Treatment of comorbid sleep disorders and posttraumatic stress disorder in U.S. active duty military personnel: A pilot randomized clinical trial

Abstract

Insomnia and nightmares are common in patients with posttraumatic stress disorder (PTSD). They are associated with worse psychological and physical health and worse PTSD treatment outcomes. In addition, they are resistant to PTSD treatments, which do not typically address sleep disorders. Cognitive behavioral therapy for insomnia and nightmares (CBT-I&N) and cognitive processing therapy (CPT) for PTSD are first-line treatments, but limited evidence exists guiding the treatment of individuals with all three disorders. The current study randomized U.S. military personnel (N = 93) to one of three conditions: CBT-I&N delivered before CPT, CBT-I&N delivered after CPT, or CPT alone; all groups received 18 sessions. Across groups, participants demonstrated significantly improved PTSD symptoms. Because the study was terminated prematurely due to challenges with recruitment and retention, it was underpowered to answer the initially intended research questions. Nonetheless, statistical findings and relevant clinically meaningful changes were observed. Compared to participants who received CPT alone, those who received CBT-I&N and CPT, regardless of sequencing, demonstrated larger improvements in PTSD symptoms, d = −0.36; insomnia, d = −0.77; sleep efficiency, d = 0.62; and nightmares, d = −.53. Compared to participants who received CBT-I&N delivered before CPT, those who received CBT-I&N delivered after CPT demonstrated larger improvements in PTSD symptoms, d = 0.48, and sleep efficiency, d = −0.44. This pilot study suggests that treating comorbid insomnia, nightmares, and PTSD symptoms results in clinically meaningful advantages in improvement for all three concerns compared to treating PTSD alone.

Insomnia and nightmares are common among U.S. military personnel seeking treatment for posttraumatic stress disorder (PTSD), reported by 93% and 68% of these individuals, respectively (Pruiksma et al., 2016; Taylor, Pruiksma, Hale, et al., 2020). The symptoms and sequelae of PTSD are also common consequences of insomnia, nightmares, and other sleep disorders (American Psychiatric Association [APA], 2013). Insomnia and nightmares have been shown to improve, but generally do not remit, after evidence-based PTSD treatment (Haynes et al., 2020; Larsen et al., 2019; Pruiksma et al., 2016; Schnurr & Lunney, 2019; Taylor, Pruiksma, Hale, et al., 2020), with as many as 50%–73% of service members continuing to report nightmares and/or insomnia following treatment (Pruiksma et al., 2016; Taylor, Pruiksma, Hale, et al., 2020). This may be partially attributable to a lack of focus on addressing sleep symptoms in most first-line PTSD treatments. In addition, baseline insomnia and nightmare severity may predict poorer PTSD treatment response (Belleville et al., 2011; Taylor, Pruiksma, Hale, et al., 2020).

Given that PTSD, insomnia, and nightmares demonstrate significant symptom overlap, it is plausible that comorbid PTSD and sleep disorders have joint culpability in symptom severity. If so, successfully treating insomnia and nightmares along with PTSD should improve both the overall functioning and quality of life of service members with PTSD. Insomnia and nightmares negatively impact the rapid eye movement (REM) and slow-wave sleep (SWS) stages (Backhaus et al., 2006; Simor et al., 2012), which are important for learning, memory consolidation, emotional processing, and adaptation to stress (Backhaus et al., 2006; Pace-Schott et al., 2009; Silvestri, 2005; Silvestri & Root, 2008; Walker & van der Helm, 2009). In this way, insomnia and nightmares may impede PTSD treatments (Straus et al., 2018; Walker & van der Helm, 2009). Furthermore, adequate sleep promotes the extinction of conditioned fear in healthy humans (Pace-Schott et al, 2009; Straus et al., 2017). Thus, improved sleep should result in more efficient cognitive and emotional processing of memories, thought to be one mechanism of action for PTSD treatments (Cahill et al., 2009; Foa & Kozak, 1986).

Although cognitive behavioral therapies (CBTs) for PTSD (Cahill et al., 2009), insomnia (Morgenthaler et al., 2006; Mysliwiec et al., 2020), and nightmares (Morgenthaler et al., 2018) are considered first-line treatments for each respective condition, and sleep treatments have been found to be effective among individuals with PTSD (Miller et al., 2020), few studies have examined approaches for sequencing comorbid sleep disorders and PTSD. Galovski et al. (2016) examined the effectiveness of sleep-directed hypnosis compared to self-monitoring prior to starting cognitive processing therapy (CPT) for PTSD in a sample of 108 civilian women. The authors found that sleep-directed hypnosis improved insomnia symptoms and global sleep quality but did not augment gains in PTSD treatment. In another study, Pigeon et al. (2022) compared CBT for insomnia (CBT-I) to attention control prior to starting CPT in 110 civilian participants exposed to interpersonal violence who had PTSD, depression, and insomnia. After treatment, participants who received CBT-I plus CPT demonstrated larger reductions in symptoms of insomnia, depression, and PTSD, indicating the sequential delivery of CBT-I followed by CPT had significant effects on all three compared to attention control followed by CPT. Finally, in a pilot study of 22 military veterans and active duty service members with combat-related PTSD, Walters et al. (2019) found that adding CBT-I and imagery rehearsal therapy (IRT) for nightmares after prolonged exposure therapy (PE) resulted in large improvements in sleep and modest nonsignificant improvements in daytime PTSD symptoms compared to the addition of supportive care therapy to PE. As there is limited scientific research on how best to sequence leading treatments for PTSD, insomnia, and nightmares, studies are needed to replicate these findings and determine the optimal treatment sequence.

A strong argument can be made that sleep disturbances should be treated before PTSD (Miller et al., 2020). Treating sleep first could (a) alleviate sleep deprivation, thus improving cognitive functioning and emotional regulation and helping the patient to better process their traumatic experience; (b) be more acceptable than PTSD treatment for active duty military and veterans; (c) demystify the therapy process, allowing for an easier transition and more confidence in PTSD treatments (Baddeley & Gros, 2013); and (d) increase a patient’s rapport with their provider before processing trauma-related material as a part of the PTSD treatment.

Conversely, a strong argument can also be made for treating PTSD before treating sleep disturbances. Improving PTSD first could (a) reduce hyperarousal symptoms that may be undermining sleep; (b) be more efficient, allowing providers to focus on treating residual sleep problems that remain after PTSD treatment as needed, potentially further improving overall PTSD outcomes by addressing residual sleep-related concentration, irritability, and anhedonia problems; (c) be more acceptable to trauma therapists who are accustomed to addressing trauma and may be reticent to treat sleep before PTSD without an evidence base to guide them.

The primary objective of this study was to determine if adding CBT for insomnia and nightmares (CBT-I&N) to CPT (i.e., CBT-I&N and CPT, regardless of sequence) for combat-related PTSD resulted in larger reductions in PTSD symptoms than CPT alone for individuals with all three problems. The secondary objective was to determine whether the sequencing of CBT-I&N before or after CPT resulted in differential effects on PTSD symptom reduction. The third objective was to explore the impact of providing CBT-I&N before or after CPT on comorbid insomnia and nightmares.

We hypothesized that (a) adding CBT-I&N to CPT would result in larger reductions in PTSD symptom severity than CPT alone, (b) adding CBT-I&N before CPT would result in larger reductions in PTSD symptom severity than adding CBT-I&N after CPT, and (c) similar effects as would be seen relative to insomnia symptoms and the number of reported nightmares.

METHOD

A complete description of the methods used in the current study can be found in a methods paper (Taylor et al., 2020). The methods relevant to the present study are summarized briefly in this section.

Participants

Study participants were 93 active duty military and recently discharged veterans living near Fort Cavazos, Texas (USA) who had comorbid PTSD, insomnia, and nightmares. Participants were recruited from October 2016 to July 2018.

Inclusion criteria

Research participants met the following inclusion criteria: age 18–65 years; chronic or persistent insomnia according to Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; APA, 2013) criteria; chronic or persistent nightmare disorder according to the DSM-5 criteria; PTSD according to DSM-5 criteria; active duty military or recently discharged veteran status and eligibility for military medical treatment; willing to refrain from new behavioral health or medication treatment for sleep, PTSD, or nightmares during study participation; and intent to remain in the geographic area for 5 months following the first assessment.

Exclusion criteria

Exclusion criteria were having returned from a deployment within the past 3 months; current suicide or homicide risk meriting crisis intervention; the inability to speak and read English; a history of a moderate-to-severe traumatic brain injury; pregnancy; sleep efficiency greater than 85%; serious ongoing mental health symptoms (e.g., related to bipolar disorder, psychosis); or being currently engaged in evidence-based psychotherapy for PTSD (i.e., PE or CPT), insomnia, or nightmares (e.g., CBT for insomnia and/or nightmares), as assessed using self-report or medical record review. Active duty participants had the option of having the research staff speak with their command to ensure they were afforded time to participate in the study.

Procedure

Study design and approval

Figure 1 depicts a flow chart illustrating study design and recruitment. This is one of the few studies to fully follow the Recommendations for a Standard Research Assessment of Insomnia (Buysse et al., 2006) through our use of “standard research assessments and reporting standards in [three] key areas: diagnosis of insomnia and comorbid conditions, assessment of sleep and insomnia severity, and assessment of waking correlates and consequences of insomnia” (p. 1167). Assessments were completed at baseline; posttreatment; and 1-, 3-, and 6-months posttreatment. A limited number of measures were also administered during treatment. This study used a three-arm randomized design (Figure 1) to evaluate the efficacy of the following 18-session treatments delivered over 12 weeks: six sessions of weekly CBT-I&N delivered before 12 sessions of twice-weekly CPT (i.e., CBT-I&N before CPT), six sessions of weekly CBT-I&N delivered after 12 sessions of twice-weekly CPT (i.e., CBT-I&N after CPT), or 12 sessions of twice-weekly CPT followed by six sessions of weekly CPT (i.e., CPT alone). Study procedures were approved and monitored by the Institutional Review Board (IRB) at the University of Texas Health Science Center at San Antonio, serving as the primary IRB for engaged institutions, and Duke University. The STRONG STAR Data Safety and Monitoring Board monitored the conduct of the study. First consent was obtained in September 2016, and the last follow-up assessment was conducted in June 2018.

FIGURE 1.

CONSORT flowchart.

Note: PTSD = posttraumatic stress disorder; CBT-I&N = cognitive behavioral therapy for insomnia and nightmares; CPT = cognitive processing therapy; tx = treatment; wk = week. ^aReflects voluntary reasons such as participant request and lost contact. ^bReflects involuntary reasons such as military deployment or primary investigator discretion.

Screening, baseline assessment, and sleep assessments

Interested individuals first completed a brief telephone screening during which the basic study inclusion and exclusion criteria were reviewed. Eligible individuals were then invited to attend an in-person meeting to provide informed consent and complete the first portion of the baseline assessment, which was conducted in three parts over several days.

After providing informed consent, participants completed self-report questionnaires and structured diagnostic interviews for PTSD, including the PTSD Checklist for DSM-5 (Weathers, Litz, et al., 2013) PCL-5 and the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5; Weathers, Blake, et al., 2013), as well as for insomnia and nightmares, which were assessed using the Structured Clinical Interview for DSM-5 Sleep Disorders module (SCISD; Taylor, Wilkerson, et al., 2018). Diagnostic interview administration was monitored monthly for calibration and interrater reliability (Barnes et al., 2018). Participants who met the study diagnostic criteria were given an activity monitoring device, the Actiwatch Spectrum (Respironics, 2005), to wear and a sleep diary to complete over the following week. Upon verifying sleep efficiency of less than 85% per the sleep diary, participants completed one night of ambulatory, in-home sleep polysomnography (PSG) monitoring to document sleep architecture. Participants were then referred to the Sleep Disorders Center at Carl R. Darnall Army Medical Center for a clinically indicated, diagnostic PSG test to diagnose underlying sleep disorders, such as sleep apnea or periodic limb movement disorder, which could aggravate insomnia, nightmares, and PTSD.

Random assignment

Participants who met all the inclusion and exclusion criteria following the baseline screening were randomized using statistician-generated permuted blocks of 3, 6, and 9 through a web-based interface. Assessors and primary investigators were blinded to treatment group assignments and participants were reminded to keep assessors blinded. All interventions were conducted in an individual format in a military treatment facility.

Interventions

PTSD.

All participants received CPT for PTSD (Resick et al., 2016), a structured cognitive therapy in which patients are taught to examine their thoughts about their traumatic experiences and the implications for themselves, others, and the world. If a patient determines the thoughts to be inaccurate or unhelpful, through the therapists’ Socratic questioning and progressive worksheets, the individual can develop more balanced and fact-based thoughts to practice. The version of CPT used was validated in active duty service members and veterans and consists of 12 twice-weekly, 1-hr sessions (Resick et al., 2015, 2017). In the CPT alone group, participants received six additional weekly sessions. During the additional sessions, individuals continued practicing skills they learned in the 12-session protocol and working on problematic beliefs that had not been resolved, using appropriate worksheets, similar to a recent variable-length CPT protocol (Resick et al., 2021).

Sleep.

Participants in the two groups that included sleep intervention (i.e., the CBT-I&N before and after groups) received six sessions of weekly CBT-I&N using a manualized protocol developed for this study. The CBT-I&N protocol was a combination of CBT-I and exposure, relaxation, and rescripting therapy (ERRT; Davis & Wright, 2006; Pruiksma et al., 2020) for nightmares, both of which have been validated in active duty service members (Pruiksma et al., 2020; Taylor, Peterson et al., 2017, 2018). Session 1 of the combined sleep treatment provided psychoeducation about sleep and lifestyle habits that may be perpetuating insomnia. The behavioral model of insomnia was adapted to include nightmares. Stimulus control and sleep restriction were explained using the baseline sleep diary. In each subsequent session, stimulus control practice was reviewed and the sleep window was titrated. Session 2 provided psychoeducation about nightmares and introduced progressive muscle relaxation. Session 3 included nightmare exposure exercises that provided the rationale for confronting, rather than avoiding, nightmares. The nightmare exposure exercises introduced during Session 3 followed those used in ERRT protocols (Davis & Wright, 2006) and instructed participants to identify a target nightmare, write a detailed description of the nightmare, read the account aloud in session, and identify the main trauma-related themes in the nightmare (i.e., safety, trust, power and control, esteem, and intimacy); these trauma-related themes are the same as those discussed in CPT. The process of rescripting the nightmare according to the identified theme was introduced, and participants were encouraged to think about possible ideas for rescripting before the next session. Next, a deep breathing exercise was introduced. Session 4 included having the patient write a new dream according to the themes identified that they could then imagine and picture (i.e., imagery rehearsal) prior to going to sleep, and the session ended with a deep breathing exercise. In Session 5, participants had the opportunity either to continue working on rescripting their first dream or to engage in the written exposure exercise for a second nightmare. Sleep hygiene was then reviewed. Session 6 included a review of progress and relapse prevention.

Therapist training and treatment adherence

Therapists first completed a 2-day CPT workshop and a 1-day CBT-I&N workshop. They then treated at least two patients with PTSD, insomnia, and nightmares under supervision, with fidelity assessment, prior to treating consented study participants. The supervisors met weekly with the study therapists. Therapist treatment adherence and competence were determined by independent raters who were trained in CPT or CBT-I&N and who were not otherwise involved in the project.

Measures

The measures and schedule of administration used in the current study are listed in Table 1, with greater detail on those not used in the current study found elsewhere (Taylor et al., 2020). Outcome measures specific to the current study are included herein.

TABLE 1.

Schedule of MEASURES

Measure	Baseline	Interim	Posttreatment
			1 week	1 month	3 months	6 months
Demographic information
Demographic and military service characteristics	X
PTSD measures
Clinician-Administered PTSD Scale (CAPS-5)	X			X	X	X
PTSD Checklist for DSM-5 (PCL-5)	Xa,b	X	Xc	Xc	Xc	Xc
Sleep measures
Structured Clinical Interview for DSM-5 for Sleep Disorders	X				X	X
Sleep diary and nightmare log	Xb	X	X
Insomnia Severity Index (ISI)	Xa,b	X	Xc	Xc	Xc	Xc
Diagnostic polysomnography	X5	X	X

Note: DSM-5 = Diagnostic and Statistical Manual of Mental Disorders (5th ed.); interim = interim assessment between treatment phases; PTSD = posttraumatic stress disorder;

^aAlso given once weekly during the course of cognitive processing therapy (CPT).

^bAlso given once weekly during the course CBT for insomnia and nightmares (CBT-I&N).

^cIncluded in the abbreviated assessment battery that could be administered over the phone at the posttreatment assessments if the participant no longer lived in the area.

PTSD symptoms

The PCL-5 (Weathers, Litz, et al., 2013) is a highly reliable and valid 20-item, self-report measure that assesses how much respondents have been bothered by the DSM-5 symptoms of PTSD during the past month. Items are rated on a scale of 0 (not at all) to 4 (extremely), with higher scores indicating higher symptom levels. In the present study, Cronbach’s alpha for the full scale was .90.

Sleep problems

Insomnia and nightmares.

The SCISD (Taylor, Wilkerson, et al., 2018) is a semistructured interview designed to diagnose certain sleep disorders according to DSM-5 criteria and screen for other sleep disorders that require an objective overnight sleep study for a diagnosis (e.g., obstructive sleep apnea [OSA]). The SCISD was conducted to screen for OSA and assess the inclusion criteria for insomnia and nightmare disorders.

Insomnia severity.

The Insomnia Severity Index (ISI; Morin, 1993) is a seven-item, self-report measure that is used to assess perceived insomnia severity, with items related to difficulty falling and staying asleep as well as daytime dysfunction over the past 2 weeks. Items are rated on a scale of 0 to 4, with higher scores indicating more severe insomnia. In the present sample, Cronbach’s alpha for the ISI was .80.

Sleep efficiency and nightmares.

Participants tracked the parameters of their sleep (e.g., bedtime, sleep onset, waketime) using the Consensus Sleep Diaries (Carney et al., 2012) each morning for 7 consecutive days. We added a question to capture nightmare frequency (e.g., “How many nightmares woke you?”). Participants’ mean sleep efficiency was calculated as total sleep time/time in bed * 100.

Data analysis

Primary and secondary analyses

To test the primary hypothesis that providing both CBT-I&N and CPT, regardless of sequence, would be more effective in reducing PTSD symptoms than simply adding more sessions of CPT (i.e., CPT alone), the primary analysis compared the combined (i.e., collapsed) results of the two groups that received CBT-I&N (i.e., CBT-I&N before CPT and CBT-I&N after CPT) to the group that received CPT alone in a repeated-measures mixed-effects regression model with PTSD severity (i.e., PCL-5 total score) as the outcome variable. To test the secondary hypothesis that providing CBT-I&N before CPT would be more effective in reducing PTSD symptoms than providing CBT-I&N after CPT, the secondary analysis compared the two groups that received CBT-I&N to each other (CBT-I&N before vs. CBT-I&N after) using the same repeated-measures mixed-effects regression model design as conducted for the primary hypothesis. In addition, contingency table analyses were performed using PCL-5 scores, ISI scores, and the number of reported nightmares per week (i.e., from the sleep diary) to compare the treatment groups on response. To define treatment response, reliable change index (RCI) analyses were used to determine treatment response: a reduction of 10 points or more on the PCL-5 indicated PTSD treatment response; a reduction of 6 points or more on the ISI indicated insomnia treatment response; a decrease of 50% or more in the number of reported nightmares per week indicated nightmare treatment response. We determined remission status as a PCL-5 score of 32 or below, as recommended by Weathers et al. (2013), for PTSD; an ISI score below 15 (i.e., subthreshold insomnia) for insomnia; and zero nightmares per week for nightmares. The dependent variable was the comparison between scores on measures of PTSD, insomnia, and nightmares collected at baseline and scores collected at the last available assessment. Analyses were not continued through follow-up due to high dropout. To address concerns about Type I error, principal analyses were restricted to the stated primary hypotheses and outcomes with exploratory analyses identified as such. Effect sizes were provided for all analyses, and all statistical tests were 2-tailed at an alpha level of .05.

A series of mixed-effects regression analyses were performed to examine changes in weekly ISI scores (Morin, 1993) as well as sleep efficiency and the number of nightmares reported per week, as assessed using sleep diaries. Nightmares were measured using diary data to increase power and because nightmares reported prospectively are likely more accurate than nightmares reported retrospectively.

Analyses were intent-to-treat and used all data from randomized participants regardless of the extent of their engagement in treatment or follow-up. The analyses were performed using the PROC MIXED in SAS (Version 9.4) and represent population-averaged (i.e., marginal) regression models with repeated measures. The fixed design effects were treatment arm, visit, and their interaction, specifying either autoregressive moving average or first-degree autoregressive covariance structure based on information criteria provided by the software. Although these models produce general multidegrees of freedom F tests of main and interaction effects, the specific hypothesis tests were done using planned contrasts, such as pairwise comparisons of pre–post change based on differences of estimated least-square means. The SAS MIXED procedure estimates model parameters using restricted maximum likelihood estimates. These are valid under the commonly accepted assumption that data are missing at random, and do not require imputation of missing data. In general, these likelihood-based parameter estimates are virtually identical to those obtained with multiple imputation and simpler to work with because they do not depend on multiple, randomly generated datasets. The reported hypothesis tests were a priori–planned t tests and provide contrasts of the regression-based estimates of group means. Baseline scores were compared to the scores collected at the last assessment point, to account for participant dropout, up to posttreatment as the dependent variables. Effect sizes were provided for all analyses, and all statistical tests were two-tailed at an alpha level of .05.

Dropout

Kaplan–Meier survival analysis was used to model time to all-cause discontinuation, excluding those participants who completed all scheduled sessions. The last session attended was used as time-to-dropout, or time to stopping study participation due to military reasons such as deploying, changing stations, or separating from the military. Differences between treatment arms were evaluated using the log-rank test. The same statistical approach was used to evaluate predictors of discontinuation with univariate log-rank tests.

Statistical power

Original sample size calculations were made using RMASS2 software for longitudinal data (Hedeker et al., 1999) to ensure sufficient power to answer the secondary hypothesis that adding CBT-I&N before CPT would result in larger reductions in overall PTSD symptom scores (i.e., PCL-5) than CBT-I&N delivered after CPT. This was done because this important pairwise comparison was a less powerful analysis than the overall three-group comparison. A power analysis using repeated measures with pre- and posttreatment assessments indicated that 222 participants were needed (n = 74 per group). The analysis plan assumed a medium effect (i.e., Cohen’s d = 0.50), a power of .80, and a significance threshold set at p = .05 (two-sided Type I error) and allowed for 25% attrition at posttreatment.

The study experienced significant recruitment challenges. At the start of the research in 2016, a military-wide force reduction was initiated to reduce the overall number of active duty military personnel. This resulted in reductions in PTSD and sleep disorders in military personnel returning from deployments. In addition, changes in PTSD criteria (Resick et al., in press) from the DSM-IV (APA, 2000) were used to estimate recruitment rates based on DSM-5 criteria (APA, 2013), and the requirement that participants also had to meet the full DSM-5 diagnostic criteria for insomnia and nightmare disorders resulted in enrollment rates that were lower than expected based on previous STRONG STAR consortium PTSD clinical trials. Finally, the attrition rates were higher than expected. Because of these difficulties, the government steering committee recommended terminating the study before the planned participant recruitment had been accomplished. The results presented here are based on data collected from 93 participants who were enrolled and randomized at the time recruitment was terminated. Although the analyses are underpowered, we believe it is both ethical and useful to publish these data and estimated effect sizes to inform future study designs in an area where little data exist to guide evidence-based practice.

RESULTS

Participant flow and recruitment

Table 2 shows the baseline demographic characteristics for the total sample and each group. There were no significant baseline differences between groups. As depicted in Figure 1, 93 participants met the study criteria and were randomized into the CBT-I&N before CPT (n = 31), CBT-I&N after CPT (n = 31), or CPT alone (n = 31) conditions.

TABLE 2.

Baseline demographic, mental health, psychosocial, and medical health characteristics

	Total (N = 93)		CBT-I&N before CPT (n = 31)		CPT alone (n = 31)		CBT-I&N after CPT (n = 31)
Characteristic	M	SD	M	SD	M	SD	M	SD
Age	36.20	7.49	36.29	7.52	36.10	7.21	36.22	7.97
Months in service	175.40	85.00	174.60	78.20	181.60	81.60	169.90	96.40
	n	%	n	%	n	%	n	%
Male gender	68	73.1	22	71	20	65	26	84
Race
White	42	45.3	13	42.0	14	45.2	15	48.4
African American	32	34.4	12	38.7	11	35.5	9	29.0
Other	19	20.4	6	19.3	6	19.3	7	22.6
Hispanic/Latino	22	23.7	5	16.1	5	16.1	12	38.7
Educational attainment
High school/GED	12	12.0	4	12.9	2	6.4	6	19.3
Some college/associate degree	59	63.4	20	64.5	19	61.2	20	64.5
College/graduate degree	22	25.8	7	22.6	10	32.2	5	16.1
Married/cohabitating	65	70.0	21	67.7	21	67.7	23	74.1
Military gradea
E-1–E-6	62	66.7	24	77.4	14	45.1	25	80.6
E-7–E-9	27	29.0	6	19.3	16	51.6	5	16.1
WO/officer	4	4.3	2	6.4	1	3.2	1	3.2
Military status
Active duty	86	92.5	28	90.3	28	90.3	30	96.8
Veteran	6	6.4	2	6.4	3	9.7	1	3.2
National Guard	1	1.0	1	3.2	0	0.0	0	0.0
History of head injuries with persistent symptoms	51	54.8	18	58.0	18	58.0	15	48.4

Note. For display purposes only, some small cell sizes were combined or are not displayed for race, educational attainment, and marital status variables; test statistics and p values are based on original cell sizes. CBT-I&N = cognitive behavioral therapy for insomnia and nightmares; CPT = cognitive processing therapy; WO = warrant officer.

^aE-1–E-3: junior enlisted personnel, E-4–E-6: junior noncommissioned officers, E-7–E-9: senior noncommissioned officers.

Dropout

As shown in Figure 1, 23.7% (n = 22) participants dropped out of treatment for voluntary reasons (e.g., participant request due to life circumstances or lost contact), and 51.6% (n = 48) discontinued for involuntary reasons, such as military deployment or separation or at the discretion of the research team. The dropout analysis was nonsignificant, log-rank χ²(2, N = 93) = 1.35, p = .509, indicating there were no differences between treatment arms with respect to all-cause discontinuation. There were also no significant baseline predictors of discontinuation.

Primary and secondary analyses

Although analyses were nonsignificant, combined CBT-I&N and CPT outperformed CPT alone on all measures (i.e., PTSD symptoms, insomnia, sleep efficiency, nightmares) to varying degrees, |ds| = 0.36–0.77, Mdn d = 0.55 (see Table 3).

TABLE 3.

Mixed linear model analyses of posttraumatic stress disorder and sleep outcomes

Variable groupa	B	Week													PT	Time			CBT-I&N+CPT vs. CPT alone			Sleep before vs. sleep after
		1	2	3	4	5	6	I	7	8	9	10	11	12		Δ	p	d	t(2)	p	d	t(2)	p	d
n	93	83	76	72	65	54	48	55	48	41	40	36	33	31	27
PCL-5																			−0.83	.404	−0.36	0.94	.347	0.48
Sleep before	53.7	49.5	47.0	43.9	41.0	39.3	39.0	40.6	38.6	39.0	41.5	38.1	38.3	36.2	40.3	−13.3	.006	−1.02
CPT alone	47.8	46.3	47.2	42.3	41.3	41.7	42.5	41.1	39.8	39.3	38.2	37.3	36.4	35.4	36.1	−11.7	.011	−0.90
Sleep after	53.0	51.2	52.4	49.8	45.0	43.3	42.1	40.7	38.6	35.1	35.9	32.3	32.6	31.3	33.3	−19.7	< .001	−1.51
ISI																			−1.73	.083	−0.77	−0.85	.395	−0.43
Sleep before	22.2	19.4	17.4	16.2	14.9	13.9	13.3	14.1	13.6	14.0	14.3	15.1	14.3	14.3	14.6	−7.6	< .001	−1.67
CPT alone	19.7	19.6	19.5	19.1	19.0	19.9	18.9	18.6	17.7	17.6	17.3	17.2	18.0	17.2	16.6	−3.1	.062	−0.68
Sleep after	22.5	21.9	21.3	21.9	21.5	19.3	19.1	20.1	19.7	18.3	17.0	16.2	16.5	16.2	16.9	−5.6	< .001	−1.24
Sleep Diary
SE																			1.33	.189	0.62	−0.79	.434	−0.44
Sleep before	72.4							90.2							80.7	−8.34	.071	.73
CPT alone	70.9							80.7							74.6	−3.73	.390	.33
Sleep after	69.4							78.2							82.7	−13.28	.003	1.17
Number of NM																			−1.43	.156	−0.53	0.32	.747	0.15
Sleep before	5.2							4.1							3.3	−1.96	.196	−0.45
CPT alone	5.9							4.4							6.0	0.05	.959	0.02
Sleep after	5.3							4.6							2.7	−2.60	.053	−0.59

Note: CBT-I&N= cognitive behavioral therapy for insomnia and nightmares; CPT = cognitive processing therapy; BL = baseline; I = interim; PT = posttreatment; SE = sleep efficiency; NM= nightmares; ISI = Insomnia Severity Index; PCL-5 = PTSD = Posttraumatic Stress Disorder (PTSD) Checklist for DSM-5.

^aFor the “sleep before” group, CBT-I&N was delivered before CPT; for the “sleep after” group, CBT-I&N was delivered after CPT.

^bNumber of NM per week, assessed using the sleep diary and nightmare log.

PTSD

All three treatment groups demonstrated significant reductions in PTSD symptoms between baseline and 1-week posttreatment (i.e., PCL-5 score reduction), ds = −0.90–1.51, p < .001–p = .011. The Group (i.e., CBT-I&N and CPT vs. CPT alone) × Time (i.e., pretreatment to posttreatment) interaction (Figure 2) was nonsignificant, d = −0.36, p = .404. Similarly, the analyses conducted to determine if sequencing CBT-I&N before or after CPT would result in differential pretreatment-to-posttreatment effects on PTSD symptom reduction were also nonsignificant, d = 0.48, p = .347. Similar results were observed for the CAPS-5 but with fewer data points due to significant attrition before posttreatment assessments, as the CAPS-5 was only administered at baseline and 1-month follow-up.

FIGURE 2.

Primary Posttraumatic Stress Disorder (PTSD) Checklist for DSM-5 (PCL-5) outcomes comparing cognitive processing therapy (CPT) with cognitive behavioral therapy for insomnia and nightmares added (CBT-I&N averaged and CPT) Groups, combined versus the CPT alone group.

Note: CPTinAVG = Average across both groups that included CPT and CBT-I&N, BL = baseline; W = week; Int = interim; 1WFU = 1-week follow-up.

Sleep

Participants in the two groups that included CBT-I&N demonstrated significant reductions in insomnia symptoms (i.e., ISI score) from baseline to 1-week posttreatment, ds = −1.24–−1.67, ps < .001, but those in the CPT alone group did not, d = −0.68, p = .062. However, the Group × Time interaction was again nonsignificant, d = −0.77, p = .083, as were the analyses conducted to determine whether CBT-I&N sequencing relative to CPT delivery would result in differential pretreatment-to-posttreatment effects, d = −0.43, p = .395.

Similar results were seen when using the sleep diary (i.e., SE) whereby CBT-I&N delivered after CPT demonstrated a significant treatment effect, d = 1.17, p = .003, but participants in the CPT alone group did not show significant improvement, d = −0.33, p = .390; CBT-I&N delivered before CPT did not show a significant treatment effect, d = 0.73, p = .071. Again, the two Group × Time interactions were nonsignificant, CBT-I&N and CPT versus CPT alone: d = 0.62, p = .189; CBT-I&N delivered before or after CPT, d = −0.44; p = .434.

With regard to the number of reported nightmares per week, no group demonstrated significant treatment effects. Again, the interaction analyses were both nonsignificant (see Table 3).

Response and remission

As shown in Table 4, there were no significant between-group differences in either response or remission rates for PTSD, insomnia, or nightmares.

TABLE 4.

Response and remission at last visit through posttreatment

Variable	CBT-I&N before CPT (n = 31)		CPT alone (n = 31)		CBT-I&N after CPT (n = 31)		χ2(2)	p
	n	%	n	%	n	%
PTSDa
Response	13	41.9	16	51.6	16	51.6	0.78	.679
Remission	10	32.3	12	38.7	11	35.5	0.28	.869
Insomniab
Response	14	45.2	6	19.3	9	29.0	4.91	.086
Remission	11	35.5	11	35.6	5	16.1	3.76	.153
Nightmaresc
Response	8	25.8	8	25.8	5	16.1	0.61	.737
Remission	6	19.4	7	22.6	5	16.1	0.41	.813

Note: Response and remission for posttraumatic stress disorder (PTSD) and insomnia were determined using the reliable change index (RCI). Data are based on PTSD Checklist for DSM-5 (PCL-5) data from the last visit through 1-week follow-up. The one-tailed RCI, used here, cuts off the amount of improvement that would be expected to occur no more than 5% of the time by chance due to unreliability if no change actually occurred. CBT-I&N = cognitive behavioral therapy for insomnia and nightmares; CPT = cognitive processing therapy; PTSD = posttraumatic stress disorder. df for chi-square = 2.

^aResponse was defined using the RCI as a PTSD Checklist for DSM-5 (PCL-5) score improvement of 10 points or more (one-tailed). Remission was defined as a PCL-5 total score of less than 32.

^bResponse was defined using the RCI as an Insomnia Severity Index (ISI) score improvement of 6 points or more (one-tailed). Remission was defined as an ISI total score of less than 15.

^cResponse was defined as a reduction of 50% or more in the number of nightmares per week. Remission was defined as zero nightmares per week.

DISCUSSION

This pilot study aimed to examine the potential benefit of adding sleep-specific treatments (i.e., CBT-I&N) before or after CPT to explore whether sleep interventions could enhance PTSD and sleep outcomes given that all participants received state-of-the-science PTSD treatment. As expected, all groups had significantly improved PTSD symptoms. Although the analyses were nonsignificant, combined CBT-I&N and CPT outperformed CPT alone on all measures (i.e., PTSD symptoms, insomnia, sleep efficiency, nightmares) to varying degrees (|ds| = 0.36–0.77, Mdn d = 0.55). In addition, adding CBT-I&N after CPT produced additional, nonsignificant medium effects over delivering CBT-I&N before CPT on PTSD symptoms and sleep efficiency.

Although the results were nonsignificant, these findings are congruent with the existing evidence (Walters et al., 2019) that a comprehensive approach to treating patients with insomnia and nightmares in addition to PTSD should include treatment for all three problems. PTSD treatment alone is not a sufficient treatment for patients diagnosed with comorbid PTSD, insomnia, and nightmare disorder (Haynes et al., 2020; Larsen et al., 2019; Pruiksma et al., 2016; Schnurr & Lunney, 2019; Taylor, Pruiksma, Hale, et al., 2020). The sequencing results for PTSD outcomes favored treating insomnia and nightmares after treating PTSD. It is possible that treating PTSD first reduced the hyperarousal symptoms that can impede sleep, making the CBT-I&N component more effective.

The lack of benefit from CPT alone on insomnia severity and sleep efficiency was expected and replicated findings from previous studies (Galovski et al., 2016; Haynes et al., 2020; Larsen et al., 2019; Pruiksma et al., 2016; Schnurr & Lunney, 2019; Walters et al., 2019). The results demonstrating that the groups that included CBT-I&N outperformed the CPT alone group with respect to improvements in insomnia and nightmares also replicated the findings from a previous pilot study (N = 23) that added CBT-I&N after prolonged exposure therapy (Walters et al., 2019). Although the present results did not demonstrate statistically significant Group × Time differences in insomnia or PTSD symptom reduction for the treatments with added CBT-I&N, observed mean changes were larger for the participants in the group that received CBT-I&N before CPT. This parallels findings reported by Pigeon et al. (2022), who observed that CBTi followed by CPT was significantly better at improving symptoms in these domains than attention control followed by CPT in a sample of 110 survivors of interpersonal violence. It is important to note that four sessions of attention control followed by 12 of CPT is likely not as active as 12 sessions of CPT followed by six additional sessions of CPT, as was used in the current study, which may explain why our interaction effects did not reach significance.

Similarly, there were no significant differences between groups with regard to response or remission rates. This is not surprising considering the study was not adequately powered to find significant differences in continuous measures of these constructs, meaning that dichotomizing would only reduce the power even further. Therefore, these results should be interpreted with caution.

Attrition was likely influenced by significant troop draw-downs during the trial, which resulted in service members unexpectedly and quickly being able to leave active duty service. This likely resulted in many needing to drop out of the study, as did the intensive assessment battery and process (e.g., actigraphy, sleep diaries, in-home overnight PSG, and blood draws), and the long course of treatment (i.e., 18 sessions over approximately 12 weeks). We were able to contact some participants who dropped out to ascertain their reasons for stopping their participation. Reported reasons for voluntary dropout included surgeries, being busy with school in addition to work and family responsibilities, being unable to complete 12 weeks of treatment within 18 weeks, opting for intensive inpatient PTSD treatment instead of outpatient treatment, or feeling they had received enough treatment and were satisfied with the improvements achieved. Involuntary reasons for dropout included being chaptered out of the military, an unexpected military deployment, a permanent change of station to another duty assignment, and retiring sooner than expected.

The primary limitation of the current study was that it was underpowered to find significant results. As mentioned, although originally powered to find Cohen’s d effect sizes of .50 or larger for each of the primary outcomes, the study was terminated prematurely due to government oversight concerns about recruitment and retention. However, it would be unethical to subject participants to the time and risks involved in research and not provide the promised benefits (i.e., data supporting evidence-based practice decisions). Thus, despite the nonsignificant findings and risk of publication bias, it is important to publish these data, with corresponding clinical effect sizes, to inform the design of future research studies and provide important pilot data in an area where little data exist to guide evidence-based practice. The sample was also limited in terms of diversity with regard to gender because almost three-quarters of the participants were men.

As PTSD, insomnia, and nightmares have such high levels of comorbidity, future studies need to investigate more efficient ways to combine the therapies for the three disorders to reduce patient time and financial burden and improve retention. Some possibilities include massed or accelerated treatment formats or those that use fewer sessions, such as written exposure therapy for PTSD (Sloan & Marx, 2019). In addition, retention might also be improved by offering at least part of the intervention via teletherapy, as travel time to and from the clinic can be substantial, and some service members were transferred to new bases during the study, which resulted in dropout. Moreover, without a formal analysis, there appears to have been a higher dropout rate in the second phase of the CPT alone group and possibly the group that received CPT before CBT-I&N, future research should seek to replicate this finding to understand such dropout. Finally, retention may be improved by focusing on only the most important and least intensive assessments (i.e., self-report PTSD and sleep questionnaires), rather than structured clinical interviews, and saving mechanistically focused measures of blood and sleep (i.e., polysomnography and actigraphy) biomarkers for later studies.