Sleep disorder symptoms are associated with greater posttraumatic stress and anger symptoms in US Army service members seeking treatment for posttraumatic stress disorder

Abstract

Study Objectives:

Characterize associations between sleep impairments and posttraumatic stress disorder (PTSD) symptoms, including anger, in service members seeking treatment for PTSD.

Methods:

Ninety-three US Army personnel recruited into a PTSD treatment study completed the baseline assessment. State-of-the-science sleep measurements included 1) retrospective, self-reported insomnia, 2) prospective sleep diaries assessing sleep patterns and nightmares, and 3) polysomnography measured sleep architecture and obstructive sleep apnea-hypopnea severity. Dependent variables included self-report measures of PTSD severity and anger severity. Pearson correlations and multiple linear regression analyses examined if sleep symptoms, not generally measured in PTSD populations, were associated with PTSD and anger severity.

Results:

All participants met PTSD, insomnia, and nightmare diagnostic criteria. Mean sleep efficiency = 70%, total sleep time = 5.5 hours, obstructive sleep apnea/hypopnea (obstructive sleep apnea-hypopnea index ≥ 5 events/h) = 53%, and clinically significant anger = 85%. PTSD severity was associated with insomnia severity (β = .58), nightmare severity (β = .24), nightmare frequency (β = .31), and time spent in Stage 1 sleep (β = .27, all P < .05). Anger severity was associated with insomnia severity (β = .37), nightmare severity (β = .28), and obstructive sleep apnea-hypopnea during rapid eye movement sleep (β = .31, all P < .05).

Conclusions:

Insomnia and nightmares were related to PTSD and anger severity, and obstructive sleep apnea-hypopnea was related to anger. Better assessment and evidence-based treatment of these comorbid sleep impairments in service members with PTSD and significant anger should result in better PTSD, anger, and quality-of-life outcomes.

Clinical Trials Registration:

Registry: ClinicalTrials.gov; Name: Treatment of Comorbid Sleep Disorders and Post Traumatic Stress Disorder; Identifier: NCT02773693; URL: https://clinicaltrials.gov/ct2/show/NCT02773693.

Citation:

Miles SR, Pruiksma KE, Slavis D, et al. Sleep disorder symptoms are associated with greater posttraumatic stress and anger symptoms in US Army service members seeking treatment for posttraumatic stress disorder. J Clin Sleep Med. 2022;18(6):1617–1627.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Previous research has been limited by highly sensitive (eg, single self-report measures) but largely nonspecific measures of sleep disorder symptoms when examining posttraumatic stress disorder (PTSD) samples. This paper examined how well-defined symptoms of sleep disorders were related to self-report measures of PTSD and anger symptoms in service members seeking treatment for PTSD.

Study Impact: PTSD severity was related to greater insomnia severity, more severe nightmares, more frequent nightmares, and more time in Stage 1 sleep. Anger severity was related to greater insomnia severity, greater nightmare severity, and a higher apnea-hypopnea index during rapid eye movement sleep. Treatments for PTSD and anger symptoms should be augmented with treatments that target sleep disorders.

INTRODUCTION

Posttraumatic stress disorder (PTSD) is often difficult to treat, particularly in active-duty military populations. Two of the most distressing PTSD symptoms include sleep impairments and anger, which can lead to aggression. Unfortunately, many patients continue to report distress, including continuing sleep problems^1–5 and dysregulated anger,⁶ after evidence-based psychotherapies for PTSD. Finding targets to improve the precision, and thus effectiveness, of evidenced-based treatments for PTSD is a high priority.

Sleep impairments are included as diagnostic criteria for PTSD in the Diagnostic and Statistical Manual for Mental Disorders, fifth edition (DSM-5).⁷ Specifically, the intrusions cluster of PTSD symptoms includes “recurrent distressing dreams in which the content and/or affect associated with the dream are related to the traumatic event(s)” (p. 271) and the hyperarousal cluster includes “difficulty falling or staying asleep or restless sleep”⁷ (p. 271–2). Self-reports of these symptoms may be proxies for independent comorbid sleep-related disorders (ie, nightmare and insomnia disorders), which may interact with other symptoms of PTSD. For example, “difficulty falling or staying asleep” may indicate insomnia, which is also associated with other PTSD symptoms, such as strong negative feelings, loss of interest in activities one used to enjoy, and difficulty concentrating. These sleep impairments often do not remit after PTSD treatment.^1–5 In one study of US Army personnel with PTSD in a clinical trial of evidence-based PTSD treatment, > 72% continued to meet criteria for difficulty falling or staying asleep and > 50% continued to report nightmares at posttreatment.^3,5 However, it should be noted nightmares are likely to be reduced after successful PTSD treatment (ie, in those patients who respond to treatment and achieve overall symptom reduction) but trouble falling and/or staying asleep persists even in treatment responders.^2–8 For service members and veterans, baseline severity of sleep disorders may predict more intractable PTSD, requiring more treatment sessions.⁹

Another sleep condition that may present as symptoms of PTSD is obstructive sleep apnea-hypopnea (OSAH) syndrome. OSAH is characterized by the upper airway collapsing, causing full (ie, apnea) or partial cessation (ie, hypopnea) of ventilation during sleep. Up to 63% of service members and veterans with PTSD also meet diagnostic criteria for OSAH.^5,10 Not surprisingly, self-reported OSAH does not respond to PTSD treatments and may place patients at risk for less symptom reduction when participating in evidence-based psychotherapies for PTSD.^5,11

Studies that examined if severity of OSAH was related to severity of PTSD using polysomnography (PSG) evaluations have found mixed results, likely due to the variance in samples that were studied and measurement strategies that were used. Small studies of veterans have found associations between OSAH and greater PTSD symptom severity.^5,12,13 Alternatively, Mysliwiec and colleagues¹⁴ found that, in a sample of active-duty service members reporting to a military sleep disorders clinic, OSAH was only associated with greater self-reported PTSD symptoms among those with comorbid insomnia. Another study found no differences in OSAH severity (apnea-hypopnea index [AHI]) in a comparison of veterans with mild to moderate PTSD severity and veterans with severe to very severe PTSD severity.¹⁵ These inconsistent results may be due to the restriction of range in OSAH and AHI, because the patients were referred for a sleep evaluation and most, or all, had OSAH.^12,15,16

Finally, a recent meta-analysis confirmed abnormal polysomnographic sleep architecture in patients with PTSD.¹⁶ Increased PTSD severity was associated with decreased percentage of slow-wave sleep and decreased sleep efficiency. Compared to healthy controls, patients with PTSD had decreased total sleep time, slow-wave sleep, and sleep efficiency and increased wake after sleep onset. Younger patients (age < 30 years) with PTSD also spent a smaller percentage of time in rapid eye movement (REM) compared to controls. Patients younger than 40 years of age with PTSD had increased wake after sleep onset when compared to controls. These results suggest multiple components of sleep architecture are distinctly impaired among patients with PTSD, particularly among those who are younger.

Anger and subsequent aggression are also problematic symptoms of PTSD that linger after successful treatment.⁶ These PTSD symptoms are captured as the following criteria: “persistent negative emotional state (eg, fear, horror, anger, guilt, or shame)”^7(p272) and “irritable behavior and angry outbursts (with little or no provocation) typically expressed as verbal or physical aggression toward people or objects”⁷ (p. 272). Service members and veterans with PTSD report greater levels of anger than civilians with PTSD.¹⁷ Anger is one of the most problematic symptoms of PTSD, and one study found that the majority of service members continue to report significant anger after PTSD treatment.⁶

Given anger and aggression can result in violence within the family unit, legal charges,¹⁸ and suicide,¹⁹ finding effective treatments that address anger are important. Several studies suggest that short sleep (ie, sleep deprivation), which can be a result of OSAH, insomnia, or nightmare disorders, causes significant increases in negative moods (eg, anger), emotional dysregulation, and impaired social functioning in the general population.^20,21 Moreover, younger people (age < 30 years) appear to be more vulnerable to the effects of sleep deprivation on increasing anger when compared to older adults.^20,22 This is concerning considering that the average age of enlisted active duty service members is 27 years, with 73% of enlisted service members being younger than 30 years.²³ In non-PTSD samples, studies show sleep duration and REM sleep are associated with reduction in aversive emotions and increases in positive emotions,²⁴ as well as improved emotion regulation.²⁵ In summary, there is some limited research showing that in patients with insomnia,^26,27 nightmares,²⁸ and OSAH²⁹ all are associated with greater dysregulation of anger symptoms. Because sleep disorders are related to increased anger,²¹ it allows for the possibility that improving sleep may be a novel way to reduce dysregulated anger in those with PTSD.

The current study sought to increase the understanding of how sleep problems and architecture relate to PTSD symptoms, including anger, by analyzing baseline data of service members seeking treatment for PTSD. Sleep problems were evaluated in a state-of-the-art and multidimensional manner in the parent trial (ie, validated sleep disorder questionnaires, daily self-reported assessment, and objective PSG).³⁰ Consequently, we aimed to examine the association between PTSD and anger symptoms with retrospective self-reported insomnia (Aim 1), prospective sleep diary measures (Aim 2), and sleep architecture measured by PSG (Aim 3). We hypothesized that greater insomnia severity would be associated with greater PTSD and anger symptom severity (Aim 1). The other aims were exploratory, without a priori hypotheses.

METHODS

Procedures

Details of the parent trial design and methodology are published elsewhere.³⁰ In brief, the study was approved by the local Institutional Review Board and the VA Research and Development Committee; the US Army Medical Research and Development Command Human Research Protection Office reviewed and monitored the trial’s regulatory approvals. The parent study conformed to all state and federal research regulations. The study was conducted in association with the South Texas Research Organizational Network Guiding Studies on Trauma and Resilience and the Consortium to Alleviate PTSD³¹ at the Carl R. Darnall Army Medical Center, located on the Fort Hood military installation in Killeen, Texas. Participants were primarily recruited through referrals from behavioral health providers and from the Carl R. Darnall Army Medical Center Sleep Disorders Clinic between October 2016 and July 2018.

Participants provided informed consent prior to data collection and completed a baseline assessment that included clinical interviews and self-report measures. All diagnostic clinical interviews were conducted by master’s-level or doctoral-level trained independent evaluators who regularly attended calibration exercises. Participants were then asked to complete 1 week of sleep diaries to assess for sleep efficiency. All participants underwent an in-lab PSG to diagnose underlying sleep disorders, such as OSAH. Some of the PSGs (19%) were split-night studies during which the participant was fitted for continuous positive airway pressure if OSAH was detected. If participants had completed a diagnostic PSG within the previous 2 years, they were only referred again if there was a change warranting re-evaluation (eg, significant weight gain). Otherwise, their diagnostic PSG information was retrieved from the electronic medical record.

Participants

Participants were active-duty service members or recently discharged veterans diagnosed with PTSD assessed by the Clinician Administered PTSD Scale for DSM-5³² and diagnosed with insomnia disorder and nightmare disorder assessed by the Structured Clinical Interview for DSM-5 Sleep Disorders.³³ Insomnia disorder was also confirmed with a sleep diary, and participants were excluded if their sleep efficiency > 85%. Participants were between 18 and 65 years of age, had experienced at least 1 deployment in support of combat operations after September 11, 2001 (ie, Operation Iraqi Freedom, Operation Enduring Freedom, Operation New Dawn), were eligible for care on base, and planned to be in the area for at least 5 months. Exclusion criteria included having returned from a deployment < 3 months ago, suicidal or homicidal risk that required immediate intervention, not speaking English, moderate/severe brain damage, pregnancy, serious mental illness such as bipolar or psychotic disorders, and current engagement in psychotherapy for PTSD, insomnia, or nightmares. All participants agreed not to engage in behavioral health treatment, including changes in medications for PTSD, insomnia, or nightmares outside the study.

Measures and measurements

Demographics

Demographics and military status information was collected via self-report.

Insomnia Severity Index

The Insomnia Severity Index (ISI)³⁴ is a 7-item, self-report measure that assesses perceived severity of insomnia with difficulty falling and staying asleep, as well as daytime dysfunction over the past 2 weeks. Item scores range from 0 to 4, with total scores of 15 or greater indicating clinically significant insomnia (0–7 = absence of insomnia; 8–14 = subthreshold insomnia; 15–21 = moderate insomnia; and 22–28 = severe insomnia).³⁴ In this study, the internal consistency of the ISI was α = 0.80.

Sleep diary and nightmare diary

To assess self-reported sleep patterns, participants tracked the parameters of their sleep (eg, bedtime, sleep onset, waketime) with the Consensus Sleep Diary³⁵ for 7 days. We added 3 questions: 1) “Do your military duties end after 2100 or begin before 0600?” 2) “How many nightmares woke you?” 3) “Rate overall severity of nightmares (0 = Not at all, 1 = Slightly, 2 = Moderately, 3 = Very much, 4 = Extremely, NA = not applicable).” Participants completed daily diary entries each morning with an estimate of their sleep the night prior. Derived mean variables included total sleep time, sleep efficiency (total sleep time/time in bed * 100), sleep quality (0 = Very Poor, 1 = Poor, 2 = Fair, 3 = Good, 4 = Very Good), and nightmare frequency and severity.

Polysomnography

A diagnostic in-lab PSG was conducted by a registered polysomnographic technician. The fully attended Level 1 PSGs were conducted in accordance with the American Academy of Sleep Medicine–recommended procedures,³⁶ using the standard parameters of electroencephalogram from F1, F2, C3, C4, O1, and O2; left and right electrooculogram; chin and leg electromyogram; electrocardiogram (modified lead II); thermistor and nasal pressure; and oximetry, breathing effort. All studies were interpreted by qualified physicians, and all staff who scored and interpreted the PSGs were masked to other sleep assessments. PSG indices included AHI total and AHI during REM. AHI < 5 events/h are considered normal; scores 5–14 events/h = mild OSAH; scores 15–29 events/h = moderate OSAH; AHI ≥ 30 events/h = severe OSAH.³⁶ Other PSG indices included percentage of time spent in each sleep stage (N1, N2, N3, and REM). Percent of the night spent in each sleep stage was calculated by taking number of minutes in each stage divided by total sleep time (in minutes). Normal percentages of sleep time spent in N1 = 2–5%, in N2 = 45–55%, N3 = 5–20%, and REM = 20–25%.³⁷

PTSD Checklist for DSM-5

The PTSD Checklist for DSM-5 (PCL-5)³⁸ is a reliable and valid 20-item, self-report measure that assesses the DSM-5 symptoms of PTSD. Scoring is based on how much the individual is bothered by the symptoms during the past month on a scale from 0 (not at all) to 4 (extremely). Scores at or above 31 suggest a probable PTSD diagnosis in service members.³⁹ Findings between PCL-5 and sleep measures (described below) were similar regardless of inclusion or exclusion of the 2 PCL-5 sleep items (“Repeated, disturbing dreams of the stressful experience?” and “Trouble falling or staying asleep?”). Therefore, we kept these 2 items in the results. In this study, the internal consistency of the PCL-5 with all 20 items was α = 0.90.

Dimensions of Anger Reactions-5

The Dimensions of Anger Reactions-5 (DAR-5)⁴⁰ is a 5-item, short-form version of the original Dimensions of Anger Reactions Scale.⁴¹ It assesses anger frequency, intensity, duration, aggression, and interference with social functioning. Participants indicate the degree to which each of 5 items describes their feelings or behavior over the last 4 weeks, from 1 (“none or almost none of the time”) to 5 (“all or almost all of the time”). The items are summed for a score ranging from 5 to 25; a cut-point of 12 is recommended to indicate a level of anger that warrants clinical attention.⁴⁰ Current sample internal consistency was α = 0.87.

Data analyses

All analyses were run in the statistical program R version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria). Descriptive statistics were used to characterize the demographics of the sample. Pearson correlations with 95% confidence intervals were used to examine relationships between PSG variables and PCL-5 and DAR-5. For Aim 1, multiple linear regression analyses were run using R and the R package sjPlot with retrospective sleep measures (ISI) and demographic covariates (all entered simultaneously) with PCL-5 and DAR-5. For Aim 2, multiple linear regression analyses were run with prospective sleep diary metrics (eg, weekly mean of sleep efficiency, number of awakenings, sleep quality, total sleep time nightmare frequency and severity) as simultaneous independent variables, and PCL-5 and DAR-5 scores were used as the dependent variables. Finally, for Aim 3, multiple linear regression analyses with PSG sleep architecture metrics (eg, total AHI, REM AHI, arousal index, and percent in each sleep stage) were the independent variables and the PCL-5 and DAR-5 were the dependent variables. REM AHI was examined in addition to total AHI variable based on previous literature linking it to greater emotional distress.⁴² Each sleep stage, total AHI, and REM AHI were examined in separate regression models to avoid multicollinearity. For all regression models, variance inflation factors were calculated to examine potential multicollinearity between variables.

RESULTS

Ninety-three service members who had PTSD, insomnia disorder, and nightmare disorder completed the baseline assessment and made up the current sample. The sample was mostly men (73%), with a mean age of 35.9 years (standard deviation [SD] = 8.4). Almost half (45%) identified as White, followed by 34% who identified as African American. The majority (87%) had completed at least some college. Most participants (94%) were in the noncommissioned officer grades of Enlisted 4, 5, and 6 (66%). Participants had served an average of 14.6 years (SD = 7.1). Additional demographic and military variables are presented in Table 1.

Table 1.

Demographic characteristics of service members (n = 93).

	Values
Age, years	36.25 (7.53)
Years in military	14.61 (7.08)
Number of deployments	2.60 (1.08)
Sex, female	25 (27%)
Hispanic ethnicity	22 (24%)
Race
White	42 (45%)
African American	32 (34%)
Other	9 (10%)
Native American	6 (7%)
Pacific Islander	3 (3%)
Asian American	1 (1%)
Marital status
Married/living with partner	69 (74%)
Divorced	16 (17%)
Single/not living with partner	8 (9%)
Education
Some college	45 (48%)
4-year college degree	18 (19%)
Associate degree	14 (15%)
High school diploma	11 (12%)
Master’s degree	4 (4%)
GED	1 (1%)
Military status
Active duty	86 (93%)
National Guard	1 (1%)
Retired	6 (7%)
Grade
Junior enlisted E1-E3	1 (1%)
Noncommissioned officer E4-E6	61 (66%)
Senior noncommissioned officer E7-E9	27 (29%)
Warrant officer WO1-O5	4 (4%)
Branch = Army	93 (100%)

Values presented as mean (standard deviation) or n (%). Percentages reflect valid percentages (ie, missing data excluded from the total in calculations). E = enlisted, GED = General Educational Diploma, O = officer, WO = warrant officer.

As expected with the study selection criteria, the average scores on the ISI (mean [M] = 21.45, SD = 4.55) and PCL-5 (M = 51.48, SD = 13.06) exceeded the clinical cutoffs of 15 and 31, respectively, indicating the group had significant insomnia and PTSD symptoms (Table 2). Diary-assessed mean sleep efficiency was 70%. Participants reported an average of 5.5 (SD = 4.4) nightmares per week and 0.9 (SD = 0.7) nightmares per night. Mean sleep diary-assessed total sleep time was 5.5 hours and was consistent with PSG-assessed total sleep time of 5.4 hours. During the overnight PSG study, the group spent a slightly higher than average percent of time in Stage 3 (N3) sleep (25.5% vs 5%–20%) and slightly lower than average percent of time spent in REM (18.3% vs 20%–25%) than the general population, possibly due to the participant being woken up during the sleep study if positive OSAH results were found. More than half the sample (53%) met criteria for OSAH (ie, apnea-hypopnea index ≥ 5 events/h). PSG measured AHI was higher during REM compared to during the total sleep period (M = 14.8; M = 10.5, respectively). Eighty-five percent of the sample exceeded the clinical cutoff of 12 for anger difficulties. Additional sleep, PTSD, and anger descriptive statistics are presented in Table 2.

Table 2.

Descriptive statistics of sleep, PTSD, and anger (n = 93).

	Values
Insomnia Severity Index	21.45 (4.55) [8–28]
Sleep diary (means across 1 week)
Total sleep time (hours)	5.50 (1.30) [1.09–9.03]
Sleep efficiency	70.42 (11.73) [17.14–85.67]
Sleep quality (0 = very  poor; 4 = very good)	1.55 (0.59) [0–3]
Nightmare frequency  (per night)	0.87 (0.72) [0–4]
Nightmare frequency  (per week)	5.53 (4.36) [0–21]
Nightmare severity (0 = not  at all; 4 = extremely)	2.10 (0.90) [0–4]
Polysomnography
Total sleep time (hours)	5.43 (1.74) [1.3–7.58]
Architecture
N1%	5.49 (4.89) [0–24.7]
N2%	51.45 (13.38) [14.7–79.6]
N3%	25.46 (15.37) [1.4–76.9]
REM%	18.32 (10.92) [0–44.1]
Sleep-disordered breathing
AHI – Total (events/h)	10.51 (14.66) [0–92.9]
AHI – REM	14.77 (19.21) [0–101.1]
AHI – Total ≥ 5	40 (53%)
Arousal index	17.16 (13.34) [4–89.5]
PTSD
CAPS-5	35.72 (7.23) [18–51]
PCL-5	51.48 (13.06) [21–80]
PCL-5 (minus insomnia and  nightmare items)	45.35 (12.16) [19–72]
Anger
DAR-5	17.41 (5.31) [5–25]

Values are presented as mean (standard deviation) [range] or n (%). Percentages reflect valid percentages (ie, missing data excluded from the total in calculations). AHI = apnea-hypopnea index, CAPS-5 = Clinician Administered PTSD Scale for the DSM-5, DAR-5 = Dimensions of Anger Scale-5, N1% = percentage of time spent in Stage 1 sleep, N2% = percentage of time spent in Stage 2 sleep, N3% = percentage of time spent in Stage 3 sleep, PCL-5 = PTSD Checklist for DSM-5, PTSD = posttraumatic stress disorder, REM% = percentage of time spent in rapid eye movement sleep.

Table 3 displays the Pearson correlations between PSG variables and PCL-5 and DAR-5. PTSD was positively correlated with percent of time spent in N1 sleep (r = .30, P < .05) and negatively correlated with total sleep time (r = −.24, P < .05). Anger symptoms were positively correlated with REM AHI (r = .30, P < .05).

Table 3.

Correlations with confidence intervals between PSG variables with PCL-5 and DAR-5.

Variable	1	2	3	4	5	6	7	8	9	10	11
1. Total AHI
2. REM AHI	.63** [.46, .76]
3. Arousal Index	.81** [.72, .88]	.52** [.32, .68]
4. N1%	.14 [−.10, .36]	−.01 [−.25, .23]	.25* [.02, .46]
5. N2%	.03 [−.20, .26]	.06 [−.18, .30]	.16 [−.08, .38]	.01 [−.23, .24]
6. N3%	.15 [−.09, .37]	-.10 [−.33, .15]	−.01 [−.25, .22]	−.20 [−.41, .04]	−.74** [−.83, −.61]
7. REM%	−.33** [−.52, −.10]	.08 [-.16, .31]	−.33** [−.52, −.10]	−.21 [−.43, .02]	−.23 [−.44, .00]	−.37** [−.56, −.15]
8. TST	−.65** [−.77, −.50]	−.27* [−.48, −.03]	−.56** [−.70, −.37]	−.24* [−.45, −.00]	.07 [−.16, .30]	−.40** [−.58, -.18]	.54** [.35, .69]
9. SE	−.17 [−.39, .07]	−.00 [−.24, .24]	−.34** [−.54, −.12]	−.40** [−.58, −.18]	−.27* [−.47, −.03]	.10 [−.14, .33]	.39** [.17, .57]	.47** [.26, .63]
10.WASO	−.13 [−.36, .11]	−.05 [−.29, .19]	−.09 [−.32, .15]	−.01 [−.24, .23]	.36** [.14, .55]	−.30* [−.50, −.07]	−.02 [−.25, .22]	.07 [−.17, .30]	−.31** [−.51, −.08]
11. DAR-5	.20 [−.03, .40]	.26* [.03, .47]	.14 [−.10, .36]	−.03 [−.26, .21]	.11 [−.13, .34]	−.15 [−.37, .08]	.12 [−.11, .34]	−.18 [−.40, .05]	−.04 [−.27, .20]	−.13 [−.36, .11]
12. PCL-5	.14 [−.09, .36]	.04 [−.20, .27]	.18 [−.05, .40]	.30* [.07, .49]	.07 [−.17, .30]	−.08 [−.31, .16]	−.04 [−.27, .20]	−.24* [−.44, -.01]	−.17 [−.39, .06]	−.10 [−.33, .14]	.46** [.28, .61]

Values in square brackets indicate the 95% confidence interval for each correlation. “n” varies (n = 68 to 93) for correlations. The confidence interval is a plausible range of population correlations that could have caused the sample correlation.⁵¹ *P < .05. **P < .01. AHI = apnea-hypopnea index, DAR-5 = Dimensions of Anger Scale-5, N1% = percentage of time spent in Stage 1 sleep, N2% = percentage of time spent in Stage 2 sleep, N3% = percentage of time spent in Stage 3 sleep, PCL-5 = PTSD Checklist for DSM-5, PSG = polysomnography, REM% = percentage of time spent in rapid eye movement sleep, SE = sleep efficiency, TST = total sleep time, WASO = wake after sleep onset.

Aim 1: Examining relationships of retrospective self-report insomnia symptoms with PTSD and anger symptoms

Insomnia symptoms were associated with greater PTSD symptoms (β = 0.58, P < .01) and greater anger symptoms (β = 0.39, P < .01) (Table 4).

Table 4.

Multiple linear regressions with retrospective self-report insomnia symptoms and PTSD and anger symptoms (n = 84).

	PCL-5				DAR-5
Predictors	Estimates	Std Beta	95% CI	P	Estimate	Std Beta	95% CI	P
Intercept	24.32	0.00	3.94–44.70	.020	15.19	0.00	5.53–24.85	.002
ISI	1.64	0.58	1.12–2.15	< .001	0.43	0.37	0.19–0.67	.001
Age	−0.07	−0.04	−0.38–0.24	.647	−0.14	−0.21	−0.29–0.00	.055
Female	−1.60	−0.05	−7.31–4.11	.578	−0.55	−0.04	−3.26–2.15	.686
Enlisted	−3.82	−0.06	−14.77–7.13	.490	−2.31	−0.09	−7.50–2.88	.378
#Deploy	−0.63	−0.05	−2.82–1.56	.569	0.12	0.02	−0.92–1.16	.820
R2/R2 adjusted	0.359/0.318				0.177/0.124

Enlisted is coded as 0 = warrant officer (WO1-O5), 1 = enlisted (E1-E9), Female is coded as 0 = male, 1 = female, #Deploy = number of deployments. Values that are bolded indicate statistically significant results. CI = confidence interval, DAR-5 = Dimensions of Anger Scale-5, ISI = Insomnia Severity Index, PCL-5 = PTSD Checklist for DSM-5, Std = Standardized.

Aim 2: Examining relationships of prospective sleep diary measures with PTSD and anger symptoms

Greater mean nightmare frequency (β = 0.26, P = .0017) and greater mean nightmare severity (β = 0.24, P = .0042) were associated with greater PTSD symptoms. Greater nightmare severity (β = 0.30, P = .015) was associated with greater anger symptoms (Table 5).

Table 5.

Multiple linear regressions with mean prospective sleep diary measures associating with PTSD and anger severity (n = 88).

	PCL-5				DAR-5
Predictors	Estimates	Std Beta	95% CI	P	Estimates	Std Beta	95% CI	P
Intercept	48.78	0.00	29.09–68.47	< .001	7.81	0.00	−0.85–16.47	.076
TST	−0.07	−0.01	−2.57–2.42	.954	0.30	0.07	−0.80–1.39	.592
SE	−0.05	−0.05	−0.34–0.23	.715	0.08	0.17	−0.05–0.20	.219
SQ	−2.96	−0.13	−7.58–1.65	.205	−1.18	−0.13	−3.21–0.86	.253
Nightmare severity	3.44	0.24	0.14–6.74	.042	1.81	0.30	0.36–3.27	.015
Nightmare frequency	4.87	0.26	0.89–8.85	.017	0.46	0.06	−1.29–2.21	.602
R2/R2 adjusted	0.234/0.187				0.130/0.077

Values that are bolded indicate statistically significant results. All sleep diary variables represent the mean values across 7 days. R² indicates the total percentage of variance explained; R² adjusted (adj.) indicates the total percentage of variance explained, adjusted for the number of independent variables in the model. CI = confidence interval, DAR-5 = Dimensions of Anger Reactions-5, PCL-5 = PTSD Checklist for DSM-5, SE = sleep efficiency, SQ = sleep quality, Std = standardized, TST = total sleep time.

Aim 3: Examining relationships of diagnostic PSG measures predicting PTSD and anger symptoms

A greater percentage of time spent in Stage 1 (N1) sleep was associated with greater PTSD symptoms (β = 0.27, P = .028). None of the above PSG results changed when we substituted REM AHI for total AHI (Table 6). Greater REM AHI was associated with greater anger symptoms (β = 0.31 to 0.29; Table 7) when percentage of REM sleep was not included in the model. Across all regression models, all variance inflation factor values were < 4, suggesting that variables were not highly multicollinear.

Table 6.

Multiple linear regressions with total AHI and PSG indices associating with PTSD severity (n = 71) and REM AHI and PSG associating with PTSD severity (n = 66).

PCL-5					PCL-5
Predictors	Est.	β	95% CI	P	Predictors	Est.	β	95% CI	P
Intercept	46.08	−0.00	40.46–51.70	< .001	Intercept	44.91	−0.00	39.30–50.53	< .001
Total AHI	0.06	0.07	−0.29–0.41	.730	REM AHI	−0.02	−0.03	−0.21–0.18	.860
Arousal Index	0.06	0.06	−0.34–0.45	.733	Arousal Index	0.12	0.13	−0.15–0.40	.374
N1%	0.71	0.27	0.08–1.34	.028	N1%	0.74	0.28	0.08–1.39	.028
N2%					N2%
N3%					N3%
REM%					REM%
R2/R2 adj.	0.102/0.062				R2/R2 adj.	0.113/0.070
Intercept	46.67	0.00	34.32–59.02	< .001	Intercept	45.85	−0.00	32.70–59.00	< .001
Total AHI	0.03	0.03	−0.34–0.40	.874	REM AHI	−0.05	−0.08	−0.26–0.15	.588
Arousal Index	0.14	0.15	−0.26–0.55	.485	Arousal Index	0.21	0.22	−0.07–0.49	.133
N1%					N1%
N2%	0.04	0.04	−0.19–0.28	.725	N2%	0.04	0.04	−0.21–0.29	.740
N3%					N3%
REM%					REM%
R2/R2 adj.	0.036/−0.008				R2/R2 adj.	0.042/−0.004
Intercept	50.65	−0.00	42.75–58.55	< .001	Intercept	50.70	−0.00	43.33–58.07	< .001
Total AHI	0.05	0.06	−0.32–0.43	.778	REM AHI	−0.07	−0.10	−0.27–0.13	.498
Arousal Index	0.13	0.13	−0.28–0.53	.532	Arousal Index	0.23	0.24	−0.04–0.50	.100
N1%					N1%
N2%					N2%
N3%	−0.07	−0.09	−0.28–0.14	.494	N3%	−0.12	−0.13	−0.33–0.10	.290
REM%					REM%
R2/R2 adj.	0.041/−0.002				R2/R2 adj.	0.058/0.012
Intercept	48.93	−0.00	40.41–57.45	< .001	Intercept	47.46	−0.00	37.93–56.99	< .001
Total AHI	0.02	0.02	−0.35–0.38	.930	REM AHI	−0.06	−0.09	−0.27–0.15	.580
Arousal Index	0.16	0.16	−0.24–0.56	.430	Arousal Index	0.23	0.24	−0.08–0.53	.148
N1%					N1%
N2%					N2%
N3%					N3%
REM%	−0.01	−0.01	−0.32–0.29	.932	REM%	0.02	0.01	−0.34–0.37	.919
R2/R2 adj.	0.034/−0.009				R2/R2 adjusted	0.041/−0.006

Values that are bolded indicate statistically significant results. R² indicates the total percentage of variance explained; R² adjusted (adj.) indicates the total percentage of variance explained, adjusted for the number of independent variables in the model. AHI = apnea-hypopnea index, CI = confidence interval, Est = estimate, N1% = percentage of time spent in Stage 1 sleep, N2% = percentage of time spent in Stage 2 sleep, N3% = percentage of time spent in Stage 3 sleep, PCL-5 = PTSD Checklist for DSM-5, REM = rapid eye movement, REM% = percentage of time spent in REM sleep.

Table 7.

Multiple linear regressions with total AHI and PSG indices associating with anger (n = 71) and REM AHI and PSG indices associating with anger (n = 66).

DAR-5					DAR-5
Predictors	Est.	β	95% CI	P	Predictors	Est.	β	95% CI	P
Intercept	17.16	−0.00	14.69–19.62	< .001	Intercept	16.28	−0.00	13.88–18.68	< .001
Total AHI	0.11	0.28	−0.05–0.26	.171	REM AHI	0.09	0.31	0.01–0.17	.038
Arousal Index	−0.03	−0.08	−0.21–0.14	.695	Arousal Index	−0.01	−0.04	−0.13–0.10	.813
N1%	−0.05	−0.05	−0.33–0.22	.704	N1%	−0.00	−0.00	−0.28–0.28	.987
N2%					N2%
N3%					N3%
REM%					REM%
R2/R2 adj.	0.050/0.007				R2/R2 adj.	0.084/0.039
Intercept	14.56	−0.00	9.37–19.76	< .001	Intercept	15.36	−0.00	9.96–20.77	< .001
Total AHI	0.12	0.33	−0.03–0.28	.115	REM AHI	0.09	0.31	0.01–0.17	.034
Arousal Index	−0.06	−0.15	−0.23–0.11	.473	Arousal Index	−0.02	−0.04	−0.13–0.10	.761
N1%					N1%
N2%	0.05	0.12	−0.05–0.15	.310	N2%	0.02	0.04	−0.08–0.12	.716
N3%					N3%
REM%					REM%
R2/R2 adj.	0.063/0.021				R2/R2 adj.	0.086/0.041
Intercept	19.09	−0.00	15.81–22.38	< .001	Intercept	17.01	−0.00	13.96–20.06	< .001
Total AHI	0.15	0.39	−0.01–0.30	.065	REM AHI	0.09	0.29	0.00–0.17	.043
Arousal Index	−0.08	−0.18	−0.24–0.09	.374	Arousal Index	−0.01	−0.03	−0.12–0.10	.835
N1%					N1%
N2%					N2%
N3%	−0.08	−0.21	−0.16–0.01	.085	N3%	−0.03	−0.08	−0.12–0.06	.506
REM%					REM%
R2/R2 adj.	0.089/0.049				R2/R2 adj.	0.090/0.046
Intercept	14.88	−0.00	11.33–18.43	< .001	Intercept	15.39	−0.00	11.48–19.30	< .001
Total AHI	0.12	0.33	−0.03–0.28	.105	REM AHI	0.08	0.28	−0.00–0.17	.064
Arousal Index	−0.03	−0.07	−0.20–0.14	.718	Arousal Index	0.00	0.00	−0.13–0.13	.991
N1%					N1%
N2%					N2%
N3%					N3%
REM%	0.10	0.19	−0.03–0.22	.137	REM%	0.04	0.07	−0.11–0.19	.593
R2/R2 adj.	0.079/0.038				R2/R2 adj.	0.088/0.044

Values that are bolded indicate statistically significant results. R² indicates the total percentage of variance explained; R² adjusted (adj.) indicates the total percentage of variance explained, adjusted for the number of independent variables in the model. AHI = apnea-hypopnea index, CI = confidence interval, DAR-5 = Dimensions of Anger Reactions-5, Est = estimate, N1% = percentage of time spent in Stage 1 sleep, N2% = percentage of time spent in Stage 2 sleep, N3% = percentage of time spent in Stage 3 sleep, REM = rapid eye movement, REM% = percentage of time spent in REM sleep.

DISCUSSION

We evaluated the associations between retrospective self-report insomnia, prospective daily sleep diaries, and PSG indices with PTSD symptom severity and anger severity in 93 PTSD treatment-seeking service members with comorbid PTSD, nightmares, and insomnia disorder. As expected for the study selection criteria, PTSD severity (M = 51.48, SD = 13.06), insomnia severity (M = 21.45, SD = 4.55), and nightmare frequency per week (5.5/wk; 0.9/night) were all high. In addition, average sleep duration was short based on sleep diary (5.5 hours) and PSG (5.4 hours). Fifty-three percent of participants had OSAH (ie, AHI ≥ 5 events/h), indicating a sleep apnea diagnosis, which is higher than the 20%–33% found in previous studies that examined veterans in PTSD treatment.¹¹ Yet, the observed rate of OSAH was consistent with a recent meta-analysis, suggesting that 63% of veterans and service members with PTSD have probable OSAH.¹⁰ Our high OSAH incident rate is likely due to the recruitment of participants from a sleep disorders clinic, which is similar to many of the samples in the meta-analysis.¹⁰

Insomnia symptoms (ie, ISI), the number and severity of nightmares (from sleep diary), and the percent time spent in N1 sleep (from PSG) were each positively associated with PTSD symptom severity. The insomnia finding is consistent with other clinical⁵ and epidemiological findings.²⁶ The nightmare finding is also similar to prior studies.^3,5,8 Because this study was cross-sectional, the results may be due to unspecified third variables, and the direction of the relationship (and whether it is reciprocal) was unable to be tested. Some studies have shown that PTSD treatment does not alleviate insomnia and only partially alleviates nightmare disorder, particularly in service members and veterans.^3,8 In contrast, studies of cognitive-behavioral treatments for nightmares (which include a range of insomnia-specific treatment components) typically find significant decreases in PTSD symptoms outside of nightmares and insomnia,⁴³ although this was not found in a sample of Vietnam veterans treated with group therapy.⁴⁴

More time spent in N1 sleep per night was also associated with greater PTSD symptom severity. This finding is inconsistent with a meta-analysis that failed to find a difference between PTSD and controls in the percent of time spent in N1.¹⁶ However, this same meta-analysis found that PTSD was associated with reduced sleep efficiency, slow-wave sleep, and total sleep time, and greater wake after sleep onset, all of which may partially be attributed to increased sleep fragmentation and more time spent in N1 sleep. Discrepancies between our study and this meta-analysis may also be attributable to PSG score differences. Most studies in the meta-analysis used Rechtschaffen and Kales PSG scoring rules, whereas we used the American Academy of Sleep Medicine scoring recommendations.^16,45

It is unclear if sleep architecture is changed by PTSD or if abnormal sleep architecture is a predisposing factor for PTSD. The high prevalence of OSAH found in our cohort could be related to more sleep interruptions in deeper stages of sleep, resulting in awakenings and more N1 sleep, which may exacerbate PTSD symptoms. Longitudinal studies are needed to understand this relationship. If abnormal sleep architecture precedes PTSD symptom development, it may be a possible biomarker that helps identify those most at risk of developing PTSD after trauma exposure.

In this treatment-seeking sample of service members, all met PTSD criteria and more than 53% exceeded the OSAH PSG cutoff. Consequently, we may have had a problem with limited range of PTSD and OSA. Future research should examine if OSAH and prescribed continuous positive airway pressure use are related to PTSD severity in larger samples that contain more variance in PTSD and OSAH symptoms. Continuous positive airway pressure use has been associated with reductions in PTSD symptoms even without PTSD-focused treatment.⁴⁶

Strikingly, 85% of our sample exceeded the cutoff for clinical levels of anger difficulties warranting further assessment and treatment by a mental health clinician. This is concerning, because anger is linked to a wide range of negative psychological, physical, legal, and health outcomes.^17,47,48 Insomnia, nightmare severity, and higher REM AHI were positively associated with anger symptoms. Service members who have REM sleep interrupted by apneas and hypopneas may experience greater anger symptoms, which is consistent with previous literature demonstrating that achieving REM sleep is related to fewer and less intense negative emotions.⁴² Finding ways to reduce anger is a high priority for service members and their families. While traditional treatments for anger problems typically involve cognitive-behavioral therapies,⁴⁹ our results suggest that a more holistic approach may be warranted for those with PTSD and anger problems. Assessing and providing treatment for insomnia, nightmares, and OSAH may be a novel method to reduce anger in service members with PTSD. Additionally, service members may be more willing to initiate treatment for sleep disorders (ie, insomnia, nightmares, OSAH) than for anger or PTSD, because there is less stigma associated with sleep conditions. In turn, alleviating distress related to sleep disorders may make the service member more willing to engage in PTSD and anger therapies.

This study includes several limitations. As stated, restriction of range of PTSD, insomnia, and nightmare symptoms may have led to a Type-II error (failing to find a statistical relationship when one actually exists). These results may not generalize to other service members or to service members with only 1 of the conditions studied. Future research would benefit from examining a continuum of PTSD severity in relation to sleep, rather than only the most severe cases. PSG studies were completed over a single night in an unfamiliar environment, so the results may not reflect typical sleep architecture. The data were cross-sectional and cannot infer causality. Future longitudinal and experimental research should explore bidirectional associations between PTSD, anger, and sleep disorder symptoms, which may lead to potential biomarkers for PTSD and anger.

Notwithstanding these threats to internal and external validity, this study was noteworthy because it used multiple state of the science methods to comprehensively examine the associations between sleep problems with PTSD and anger symptoms. Using a variety of methods to assess sleep is important because subjective reports of sleep do not always match objective measurements,⁵⁰ especially regarding sleep efficiency and wake after sleep onset. Conversely, it is also unclear if objective measurements truly capture the effect of disturbed sleep on subjective well-being.⁵⁰

Insomnia and nightmares are often regarded as independent treatment targets. However, our results suggest providers should treat these conditions if they want to reduce PTSD symptoms and anger, 2 conditions difficult to reduce with current treatment methods.^1,49 Insomnia can be easily assessed with brief, free self-report measures (eg, the ISI) that can be given by providers in a variety of clinics. Cognitive behavioral therapy for insomnia and nightmares should be offered to patients who endorse these problems. Although total AHI was not related to PTSD in our sample, REM-related AHI was related to higher levels of anger. In addition, increased N1 sleep, perhaps because of OSAH, was related to greater PTSD severity. Thus, as mentioned above, patients with PTSD and anger should be fully assessed for OSAH and provided treatment with continuous positive airway pressure, which should improve daytime function, energy, and possibly motivation to complete PTSD and anger therapies.

ABBREVIATIONS

AHI

apnea-hypopnea index

CI

confidence interval

DAR-5

Dimensions of Anger Reactions-5

DSM-5

Diagnostic and Statistical Manual for Mental Disorders, fifth edition

ISI

Insomnia Severity Index

M

mean

N1%

percentage of time spent in stage 1 sleep

N2%

percentage of time spent in stage 2 sleep

N3%

percentage of time spent in stage 3 sleep

OSAH

obstructive sleep apnea-hypopnea

PCL-5

PTSD Checklist for DSM-5

PSG

polysomnography

PTSD

posttraumatic stress disorder

REM

rapid eye movement

SD

standard deviation

DISCLOSURE STATEMENT

All authors have seen and approved this manuscript. Work for this study was performed at Carl R. Darnall Army Medical Center, located on the Fort Hood military installation in Killeen, Texas. This study was funded by Consortium to Alleviate PTSD (CAP) award numbers W81XWH-13-2-0065 from the US Department of Defense, Defense Health Program, Psychological Health and Traumatic Brain Injury Research Program (PH/TBI RP), and I01CX001136-01 from the US Department of Veterans Affairs, Office of Research & Development, Clinical Science Research & Development Service. The views expressed herein are solely those of the authors and do not reflect an endorsement by or the official policy or position of the US Army, the Department of Defense, the Department of Veterans Affairs, or the US Government. The data from this study are maintained at the University of Texas Health Science Center at San Antonio in the STRONG STAR Repository. Requests for access to the data can be emailed to repository@strongstar.org. The authors report no conflicts of interest.