Randomized controlled trial of cognitive refocusing versus stimulus control treatment for college insomnia: Feasibility of a brief, electronic-based, and peer-led approach

Abstract

Objective:

Performance of Cognitive Refocusing Treatment for Insomnia (CRT-I) relative to stimulus control treatment (SCT) remains unknown among college students. This pilot trial compared single-session, electronic-based, peer-led CRT-I to SCT, and as well as awareness-based (AC) and no-treatment (NTC) controls.

Participants:

College students (N=82; M_age=18.59 [SD=0.78]; 58% female; 61% White; 16% Hispanic) with insomnia symptoms were randomly assigned to CRT-I (n=12), SCT (n=14), and AC (n=14) conditions, or non-randomly recruited to NTC (n=42).

Methods:

All participants completed baseline and one-month follow-up surveys, and reported daily task enactment (except NTC).

Results:

Feasibility ratings were comparable across conditions. Within-group treatment effects revealed greater improvements in (a) insomnia symptom severity among CRT-I (d=1.13) and SCT (d=1.66) groups relative to AC (d=0.90) and (b) pre-sleep cognitive arousal among CRT-I (d=0.94) and SCT (d=1.42) groups relative to AC (d=0.75). Conclusions: Brief, electronic-based, peer-led CRT-I and SCT interventions appear feasible and potentially efficacious for college insomnia.

Introduction

College students commonly endorse insomnia symptoms,^1–3 which include subjective complaints of difficulty falling and/or staying asleep.⁴ As many as 9% of college students meet clinical criteria for insomnia⁵ and over 40% report insomnia symptoms such as difficulty falling asleep and/or sub-optimal sleep efficiency.⁶ College sleep problems are associated with substantial short-term consequences, including depressed mood and suicidal ideation/behaviors^7,8 and compromised academic attainment.⁹ Insomnia symptoms are also a risk factor for drastic long-term mental and physical health problems, including psychiatric diagnoses,¹⁰ cardiovascular and metabolic disease risk, neurocognitive dysfunction, and all-cause mortality.¹¹ College students are a specific population of interest to insomnia research given uniquely elevated symptom prevalence and novel considerations for treatment (such as sleep scheduling limitations, substance use, and shared and/or suboptimal environmental sleep conditions, which may pose challenges for standard treatments).^12,13

Treatment modality of college insomnia intervention: Considerations for feasibility

Three primary considerations for feasibility¹⁴ of college insomnia intervention include the prioritization of brief, electronic-based, and peer-led intervention. First, brief college insomnia intervention is indicated to accommodate the busy nature of the residential college context,¹⁵ which may function as a barrier to accessibility and adherence of more time-intensive interventions. Further, brief intervention is aligned with the movement toward stepped care models of college mental health services¹⁶ to accommodate increasing utilization.¹⁷ Second, electronic-based interventions which may be delivered online and/or practically incorporated into college primary care¹⁸ may further help overcome barriers to treatment access (e.g., financial, cultural)¹⁹ and accommodate the increasingly web-based nature of the college learning environment.²⁰ Previous research in college students indeed demonstrates the efficacy of both brief (e.g., SCT, sleep restriction treatment, and sleep hygeine psychoeducation^21,22) and electronic-based (e.g., Cognitive Behavioral Therapy for Insomnia [CBT-I] delivered in a presentation-based, workshop format²³ or entirely online²⁴) insomnia intervention. Third, consistent with role theory, youth may learn more effectively from peers of the same generation (as opposed to individuals who are older and perceived as authority figures).²⁵ As such, peer interventionists are widely examined and utilized in college health behavior change interventions, such as substance use (e.g., risky drinking),²⁶ sexual health,²⁷ and nutrition and physical activity²⁸ behavioral targets. Extensive literature on peer-led intervention for college health risk behaviors suggests that peer-led intervention may also mitigate burden to healthcare services and optimize treatment access and outcomes.²⁹ In summary, feasibility studies of single-session, electronic-based, and peer-led college insomnia intervention are exigent to justify ongoing research into treatment efficacy.¹⁴

Cognitive theory and intervention for insomnia

Cognitive theory of insomnia^30,31 posits that pre-sleep cognitive arousal (i.e., state-dependent thought content occurring as an individual falls asleep, such as worry)³² is associated with physiologic and/or emotional arousal, thereby contributing to the development and maintenance of insomnia symptoms over time. Indeed, as many as 90% of adults endorse and attribute their insomnia symptoms to pre-sleep cognitive arousal.³³ Pre-sleep cognitive arousal is also associated with insomnia among college students, specifically.³⁴ The most commonly cited cognitive sleep interferences among college students include worry and academic stress.^35,36 Indeed, college students prevalantly endorse worry and self-critical coping strategies for managing intrusive pre-sleep thoughts, and both strategies are associated with more severe insomnia symptoms.³⁴ Further, a recent study³⁷ found that college students endorsed maladaptive metacognitions about sleep, including unrealistic expectations about sleep and worry about insomnia and its effects.³⁸ Notably, although pre-sleep somatic arousal (e.g., racing heart, muscle tension)³² has additionally been indicated as a risk factor for college insomnia,³⁹ people with insomnia have consistently cited pre-sleep cognitive arousal to cause their sleep problems.^33,40

Cognitive Refocusing Treatment for Insomnia (CRT-I)

Emerging literature demonstrates preliminary efficacy of Cognitive Refocusing Treatment for Insomnia (CRT-I) among college students using a single-session, in-person, provider-led modality.⁴¹ CRT-I directly intervenes on pre-sleep thought content by replacing emotionally/physiologically activating cognitions with a self-identifed topic of mental activity, thus regulating arousal and promoting sleep onset.^41,42 CRT-I is considered distinct from cognitive therapy for insomnia, which focuses on restructuring maladaptive beliefs about sleep (for example, the belief that 8 hours of unbroken sleep is needed every night for adequate diurnal functioning).³¹ The individualized approach of CRT-I is reasoned to optimize attractiveness and motivation for implementation, thereby maximizing treatment effect.^41,42 Indeed, CRT-I performed better than sleep hygiene education in a controlled study of college students, demonstrating efficacy of CRT-I delivered in a one-session format.⁴¹

It remains unknown how CRT-I performs compared to stimulus control treatment (SCT).^12,43 SCT may be a relevant comparator to CRT-I given its contrasting behavioral (versus cognitive) focus, as well as demonstrated efficacy among college students²¹ and comparable feasibility of a brief delivery format.⁴³ SCT aims to re-establish learned associations between sleeping spaces and sleep practices using strategies such as getting up at the same time every morning, minimizing diurnal napping, avoiding arousing activities in the bedroom, going to bed only when tired, getting out of bed and engaging in a quiet activity in another room if unable to intiate sleep within 15– 20 minutes, and repeating these tasks as needed at bedtime and following night-time waking.⁴⁴ However, a feasibility study demonstrated that most college students do not adhere to all of the varied components of SCT on a daily basis,²² necessitating ongoing investigation of simplified brief interventions with potentially-superior feasibility for at-home enactment in this population.

Current study

This pilot randomized controlled trial tested the feasibility and efficacy of two single-session, electronic-based, peer-led insomnia interventions, including CRT-I and SCT, as well as an awareness-based control (AC; consistent with prior studies of college sleep intervention⁴³). Thus, the current design represents novel evidence-based considerations to optimize implementation feasibility for college students, building on a previous study investigating a single-session, in-person, provider-led college CRT-I intervention.⁴¹ University students provided ratings of insomnia severity and pre-sleep arousal at baseline and at 1-month follow-up, and provided baseline treatment feasibility ratings and interim daily reports of treatment implementation for 30 days. A no-treatment control (NTC) group recruited non-randomly also completed baseline and follow-up assessments for comparison. It was hypothesized that, compared to SCT participants, participants receiving CRT-I would report greater perceived treatment feasibility and at-home treatment enactment, as well as reduced insomnia severity and levels of pre-sleep cognitive arousal. Further, it was hypothesized that CRT-I and SCT groups would out-perform both control groups in terms of reduced insomnia severity and pre-sleep cognitive arousal.

Method

Participants

Participants were recruited from undergraduate introductory psychology classes at a private university in the northeastern United States and compensated with research credit to satisfy a class requirement. Thus, this sample is not considered treatment-seeking, although all participants endorsed interest in an intervention that “helps you initiate sleep or stay asleep.” Participants were enrolled over 4 consecutive semesters spanning two academic years (2013–2015) and were randomly assigned to one of three groups: (a) cognitive refocusing treatment for insomnia (CRT-I), (b) stimulus control treatment (SCT), or (c) awareness-based control (AC). A block randomization procedure using a computer-based random number generator of 3 numbers per set was used to achieve equivalent sample sizes between groups, consistent with recommendations.⁴⁵ A no-treatment control group (NTC) was subsequently recruited non-randomly over 6 consecutive semesters spanning three academic years (2015–2018) for comparison.

To be included in the study, participants were required to: (a) be 18 years of age or older; (b) be enrolled as an undergraduate college student; (c) endorse difficulty sleeping; (d) report sleep onset latency and/or wake after sleep onset of greater than 30 minutes at least three nights per week for at least 1 month, consistent with quantitative⁴⁶ criteria for insomnia; (e) report “difficulty turning off your mind at bedtime,” and thus a candidate for a cognitive insomnia intervention; and (f) endorse interest in an intervention that “helps you initiate sleep or stay asleep.” In order to exclude students with sleep problems due to nightmare- or breathing-related sleep disorders (versus insomnia), students were ineligible to participate if they endorsed (a) frequent nightmares (i.e., more than two nights per week) or an interest in treatment to reduce the frequency of nightmares, and/or (b) common indicators of breathing-related sleep disturbance (i.e., heavy snoring or general sleepiness during the day or difficulty breathing or gasping for breath during sleep). Students were also ineligible if they reported concurrent insomnia treatment. Notably, data from the larger recruitment pool is unavailable, precluding calculation of percentage of individuals omitted for aforementioned criteria. Of the original sample of 91 participants, those who reported definite morning (scores 70–86; n=0) or definite evening (scores 16–30; n=9) circadian preference per standard cutoffs using the Morningness-Eveningness Questionnaire⁴⁷ were excluded from analyses to control for potential circadian rhythm/phase disorders; students who did not provide data on this additional exclusion criteria (n=8) were retained for analyses.

Thus, the final analytic dataset (see Figure 1) consisted of 82 participants (M_age=18.59 [SD=0.78; range: 18–22]; 57% female; 57% White, 18% Asian, 6% Black or African American, 6% Other or “I don’t know,” 4% Multiracial, 1% American Indian or Alaska Native [7% (n=6) did not indicate race]; 16% Hispanic). Of the 82 participants at Baseline, participants that did not provide data at the one-month Follow-Up (i.e., attrition; n=3 [4%]; n=2 SCT and n=1 AC) were retained in mixed modelling analyses (see Data analytic strategy).

Figure 1.

Study flowchart. CRT-I, SC, and AC data collected 2013–2015; NTC data collected 2015–2018. ^aNote that, due to recruitment via undergraduate participant pool with automatic eligibility screening, it is not possible to determine the total number of students assessed for eligibility. ^bExcluded for definite circadian preference. ^cReason for attrition at follow-up unknown.

Procedure

All procedures were approved by the Institutional Review Board at the university. This exploratory pilot feasibility study used a 4 × 2 repeated measures design to explore the interaction effects of group (CRT-I, SCT, AC, and NTC) by time (baseline and follow-up) on insomnia symptom severity and pre-sleep arousal. Participants in CRT-I, SCT, and AC groups met individually with a peer interventionist at the university laboratory for one session, during which they provided informed consent, completed online baseline surveys, and received the intervention. Specifically, participants received a handout to take home and viewed a 15–30-minute electronic slideshow-based presentation detailing intervention rationale and procedures. Participants were encouraged to ask questions of the peer interventionist to clarify the intervention instructions both during and after the intervention as needed. Peer interventionists were advanced undergraduate psychology majors trained by the principal investigator (PI) and assigned non-randomly per scheduling availability. Training included readings, extensive observed peer-based role plays, and all interventionists received weekly supervision conducted by the PI. Interventionists delivered scripted content using predetermined checklists to guide a standardized protocol. Participants then provided daily reports of treatment implementation for 30 days and a one-month follow-up survey, all completed online at a place of their choosing. Assessment procedures for the non-randomized NTC group were identical (i.e., consented in the lab and completed the same online baseline and 1-month follow-up surveys), except daily treatment implementation reports were non-applicable.

Cognitive Refocusing Treatment for Insomnia (CRT-I).

The handout and presentation highlighted the importance of shifting physiologically and emotionally arousing thoughts to non-arousing thoughts before bed to improve sleep.^41,42 After viewing the electronic slideshow-based presentation, participants collaborated with the peer interventionist to identify three topics to think about at bedtime that were both non-arousing (i.e., devoid of emotion-laden, negative, exciting, or worrisome content) and compelling enough to maintain attention (e.g., mentally completing puzzles or reciting song lyrics). Participants were instructed to focus attention on one of these topics at bedtime and upon waking during the sleep period, refocusing as needed if/when distraction occurred. If the initial selected topic failed to maintain their attention or otherwise elicited emotional or physiologic arousal, participants were encouraged to switch topics, then focus on the same topic consistently before bed to develop learned associations between specific thought content and sleep.

Stimulus Control treatment (SCT).

The handout and presentation highlighted the importance of learning to associate the bed and bedroom with sleep (as opposed to physiologically and emotionally arousing thoughts).^44,48 Participants were instructed to follow six rules: (a) use bedroom only for sleep and sex; (b) go to bed only when sleepy; (c) if unable fall asleep within 15–20 minutes, leave the bed and do something in another room; (d) if unable to fall asleep quickly upon returning to bed, repeat previous rule as many times as needed; (e) leave the bed at the same time every day; (f) avoid napping. After viewing the electronic slideshow-based presentation, participants collaborated with the peer interventionist to identify two relaxing activities to do when unable to sleep after leaving the bedroom (e.g., reading).

Awareness-based control (AC).

The AC group was included to control for the participation burden (e.g., time and attention) and treatment-related expectations associated with intervention receipt and enactment (per recommendations for controlled study designs⁴⁹), as well as measurement reactivity and fatigue effects potentially associated with daily reporting⁵⁰ over the one-month assessment period. The handout and presentation for this condition highlighted the importance of having an awareness of the present moment in order to prevent physiologically and emotionally arousing thoughts before bed.⁵¹ Participants were instructed to “stay in the moment” by focusing on the physical sensation of their breathing (e.g., abdomen rising/falling) and/or their body (e.g., clothes touching skin). After viewing the electronic slideshow-based presentation, participants collaborated with the peer interventionist to practice noticing secondary thoughts (i.e., “wandering mind”) and refocusing on the present moment as needed by utilizing aforementioned strategies. Participants were also advised not to try to fall asleep, but rather allow their sleep to onset naturally. These instructions are notably briefer than multicomponent mindfulness interventions utilized in college insomnia trials⁴³ (for example, including imagery, deep breathing, and insight meditation⁵²). As such, this intervention was intended to omit meaningful treatment ingredients, thereby serving as a useful control comparison to the CRT-I treatment (which by contrast prescribed active thought modification).

Procedures to optimize feasibility

To optimize feasibility and account for predictable heterogeneity in treatment implementation, several recommended strategies⁵³ were used to induce and assess specific processes of treatment delivery, receipt, and enactment (i.e., compliance).⁵⁴ Treatment delivery was (a) induced by providing intensive, ongoing training to all peer interventionists and adhering to manualized procedures, and (b) assessed by asking participants to rate their peer interventionist at the end of their baseline appointment (notably, in order to avoid social desirability bias, participants were reassured that the specific interventionist would not view these responses). Treatment receipt was (a) induced during the baseline appointment by providing participants with aforementioned handouts of key points, using verbal prompts to encourage attendance to key treatment ingredients and quizzing participants to confirm understanding, and (b) assessed by asking participants to rate their confidence in enacting treatment. Treatment enactment was both induced and assessed by soliciting daily self-reports of treatment enactment for 30 days following the baseline appointment.

Measures

Insomnia severity

The 7-item Insomnia Severity Index (ISI)⁵⁵ assessed severity of insomnia symptoms at baseline and follow-up. Participants indicated perceived severity of insomnia symptoms (e.g., difficulty falling asleep, waking up too early) using a 5-point Likert scale ranging from 0 (none) to 4 (very severe). Sum scores (possible range=0–28; α=.64-.80) were used for analyses, with scores ranging from 0 to 7 indicating no clinically significant insomnia, 8–14 indicating sub-threshold insomnia, 15–21 indicating moderate clinical insomnia, and 22–28 indicating severe clinical insomnia.⁵⁶

Pre-sleep arousal

The 16-item Pre-Sleep Arousal Scale³² was administered at baseline and follow-up to assess participants’ typical state of arousal before sleep. Participants rated the severity of certain behaviors that may elicit cognitive (e.g., worry about falling asleep) and somatic (e.g., heart racing, pounding, or beating irregularly) pre-sleep arousal based on a 6-point scale ranging from 1 (not at all) to 5 (extremely). Sum scores of cognitive (possible range=8–40; α=.80-.91) and somatic (possible range=8–40; α=.68-.82) pre-sleep arousal subscales were used for analyses, with higher scores indicating a greater degree of arousal before bed. Ancillary analyses using the 5-item cognitive subscale Pre-Sleep Arousal Scale⁵⁷ yielded internal reliability and treatment effect size patterns across groups consistent with main results using the traditional full scale, which was thus retained for main analysis herein.

Morning circadian preference

The 19-item Morningness-Eveningness Questionnaire⁴⁷ assessed participants’ circadian preference at baseline, including tiredness during the day and preferred times to do diverse activities and to go to sleep. Sum scores (possible range=31–69; α=.76) were used for analyses, with a higher score indicating a morning (versus evening) circadian preference. Standard cutoffs were used to identify definite morning (scores 70–86) and definite evening (scores 16–30) circadian preference.⁴⁷

Inadequate sleep hygiene

The 19-item Sleep Behaviors Questionnaire⁵⁸ assessed inadequate sleep hygiene (based on The International Classification of Sleep Disorders-II criteria⁵⁹) at baseline and follow-up. Respondents indicated the average number of days per week that they engaged in potential sleep-impeding behaviors during the previous month, including specific domains of improper sleep scheduling, use of sleep-disrupting products, arousing behaviors near bedtime, and use of the bed for activities other than sleep. Sum scores (possible range=0–133) were used for analyses, with higher scores indicating worse sleep hygiene.

Depression severity

The 8-item Patient Health Questionnaire (PHQ-8)⁶⁰ assessed depression symptom severity at baseline. Participants reported how often they experienced major depressive symptoms during the previous 2 weeks, with response options based on a 4-point Likert scale ranging from 0 (not at all) to 3 (nearly every day). Sum scores (possible range: 0–24; α=.73-.84) were used for analyses, with scores of ≥10 indicating likely moderate depression symptoms.⁶⁰

Anxiety severity

The 7-item Generalized Anxiety Disorder self-report scale (GAD-7)⁶¹ assessed anxiety symptom severity at baseline. Participants reported how often they were bothered by anxiety symptoms during the previous 2 weeks, with response options based on a 4-point Likert scale ranging from 0 (not at all) to 3 (nearly every day). Sum scores (possible range: 0–21; α=.85-.90) were used for analyses, with scores of ≥10 indicating likely moderate anxiety symptoms.⁶¹

Time of year

Academic semester of participation (1=Fall, 0=Spring) and participation over a major university break (i.e., week-long Fall and Spring breaks; 1=yes, 0=no) were assessed given predictable changes in sleep.³⁵ Notably, participation was restricted to the Fall and Spring semesters, so no students participated over Winter or Summer university breaks.

Demographics

Age, female gender (1 vs. 0), White race (1 vs. 0), Hispanic ethnicity (1 vs. 0) and first-year class (1 vs. 0) were assessed at baseline given demonstrated associations with insomnia.⁶²

Feasibility (CRT-I, SC, and AC conditions)

Delivery and receipt.

Four items from the Treatment Evaluation Questionnaire⁶³ assessed treatment credibility and acceptability on Day 1 of daily surveys. Participants rated the extent to which they (a) felt the treatment was logical and reasonable, (b) liked and had confidence in the interventionist, (c) had confidence that the treatment would effectively decrease insomnia, and (d) would recommend the intervention to a friend with a similar problem. The items were measured on a Likert scale ranging from 0 (strongly disagree) to 5 (not sure) to 10 (strongly agree).

Each group additionally completed an investigator-developed survey of procedural knowledge specific to the treatment received, including 6 true-false items specific to the treatment received. Example items for CRT-I include: (a) “Unwanted thoughts can be replaced by an engaging and interesting thought-related task”; (b) “Thinking about your upcoming exam is generally beneficial for your sleep.” Example items for SCT include: (a) “It is good for your sleep to avoid the bed and bedroom when you are not sleeping”; (b) “Napping during the day is beneficial for your night-time sleep.” Example items for AC include: (a) “Focusing on the breath can be a helpful tool to stay in the present moment”; (b) “If you are having trouble sleeping, you should focus on trying to fall asleep.” A count score of the number of items answered correctly (possible range: 0–6) was calculated for descriptive purposes, with greater scores indicating a greater level of procedural knowledge following baseline treatment session.

Enactment.

Investigator-developed daily waking surveys assessed enactment (1=yes, 0=no) of each component of the assigned treatment the previous night. For CRT-I, three items assessed pre-sleep task enactment, night-time waking, and night-time waking task enactment. For SCT, six items assessed using bed only for sleep and sex, trying to initiate sleep only when sleepy, avoiding diurnal napping, leaving bed at same time each day, leaving bed and doing something in another room when unable to sleep for >20 minutes, and repeating aforementioned task enactment as much as needed to (re)initiate sleep. For AC, seven items assessed attempting to stay in the present moment, focusing on breath to “stay in the present moment,” wandering mind at bedtime, focusing on breath or physical sensation to come back to the present moment, night-time waking, and night-time waking task enactment.

Data analytic strategy

Analyses were conducted using SPSS version 26.⁶⁴ Group comparisons of baseline variables were conducted using one-way ANOVA (and Tukey HSD post-hoc tests) for continuous variables and non-parametric Kruskal-Wallis test (with Mann-Whitney U post-hoc tests) for dichotomous variables. Descriptive statistics, Mean (SD) for continuous variables and n (%) for categorical variables, were used to evaluate feasibility. Separate repeated measures mixed models examined group (CRT-I, SCT, AC, and NTC) by time (Baseline versus Follow-Up) main effects and interaction effects on change in three outcome variables (i.e., treatment effect): insomnia severity, cognitive pre-sleep arousal, and somatic pre-sleep arousal. A mixed modelling approach was selected for current repeated measures analyses due to its accommodation of the multilevel data structure (i.e., Baseline and Follow-Up observations nested within individuals). Further, this analytic approach incorporates all available data, such that participants missing on Baseline or Follow-Up data were not excluded from analyses).⁶⁵ All predictors were modelled as fixed effects using compound symmetry covariance structure and restricted maximum likelihood (REML) missing data procedure to accommodate the current small sample size.⁶⁶ Covariates included relevant baseline characteristics not equivalent between groups at baseline. Cohen’s d⁶⁷ was used to assess within-group clinical effect using adjusted means and standard deviations, accounting for dependence of means (i.e., correlation).⁶⁸ Ancillary analysis replicated main mixed models omitting participants non-randomly recruited to the NTC condition (that is, among n=40 participants who were randomly assigned to CRT-I, SCT, or AC conditions) to investigate whether treatment effects patterns were influenced by this lack of randomization.

Results

Baseline characteristics and covariate selection

Complete baseline descriptive statistics are presented in Table 1. Participants reported mild-to-moderate clinical insomnia symptoms (M=13.67 [SD=3.62]) and 43% surpassed the threshold for identifying insomnia disorder (ISI score≥15).^69,70 Though the sample was composed of mostly underclass university students (67% first-year), overall participants demonstrated concurrent mental health characteristics consistent with adult community samples complaining of insomnia.⁶² On average, participants endorsed mild-to-moderate depression symptoms (M=9.22 [SD=4.06]) and 45% surpassed the threshold for probable major depression (PHQ-8 score≥10).⁶⁰ Participants endorsed mild anxiety symptoms (M=7.72 [SD=4.37]) and 31% surpassed the threshold for probable generalized anxiety disorder (GAD-7 score≥10).⁶¹

Table 1.

Baseline Demographic and Sleep Characteristics and Group Comparisons

	Full Sample (N = 82)		CRT-I (n = 12)	SCT (n = 14)	AC (n = 14)	NTC (n = 42)	Group Comparison	Post-Hoc Comparisons
Variable (possible range)	Records (%)	M (SD) or %	M (SD) or %	M (SD) or %	M (SD) or %	M (SD) or %
Age	78 (93%)	18.59 (0.78)	18.20 (0.42)	18.75 (0.62)	18.43 (0.65)	18.69 (0.90)	F(3,74) = 1.46
Female Gender (1 vs. 0)	80 (98%)	57%	58%	57%	43%	62%	χ2(3) = 1.76
White Race (1 vs. 0)a	76 (93%)	57%	58%	29%	43%	71%	χ2(3) = 7.67
Hispanic Ethnicity (1 vs. 0)	66 (80%)	16%	8%	36%	21%	10%	χ2(3) = 5.03
First-Year Students (1 vs. 0)	78 (93%)	67%	67%	57%	71%	69%	χ2(3) = 0.56
Time of Year at Study Participation
Fall (vs. Spring) Semester	82 (100%)	68%	67%	50%	57%	79%	χ2(3) = 4.97
Completed over major university break (1 vs. 0)	82 (100%)	59%	50%	57%	50%	64%	χ2(3) = 1.35
Depression Severity (0–24)	81 (99%)	9.22 (4.06)	9.27 (3.44)	9.14 (3.63)	9.71 (3.83)	9.07 (4.51)	F(3,77) = 0.09
Anxiety Severity (0–21)	79 (96%)	7.72 (4.37)	7.09 (4.48)	6.21 (2.99)	9.38 (4.65)	7.88 (4.58)	F(3,75) = 1.29
Morning Circadian Preference (16–86)	74 (90%)	42.81 (7.98)	45.00 (6.75)	41.75 (8.50)	40.83 (6.78)	43.22 (8.47)	F(3,70) = 0.57
Insomnia Severity Index (0–20)	81 (99%)	13.67 (3.62)	14.45 (4.18)	13.00 (2.57)	15.21 (3.19)	13.17 (3.82)	F(3,77) = 1.48
Inadequate Sleep Hygiene (0–133)	80 (98%)	49.36 (13.11)	47.82 (10.85)	41.57 (11.75)	48.46 (12.64)	52.64 (13.39)	F(3,76) = 2.77*	SCT < NTC**
Cognitive Pre-Sleep Arousal (8–40)	80 (98%)	25.55 (5.93)	26.82 (5.98)	24.86 (5.89)	23.65 (6.03)	26.10 (5.92)	F(3,76) = 0.83
Somatic Pre-Sleep Arousal (8–40)	81 (99%)	14.94 (4.49)	13.82 (5.42)	14.57 (3.61)	15.43 (2.90)	15.19 4.98)	F(3,77) = 0.35

Note. CRT-I = cognitive refocusing treatment; SCT = stimulus control treatment; AC = awareness-based control; NTC = no-treatment control. Group comparisons and post-hoc analyses were conducted using one-way ANOVA (and Tukey’s HSD) tests for continuous variables and non-parametric Kruskal-Wallis (and Mann-Whitney U) tests for dichotomous variables.

^aNote that group comparison of complete racial categories is precluded by low endorsement in the current sample (i.e., 18% Asian, 6% Black or African American, 6% Other or “I don’t know,” 4% Multiracial, 1% American Indian or Alaska Native [7% (n=6) did not indicate race]).

^*p < .05.

^**p < .01.

Groups were comprised as follows: n=12 (15%) Cognitive Refocusing Treatment for Insomnia, n=14 (17%) Stimulus Control Treatment, n=14 (17%) Awareness-Based Control, and n=42 (51%) No-Treatment Control. Equivalence was not achieved at p<.05 on baseline inadequate sleep hygiene (F[3,76]=2.77, p=.047), which was thus selected as a covariate for subsequent repeated measures comparisons. Specifically, post hoc group comparisons indicated that NTC participants reported significantly worse baseline sleep hygiene compared to the SCT participants (M_difference=11.07 [SE=3.92], p=.01). Groups did not differ on any other participant characteristics, mental health, or other sleep variables (ps=.05-.97).

Intervention feasibility (CRT-I, SCT, and AC)

Descriptive information for baseline acceptability/credibility ratings and procedural knowledge is presented in Table 2. On average, participants from all three groups agreed that they (a) felt the treatment was logical and reasonable, (b) liked and had confidence in the interventionist, (c) had confidence that the treatment would effectively decrease insomnia, and (d) would recommend the intervention to a friend with a similar problem.

Table 2.

Feasibility Characteristics by Treatment Group: Baseline Acceptability and Credibility Ratings and Percentage of Daily Treatment Task Enactment at Bedtime and Night-Time Waking

CRT-I (n = 12)			SCT (n = 14)			AC (n = 14)
Baseline Acceptability and Credibility Ratings
Item (possible range)	M (SD)	range	Item (possible range)	M (SD)	range	Item (possible range)	M (SD)	range
Treatment is Logical/Reasonable (0–10)	8.25 (1.22)	6–10	Treatment is Logical/Reasonable (0–10)	8.86 (1.23)	7–10	Treatment is Logical/Reasonable (0–10)	7.86 (1.03)	6–10
Confidence in Interventionist (0–10)	9.42 (1.16)	7–10	Confidence in Interventionist (0–10)	9.50 (0.85)	7–10	Confidence in Interventionist (0–10)	9.07 (1.54)	5–10
Confidence in Treatment (0–10)	7.33 (1.37)	5–9	Confidence in Treatment (0–10)	7.86 (1.56)	5–10	Confidence in Treatment (0–10)	6.43 (0.94)	5–8
Would Recommend (0–10)	7.83 (1.90)	4–10	Would Recommend (0–10)	8.57 (1.60)	5–10	Would Recommend (0–10)	7.36 (1.60)	3–10
Procedural Knowledge (0–6)	5.83 (0.39)	5–6	Procedural Knowledge (0–6)	5.79 (0.43)	5–6	Procedural Knowledge (0–6)	5.92 (0.28)	5–6
Daily Bedtime Enactment
Treatment Task (n = 9; 232 daily observations)		n (%)	Treatment Task (n = 8; 156 daily observations)		n (%)	Treatment Task (n = 9; 197 daily observations)		n (%)
Refocused thought content		178 (77%)	Left bed at consistent time		88 (56%)	Made an effort to stay in the present moment		143 (73%)
			Avoided diurnal napping		129 (83%)	Focused on breath to stay in present moment		136 (69%)
			Used bed only for sleep and sex		111 (71%)	Thoughts wandered (114 daily observations)		114 (58%)
			Tried to sleep only when sleepy		145 (93%)	Focused on breath to return to present momentb		85 (75%)
			Sleep onset latency >20 min (87 daily observations)		87 (56%)
			Left the bedrooma		37 (43%)
			Repeated SCT instructions as neededa		30 (34%)
Subset of Night-time Waking Enactment
Treatment Task		n (%)	Treatment Task		n (%)	Treatment Task		n (%)
Night-time waking (regardless of duration)		86 (37%)	Night-time waking (>20 minutes)		16 (10%)	Night-time waking (regardless of duration)		55 (28%)
Refocused thought contentc		42 (49%)	Left the bedroomc		6 (38%)	Made an effort to stay in the present momentc		43 (78%)
			Repeated SCT instructions as neededc		6 (38%)	Focused on breath to stay in present momentc		38 (69%)

Note. CRT-I = cognitive refocusing treatment; SCT = stimulus control treatment; AC = awareness-based control. Participants from each group provided daily data on treatment enactment for 30 consecutive days (maximum) following treatment delivery/receipt at the baseline appointment. Percentages represent the proportion of available daily observations in which participants endorsed positive treatment task enactment. Data regarding night-time waking enactment represents a subset of daily observations in which participants reported waking after sleep onset.

^aPercentage among subset of SCT observations of sleep onset latency greater than 20 minutes.

^bPercentage among subset of AC observations of wandering thoughts.

^cPercentage among subset of observations of nighttime waking enactment.

Further, participants from all three groups demonstrated procedural knowledge. Post-hoc group comparison tests from one-way ANOVA tests revealed significant group differences in confidence that treatment will effectively decrease insomnia (F[2,37]=4.23, p=.02), such that AC participants endorsed lesser confidence compared to SCT (p=.01) and marginally CRT-I (p=.09) participants. Marginal group differences were also observed in belief that the treatment is reasonable and logical for insomnia (F[2,37]=2.64, p=.09), such that SCT participants endorsed greater belief compared to AC participants (p=.03). No significant differences were observed on any other acceptability/credibility items or procedural knowledge across treatment groups (ps=.18-.63).

Percentages of daily treatment task enactment over the 30-day assessment period, both at bedtime and night-time waking, are presented by group in Table 2. Enactment of various treatment tasks ranged from 34–93% across groups, reflecting variability in at-home practice in the natural college environment. For CRT-I participants specifically, proportion of enactment for refocused thought content was 77% at bedtime and 49% at nighttime waking. For SCT participants, proportion of enactment ranged from 56–83% for diurnal behavioral recommendations, 34–93% at bedtime, and 38% at nighttime waking. For AC participants, proportion of enactment ranged from 69–75% at bedtime and 69–78% at nighttime waking.

Treatment effects (repeated measures mixed models)

Insomnia severity

Adjusted mean scores for baseline and follow-up insomnia severity and effect size are presented in Table 3. Analyses showed Group x Time interaction for insomnia severity (F[3,76]=3.12, p=.03) after controlling for baseline inadequate sleep hygiene, indicating that the change in insomnia severity over time differed by intervention group. Specifically, fixed Group x Time interaction effects from mixed models revealed that CRT-I (b=3.07, SE=1.40, 95% CI [0.28, 5.86], p=.03; d=1.13), SCT (b=3.00, SE=1.35, 95% CI [0.32, 5.68], p=.03; d=1.66), and AC (b=2.57, SE=1.34, 95% CI [−0.10, 5.25], p=.06; d=0.90) groups experienced stronger treatment effects relative to NTC participants (reference group; d=0.45). Ancillary mixed modelling omitting NTC participants likewise yielded comparable treatment effects across conditions on insomnia symptom severity (F[2,35]=0.06, p=.95).

Table 3.

Adjusted Descriptive Statistics, Effect Sizes, and Repeated Measures Significance Testing (Group by Time) at Baseline and Follow-Up

	Insomnia Severity Index			Cognitive Pre-Sleep Arousal			Somatic Pre-Sleep Arousal
	Baseline	Follow-Up	d	Baseline	Follow-Up	d	Baseline	Follow-Up	d
Group	M (SD)	M (SD)		M (SD)	M (SD)		M (SD)	M (SD)
CRT-I (n = 12)	14.45 (4.18)	9.50 (5.44)	1.13	26.82 (5.98)	19.42 (8.43)	0.94	13.82 (5.42)	11.50 (2.81)	0.45
SCT (n = 14)	13.00 (2.57)	7.92 (4.29)	1.66	24.86 (5.89)	16.67 (5.79)	1.42	14.57 (3.61)	12.50 (4.40)	0.57
AC (n = 14)	15.38 (3.25)	11.15 (4.69)	0.90	23.08 (5.88)	20.31 (8.41)	0.41	15.31 (2.98)	13.69 (3.66)	0.49
NTC (n = 42)	13.17 (3.82)	11.33 (4.79)	0.45	26.10 (5.92)	21.19 (6.57)	0.75	15.19 (4.98)	13.69 (5.02)	0.33
Significance Testing	F(3,76)= 3.12*			F(3,77)= 1.68			F(3,78)= 0.20
Fixed Effects	NTC < CRT-I,* SCT,* AC†

Note. N = 82. CRT-I = cognitive refocusing treatment; SCT = stimulus control treatment; AC = awareness-based control; NTC = no-treatment control. Repeated measures mixed models yielded (a) significance testing of group by time interaction representing mean differences between baseline and follow-up assessments within-participants after controlling for baseline inadequate sleep hygiene, and (b) group comparisons representing estimates of fixed effects.

^*p < .05.

^†p < .10.

Pre-sleep arousal

Adjusted mean scores for baseline and follow-up pre-sleep cognitive and somatic arousal and effect sizes are presented in Table 4. Analyses showed a nonsignificant Group x Time interaction for cognitive pre-sleep arousal (F[3,77]=1.68, p=.18) after controlling for baseline inadequate sleep hygiene, although CRT-I (d=0.94) and SCT (d=1.42) groups experienced greater improvements in pre-sleep cognitive arousal reduction relative to AC (d=0.41) and NTC (d=0.75) groups. Analyses showed relatively weaker Group x Time interaction for somatic pre-sleep arousal (F[3,78]=0.20, p=.90) after controlling for baseline inadequate sleep hygiene within all groups (ds=0.33–0.57; see Table 3). Ancillary mixed modelling omitting NTC participants likewise yielded comparable treatment effects across conditions on cognitive (F[2,36]=2.44, p=.10) and somatic (F[2,35]=0.52, p=.60) pre-sleep arousal.

Discussion

This pilot randomized controlled trial tested the feasibility and efficacy of two single-session, electronic-based insomnia interventions, including CRT-I and SCT, as compared to an awareness-based control (AC) and non-randomized no-treatment control (NTC) group. Participants in CRT-I, SCT, and AC conditions endorsed comparable credibility/acceptability of intervention. All active conditions demonstrated reduction of insomnia symptom severity and pre-sleep cognitive arousal (but not somatic arousal) at one-month follow-up, with strongest effects observed for CRT-I and SCT groups. Although results are interpreted with caution due to the small sample and non-randomized NTC condition, these findings provide preliminary support for single-session CRT-I and SCT for college insomnia that is novel in terms of its electronic-based and peer-facilitated modality.

Feasibility

Overall, participants in the CRT-I condition endorse levels of treatment acceptability and credibility comparable to SCT and AC, highlighting CRT-I as conducive to this treatment modality among college students. Notably, the SCT intervention demonstrates strength over AC in terms of belief that the treatment is reasonable/logical for insomnia and confidence that the treatment will effectively decrease insomnia symptoms, providing support for credibility of this brief, electronic, 1-session intervention. Confidence in the peer interventionist was also strong, ranging from 90–95% across CRT-I, SCT, and AC conditions, demonstrating promising support for a peer-led college insomnia intervention.

Enactment of treatment recommendations was moderate to high in both the CRT-I and SCT groups. CRT-I involves relatively simple instructions and 77% of individuals engaged in the techniques at bedtime. In contrast, SCT entails various elements, thus it was hypothesized that adherence would be compromised due to students’ flexible schedules and suboptimal environments. Findings show mixed support for this position such that, while students are relatively adherent to recommendations such as staying away from the bed when not sleeping, avoiding napping, and going to bed when sleepy, they are only moderately adherent to recommendations for a consistent wake up time and leaving the bed when unable to sleep. These findings are consistent with a previous study finding that students adhered to, on average, 4–5 of their 6 stimulus control recommendations, and show moderate to strong adherence to SCT in this population.²²

The current findings supporting feasibility of brief, electronic, peer-led interventions are timely and relevant to the field of behavioral sleep medicine, suggesting solutions for cost-effective provision of accessible treatment for college sleep. Historically, the majority of college insomnia intervention has been intensive (i.e., multi-session), in-person, and expert-facilitated.^12,43 However, a recent meta-analysis highlighted comparable efficacy across diverse psychological interventions and delivery modalities for college sleep,¹³ suggesting that brief and self-help modality interventions performed comparably to more intensive and traditional interventions (e.g., CBT-I administered in individual therapy). Indeed, the rapidly-changing nature of the treatment landscape (e.g., push for electronic/telehealth intervention necessitated by the COVID-19 pandemic,⁷¹ including sleep treatment⁷²) and college learning environment (e.g., mainstreaming of online education⁷³ and transition to remote learning necessitated by the COVID-19 pandemic⁷⁴) necessitates ongoing feasibility evaluation and flexible adaptation of standard insomnia intervention. As a field, it is imperative to continue investigating comparison of low-cost and minimally-invasive college insomnia intervention and should examine the most efficient utilization of limited resources in terms of time, cost, and clinician expertise.

Treatment Effects

Based on existing empirical support for the efficacy of CRT-I,^41,42 cognitive theory of insomnia,^30,31 and college-specific environmental and lifestyle barriers to adhering to complex intervention protocols,¹² it was hypothesized that CRT-I would out-perform SCT and an awareness-based control (AC) in terms of insomnia symptom severity and cognitive (but not somatic) pre-sleep arousal. Contrary to hypotheses, CRT-I performed comparable to SCT and AC groups for improving insomnia severity and comparable to SCT for improving cognitive pre-sleep arousal. Notably, although it was hypothesized that improvements in pre-sleep arousal would yield subsequent improvements in sleep, it is alternatively possible that improvements in pre-sleep arousal are an artefact of improved sleep.

These novel findings suggest a common mechanism by which these single-session, electronic-based interventions intervene on both pre-sleep thoughts and thus insomnia symptoms. Both cognitive and behavioral interventions resulted in comparable results in this study, suggesting that behavioral interventions that do not directly target cognitive arousal may nonetheless reduce it. Consistent with previous research highlighting pre-sleep arousal as a mediator of insomnia intervention (e.g., CBT-I),⁷⁵ current findings support the theorized risk mechanism involving cognitive arousal underlying effects of brief intervention (both cognitive and behavioral) on clinical outcomes in college insomnia. Behavioral models of insomnia (such as stimulus control⁴⁸ and 3P/Speilman⁷⁶ models) highlight operant conditioning processes, such that repeatedly pairing sleep-related stimuli with wakefulness contributes to a perpetuating factor of conditioned wakefulness or arousal.⁷⁷ The neurocognitive model of insomnia posits that pre-sleep cortical arousal (behavioral intervention target) may actually be an analogue of pre-sleep cognitive arousal (cognitive intervention target),⁷⁸ suggesting that behavioral and cognitive insomnia interventions may act on a common perpetuating factor (i.e., pre-sleep cognitive activity). Continued investigation is needed to explicate mechanisms underlying treatment efficacy involving cognitive pre-sleep arousal to highlight active ingredients of the diverse brief, electronic, and peer-led interventions assessed in the current study.

Individuals in the CRT-I treatment group demonstrated a 5-point reduction in ISI, which is comparable to limited existing single-session college insomnia intervention.⁷⁹ Symptom reduction observed herein is only slightly less than literature recommendations for a 6-point reduction to represent a clinically meaningful improvement in insomnia symptoms among clinical adult samples,⁸⁰ again highlighting promise for continued research in larger samples. Several explanations may exist regarding comparable treatment effects across conditions on insomnia symptom severity, including the awareness-based control condition. First, findings may demonstrate the true potency of a brief intervention, regardless of content. One alternative explanation is that the awareness-based control condition utilized in this study involved active treatment ingredients related to mindfulness, and thus cognitive arousal. For example, previous literature has demonstrated the effects of pre-sleep mindfulness-based intervention⁸¹, and these findings may likewise indicate the potential benefits of a brief mindfulness-focused treatment. A second alternative explanation for this finding is placebo effect. That is, because the awareness-based control condition was framed in terms of anticipated sleep-related benefits, participants believed that it did indeed improve their sleep. Replication is needed among larger samples of college students to replicate findings and assess potential differences across diverse evidence-based sleep interventions delivered in this brief, electronic, and peer-led modality.

Current results extend findings from Gellis (2013)⁴¹ by comparing a single-session CRT-I intervention to a “gold standard” behavioral alternative intervention (SCT) and multiple control groups using an electronic-based approach, as compared to demonstrating supplemental utility over sleep hygiene (SH) psychoeducation (i.e., CRT-I + SH) using traditional “pencil-and-paper” modality. Although CRT-I intervention in this study demonstrated comparable treatment effects in terms of insomnia severity and cognitive pre-sleep arousal, a novel contribution of this study is that it demonstrates comparable performance of electronic-based CRT-I (without supplemental SH) with SCT (and AC). Also consistent with Gellis (2013),⁴¹ treatment effects for CRT-I in the current study are lower than a previous study using a 4-session CRT-I among veterans,⁴² despite controlling for other sleep disorders (via exclusion from analytic sample). Thus, it remains plausible that more sessions of CRT-I contributes to a greater treatment effect and ongoing research is needed to determine the ideal balance of treatment brevity and effect.

Strengths and limitations

This study benefitted from several methodological strengths. First, this study examined single-session, electronic-based, and peer-led interventions, which represents an incremental contribution to the literature and speaks to tailoring to the current needs of college student population. Second, this study examined day-to-day treatment enactment within the natural college environment, thus providing ecological validity and valuable insight into treatment effectiveness (versus strictly experimental efficacy). Third, several strategies were implemented to induce and assess treatment implementation, including specific processes of delivery, receipt, and enactment (i.e., compliance).⁵⁴ Consistent with strategies utilized herein, future treatment enactment among college students may be optimized by targeting and assessing initial treatment receipt (e.g., handouts of key points, verbal prompts to encourage attendance to key treatment ingredients, quizzing participants to confirm understanding, and rating confidence in enacting treatment) as well as soliciting daily self-reports for external accountability.^53,54 Fourth, inclusion of the awareness-based control group enabled superior experimental control over diverse intervention and assessment conditions^49,50 (but see also Limitations, below). Lastly, this study assessed and statistically controlled for completion of treatment over major university breaks as a potential confound to treatment effects.

Notwithstanding novel and significant contributions to the literature, limitations of the current study must be considered. First, data were drawn from a small pilot sample of predominantly White and first-year sample at a private, north-eastern university during the fall academic semester; thus, generalizability to more heterogeneous samples throughout the academic year remains uncertain. Relatively low internal reliability demonstrated in the current pilot study across measures may be attributable to the small sample size,⁸² further supporting replication in larger samples. Further, although participants reported mental health problems consistent with clinical insomnia samples, it remains uncertain if these results of non-treatment-seeking college students would generalize to a community-based clinical sample. Second, although students who reported concurrent insomnia treatment during the assessment period were excluded from analysis, information on concurrent psychotherapy and medication usage was not assessed; thus, it is uncertain whether extraneous intervention may have supported or dampened effects of the current sleep interventions. Third, despite comparable treatment effects across conditions observed herein, the current inclusion criteria targeting candidacy for a cognitive insomnia intervention (that is, “difficulty turning off your mind at bedtime”) may inadvertently confer bias preferencing cognitive (versus behavioral) interventions, thereby necessitating replication is more generalized samples. Fourth, although every effort was made to recruit a control sample of comparable demographics as intervention groups, the current control group was recruited non-randomly at a later date and thus results may be influenced by time contexts. In addition, self-report measures may be vulnerable to memory error and/or social desirability bias. Lastly, current results of insomnia symptom severity may only generalize to insomnia-specific sleep problems (versus other sleep problems and adverse experiences).

Future directions

Results of this study may inform future research. First, to investigate generalizability of current findings, replication is warranted with (a) more heterogeneous college samples (e.g., race/ethnicity, socioeconomic status) across both fall and spring semesters and (b) community-based clinical insomnia sample. Second, to minimize confounding influences, future replications should assess concurrent mental health intervention and treatment fidelity. Third, future investigations may use objective sleep assessment (e.g., actigraphy) to assess convergent validity with participant self-reports. Fourth, research with multiple follow-up assessments is needed to explicate the temporal ordering of observed treatment outcomes using mediation path analysis (i.e., bedtime cognitive arousal as a mediator of treatment effects on college insomnia symptoms), which may inform clinical treatment recommendations. Fifth, future studies might assess efficacy of strategies targeting observed predictors of adherence to brief behavioral insomnia intervention among college students to potentially improve task enactment rates, such as discouraging alcohol intake, enhancing treatment-related self-efficacy, and monitoring and providing feedback on sleep, early in treatment.²² Sixth, regarding peer-based intervention methods, future studies may additionally (a) randomly assign peer interventionists across treatment groups, (b) investigate and control for interventionist effects, and (c) develop recordings so that an independent coder may confirm that the correct intervention was delivered. Lastly, continued research is needed comparing CRT-I delivery durations to further justify the use of single-session modality.

Conclusions

This pilot randomized clinical trial furthers the literature on CRT-I for college students by assessing feasibility and exploring treatment effects (compared to stimulus control treatment and control groups) using a single-session, electronic-based, peer-facilitated approach. Results highlight comparable perceived feasibility and treatment effect across both cognitive and behavioral intervention. Results also suggest that a single-session, electronic-based, peer-led intervention targeting pre-sleep arousal may be efficacious in reducing insomnia symptom severity and pre-sleep cognitive arousal over one month, regardless of the specific intervention. Although replication is needed, current preliminary findings may guide future research on mechanisms of brief insomnia intervention and ultimately guide the ongoing tailoring of clinical treatment recommendations for college insomnia.