Telephone-Guided Sleep Restriction for Insomnia: A Randomized Sleep Diary-Controlled Trial

Insomnia is a chronic problem involving difficulty falling asleep, staying asleep, and/or waking up too early, leading to daytime complaints such as fatigue, concentration problems, and impaired functioning [1]. Insomnia affects approximately 1 in 10 individuals and is linked to high societal costs and mental and physical health problems, such as depression, anxiety, and cardiovascular disease [2]. The first-line treatment for insomnia, cognitive behavioral therapy for insomnia (CBT-I), produces robust, long-term improvements in sleep, quality of life, and depressive symptoms [3‒5]. Despite its proven effectiveness and clear recommended status, most people with insomnia do not receive CBT-I [6].

Relative to the high prevalence of insomnia, the treatment capacity of CBT-I is limited [7]. CBT-I is a multicomponent treatment provided by trained therapists [6]. Currently, the number of trained CBT-I therapists is insufficient to meet the high demand, leading to access problems [8]. Instead of CBT-I, patients often receive more readily available but less effective treatment options such as sleep medication or basic sleep hygiene education [5, 9, 10]. As a result, investigating scalable formats that can improve CBT-I’s capacity remains a high priority in the field [7].

One way to improve CBT-I’s scalability is to explore remote formats that require less therapist involvement. For instance, digital forms of CBT-I can achieve insomnia improvements with fewer logistic barriers and lower costs [11]. However, digital forms of CBT-I often encounter problems with dropout [12, 13], and in general, CBT-I is most effective when therapist guidance is provided [12, 14]. Grasping the various CBT-I components remotely, with no or minimal therapist support, may be quite difficult for patients. Simplifying the content of CBT-I to match a more minimal therapist role may benefit such low-intensity solutions.

Promising in this regard is to focus on one core component of CBT-I: sleep restriction therapy (SRT). SRT is a behavioral intervention in which patients restrict their time in bed (TIB) to the actual time they spend asleep, based on averages from a personal sleep diary. This results in new bed and rise times, which patients consistently maintain, while avoiding safety behaviors like daytime napping [15]. SRT operates by limiting and regularizing the opportunity for sleep, which is hypothesized to consolidate sleep by increasing sleep pressure, decreasing arousal, and strengthening the circadian sleep-wake cycle [16, 17]. When sleep efficiency (SE) – referring to the percentage of the TIB spent asleep – increases during SRT, bedtimes are gradually extended again.

Multiple trials have shown that SRT is effective as a standalone treatment and leads to improvements in insomnia, SE, sleep onset latency (SOL), wake time after sleep onset (WASO), and depressive symptoms, with effects in the range of multicomponent CBT-I [18‒20]. Importantly, SRT has a short treatment duration and straightforward protocol requiring minimal therapist training, making SRT a potentially scalable treatment solution. This scalability potential was recently demonstrated in a large trial by Kyle and colleagues [21], who found that nurse-delivered SRT had a 95.3% probability of being cost-effective.

Building on the scalability potential of SRT, we propose delivering SRT in a remote telephone-guided format. A telephone-guided format improves accessibility by removing the need for travel time, office space, and geographical proximity to providers. Also, as patients often experience SRT as challenging, repeated therapist support calls may help maintain adherence [22, 23]. The remote format, never seeing the therapist in person, also offers a level of anonymity that may encourage patients to disclose treatment difficulties [24]. While multicomponent telephone-guided CBT-I has shown comparable efficacy to in-person CBT-I [13], SRT has only been studied with a mix of in-person sessions followed by telephone sessions (e.g., [21, 25, 26]). The efficacy of a fully remote telephone-guided SRT approach has yet to be examined.

To investigate this, we conducted a randomized controlled trial to compare telephone-guided SRT to a sleep diary control group. We aimed to see whether, in line with our hypotheses, telephone-guided SRT was superior to the control group in improving insomnia severity (primary outcome), sleep diary measures, depressive symptoms, anxiety symptoms, presleep arousal, sleep-safety behaviors, daytime sleepiness, and dysfunctional sleep-related cognitions (secondary outcomes). We measured the outcomes at baseline and posttest after 6 weeks. Participants in the SRT group repeated the outcome assessments after 3 and 6 months, while the control group received treatment after the posttest, without further assessments. Throughout the study, all participants completed daily sleep diaries.

We recruited participants through advertisements on the study website (www.insomnie.nl) and social media (Instagram and Facebook). Inclusion criteria were: (1) Insomnia Severity Index (ISI) score above the clinical cutoff for insomnia (ISI ≥10 [27]), (2) self-reported insomnia complaints, i.e., participants were awake for ≥30 min per night, for ≥3 nights per week, for ≥3 months, and experienced daytime problems related to sleep complaints, (3) SE <85% based on self-reported average sleeping times and sleep averages from a 7-day baseline sleep diary, (4) Dutch-speaking, and (5) age of ≥18 years.

We excluded participants who: (1) had received CBT-I in the past 12 months, (2) were currently in psychological treatment (started <6 months ago) or were awaiting psychological treatment, (3) had changed their psychoactive medication in the last 6 weeks, (4) reported a diagnosis of psychosis or schizophrenia, (5) reported severe depressive complaints or concrete suicidal ideation (total score on Beck Depression Inventory-II (BDI-II) ≥29, BDI-II item 9 ≥2, respectively) [28], (6) were pregnant or breastfeeding, (7) worked night shifts, or (8) had no access to internet to fill out questionnaires. Sleep medication use was allowed. Participants received no financial compensation.

All data were collected online, using Qualtrics surveys [29]. An online screener checked most of the inclusion and exclusion criteria and contained informed consent. Eligible participants who provided informed consent completed a 7-day baseline sleep diary and received a phone call to explain the study procedures. All sleep diaries during the study were completed online and programmed within Qualtrics [29]. Each morning, participants received an email invitation to complete the sleep diary, followed by up to one reminder email. Participants with an SE of >85% were excluded. All remaining participants were randomized and called again to confirm their allocated group. For the SRT group, this call continued with the first SRT session (see “Intervention”).

Participants were informed that they could withdraw from the study at any time. Participants who discontinued treatment were asked to continue the assessments, if possible. Responses on assessments were encouraged with up to three reminder emails.

This trial was approved by the internal Ethical Review Board of the University of Amsterdam (2022-CP-15342) and registered at clinicaltrials.gov (NCT05548907). Reporting follows the CONSORT guidelines (see the online suppl. material, available at https://doi.org/10.1159/000545138).

We aimed to recruit at least 120 participants for 80% power to detect a medium effect (d = 0.50, two-sided t-test, α = 0.05 [30]). We initially recruited 108 participants and added a second recruitment period to reach the target size. We used our full treatment capacity to include 39 additional participants, leading to a final sample size of 147.

Random allocation was programmed with Qualtrics [29], using a 1:1 allocation ratio in blocks of six. Due to a programming error midway through the second recruitment period, 3 participants who did not meet the sleep diary SE criterion were mistakenly included and assigned to the SRT group, creating a slight imbalance. We adjusted the allocation ratio to 3:2 to correct this, with blocks of five (with 3 participants assigned to SRT and two to control) for the last 20 participants.

Participants were allocated after the baseline sleep diary week, at the end of the pretest. After randomization, a message showed participants their group. Researchers were notified of the allocated group via an automatic email. Study personnel and coaches were not blinded to the group assignments.

Every week a short online assignment formed the basis for the coaching session. This assignment showed participants their last week’s sleep diary averages (i.e., bed- and wake time, TIB, total sleep time [TST], total wake time [TWT], and SE). Participants used this information and the restriction guidelines to suggest their coming week’s sleep window (e.g., new TIB, bedtime, and rise time), possible activities for the extra time spent awake, and any questions or comments to discuss with their coach.

The SRT included seven 10–15-min telephone sessions. In the first session, the coach explained the treatment rationale, answered questions, and reviewed the proposed sleep window from the online assignment. If participants deviated from the guidelines, the coach encouraged participants to persevere through initial discomfort for better results. The position of the sleep window (i.e., restriction at night, in the morning, or a combination) was based on patient preference. Sessions 2–6 followed a similar structure: discussing progress, providing encouragement and support for challenges (e.g., fatigue), and adjusting the sleep window. The final session focused on reviewing progress and discussing future maintenance. Participants were encouraged to avoid daytime napping and maintain consistent bed and rise times, including on weekends. The protocol did not include formal stimulus control, but coaches were allowed to suggest briefly getting out of bed if participants described lying wide awake for long periods.

Coaches (n = 10) were recruited from the Department of Clinical Psychology at the University of Amsterdam. We recruited both staff members and graduate students. Of the SRT group, 60% of participants were coached by staff members and 40% by graduate students. Coaches received brief training in the SRT protocol during a 2-h session guided by the last author (JL), a certified psychologist and sleep expert. Coaches also attended weekly 1-h supervision sessions, led by JL, to discuss sleep diaries and treatment progress.

The sleep diary control group kept a sleep diary for 7 weeks and completed the outcome assessments. Like the SRT group, the control group received automatic weekly messages summarizing their sleep diary averages. After the posttest, the control group received an online CBT-I workbook called iSleep [31], with the option to receive telephone-guided sleep restriction.

The primary outcome was insomnia severity, measured with the ISI [32]. The ISI includes 7 items on sleeping difficulties and experienced consequences, rated on a 5-point scale (total score range: 0–28) with higher scores reflecting more severe insomnia. The ISI has demonstrated adequate reliability and validity as an online questionnaire [32, 33]. An ISI reduction of eight or more is considered a clinically relevant change, and a posttreatment ISI score below eight full remission [27].

Sleep-related arousal was measured with the Pre-Sleep Arousal Scale (PSAS) [34]. The PSAS includes 16 items scored on a 5-point scale (total score range: 16–80), with higher scores indicating more arousal. The PSAS has shown good internal consistency and convergent and discriminant validity in a community sample [35].

Sleep-related safety behavior was measured with the Sleep-Related Behaviors Questionnaire (SRBQ) [36]. The SRBQ includes 32 items scored on a 5-point scale (total score range: 0–128), with higher scores indicating more safety behaviors. The SRBQ has shown good internal consistency and can discriminate normal sleepers from those with insomnia [36].

Sleep-related dysfunctional cognitions were measured with the Dysfunctional Beliefs and Attitudes about Sleep Scale short form (DBAS-16) [37], which includes 16 items rated on an 11-point agreement scale (total score range: 0–160), with higher scores indicating more dysfunctional cognitions. The DBAS-16 has good psychometric properties [38].

Daytime sleepiness was measured with the Epworth Sleepiness Scale (ESS) [39]. In the ESS the respondent rates 8 items on the likelihood of dozing off during various daytime situations (4-point scale, total score range: 0–24). Higher scores indicate more daytime sleepiness. The ESS has adequate internal consistency [40].

Anxiety symptoms were measured with the Hospital Anxiety and Depression Scale (HADS-A) [41]. The HADS-A consists of 7 items rated on a 4-point scale, with higher scores indicating more anxiety (total score range: 0–21). The HADS-A has shown good reliability and validity as a symptom screener [42].

Depressive symptoms were measured with the Patient Health Questionnaire (PHQ-9) depression scale [43], in which 9 items are scored on a 4-point scale (total score range: 0–27), with higher scores indicating more depression. The PHQ-9 has shown strong validity and reliability [43].

We used questions from the Consensus Sleep Diary (CSD) [44] to measure weekly averages of TST (min), SOL (min), WASO (min), TWT (min), TIB (min), rise time and bedtime variability (RTv and BTv, min), and SE (%).

At the posttest, the SRT group rated their satisfaction with the therapy (7-point scale from very unsatisfied to very satisfied, with room to describe), whether they would recommend the treatment to others (yes/maybe/no, with room to describe), the support provided by their coach (grade from 1 to 10), and whether they missed anything during this support (room to describe).

The posttest and follow-ups also contained items on uptake of other sleep or psychological treatment during the trial (yes/no, room to describe), use of sleep medication during the trial (average days used per week, with room to describe), use and change in other psychoactive medication (yes/no, room to describe), and adverse events (yes/no, room to describe) on (1) unpleasant events related to fatigue or sleepiness, (2) falling, and (3) accidents [22]. In case an adverse event was reported, this was followed by items on severity (7-point scale from not severe at all to very severe), experienced negative consequences (yes/no), and hospitalization (yes/no).

Treatment effects on primary and secondary outcomes were analyzed based on the intention-to-treat principle. Estimated marginal means were calculated with linear mixed-effect regression models (LMMs), using R packages lme4 [45] and emmeans [46]. Between-group effects at posttest were based on models that included predictors of time (pre vs post), group (control vs SRT), and a time*group interaction, including a random intercept. Within-group effects from pre to follow-ups for the SRT group were based on models with two separate time predictors indicating the follow-up moment (pre vs FU1; pre vs FU2), including a random intercept. The LMM analyses used an unstructured covariance structure and fitting was based on maximum likelihood. All analyses used two-sided testing with an alpha of 0.05.

We used Chi-squared tests to examine differences in the frequency of adverse events between the SRT and control group at posttest. Similarly, we used Chi-squared tests to examine group differences in dropout frequency, defined for the SRT group as dropout during the SRT intervention and for the control group as dropout from completing the sleep diary. Additionally, we used t-tests to evaluate whether participants who dropped out of SRT or the sleep diary had different pretest scores than participants who did not. For nonnormally distributed pretest scores, we used the Mann-Whitney U tests instead of t-tests.

Rise time and bedtime variability were calculated as the standard deviation of the mean rise time and bedtime per participant per assessed sleep diary week. We defined outliers as baseline outcome scores greater than 3.29 standard deviations from the mean. There were 14 outlier scores across all baseline variables, which stemmed from 7 different participants and concerned the baseline variables PHQ score (1), SE (3), TST (1), SOL (2), WASO (2), TWT (2), RTv (1), and BTv (2). All reported results in the main article are based on the full dataset. The online supplementary material contains the results of the main analysis without outliers, where for participants with a baseline outlier on a specific variable, scores for that variable at all time points were excluded.

The proportional group differences of participants achieving a clinical response (ΔISI ≥8) or clinical remission (ISI post <8) were analyzed with Chi-squared tests. We calculated between-group Cohen’s d effect sizes with the mean pre-post change in the SRT group minus the mean pre-post change in the control group, divided by the pooled pretest standard deviation [47]. We calculated within-group Cohen’s d effect sizes with the pre-post change (within the SRT or the control group), pre-FU1, or pre-FU2 change (within the SRT group), divided by the respective group’s pretest standard deviation. All effect size calculations were based on the LMM-estimated mean scores and estimated standard deviations (with the SD calculated from the model’s $SE*n$⁠).

Recruitment took place in September 2022 and March 2023. The last follow-up was completed in November 2023. Of the 406 screened participants, we randomized 147 to either the sleep diary control group (n = 71) or the SRT group (n = 76). See Figure 1 for the reasons for attrition and the full participant flow. In the sleep diary control group, 65 participants (91.5%) provided data at posttest. In the SRT group, 62 (81.6%) participants provided data at posttest, 60 (78.9%) at the 3-month follow-up, and 45 (59.2%) at the 6-month follow-up.

The final sample included 147 adults (112 females, 76.2%), with an age range of 19–74 years (M = 46.78, SD = 14.51). Of the sample, 89.8% were born in the Netherlands. In the past month, 38.1% had taken sleep medication and 10.2% other psychoactive medication. The mean BDI-II score was 16.22 (SD = 6.76, range: 2–28), consistent with a mild to moderate level of depressive complaints. The mean baseline ISI score was 18.51 (SD = 3.35, range: 10–18). Table 1 shows the demographic characteristics by group. Demographic characteristics did not significantly differ between the groups at baseline.

The seven SRT sessions took place within a 6-week timeframe. We defined treatment dropout as informing study staff of a decision to withdraw from treatment or withdrawing without notice (i.e., loss of contact despite repeated contact attempts). We defined treatment completion as attending at least 5 sessions, meaning that missing up to two sessions was allowed (e.g., due to illness) as long as participants did not withdraw from treatment. The SRT treatment was completed by 57 (75%) of the SRT participants. For treatment completers, the mean total duration of all telephone sessions was 74.64 min (SD = 29.89, range: 33–136), with a mean number of 6.37 sessions (SD = 0.79, range: 5–7) lasting on average 11.4 min per session (SD = 4.26, range: 5–25). On average, treatment completers filled out 5.07 (72.4%) of the 7 weekly online assignments.

Of the SRT participants, 19 (25%) dropped out of treatment. The most common reasons for dropout during SRT were physical illness (n = 4) or experiencing SRT as too challenging (n = 3). For treatment dropouts who completed at least the first session (n = 15), the mean total duration of all telephone sessions was 20.67 min (SD = 6.12, range: 10–27), with a mean number of 1.66 sessions (SD = 0.82, range: 1–3) lasting on average 17.17 min per session (SD = 7.99, range = 10–27). On average, treatment dropouts filled out 1.50 (21.4%) of the 7 weekly online assignments. Of the control participants, 10 (15%) dropped out of the sleep diary. The incidence of dropout did not statistically differ between the groups, χ²(1) = 2.12, p = 0.15. None of the baseline values were related to dropout.

Of the SRT participants who completed the posttest, 46 (74.2%) would recommend the treatment to others, 4 (6.5%) would not, and 11 (17.8%) would maybe recommend the treatment. On a 7-point scale treatment satisfaction was rated a mean of 5.29 (SD = 1.58, range: 1–7), which can be interpreted as quite satisfied. The support provided by the SRT coach was rated a mean of 8.30 out of 10 (SD = 1.28, range: 6–10), with 4 SRT participants (6.5%) reporting to have missed something in their coach’s support.

At pretest, ISI scores did not significantly differ between the groups, M_DIFF = 0.14, 95% CI: [1.28, 1.57], p = 0.84. At posttest, ISI scores were significantly lower in the SRT group compared to the control group, with an estimated adjusted mean difference of −6.52, 95% CI: [−8.03, −5.01], p < 0.001, d = −1.52, corresponding to a large effect size. Within the SRT group, insomnia severity decreased from pretest to 3-month follow-up, ISI M_DIFF = −8.25, 95% CI: [−9.47, −7.03], p < 0.001, d = −1.81. Similar results were observed for the within-group difference from pretest to 6-month follow-up, ISI M_DIFF = −8.79, 95% CI: [−10.14, −7.43], p < 0.001, d = −1.93. Both the within and between-group differences on the ISI were large in effect. Figure 2 depicts the ISI scores for both groups per time point.

The LMM-estimated means of all outcomes are reported in Table 2 (pre-post) and Table 3 (follow-up). The coefficients of the LMM models, the observed means, and the observed standard deviations are reported in the online supplementary material.

Compared to the sleep diary control group at posttest, the SRT group reported significantly lower presleep arousal, PSAS M_DIFF = −9.02, 95% CI: [−12.59, −5.45], p < 0.001, d = −0.97), lower sleep safety behavior SRBQ (SRBQ M_DIFF = −12.52, 95% CI: [−18.75, −6.30], p < 0.001, d = −0.81), lower dysfunctional beliefs (DBAS M_DIFF = −15.75, 95% CI: [−24.70, −6.80], p < 0.001, d = −0.80), lower depression symptoms (PHQ-9 M_DIFF = −3.59, 95% CI: [−4.95, −2.23], p < 0.001, d = −0.73), and lower anxiety symptoms (HADS-A M_DIFF = −1.58, 95% CI: [−2.85, −0.32], p = 0.01, d = −0.53), all indicating medium or large effects.

Sleep diary outcomes at posttest were significantly lower for the SRT group compared to the sleep diary control for SOL (M_DIFF = −31.31, 95% CI: [−45.21, −17.40], p < 0.001, d = −0.62), wake after sleep onset (WASO M_DIFF = −44.90, 95% CI: [−62.98, −26.82], p < 0.001, d = −1.13), and TWT (M_DIFF = −75.66, 95% CI: [−99.54, −51.77], p < 0.001, d = −1.18), while SE was higher for the SRT group than the control group (SE M_DIFF = 13.50, 95% CI: [9.06, 17.94], p < 0.001, d = 1.11). For all these secondary outcomes, improvements were maintained at the follow-up moments, reflected in medium-to-large or large within SRT group effects from pretest compared to the 3- or 6-month follow-up. There was no between-group difference in TST at posttest (TST M_DIFF = 8.29, 95% CI: [16.08, 32.65], p = 0.50, d = 0.07). However, TST significantly increased within the SRT group from pretest compared to 3-month (TST M_DIFF = 62.45, 95% CI: [46.93, 77.98], p < 0.001, d = 0.84) and 6-month follow-ups (TST M_DIFF = 62.05, 95% CI: [45.16, 78.93], p < 0.001, d = 0.84). Similarly, there was no between-group difference in daytime sleepiness at posttest (ESS M_DIFF = −0.58, 95% CI: [−2.20, 1.03], p = 0.48, d = 0.03) but within the SRT group, daytime sleepiness decreased from pretest compared to 3-month (ESS M_DIFF = −2.33, 95% CI: [−3.22, −1.44], p < 0.001, d = −0.56) and 6-month follow-ups (ESS M_DIFF = −2.06, 95% CI: [−3.05, −1.07], p < 0.001, d = −0.50).

In line with adherence to sleep restriction, TIB was significantly lower for the SRT group compared to the sleep diary control group (TIB M_DIFF = −66.51, 95% CI: [−83.48, −49.53], p < 0.001, d = −1.56). This was also the case for variability in bedtime (BTv M_DIFF = −24.91, 95% CI: [−35.51, −14.32], p < 0.001, d = −0.64) and variability in rise time (RTv M_DIFF = −24.33, 95% CI: [−33.64, −15.01], p < 0.001, d = −0.82). Within the SRT group relative to pretest, TIB was still lower at 3-month follow-up (TIB M_DIFF = −21.48, 95% CI: [−31.79, −11.17], p < 0.001, d = −0.49) and 6-month follow-up (TIB M_DIFF = −31.45, 95% CI: [−42.65, −20.24], p < 0.001, d = −0.72). Similarly, bedtime variability was still lower at the 3-month follow-up (BTv M_DIFF = −16.67, 95% CI: [−25.66, −7.68], p < 0.001, d = −0.64) and the 6-month follow-up (BTv M_DIFF = −12.09, 95% CI: [−21.58, −2.59], p = 0.01, d = −0.46). Rise time variability, however, did not significantly differ within the SRT group relative to pretest at the 3-month follow-up (RTv M_DIFF = −5.23, 95% CI: [−12.01, 1.56], p = 0.13, d = −0.22) or the 6-month follow-up (RTv M_DIFF = −5.67, 95% CI: [−12.07, 0.72], p = 0.08, d = −0.24).

Of all participants randomized to SRT, 36 participants (47.4%) showed a clinical treatment response (ISI reduction of ≥8) from pretest to posttest, compared to 6 participants randomized to the control group (8.5%), χ²(1) = 25.34, p < 0.001. Of all participants randomized to SRT, 33 (43.4%) showed a treatment response relative to the pretest at the 3-month follow-up, and 29 (38.2%) at the 6-month follow-up.

At posttest remission (ISI score <8) was achieved by 23 (30.3%) participants randomized to SRT, versus 2 (2.8%) participants randomized to the control group, Fisher’s exact test p < 0.001. Of all participants randomized to SRT, 24 (31.6%) achieved insomnia remission at the 3-month follow-up, and 17 (22.4%) at the 6-month follow-up. Note that these response and remission rates are conservative estimates, as the percentages consider all participants who did not fill out posttest or follow-up questionnaires as clinical nonresponders.

At posttest, adverse events were reported 19 times in the SRT group and 22 times in the control group. The total incidence of adverse events did not statistically differ between the groups, χ²(1) = 0.39, p = 0.53.

Fatigue was reported by 15 participants (24.2%) in the SRT group and 12 (18.5%) in the control group. Mean fatigue severity ratings were 4.73 (±1.28) and 5.00 (±1.75), respectively (rated on a 7-point scale, with higher scores indicating more fatigue). Eleven SRT participants (17.7%) and 8 control participants (12.3%) reported experiencing negative consequences of fatigue. Falling was reported by 4 SRT participants (6.5%) and 9 control participants (13.8%). The mean severity of falling was rated 2.50 (±1.73) and 2.55 (±1.13), respectively (7-point scale, with higher scores indicating more severity). Two SRT participants (3.2%) and 6 control participants (9.2%) reported negative consequences of falling. For one control participant (1.5%), falling led to hospitalization. Additionally, one control participant (1.5%) reported being involved in a quite severe accident, which led to negative consequences but no hospitalization. No SRT participant reported the event of an accident or hospitalization.

A full description of treatment satisfaction, medication use, and other sleep or psychological treatment uptake during the trial is provided in the online supplementary material. There were no significant group differences in medication use or uptake of other treatments since the start of the trial, as reported at posttest.

In this study, we tested the efficacy of SRT in a fully remote telephone-guided format. Relative to a sleep diary control group, insomnia severity improved after SRT with a large effect. At both the 3- and 6-month follow-ups, we also found large within-group effects on insomnia severity relative to before treatment for the SRT group. Based on intention-to-treat, conservatively interpreting everyone who dropped out from the study as nonresponse, 47% of all participants randomized to SRT showed clinical improvement and 30% showed full remission of insomnia complaints. These effects and response rates are either larger than or comparable to results from trials of multicomponent CBT-I [3, 4], and SRT in a face-to-face or mixed format [18, 20, 21].

In line with the theoretical model of SRT [16], SRT increased SE and reduced sleep-related arousal, SOL, wake time after sleep onset, and sleep-related safety behavior, all with large effects compared to the sleep diary control group. These effects were maintained within the SRT group at 3- and 6-month follow-ups. As is generally found in SRT trials, daytime sleepiness and TST did not differ between the groups at posttest directly after treatment [18, 20]. This fits with findings that during the acute phase of SRT, the restricted sleep window initially leads to increased sleepiness and reduced objective TST [48]. On the follow-up measures, however, the SRT group showed a large reduction in daytime sleepiness and a large increase in TST relative to before treatment.

Even though SRT did not specifically address worry, dysfunctional beliefs, depression, or anxiety, SRT also improved these outcomes in both the short and long terms. Effects on depressive and anxiety symptoms were similar or larger compared to the small-to-moderate effects reported in systematic reviews of SRT [19] and multicomponent CBT-I [49, 50] and support the notion that mood complaints can improve when sleep improves [51].

Coaches in this trial were staff members and graduate students from a clinical psychology university department, who received a 2-h SRT training. Most coaches (60%) had no formal CBT-I background. We encountered no coach-related problems but did not formally assess whether the level of coach expertise affected the outcomes, which could be explored in future trials. Notably, a qualitative evaluation of a nurse-delivered SRT trial concluded that practice nurses who received a 4-h SRT training felt prepared to deliver SRT and indeed did so successfully [52]. Considering the shortage of CBT-I therapists the field currently faces [6, 8], SRT’s promise to expand the pool of providers by overcoming the need for formal CBT-I training offers a key advantage.

Another advantage of SRT in a telephone-guided format is its minor treatment time investment. Over the 6 weeks of SRT, coaches had an average of 75 min of telephone contact per patient completing the treatment. This total treatment time is much less than the advocated 4 to 8 sessions of in-person CBT-I (totaling 180–480 min per patient [3]) and similar to the time investment of digital guided CBT-I (an average of 5.5 weeks totaling 40–120 min per patient [53, 54]). Digital formats using automated written support have an even smaller time investment, but relatively high dropout rates have been reported for these trials (e.g., [55]).

Telephone and digital guidance SRT formats may offer distinct advantages. With digital formats, feedback can be given asynchronously, allowing patients and therapists to read and respond at their convenience. Telephone guidance, on the other hand, requires brief real-time availability of the patient and therapist, but offers direct interaction, which might help increase a sense of human connection and patient involvement during SRT [24, 56]. Comparisons of telephone-guided SRT and digital SRT with written feedback, or a combined approach, should clarify their relative benefits regarding efficacy, dropout, therapist time, and scalability. It is also worth noting that while telephone treatment is reimbursed similarly to face-to-face treatment in the Netherlands [57], reimbursement policies may differ in other countries.

This study also has limitations. Due to the minimal and remote design, we did not include clinical interviews or objective sleep measures and cannot provide details on mental comorbidity beyond depression and anxiety, or effects on sleep physiology. Additionally, we recruited participants from the general population and excluded individuals with severe depression scores, as they might require more intensive treatment. It is unclear if our findings apply to clinical samples with more comorbidity. However, promisingly, brief behavioral CBT-I interventions are generally effective in samples of insomnia with mental comorbidity [58] and a large pragmatic trial has recently shown the effectiveness of SRT (mixed in-person and telephone format) in a primary care setting [21]. Still, the (cost-)effectiveness of telephone-guided SRT will need to be tested in clinical settings. Here, the telephone format may be particularly useful as an add-on or early-phase intervention. For instance, telephone-guided SRT may be offered to patients awaiting mental health care (following the example of [59]).

Another limitation is the lack of formal treatment fidelity checks regarding the content of the telephone support calls. Although coaches were trained and supervised, the sessions were not audio recorded, so we could not formally assess how closely coaches adhered to the protocol.

A final limitation concerns dropout and nonresponse. In the SRT group, 25% of patients dropped out before completing the treatment. This dropout rate falls within the range reported for in-person multicomponent CBT-I and in-person or mixed-format SRT [21, 60, 61] and is lower than the dropout rates typically reported for digital CBT-I, which can reach up to 50% [62]. Additionally, among all SRT participants who completed the posttest (regardless of treatment dropout), 42% showed no clinical improvement in insomnia. This nonresponse rate is only slightly higher compared to the 36% rate reported in a meta-analysis of multicomponent CBT-I [3]. Still, these findings suggest that SRT, as a core element of CBT-I, may not be suitable for all patients with insomnia. For patients who do not prefer or respond to SRT, alternative treatments such as Acceptance and Commitment Therapy for insomnia [63] could be considered.

In sum, telephone-guided SRT offers an efficacious insomnia treatment that requires minimal therapist involvement and enables patients to receive care from home. Motivating patients with insomnia to adjust their bedtimes through brief weekly phone calls resulted in long-term improvements across all outcomes, sustained for up to 6 months, with effects competitive to those found after multicomponent CBT-I. Trials directly comparing the effectiveness of SRT and multicomponent CBT-I are lacking and much needed. If such head-to-head trials corroborate our findings and SRT proves noninferior to CBT-I, SRT could be an interesting, scalable first-line treatment alternative for insomnia.

We would like to thank research assistants and coaches Kim Appel, Jikke de Geus, Babs Groot, Nina Kollof, Susan Koopman, Rico Pischedda, and Ayla Polat for their valuable contributions to the study.

This study protocol was reviewed and approved by the internal Ethical Review Board of the University of Amsterdam (Approval No. 2022-CP-15342). Digital informed consent was obtained from all participants.

The authors have no conflicts of interest to declare.

An internal research grant from the University of Amsterdam supports this study. The funder had no role in the study’s design, data collection, data analysis, and reporting.

T.F.B., T.M.S., M.E., J.H.K., and J.L. conceptualized the study. M.I.L., T.M.S., M.E., A.S., and J.L. developed the treatment protocol. M.I.L., T.M.S., F.E.L., J.E.R., T.F.B., and J.L. were involved in the data collection. M.I.L., T.F.B., and J.L. performed the statistical analyses. M.I.L., T.F.B., and J.L. drafted the work, and all authors reviewed and approved the final manuscript.

The data supporting the findings of this study are not publicly available as informed consent did not explicitly include public data sharing. The data are available from the corresponding author (M.I.L.) upon reasonable request.