Brief Behavioral Treatment for Insomnia vs. Cognitive Behavioral Therapy for Insomnia: Results of a Randomized Noninferiority Clinical Trial Among Veterans

Abstract

The goal of this study was to compare a brief behavioral treatment for insomnia (BBTI), which has fewer sessions (4), shorter duration (<30–45 minutes), and delivers treatment in-person plus phone calls to cognitive behavioral therapy for insomnia (CBTI), which has 5 in-person sessions. The hypothesis was BBTI would be noninferior to CBTI. The Reliable Change Index was used to establish a noninferiority margin (NIM) of 3.43, representing the maximum allowable difference between groups on the pre-post Insomnia Severity Index change (ΔISI). Sixty-three veterans with chronic insomnia were randomized to either BBTI or CBTI and veterans in both groups had significant reductions of their insomnia severity per the ISI and improved their sleep onset latency, total wake time, sleep efficiency, and sleep quality per sleep diaries. While CBTI had a larger pre-post ΔISI, this was notsignificantly different than ΔISIBBTI and was less than the NIM. However, the 95% confidence interval of the between group pre-post ΔISI extended beyond the NIM, and thus BBTI was inconclusively noninferior to CBTI. Limitations, such as small sample size and high rate of dropout, indicate further study is needed to compare brief, alternative yet complementary behavioral insomnia interventions to CBTI. Still, evidence-based brief and flexible treatment options will help to further enhance access to care for veterans with chronic insomnia, especially in non-mental-health settings like primary care.

Chronic insomnia is a highly prevalent condition in the general population, impacting approximately 10%–22% of adults (Bramoweth & Germain, 2013). There is even greater prevalence among military service members and veterans, affecting over 25%–50% (Bramoweth & Germain, 2013; Troxel et al., 2015). Insomnia significantly impairs nighttime sleep quality, daytime function, and overall quality of life (Troxel et al., 2015). Furthermore, insomnia is a common comorbidity for psychiatric and medical disorders and an independent risk factor for developing depression, posttraumatic stress disorder (PTSD), suicidal behaviors, and cardiometabolic disorders (Hertenstein et al., 2019; Pigeon, Britton, Ilgen, Chapman, & Conner, 2012; Troxel et al., 2010). Additional burdens include increased healthcare utilization and healthcare costs (Wickwire et al., 2019).

There is a strong evidence-base for cognitive-behavioral treatment of insomnia. The recommended first-line intervention is cognitive behavioral therapy for insomnia (CBTI), a multimodal psychotherapy (Qaseem et al., 2016). CBTI typically consists of: psychoeducation about normal and abnormal sleep (e.g., sleep stages, two-process model of sleep; Borbely, Daan, Wirz-Justice, & Deboer, 2016), insomnia etiology (e.g., 3P model; Spielman, Caruso, & Glovinsky, 1987), and sleep health education; behavioral interventions, including sleep restriction (Spielman, 2011), stimulus control (Bootzin & Nicassio, 1978), and relaxation strategies (e.g., progressive muscle, autonomic, guided imagery; Borkovec & Weerts, 1976); and cognitive therapy (Harvey, 2005). While treatment can vary, CBTI is considered relatively brief, typically delivered in 4–8 sessions occurring weekly to biweekly. Treatment can be effectively delivered in person (one-on-one, groups), using telehealth (video, phone calls), online, and through workbooks (Arnedt et al., 2014; Castronovo et al., 2018; Ritterband et al., 2017; Trauer, Qian, Doyle, Rajaratnam, & Cunnington, 2015; Ulmer et al., 2018).

Within the U.S. Department of Veterans Affairs (VA), a nationwide effort to train providers to deliver CBTI has been ongoing since 2011 (Manber et al., 2013). As of June 2019, 872 providers, mainly psychologists but also psychiatrists, mental health nurse practitioners, and master’s-level mental health counselors, have been trained (Department of Veterans Affairs, Mental Health Services, 2019). Treatment outcomes from the initial cohort of VA CBTI trainees is consistent with previous randomized clinical trials (RCTs) of CBTI. Sixty percent of veterans who completed treatment achieved a treatment response with an Insomnia Severity Index (ISI) reduction of ≥8 points (pre = 20.7, post = 10.9; Cohen’s d = 2.3; Trockel, Karlin, Taylor, & Manber, 2014).

Despite this significant dissemination of training, and the clinically relevant symptom reductions achieved, accessing care can still be problematic for many veterans and barriers remain that prevent timely access to evidence-based insomnia care. Key barriers include: (a) a continued shortage of CBTI-trained clinicians to meet the needs of likely millions of veterans with an insomnia complaint; (b) CBTI is often limited to mental health clinics with the associated stigma surrounding treatment of mental health problems; (c) the persisting view among many health care providers that insomnia is merely a symptom of another disorder (e.g., depression, pain), which can limit accurate diagnosis and referral for evidence-based care; (d) a lack of patient and provider knowledge regarding CBTI; and (e) medical provider dissatisfaction with available treatment options (Ulmer et al., 2017). There are also common barriers to attend medical appointments such as distance to travel and access to transportation, work schedules, and caregiving duties. These barriers can be further exacerbated as nearly 25% of all veterans live in rural settings (~4.7 million; Department of Veterans Affairs, Office of Rural Health, 2019), which means further separation from expertise in insomnia that is often concentrated in urban settings.

Overcoming these barriers and enhancing access to insomnia care will require alternative, flexible, and complementary evidence-based treatment options. Brief Behavioral Treatment for Insomnia (BBTI) is an evidence-based adapted version of CBTI that enhances the ease of provider training and offers flexible treatment delivery (phone and in-person) of behavioral treatment components (stimulus control and sleep restriction) in a concise 4-session format, delivered over 4 to 5 weeks (Troxel, Germain, & Buysse, 2012). BBTI was developed to further enable broad delivery and dissemination of evidence-based insomnia treatments in general medical settings like primary care. BBTI, and other variants of brief CBTI, have shown moderate to large effect sizes compared to controls (e.g., ISI and PSQI; es=0.72 – 1.3) in RCTs involving community members (Buysse et al., 2011; Edinger, Wohlgemuth, Radtke, Coffman, & Carney, 2007) and military service members and veterans (Germain et al., 2014; Pigeon, Funderburk, Bishop, & Crean, 2017). Furthermore, a recent meta-analysis of CBTI, including several studies with ≤4 treatment sessions, found a large effect size (Hedges’s g = 0.98) for the reduction of the ISI from pre- to posttreatment (van Straten et al., 2018).

To move towards broader dissemination and implementation, it is important to compare BBTI directly against the gold-standard treatment, CBTI, especially considering prior research has found brief treatments, ≤4 sessions, yield similar outcomes as 8 sessions of CBTI (Edinger et al., 2007). This analysis presents the findings of the quantitative arm of the parent hybrid type 1 trial (Bramoweth, Germain, Youk, Rodriguez, & Chinman, 2018), a noninferiority RCT meant to establish that an “experimental” treatment, BBTI, is no worse than the “standard” approach, CBTI. We compared BBTI vs. CBTI with the hypothesis that BBTI will be noninferior to CBTI.

Materials and Methods

PARTICIPANTS

Veterans with chronic insomnia were recruited from a large urban VA Medical Center (VAMC) and the surrounding community to participate in the RCT. Eligible veterans were ≥18 years, met Diagnosticand Statistical Manual of Mental Disorders, 5^th edition (DSM-5) criteria for insomnia disorder (First, Williams, Karg, & Spitzer, 2015), and had an ISI ≥15 (Bastien, Vallieres, & Morin, 2001). Veterans were excluded if they had current or past bipolar, psychotic, or seizure disorder; alcohol use disorder or substance use disorder in the past 6 months; other current, severe, and unstable and/or untreated psychiatric and/or medical disorders; been hospitalized, for any reason, in the past month; untreated sleep apnea and/or other untreated noninsomnia sleep disorders; moderate to severe cognitive impairment; homeless or unstable housing; shift workers; or if female, were pregnant. This study was designed to enroll veterans who would typically be eligible to engage in CBTI at a VAMC and was similar to the guidelines as part of the VA CBTI training program (Manber et al., 2014).

PROCEDURES

All study procedures took place at VA Pittsburgh Healthcare System (VAPHS); recruitment, treatment, posttreatment, and 3-month follow-up took place May 2016 through June 2019, with 12-month follow-up through March 2020. Participants were recruited with advertisements posted at VAPHS and sleep clinics at the University of Pittsburgh, by invitation from practicing insomnia providers at VAPHS, and with postings on Craigslist.com. Eligible veterans, after a brief phone or in-person screen, were invited for an in-person appointment to sign informed consent and complete diagnostic interviews and self-report measures. Assessment of insomnia and other sleep disorders was conducted using the South Texas Research Organizational Network Guiding Studies on Trauma and Resilience (STRONG STAR) Clinical Interview for DSM-5 Sleep-Wake Disorders (Taylor et al., 2018). Risk for obstructive sleep apnea (OSA) was assessed using the STOP-BANG questionnaire (Chung et al., 2012); veterans scoring ≥5 and/or had untreated OSA were excluded and referred to the VAPHS Sleep Medicine Clinic. Cognitive impairment was assessed using the St. Louis University Mental Status Examination (SLUMS; Tariq, Tumosa, Chibnall, Perry, & Morley, 2006); veterans with scores ≤20, indicative of serious cognitive impairment, were excluded. Finally, psychiatric disorders were assessed using the Structured Clinical Interview for DSM-5, Research Version (SCID-5-RV; First et al., 2015).

Veterans were randomized to BBTI or CBTI in a 1:1 manner, in random block sizes of 2, 4, or 6, stratified by age (18–64 or ≥65) and use of a prescription sedative-hypnotic medication (yes or no). Sequentially numbered envelopes were used to assign participants based on their strata with concealment until randomization. Participants were enrolled by the Principal Investigator (ADB) with the randomization strategy developed by the study statistician (AOY) using Stata v14 and group assignment by the study coordinator (LGL).

Sample size was calculated based on noninferiority methods (Julious, 2004) and was further informed by the calculation of a noninferiority margin (NIM), the prespecified difference between treatments that demonstrates BBTI is not unacceptably worse than CBTI (Schumi & Wittes, 2011). The Reliable Change Index (RCI; Jacobson & Truax, 1991; Sloan, Marx, Lee, & Resick, 2018) measures change at posttreatment not due to error, and for this study was calculated for the “control” condition (CBTI) using VA rollout data (Trockel et al., 2014). The RCI equation uses a measure of reliability (internal consistency or test-retest reliability) for the outcome measure, in this case the ISI. The RCI for CBTI was calculated using both reliability metrics to create a range of NIMs. Using Cronbach’s alpha (α = 0.905; Morin, Belleville, Belanger, & Ivers, 2011) the RCI was 5.21 and using the test-retest reliability (r = 0.78; Savard, Savard, Simard, & Ivers, 2005) the RCI was 3.43. We chose the NIM of 3.43. Accounting for 25% attrition, based on prior CBTI and BBTI studies (Buysse et al., 2011; Germain et al., 2014; Trockel et al., 2014) the target randomization goal was 56 participants (n = 28/group) with 42 (n = 21/ group) completing posttreatment assessment to achieve 80% power.

INTERVENTIONS

Study clinicians attended a half-day training by the Principal Investigator (ADB) and study Co-Investigator (AG); CBTI and BBTI trainings were separate and study clinicians only received one training. CBTI (Manber et al., 2014) was delivered in 5 in-person sessions, weekly or bi-weekly, with a goal of completing treatment in 8 weeks. The components included psycho-education, sleep health education, stimulus control, sleep restriction, cognitive restructuring of maladaptive beliefs, and relaxation strategies as needed. Sleep diaries were completed daily throughout treatment, beginning with the baseline assessment. Increasing prescribed time in bed (TIB) occurred if sleep efficiency (SE) was ≥85%, in which case TIB was increased 15 minutes (if wanted by participant). CBTI was delivered by two licensed psychologists, one who delivered care in the mental health clinic and another in primary care.

BBTI was adapted for veterans and military service members based on the protocol developed by Troxel et al. (2012). BBTI was delivered in 4 sessions with a goal of completing treatment in 5 weeks. Sessions 1 and 3 occurred in person and Sessions 2 and 4 were brief phone calls. BBTI differed from CBTI in that it focused on the behavioral aspects of treatment (i.e., stimulus control and sleep restriction), the sessions were fewer and briefer, and delivery included both in-person and phone follow-ups. Like with CBTI, sleep diaries were completed after each session but TIB titration did not occur until Session 3 and 4 and followed the 30/30 rule: if sleep onset latency (SOL) and wake after sleep onset (WASO) were both routinely < 30 minutes, the prescribed TIB was extended by 15 minutes. BBTI was delivered by two licensed psychologists and one postdoctoral fellow under the supervision of a licensed psychologist. All BBTI sessions were delivered in the primary care setting. Neither CBTI nor BBTI treatment sessions were required to be weekly due to barriers in clinic scheduling. This was similar to current CBTI schedules at VAPHS, thus supporting clinical ecological validity.

MEASURES

See Table 1 for information on self-report measures. The primary outcome measure was the ISI (Bastien et al., 2001), administered at baseline, in-person treatment sessions, 1-week posttreatment, and the 3-month and 12-month follow-up assessments. The Consensus Sleep Diary (Carney et al., 2012) was used to collect data on sleep behaviors, including bedtimes, rise times, SOL, nighttime awakenings (NWAK), WASO, TIB, total sleep time (TST), total wake time (TWT), sleep efficiency (SE), and sleep quality (SQ). Diaries were administered for 2 weeks of baseline, during treatment, 2 weeks after the posttreatment assessment, and the 3-month and 12-month assessments. Additional sleep measures administered at baseline, posttreatment, 3-month and 12-month assessments included the Pittsburgh Sleep Quality Index (PSQI; Buysse, Reynolds, Monk, Berman, & Kupfer, 1989) as a measure of sleep quality, the Dysfunctional Beliefs and Attitudes about Sleep scale (DBAS; Morin, Vallieres, & Ivers, 2007) to assess sleep-disruptive cognitions, and the Epworth Sleepiness Scale (ESS; Johns, 1991) to assess daytime sleepiness.

Table 1.

Self-Report Measures

Measure	Items	Score Range	Desired Direction
ISI	7	0–28	↓
PSQI	18	0–21	↓
ESS	8	0–24	↓
DBAS‖	16	0–10	↓
PHQ9‡	8	0–24	↓
GAD7	7	0–21	↓
PCL5‡	19	0–76	↓
Fatigue†	6	33.4–76.8	↓
PEG	3	0–30	↓
QOL-PH†	4	16.2–67.7	↑
QOL-MH†	4	21.2–67.6	↑
WSAS	5	0–40	↓
PGIC	7	1–7	↑

Notes.

^‖average score

^‡sleep item removed

^†T scores ISI, Insomnia Severity Index

PSQI, Pittsburgh Sleep Quality Index; ESS, Epworth Sleepiness Scale; DBAS, Dysfunctional Beliefs and Attitudes about Sleep; PHQ9, Patient Health Questionnaire; GAD7, Generalized Anxiety Disorder; PCL5, Posttraumatic Stress Disorder Checklist; PEG, pain intensity, interference with enjoyment of life, and interference with general activity; QOL-PH, Quality of Life – Physical Health; QOL-MH, Quality of Life – Mental Health; WSAS, Work and Social Adjustment Scale; PGIC, Patient Global Impression of Change.

Self-report measures of psychiatric symptoms included the Patient Health Questionnaire 9 (PHQ9; Kroenke, Spitzer, & Williams, 2001) to assess depression, the Generalized Anxiety Disorder 7 (GAD7; Spitzer, Kroenke, Williams, & Lowe, 2006) to assess anxiety, and the PTSD Checklist for DSM-5 (PCL5; Blevins, Weathers, Davis, Witte, & Domino, 2015) to assess symptoms of PTSD. Additional measures included the Patient Reported Outcome Measurement Information System (PROMIS) Fatigue Scale (Reeve et al., 2007) to assess fatigue and its impact on functioning and quality of life, a brief pain scale measuring pain intensity (P), interference with enjoyment of life (E), and interference with general activity (G [PEG]; Krebs et al., 2009), the PROMIS Global Health Scale (Hays, Bjorner, Revicki, Spritzer, & Cella, 2009) to measure the impact of physical health (QOL-PH) and mental health (QOL-MH) on quality of life, the Work and Social Adjustment Scale (WSAS; Mundt, Marks, Shear, & Greist, 2002) to measure psychosocial functioning impairment due to insomnia, and the Patient Global Impression of Change (PGIC; Hurst & Bolton, 2004). Psychiatric symptom and psychosocial measures were administered at baseline, post-treatment, and 3-month and 12-month assessments.

DATA ANALYSES

Analyses were conducted with STATA v14. Baseline differences between CBTI and BBTI were analyzed using chi-square tests and t-tests. To test treatment effectiveness, an intent-to-treat (ITT, n = 63) and per-protocol (PP, n = 24) approach were used. Consistent with ITT, all individuals randomized, regardless of their completion status, were encouraged to complete the posttreatment and follow-up assessments. PP analyses were conducted on veterans in each group that met strict criteria—attending all treatment sessions within the time frame defined a priori (BBTI: n = 12, 4 sessions in 5 weeks; CBTI: n =12, 5 sessions in 8 weeks) and completing the 1-week posttreatment assessment. Linear mixed models (LMM) were used to test the main effects of treatment (CBTI vs. BBTI), time (baseline vs. posttreatment), and the interaction of Treatment × Time. Effect sizes were computed using Glass’s Δ (|M_pre − M_post|/SD_pre; Morris, 2008). Treatment response was an ISI reduction at posttreatment of ≥8 points (indicating a moderate response), and remission was a treatment response plus an ISI score ≤7 at posttreatment (indicating no clinically significant insomnia symptoms; Morin et al., 2011). To establish BBTI as noninferior to CBTI, the upper bound of the 95% confidence interval of the mean ΔISI from pre- to posttreatment (ΔISI_CBTI − ΔISI_BBTI) cannot exceed the NIM of 3.43. If the upper bound of the confidence interval exceeds the NIM, noninferiority cannot be declared (inconclusive). If the lower bound of the confidence interval exceeds the NIM, BBTI will be declared inferior to CBTI. However, should the upper bound of the confidence interval not exceed zero, BBTI will be declared superior to CBTI (Aberegg, Hersh, & Samore, 2018).

Results

PARTICIPANTS

See Figure 1 for study CONSORT and Table 2 for baseline demographics. Sixty-three veterans were randomized to treatment and 38 completed treatment and the 1-week posttreatment follow-up assessment. The dropout rate for BBTI was 29% (9/31) and for CBTI was 50% (16/32). Both groups saw the majority of dropout between baseline and Session 1. The only baseline demographic difference was that veterans randomized to CBTI had a significantly higher service connection rating than those randomized to BBTI (70% vs. 45%, p = 0.014).

FIGURE 1.

CONSORT flow chart

Table 2.

Baseline Demographics Characteristics

	All (n=63)		BBTI (n=31)		CBTI (n=32)		Statistical Findings*
Age (years), mean (sd)	55.1	(14.4)	56.8	(13.7)	53.4	(15.2)	p = 0.345
Gender, n (%)
Male	57	(90.5)	29	(93.6)	28	(87.5)	p = 0.672
Female	6	(9.5)	2	(6.4)	4	(12.5)
Race, n (%)							p=0.809
White	50	(79.4)	24	(77.4)	26	(81.3)
Black	11	(17.5)	6	(19.4)	5	(15.6)
Asian	1	(1.6)	1	(3.2)	0	(0.0)
Other	1	(1.6)	0	(0.0)	1	(3.1)
Ethnicity, n (%)							p=0.999
Hispanic/Latino	2	(3.2)	1	(3.2)	1	(3.1)
Not Hispanic/Latino	53	(84.1)	26	(83.9)	27	(84.4)
Missing	8	(12.7)	4	(12.9)	4	(12.5)
Marital Status, n (%)							p=0.168
Married	19	(30.2)	7	(22.6)	12	(37.5)
Divorced/Separated	23	(36.5)	15	(48.4)	8	(25.0)
Single/Never Married	5	(7.9)	1	(3.2)	4	(12.5)
Widowed	15	(23.8)	7	(22.6)	8	(25.0)
Other	1	(1.6)	1	(3.2)	0	(0.0)
Education (years), mean (sd)	13.3	(3.9)	13.5	(3.6)	13.1	(4.2)	p=0.743
Military Branch, n (%)							p=0.726
Army	31	(49.2)	15	(48.4)	16	(50.0)
Air Force	11	(17.5)	7	(22.6)	4	(12.5)
Marine Corps	8	(12.7)	3	(9.7)	5	(15.6)
Navy	13	(20.6)	6	(19.4)	7	(21.9)
Military Era, n (%)
Pre-Vietnam/Vietnam	28	(44.4)	14	(45.2)	14	(43.8)	p=0.866
Post-Vietnam	16	(25.4)	7	(22.6)	9	(28.1)
Persian Gulf/OEF/OIF/OND	19	(30.2)	10	(32.3)	9	(28.1)
Time in Service (years), mean (sd)	8.2	(8.5)	7.9	(7.8)	8.5	(9.2)	p=0.802
Combat							p=0.376
Yes	33	(52.4)	14	(45.2)	19	(59.4)
No	29	(46.0)	16	(51.6)	13	(40.6)
Missing	1	(1.6)	1	(3.2)	0	(0.0)
Service Connected, mean (sd)							p=0.014
%	57	(31)	45	(30)	70	(28)

Notes:

^*Based on a t-test for continuous variables or chi-square test for categorical variables (or Fisher’s Exact for cells <5).

BBTI, Brief Behavioral Treatment for Insomnia; CBTI, Cognitive Behavioral Therapy for Insomnia; OEF, Operation Enduring Freedom; OIF, Operation Iraqi Freedom; OND; Operation New Dawn.

For clinical differences at baseline, veterans in CBTI had higher rates of PTSD/Other Traumatic Stress (CBTI: n = 9, 28.1%; BBTI: n = 2, 6.5%; p = 0.043) and any psychiatric disorder (CBTI: n = 17, 53.1%; BBTI: n = 7, 22.6%; p = 0.019). More veterans randomized to CBTI met criteria for major depressive disorder (CBTI: n = 10, 31.3%; BBTI: n = 3, 9.7%), although this was not significant. Also, veterans randomized to CBTI had higher PCL5 scores (sleep item removed; CBTI = 26.4, BBTI = 16.7, p = 0.025). There were no significant differences between veterans in CBTI and BBTI on comorbid sleep disorders with the most common being OSA that was adequately managed on positive airway pressure (CBTI: n = 11, 34%; BBTI: n = 9, 29%).

TREATMENT OUTCOMES, GROUP DIFFERENCES, AND NONINFERIORITY

For the ITT analysis, per the ISI, veterans randomized to both BBTI and CBTI had significantly reduced insomnia symptoms at 1-week posttreatment (see Figure 2 and Table 3). For treatment response, 68.2% (n = 15/22) met criteria in BBTI and 76.5% (n = 13/17) in CBTI. For remission from insomnia, 31.8% (n = 7/22) met criteria in BBTI and 23.5% (n = 4/17) in CBTI. Veterans in both treatments had significant reductions on the PSQI and DBAS as well.

FIGURE 2.

Baseline, treatment session, and posttreatment ISI for BBTI and CBTI

Table 3.

Baseline and Posttreatment Sleep and Psychosocial Self-Report Outcomes by Treatment Arm

	BBTI Pre		BBTI Post				CBTI Pre		CBTI Post				ΔCBTI-ΔBBTI es
	M	(SD)	M	(SD)	Δ	es	M	(SD)	M	(SD)	Δ	es
Sleep	n=31		n=22				n=32		n=17
ISI	19.8	(3.5)	11.0	(6.5)	−8.8	2.5 ″	20.6	(3.7)	9.6	(4.9)	−11.0	3.0 ″	0.5
PSQI	12.3	(4.0)	8.4	(4.3)	−3.9	1.0 ″	12.9	(3.5)	6.8	(3.5)	−6.1	1.7 ″	0.7
ESS	8.2	(4.8)	9.4	(5.0)	1.2	0.3	9.3	(5.6)	9.2	(4.7)	−0.1	0.0	−0.3
DBAS	5.5	(1.9)	4.4	(1.6)	−1.1	0.6 ′	5.9	(1.6)	4.6	(1.9)	−1.3	0.8 ′	0.2
Psychosocial	n=31		n=22				n=32		n=17
PHQ9‡	7.4	(5.1)	5.1	(4.9)	−2.3	0.5	9.3	(5.6)	5.1	(4.3)	−4.2	0.8 ′	0.3
GAD7	6.6	(5.1)	5.0	(4.7)	−1.6	0.3	7.8	(5.1)	5.2	(3.5)	−2.6	0.5	0.2
PCL5‡	16.7	(13.2)	14.8	(13.4)	−1.9	0.1	26.4	(19.5)	14.2	(14.1)	−12.2	0.6 ′	0.5
Fatigue	61.1	(8.0)	54.2	(11.4)	−6.9	0.9 ″	62.1	(8.3)	53.4	(9.3)	−8.7	1.0 ″	0.1
PEG	15.5	(7.2)	14.6	(8.2)	−0.9	0.1	16.3	(8.6)	13.2	(8.6)	−3.1	0.4	0.3
QOL-PH	37.6	(6.5)	41.4	(8.1)	3.8	0.6 ′	37.1	(7.4)	41.8	(8.3)	4.7	0.6 ′	0.0
QOL-MH	44.1	(9.2)	45.8	(9.2)	1.7	0.2	41.4	(8.3)	45.6	(8.6)	4.2	0.5	0.3
WSAS	19.0	(11.6)	12.5	(10.1)	−6.5	0.6 ′	19.9	(8.9)	10.8	(9.1)	−9.1	1.0 ″	0.4
PGIC	1.7	(1.1)	4.7	(1.9)	3.0	2.7 ″	1.9	(1.3)	5.0	(1.6)	3.1	2.4 ″	−0.3

Notes. Δ, post – pre; p-value based on linear mixed model; es, Glass’s Δ=|M_pre – M_post|/SD_pre

^‡sleep item removed

^′p<.01

^″p<.001

BBTI, Brief Behavioral Treatment for Insomnia; CBTI, Cognitive Behavioral Therapy for Insomnia; ISI, Insomnia Severity Index; PSQI, Pittsburgh Sleep Quality Index; ESS, Epworth Sleepiness Scale; DBAS, Dysfunctional Beliefs and Attitudes about Sleep; PHQ9, Patient Health Questionnaire; GAD7, Generalized Anxiety Disorder; PCL5, Posttraumatic Stress Disorder Checklist; PEG, pain intensity, interference with enjoyment of life, and interference with general activity; QOL-PH, Quality of Life – Physical Health; QOL-MH, Quality of Life – Mental Health; WSAS, Work and Social Adjustment Scale; PGIC, Patient Global Impression of Change.

Per the sleep diaries, veterans in both groups significantly improved their SOL, TWT, SE, and SQ, and reduced their SOL and WASO by ≥50% (see Table 4). While only veterans in CBTI significantly improved their WASO, EMA, and TST, these sleep diary variables were not significantly different from veterans in BBTI at posttreatment. On psychosocial outcomes, veterans in both treatment groups had significant improvements on the PROMIS Fatigue, QOL-PH, WSAS, and PGIC (see Table 3). Only veterans in CBTI had significant improvement on the PHQ9 and PCL5 (sleep items removed), but posttreatment values were not significantly different from veterans in BBTI.

Table 4.

Baseline and Posttreatment Sleep Diary Outcomes by Treatment Arm

	BBTI Pre		BBTI Post				CBTI Pre		CBTI Post				ΔCBTI-ΔBBTI es
	M	(SD)	M	(SD)	M	Δ	es	(SD)	M	(SD)	Δ	es
Sleep Diary	n=25		n=9				n=17		n=12
SOL	51.9	(42.6)	20.8	(10.9)	−31.1	0.7 ′	50.5	(29.0)	23.5	(22.0)	−27.0	0.9 ′	0.2
WASO	39.7	(38.3)	18.5	(15.9)	−21.2	0.6	55.0	(45.5)	27.4	(33.0)	−27.6	0.6 ′	0.0
EMA	25.9	(32.2)	12.8	(20.2)	−13.1	0.4	37.9	(40.7)	10.7	(14.7)	−30.2	0.8 ′	0.4
TIB	452.0	(90.6)	449.0	(37.6)	−3.0	0.0	497.2	(94.1)	489.3	(77.7)	−7.9	0.1	0.1
TST	314.6	(91.0)	381.6	(60.0)	67.0	0.7	344.9	(58.6)	410.3	(67.6)	65.4	1.1 ″	0.4
TWT	143.2	(82.2)	67.5	(29.9)	−75.7	0.9 ′	152.3	(62.9)	78.9	(35.2)	−73.4	1.2 ″	0.4
SE	67.0	(22.5)	84.6	(7.8)	17.6	0.8 ′	70.2	(8.9)	83.9	(6.1)	13.7	1.5 ″	0.7
SQ	2.3	(0.9)	2.8	(0.8)	0.5	0.6 ′	2.4	(0.4)	3.4	(0.7)	1.0	2.5 ″	1.9

Notes. Δ, post – pre; p-value based on linear mixed model; es, Glass’s Δ=|M_pre – M_post|/SD_pre

^′p<.01

^″p<.001

BBTI, Brief Behavioral Treatment for Insomnia; CBTI, Cognitive Behavioral Therapy for Insomnia; SOL, sleep onset latency; WASO, wake after sleep onset; EMA, early morning awakenings; TIB, time in bed; TST, total sleep time; TWT, total wake time; SE, sleep efficiency; SQ, sleep quality.

For noninferiority, veterans in CBTI had a 2.2-point greater mean reduction on the ISI compared to BBTI, which was below the NIM of 3.43. However, the upper bound of the 95% confidence interval (−0.9, 5.6) exceeded the NIM, resulting in an inconclusive noninferiority determination. As a sensitivity analysis, we conducted a noninferiority analysis using an effect size of 0.5 as the NIM (ΔCBTI es – ΔBBTI es < 0.5) and found similar results. The difference was less than 0.5 but the upper bound of the confidence interval (−0.20, 1.06) exceeded the NIM.

The PP analyses of self-report measures found similar results. Sleep diaries were not analyzed due to the small sample size. Veterans in both groups had significant reductions on the ISI, PSQI, WSAS, Fatigue, and PGIC. Veterans in CBTI had additional improvements on QOL-PH and DBAS, although these were not different from BBTI at posttreatment. The PP noninferiority analysis was also inconclusive with veterans in CBTI achieving a 3.0-point greater mean reduction on the ISI than the BBTI group, which was below the NIM of 3.43. However, like the ITT analyses, the 95% confidence interval (−0.6, 7.5) still exceeded the NIM.

Discussion

CBTI is well-established as an effective treatment for chronic insomnia and is the first-line, recommended intervention for adults with insomnia (Qaseem et al., 2016). Despite this high level of evidence, CBTI remains an underutilized intervention with numerous barriers that prevent it from meeting the needs of many individuals with chronic insomnia. BBTI was developed as an alternative to CBTI, and represents an evolution of CBTI that is briefer, more flexible, and may help further increase access to insomnia care, notably in more general medical settings, such as primary care (Troxel et al., 2012). If determined to be an effective and/or noninferior treatment to CBTI, training Primary Care Mental Health Integration (PCMHI) providers to deliver BBTI, especially in the VA where these providers are common, could help greatly increase the number of trained providers and the availability of evidence-based insomnia care. By doing so, treatment will no longer be primarily siloed in specialty sleep or mental health clinics. Furthermore, training PCMHI providers can help improve insomnia education efforts for primary care providers and patients (e.g., insomnia is not always a symptom of other disorders). PCMHI providers are valuable resources about common behavioral and mental health problems, including prevalent disorders like insomnia.

The few BBTI, and abbreviated CBTI, trials to date show good efficacy with reductions of insomnia severity similar to CBTI trials (Buysse et al., 2011; Edinger et al., 2007; Germain et al., 2014; Pigeon et al., 2017). Yet, these BBTI trials have been relatively small and BBTI has not been directly compared to CBTI, an important step towards broader dissemination and implementation. In this noninferiority RCT, within-group BBTI results showed significant reduction of insomnia severity and improvements on sleep diary variables similar to prior studies of BBTI, abbreviated CBTI, and the VA CBTI rollout (Buysse et al., 2011; Edinger et al., 2007; Germain et al., 2014; Pigeon et al., 2017; Trockel et al., 2014). Despite these findings, as there was no true control group, small sample size, high rate of dropout, and unbalanced randomization, caution is required when interpreting the BBTI findings. Furthermore, while BBTI was not significantly different from CBTI on key outcomes (e.g., ISI, sleep diary, response/remission rates), noninferiority could not be established.

Both treatment groups showed significant statistical and clinical change, but there were between group differences. First, on sleep diaries, veterans in CBTI had significant improvements on all variables from pre- to posttreatment, except TIB, while those in BBTI only saw significant changes for SOL, TWT, SE, and SQ. However, between group differences were not significantly different. Also, when looking at the magnitude of change pre- to posttreatment, veterans in both groups had similar improvements in WASO (CBTI: 27.6 min; BBTI: 21.2 min) and TST (CBTI: 65.4 min; BBTI: 67.0 min). Veterans in CBTI appear to have greater improvement on SQ, including a larger effect size; however, this was not a significant difference, likely due to small sample size. Due to low completion of sleep diaries and the high variability of the data, caution, again, is advised when interpreting these findings.

The low completion rates also highlight the challenge of reliably collecting sleep diaries and the need for methods to enhance completion, perhaps with mobile applications like the VA’s CBT-i Coach or the VA’s online program, Path to Better Sleep. Furthermore, engaging in online or mobile-based interventions may be a good option for individuals who prefer a self-management approach or experience barriers engaging in person. Online interventions are also a highly scalable solution that can help improve access to evidence-based insomnia care. The various online insomnia interventions have shown good reductions of insomnia severity with moderate to large effect sizes (Zachariae, Lyby, Ritterband, & O’Toole, 2016).

Second, only those in CBTI (and not those in BBTI) had reductions in nonsleep symptoms of depression and PTSD per the PHQ9 and PCL5, respectively. The reduction of scores was likely due to higher baseline scores and greater frequency of diagnosed depression and/or PTSD, compared to those in BBTI; comorbidities were not part of the randomized block design. The reduction of symptoms could also indicate that veterans with depression and/or PTSD achieve greater benefit from CBTI (i.e., in-person delivery, more sessions, longer session duration). The severity of psychiatric symptoms may be an important factor in deciding who receives CBTI vs. BBTI. This could also indicate a floor effect for those in BBTI as their PHQ9/PCL5 scores were relatively low at baseline. Nevertheless, the interpretability of these findings is limited due to the unbalanced randomization of veterans with depression and PTSD.

Third, while not a significant difference, there was a higher rate of attrition from CBTI (50%) vs. BBTI (29%). Approximately half of dropout for both groups occurred between baseline and Session 1, with veterans in CBTI having double the dropout rate prior to and after Session 1 vs. BBTI. This could be due to the higher rates of psychiatric comorbidities among veterans randomized to CBTI, as noted above. A comparison of baseline psychosocial measures found noncompleters had significantly higher levels of depression (PHQ9), anxiety (GAD7), and PTSD (PCL5) compared to completers. Another contributing factor may be the treating clinicians’ existing caseload and related schedule availability. In the CBTI group, one of the two clinicians worked in the mental health clinic (vs. primary care) and had a larger caseload, which resulted in more challenges with rescheduling participants. This may have contributed to the higher CBTI dropout rate compared to BBTI. However, further analysis of baseline demographic, clinical, and provider characteristics is needed to identify factors linked to attrition. For comparison, the VA’s CBTI rollout trial found 24% drop out from treatment or irregular attendance (Trockel et al., 2014).

LIMITATIONS

A key limitation of this preliminary study is the relatively small sample size. While the study was adequately powered for the noninferiority methods, there was greater attrition than expected and the inconclusive noninferiority outcome indicates an underpowered comparison (Mauri & D’Agostino, 2017). Also, there is no standard approach to establishing a NIM for behavioral interventions (e.g., effect size vs. mean difference). A different approach would likely have impacted the power analysis, yielded a different sample size, and potentially different results. For example, rather than mean difference with 95% confidence intervals, the simple mean difference in pre- to posttreatment ΔISI between groups (i.e., M = 2.2), or the difference in posttreatment ISI between groups (i.e., M = 1.4) would have resulted in different outcomes. This potential for variability in noninferiority highlights the importance of establishing a more standardized approach to determining noninferiority of behavioral interventions.

Additionally, the study was conducted at a single urban VAMC, limiting generalizability of results to the broader veteran population as well as to nonveterans. Furthermore, the high rate of attrition, as discussed above, likely impacted study outcomes. However, this study was pragmatic in its delivery of insomnia care and the rate of attrition was similar to clinical practice at VAPHS. The scheduling challenges of one of the treating providers represents a real-world clinical problem that contributes to missed opportunities in delivering care. With the small sample size and higher-than-expected attrition, it is unclear how overall study findings would change had more participants remained in treatment. These findings also suggest that augmenting patient engagement strategies may improve retention and, in turn, treatment outcomes. Attrition could also be an indicator of impairing comorbid symptoms or perhaps nonclinical barriers (e.g., work conflict, travel issues). The study was also imbalanced, with veterans with comorbid psychiatric disorders randomized more frequently to CBTI, which may have impacted outcomes.

Finally, as noted above, there was a high percentage of missing sleep diaries in both groups, although those in CBTI did have a higher rate of completion from baseline to posttreatment. The low rate and differences in completion limit both the interpretability of sleep diaries and the group comparisons. Reasons for the high rate of missing data overall and the group completion differences may be due to patient-level factors (e.g., comorbidity) or perhaps clinician-level factors (e.g., encouragement of diary utilization). From a clinical perspective, this may be an advantage of BBTI’s 30/30 rule, which can integrate patient self-report in the absence of sleep diaries to titrate TIB. However, reliance on the 30/30 rule without detailed diaries limits the ability to measure change of important variables like SOL, WASO, and TST. Use of mobile- or online-based sleep diaries may improve patient diary completion, as well as provide verification of daily completion, which can further enhance delivery of treatment and outcomes. In sum, the limitations above should be taken into consideration when interpreting study findings.

CONCLUSIONS AND FUTURE RESEARCH

Veterans who received BBTI and CBTI made significant improvements across multiple domains of sleep, although study limitations (e.g., no control group, randomization imbalance, small sample size) minimize support for BBTI’s effectiveness. The brevity and flexibility of BBTI is consistent with other behavioral interventions delivered in VA PCMHI (i.e., 4–6 30-minute sessions, in-person and telehealth delivery). Yet it will be important for future studies to use larger samples with appropriate randomization and stratification of patient complexities to establish BBTI as an effective and noninferior treatment to CBTI, or perhaps to establish CBTI as the superior treatment. The inclusion of participants with comorbid psychiatric and medical disorders will be essential to clarify for whom BBTI or CBTI is most appropriate. An additional consideration is to keep BBTI session content and delivery the same but expand the time frame (e.g., 4 sessions in 8 weeks; Edinger et al., 2007), which may result in outcomes more similar to CBTI. Also, the evaluation of long-term treatment gains and potential predictors, moderators, and/or mediators of treatment outcomes will be important. Though more investigation of BBTI is necessary to determine both effectiveness and noninferiority to CBTI, there remains potential for BBTI to serve as a complementary treatment that helps increase access to evidence-based insomnia care, especially in settings like primary care.