measured sleep parameters, but also daytime sleepiness, depression, and anxiety severity [25]. Despite some findings supporting the efficacy of CBT-I in youths, it was observed that up to 25–40% of young people treated with CBT-I showed minimal treatment response or failed to achieve a remission [26], suggesting the need to consider and/or integrate alternative strategies to enhance treatment outcomes.

The Transdiagnostic Sleep and Circadian Intervention (TranS-C) is a programme that combined conventional therapeutic components from CBT-I with additional circadian-related components to address both sleep and circadian disturbances [27]. A series of studies have provided support for the TranS-C programme as a promising intervention for improving sleep and advancing the circadian phase in youths [28–30]. Specifically, in a study comparing the efficacy of TranS-C versus psychoeducation in 176 youths aged 10 to 18, TranS-C was found to result in significantly greater improvements in self-reported sleep and circadian measures at post-intervention [28]. In addition to the cognitive-behavioural approach, the use of chronobiotics has gained increasing attention as a therapeutic approach in recent years. Light is the most potent zeitgeber that entrains the endogenous circadian clock to the surrounding environment. Under controlled laboratory conditions, light has been shown to be effective in shifting the human circadian rhythm and subsequently used as a therapeutic intervention [31, 32]. In adults, bright light therapy (BLT) has been shown to improve circadian outcomes, insomnia symptoms, and depressive symptoms [33, 34]. However, the evidence of BLT in the youth population remains relatively limited. A study conducted by Gradisar et al. (2011) in adolescents with delayed sleep phase disorder, a clinical sleep disorder reflecting the extreme end of eveningness, showed moderate to large effects of CBT-I with adjunctive BLT in reducing sleep onset latency (d = 0.65–1.13), advancing bedtime and rise time (d = 0.89–1.24), and improving daytime functioning (d = 0.75–0.79) [35]. Nonetheless, previous research utilising light therapy in adolescents was mainly conducted in either mixed clinical samples [36] or those with delayed sleep phase disorder [35, 37]. It remains unclear how to address circadian issues in the context of youth insomnia, specifically whether CBT-I alone is sufficient to tackle eveningness in youths diagnosed with insomnia or if an adjunctive circadian-focused intervention, i.e. bright light therapy, could additionally address the circadian components of youth insomnia, thereby enhancing the treatment outcomes.

The primary outcome is the change in insomnia symptoms as measured by the Insomnia Severity Index (ISI). The ISI is a seven-item self-rated scale designed to measure one’s insomnia problems, satisfaction with sleep, noticeability and worries towards one’s sleep problems, and insomnia-related interference with daily life. The possible score ranges from 0 to 28, where a higher score indicates greater insomnia severity. The ISI has been validated in local youths demonstrating good construct validity and high internal consistency with a Cronbach’s alpha of 0.83 [40].

The secondary outcomes include changes in subjective sleep quality as measured by the Pittsburgh Sleep Quality Index (PSQI) [44], self-report chronotype as measured by the Munch Chronotype Questionnaire (MCTQ) [45], clinician-rated depressive symptoms as measured by the Children’s Depression Rating Scale (CDRS) [46], self-report mood symptoms as measured by the Hospital Anxiety and Depression Scale (HADS) [47], and suicidal ideation as measured by the Depressive Symptoms Inventory Suicidality Subscale (DSI-SS) [48]. Daytime functioning is subjectively assessed by the Paediatric Daytime Sleepiness Scale (PDSS) for daytime sleepiness and the Multidimensional Fatigue Inventory (MFI) for daytime fatigue [49, 50]. Secondary outcomes also include a change in the quality of life as measured by the KIDSCREEN-27 and the overall severity of clinical symptoms assessed by clinicians using the Clinical Global Impression Scale [51].

Other sleep-related outcomes will be evaluated subjectively using a sleep diary and objectively using actigraphy. Participants will be asked to complete a 7-day sleep diary while wearing a wrist actigraphy (Actiwatch Spectrum Plus, Philips Respironics). The outcomes to be analysed from the sleep diary and actigraphy are sleep onset time, wake time, total time in bed (TIB), total sleep time (TST), sleep onset latency (SOL), wake after sleep onset (WASO), and sleep efficiency (SE) across the recording period. Epoch-by-epoch actigraphy data will be extracted using cosinor and nonparametric analysis for circadian rhythm computation [52, 53]. Circadian parameters derived from cosinor analysis include the acrophase, amplitude, and MESOR, while circadian parameters derived from the nonparametric analysis include interdaily stability, intradaily variability, relative amplitude, M10 onset (10 h of maximal activity), and L5 onset (5 h of the least activity). Table 2 summarises the circadian parameters from the actigraphy computation.

Participants will also be invited to complete a series of computerised neurocognitive tasks to assess objective cognitive performance. The list of computerised neurocognitive tasks is summarised in Table 3.

Due to limited funding resources, only half of the participants in the two treatment arms will be invited to complete a two-night in-laboratory assessment at baseline and post-treatment to objectively assess one’s sleep and circadian phase. Data collection will start after more than 50% of participants have been enrolled. The first night will serve as the adaptation night, and dim-light melatonin onset (DLMO) assessment will be conducted according to the previously published protocol [54]. In brief, participants will be asked to provide a total of 17 salivary samples in 30-min intervals starting 6 h before and continuing for 2 h after the habitual bedtime under a dim light condition with the ambient light level maintained below 10 lx. The salivary melatonin specimens will then be assayed using ultra-performance liquid chromatography-tandem mass spectrometry. The clock time of DLMO, calculated with linear interpolation when the melatonin concentration exceeded and remained above an absolute threshold of 12.9 pmol/L for at least two additional samples, will be used to reflect the circadian phase [55]. On the second laboratory night, participants will complete a level I polysomnography (PSG; Grael PSG Amplifiers and Recorders, Compumedics). Electrodes and sensors will be placed according to the American Academy of Sleep Medicine Manual for Scoring of Sleep and Associated Events [56]. PSG signals will be acquired and analysed using Compumedics Profusion SLEEP4 and stored securely at the Sleep Research Clinic and Laboratory at the University of Hong Kong. PSG recordings will then be scored by registered polysomnographic technologists who are blinded from the group allocations. Polysomnography-based sleep parameters that are considered as the secondary outcomes include total time in bed (PSG-TIB), total sleep time (PSG-TST), sleep onset latency (PSG-SOL), wake after sleep onset (PSG-WASO), and sleep efficiency (PSG-SE).

The CBT-I programme consists of six weekly sessions, 90 min each. The sessions will be conducted in a group format (four to six participants per group) and facilitated by trained therapists. The group sessions are structured and adopted from a well-established protocol for treating insomnia with additional emphasis on the effect of the circadian rhythm [57]. The programme aims to address the behavioural, cognitive, and physiological factors perpetuating insomnia. Table 4 summarises the session content of the CBT-I programme.

Participants in both intervention arms will be instructed to wear a portable light therapy device for 30 min daily immediately after awakening (Re-Timer; https://www.re-timer.com↗). Re-Timer is equipped with four light-emitting diodes (LEDs; two on each side) emitting a constant green light (per manufacturer’s manual, 500 nm dominant wavelength at 506 lx). The efficacy and safety of the Re-Timer have been reported by previous research [31, 58, 59]. The CBTI + BLT group receiving an active BLT treatment will be provided with the original Re-Timer, whereas the CBTI-BLT group receiving the placebo BLT treatment will be provided with a modified Re-Timer. The modified Re-Timer has red cellophane filters and red colour cardboard paper (250 g/m²) covering the LEDs, which turns the visual light spectrum towards red colour and blocks most of the light emitted from the LEDs (< 10 lx). This level of the light emitted from the modified Re-Timer is considered insufficient to have a phase-shifting effect on the circadian rhythm [60]. During each group session, the therapist will individually review and prescribe timing for light therapy for each participant based on their sleep diary completed in the previous week, aiming to advance one’s rise time and gradually optimise the sleep schedule. The daily sleep diary will be used to record participants’ usage of the Re-Timer to analyse adherence to BLT intervention.

All intervention sessions will be delivered by clinical psychologists or doctoral students who have received extensive training in behavioural sleep medicine and circadian science. All therapists will meet regularly during the study to discuss specific issues regarding the implementation of the trial protocol and to ensure the consistency of the implementation of the treatment protocol. All therapy sessions will be video recorded, and a subset of the recordings (20%) will be randomly selected to be reviewed and closely scrutinised by the blind judges based on a list of quality checkpoints.

The likelihood of having serious adverse events during this trial is expected to be minimal. Both CBT-I and light therapy are generally well-tolerated. A systematic review of 99 studies on the randomised controlled trials of CBT-I showed no serious adverse events attributed to CBT-I [61]. Similarly, a review of 43 studies on bright light therapy showed no evidence of ocular damage due to light therapy [62]. Nonetheless, there have been reports of an increased risk of hypomania (risk ratio = 4.91, 95% CI 1.66 to 14.46), agitation, headache, blurred vision, and eye irritation in the use of bright light therapy [63]. As such, the HADS and an adverse effects checklist [64] will be used biweekly during treatment (i.e. Week 2 and Week 4) and follow-up assessment to monitor potential mood changes, the frequency of adverse events, and the related daytime impairment severity. In addition, the research clinician will evaluate any suicidal thoughts and behaviour, as well as non-suicidal self-injury, during the intervention period. Participants will be removed from the study and referred for open treatment in the case of serious adverse events (e.g. acute suicide risk or attempt).

Treatment compliance and session evaluation will be assessed by the Treatment Satisfaction Scale (TSS) and Treatment Component Adherence Scale (TCAS), both of which have been used in previous trials [65]. The daily sleep diary will be used to examine the adherence to bright light therapy, where participants will be asked to record how long and at what times they used Re-Timers each day. The six-item Credibility-Expectancy Questionnaire (CEQ) will be used to assess treatment expectancy and credibility of the intervention [66].

The target sample size was estimated based on the primary hypothesis using a priori power analysis. We expected a medium effect size (Hedges’ g = 0.30–0.47) based on the previous research [33]. Using g = 0.30 as a more conservative effect size, a total of 111 participants was estimated assuming a significance of 0.05 and an 80% power. Accounting for a 30% anticipated attrition rate, the final target sample size for the overall trial was 144 (n = 48 per arm).

The main analyses will be conducted following the intention-to-treat principle. Multilevel modelling (MLM) with restricted maximum likelihood estimation and an unstructured variance–covariance matrix will be used to analyse the main hypotheses on outcome measures. The use of MLM in analysing RCT data has the advantage of accounting for individual differences as a random effect [67–69]. MLM also has advantages over the traditional general linear model approach (e.g. mixed model repeated measures ANOVA) as it is unaffected by randomly missing data and can handle both time-dependent and independent covariates [67, 69]. Missing data for dependent variables will be assumed to be missing at random and accounted for by the maximum likelihood estimation. Two sets of MLMs will be conducted; one will include three treatment arms (CBTI + BLT vs CBTI-BLT vs WL) and two timepoints (pre-treatment vs post-treatment) as fixed effects; the second set will consist of two treatment arms (CBTI + BLT vs CBTI-BLT) and four timepoints (pre-treatment vs post-treatment vs 1-month follow-up vs 6-month follow-up) as fixed effects. The treatment × time interaction will also be included as a fixed effect on both MLMs and will be considered the treatment effect, while repeated measures will be nested within individual participants as random intercepts. A p-value of 0.05 represents statistical significance in all analyses. The p-values on outcome measures will be adjusted using the Benjamini–Hochberg procedure with a false discovery rate set as 20% to account for multiple comparisons.

For analysing the mediating effect of insomnia on the changes in mood and other daytime symptoms, the mediator (i.e. insomnia) will be added to the multilevel models for each of the primary and secondary outcomes (described above). An interim analysis is planned to be conducted when more than half of the participants have completed the 6-month follow-up assessment. The principal investigator (PI; SXL) is responsible for reviewing the interim analysis results for efficacy and futility purposes.