Sleep disturbances are common in United States military personnel. Despite their exposure to combat and trauma, little is known about nightmares in this population. The purpose of this study was to describe the prevalence and associated clinical and polysomnographic characteristics of nightmares in United States military personnel with sleep disturbances.
Retrospective review of 500 active duty United States military personnel who underwent a sleep medicine evaluation and polysomnography at our sleep center. The Pittsburgh Sleep Quality Index and the Pittsburgh Sleep Quality Index-Addendum were used to characterize clinically significant nightmares. Subjective and objective sleep attributes were compared between groups.
At least weekly nightmares were present in 31.2%; yet, only 3.9% reported nightmares as a reason for evaluation. Trauma-related nightmares occurred in 60% of those patients with nightmares. Patients with nightmares had increased sleep onset latency (SOL) and rapid eye movement (REM) sleep latency (mean SOL/REM sleep latency 16.6/145 minutes, P = .02 and P = .01 respectively) compared to those without (mean SOL/REM sleep latency 12.5/126 minutes). The comorbid disorders of depression (P ≤ .01, relative risk [RR] 3.55 [95% CI, 2.52–4.98]), anxiety (P ≤ .01, RR 2.57 [95% CI, 1.93–3.44]), posttraumatic stress disorder (P ≤ .01, RR 5.11 [95% CI, 3.43–7.62]), and insomnia (P ≤ .01, RR 1.59 [95% CI, 1.42–1.79]) were all associated with nightmares.
Clinically significant nightmares are highly prevalent in United States military personnel with sleep disturbances. Nightmares are associated with both subjective and objective sleep disturbances and are frequently comorbid with other sleep and mental health disorders.
A commentary on this article appears in this issue on page 303.
Creamer JL, Brock MS, Matsangas P, Motamedi V, Mysliwiec V. Nightmares in United States military personnel with sleep disturbances. J Clin Sleep Med. 2018;14(3):419–426.
Current Knowledge/Study Rationale: The spectrum of sleep disorders in military personnel is varied and complicated by the presence of comorbidities and psychological disorders. Currently, little is known regarding nightmares, especially trauma-related nightmares.
Study Impact: This is the first study to report on objective and subjective sleep attributes in a cohort of military personnel, determining the prevalence of nightmares and polysomnographic characteristics associated with nightmares. Clinically, this highlights the need for military and civilian health care providers to evaluate for nightmares in individuals with sleep disturbances after traumatic experiences. From a research perspective, these findings provide the basis to further address nightmares, which are associated with the pressing issues of sleep and behavioral medicine disorders and suicidality in military and veteran populations.
Nightmares are defined by repeated occurrences of extended, extremely dysphoric, and well-remembered dreams that cause clinically significant distress or impairment in social, occupational, or other areas of functioning.1 Occasional nightmares are part of a normal adaptive response to life stressors and are not pathologic, with 85% of adults reporting at least one nightmare per year.2 The prevalence of clinically significant nightmares, which occur at least once per week, is estimated between 0.9% and 6.8%.3–9 In the largest study to date (69,813 adults), the prevalence of nightmares was 3.5% in men and 4.8% in women.10 Participants from the World War II generation had a significantly higher prevalence of 7.2% in men and 7% in women. Nightmares are more common among patients with sleep disturbances, sleep disorders, depression, anxiety, and posttraumatic stress disorder (PTSD).2,9,11–15 Additionally, nightmares are associated with a heightened risk of suicidal ideation in both military and civilian populations.16,17 As sleep and psychiatric disorders are frequently reported in United States military personnel, this population may have a high prevalence of nightmare disorder (NDO).18 However, most research evaluating sleep disorders in active duty military personnel has focused on insomnia, obstructive sleep apnea (OSA), and PTSD, noting that a recent study focusing on OSA and PTSD did not evaluate nightmares.18–20
Following traumatic experiences, sleep disturbances are frequently reported, including trauma-related nightmares (TRN).21 The nightmares of trauma survivors tend to be more severe and distressing than idiopathic nightmares.1,3 Though not exclusive to PTSD, TRN have been referred to as the “hallmark” of this disorder, with rates as high as 90%.22,23 Addressing nightmares as a distinct sleep disorder following a traumatic experience is of clinical importance as patients with disturbed sleep have poor outcomes and sleep-focused therapy can lead to improvement in PTSD, anxiety, and depression.24,25
Research on nightmares in military personnel is limited, though data suggest that veterans who have left military service have a higher prevalence than the general population.10 The primary objective of this study was to characterize the prevalence of nightmares in United States military personnel with sleep disturbances. The secondary objective was to describe the clinical and polysomnography (PSG)-measured correlates of military personnel with nightmares.
This is a retrospective cohort study of 500 military personnel who were referred for evaluation to an academic military sleep disorders center between January 2016 and December 2016. Our laboratory performs approximately 2,500 PSG tests on military personnel per year. Individuals who had previously undergone PSG, those referred for postsurgical evaluation, or those who did not complete our sleep center intake questionnaire were excluded from the analysis. Groups of 50 consecutive-performed PSG tests on active duty military personnel were randomly pulled from 10 different time periods throughout the year. We evaluated the records of 106 females and 394 males (7 were excluded due to missing data elements). This resulted in a cohort of 106 females and 387 males. All patients underwent our standard diagnostic evaluation, which involves completing a self-report questionnaire consisting of the Epworth Sleepiness Scale (ESS), Insomnia Severity Index (ISI), Pittsburgh Sleep Quality Index (PSQI), questions from the PSQI-Addendum (PSQI-A), and medical and military history elements. Regarding deployment experiences, patients were asked to report whether they had deployed to Operation Iraqi Freedom, Enduring Freedom, or other deployments. To assess the use of sleep aids, patients answered a question whether they were taking the following medications: zolpidem, eszopiclone, zaleplon, trazodone, mirtazapine, or other to include over-the-counter medications. They were then classified as either taking or not taking a sleep aid. All participants underwent an attended level 1 in-laboratory diagnostic PSG. After data collection, all study variables were entered into a de-identified database prior to statistical analysis. This study was approved by the Wilford Hall Ambulatory Surgical Center institutional review board.
Demographic/biometric parameters (age, sex, deployment history, and branch of service), sleep center intake questionnaire, PSG variables, and electronic medical records (EMR) for each patient were reviewed.
Nightmares and Trauma-Related Nightmares
Self-reported reasons for sleep medicine evaluation were noted, determining if nightmares were listed as a specific symptom. A single item from the PSQI was used to assess nightmare frequency: “During the past month, how often have you had trouble sleeping because you have bad dreams?”26 A response of bad dreams at least weekly is consistent with NDO.3 The PSQI-A was used to assess nightmares related to trauma exposure and to differentiate between NDO and TRN.27 A single item from the PSQI-A was used to assess TRN: “During the past month, how often have you had trouble sleeping because you had memories or nightmares of a traumatic experience?” For the diagnosis of TRN, individuals were required to have at least weekly bad dreams on the PSQI and memories/nightmares of traumatic experiences on the PSQI-A.
The ESS was used to assess patients' sleepiness. The total ESS score ranges from 0 (less sleepy) to 24 (more sleepy). A score > 10 indicates excessive daytime sleepiness.28 Self-reported sleep efficiency (SE) was derived from PSQI item 4 by dividing reported time in bed by reported sleep time.
The ISI was used to assess insomnia symptoms with scores that range from 0 to 28. A score ≥ 15 is consistent with clinical insomnia.29
Patients with a PSG demonstrating apneas or hypopneas with an apnea-hypopnea index (AHI) > 5 events/h were rendered a diagnosis of OSA.
Mental Health Disorders and Traumatic Brain Injury
Diagnoses of interest to include depression, anxiety, PTSD, and traumatic brain injury (TBI) were obtained from each patient's EMR.
Level 1 PSG tests were performed in accordance with American Academy of Sleep Medicine (AASM) standards within an AASM accredited laboratory (Sandman Version 9.3, Embla Systems, Broomfield, Colorado, United States). PSG was performed with 16 channels, including: electrooculogram, electroencephalogram, electrocardiogram, electromyogram (submental and bilateral tibial), airflow measurements using both oronasal-thermal sensors and nasal air pressure transducers, transtracheal sounds via microphone, rib cage and abdominal movement by inductance plethysmography using thoracoabdominal belts, and continuous pulse oximetry. Studies were scored utilizing the 2012 AASM scoring guidelines with hypopneas scored as a 30% drop in the nasal pressure from baseline for at least 10 seconds and associated with an arousal and/or drop in oxygen saturation by 3%.30 PSG-measured variables, to include sleep onset latency (SOL), rapid eye movement (REM) sleep latency, total sleep time (TST), SE, sleep stages (stage N1, stage N2, stage N3, stage R sleep), wake after sleep onset (WASO), arousal index, AHI, and maximal desaturation were analyzed. All test results were reviewed and adjudicated by a board-certified sleep medicine physician.
Statistical analysis was conducted with a statistical software package (JMP Pro 12; SAS Institute; Cary, North Carolina, United States). Data normality was assessed with the Shapiro-Wilk test. Statistical analysis was based on parametric and nonparametric methods as deemed appropriate. All variables underwent descriptive analysis to describe our population in terms of demographic characteristics. Participants were then classified in two groups: nightmare disorder (NDO) and non-nightmare disorder (non-NDO), and differences between groups were assessed. For a secondary analysis, we classified patients in two groups (ie, the “NDO only” group [patients having NDO without memories or nightmares of a traumatic experience], and the group of patients in whom trauma-related nightmares, “TRN” [having both bad dreams and memories or nightmares of a traumatic experience]), were diagnosed.
An α level of .05 was used to determine statistical significance. Pairwise comparisons between groups were based on one-way analysis of variance and on the Wilcoxon rank-sum test. Fisher exact test was used for comparisons of proportions. Accounting for family-wise error, post hoc statistical significance was assessed by the Benjamini-Hochberg false discovery rate controlling procedure with q = 0.20.31 To assess the strength of the observed differences we used relative risk (RR) with 95% confidence interval (CI) and the nonparametric effect size r. Data are presented as mean (standard deviation) or percentage (number of occurrences).
Patient ages ranged from 18 to 66 years (37.7 ± 8.99) with 78.5% being male (Table 1). Participants predominantly served in the Army (45.6%) and Air Force (45.2%); 9.2% Navy/Marines. Approximately 74% of the military personnel evaluated had been deployed. The average ESS score was 12.6 ± 4.78, with 68.4% of patients reporting excessive daytime sleepiness (ESS > 10). The average ISI score was 16.5 ± 5.53, with 64.1% of patients having an ISI ≥ 15, consistent with insomnia.
Study sample characteristics.
Next, patients were classified in the NDO and non-NDO groups. Whereas only 3.9% of the 493 patients reported nightmares as a reason for sleep evaluation, 31.2% had clinically significant nightmares. The groups did not differ in terms of age, sex, or deployment history (all P > .05).
As shown in Table 1 and Table 2, the NDO and non-NDO groups did not differ in terms of ESS scores, OSA occurrence, and severity of OSA (all P > .40). Compared to those in the non-NDO group, the patients in the NDO group were more likely to have elevated ISI scores consistent with insomnia (P < .001) and the comorbid disorders of depression (P < .001), anxiety (P < .001), and PTSD (P < 0.001). These findings hold true even after we controlled for the effect of sex using the Cochran-Mantel-Haenszel test.
Sleep disorders by patient group.
Sleep disorders by patient group.
For self-reported sleep, patients in the NDO group had shorter sleep duration (P < .001) and decreased SE (P < .001). Regarding PSG variables, patients in the NDO group were characterized by longer SOL and longer REM sleep latency compared to those in the non-NDO group (Table 3). The observed PSG differences were of a small effect size.
Polysomnography-measured variables by patient group.
Polysomnography-measured variables by patient group.
We compared the NDO-only group to the TRN group (Table 4) to assess the effect of TRN. Those in the TRN group were 1.49 times more likely to have deployed. The patients with TRN had more severe insomnia symptoms and were more likely to have a diagnosis of OSA (RR = 1.25), TBI (RR = 5.90), PTSD (RR = 5.44), anxiety (RR = 2.17), and depression (RR = 2.05). Interestingly, the PSG variables of patients with TRN were characterized by increased SE, decreased WASO, and higher percentage of stage N3 sleep.
Differences between nightmare disorder only and trauma-related nightmare groups.
Differences between nightmare disorder only and trauma-related nightmare groups.
Regarding the effect of branch of service, the prevalence of NDO was higher in Army patients (38.3%) compared to Air Force patients (24.6%; Fisher exact test, P = .002). Specifically, Army patients were 1.6 times more likely to have a diagnosis of NDO compared to the Air Force patients (95% CI = 1.17–2.08). The same trend was identified for memories or nightmares of traumatic experiences. That is, compared to Air Force patients (17.7%), more Army patients (30.2%) reported memories or nightmares of traumatic experiences (Fisher exact test, P = .003; RR = 1.70, CI = 1.20–2.41).
Last, we explored the association between reported sleep aid usage and the pattern of our results. Sleep aid usage was reported by 79 of the patients (16%). After excluding these patients, analysis showed that our findings still held true without substantive differences.
To our knowledge, this is the largest study to assess clinically significant nightmares in an active duty population referred for the evaluation of sleep complaints. Overall, 31.2% of military personnel with sleep disturbances met criteria for NDO, which far exceeded the rate of patients presenting with a sleep-related concern of nightmares. This study highlights the need to evaluate nightmares in military personnel with sleep disturbances, particularly those with a history of traumatic experiences.
This is not the first study that supports the notion that individuals with nightmares do not seek clinical care for this nocturnal disorder. In a previous study, Nadorff et al. reported that most of the individuals with nightmares had not discussed this with a health care provider. In the study by Nadorff et al., those with nightmares did not believe their disorder was treatable, which may have led to this finding.32 In our study, other potential reasons why nightmares were not reported as a reason for a sleep medicine evaluation are their link with PTSD and associated stigma,33 that nightmares are expected or even perceived as normal after trauma, or health care providers have not been educated about nightmares and their associated adverse clinical outcomes.32
Our study did not allow us to definitively establish the cause of the patients' nightmares. Yet, the finding that military personnel with TRN were significantly more likely to have deployed implies that traumatic experiences from deployment likely contributed to this finding. The association of TRN with deployments is similar to findings in the study by Schreuder et al. that evaluated 223 war victims 40 years after their traumatic experience and found 56.5% had posttraumatic nightmares.34 Similarly, a recent study by Thordardottir et al. found 38% of adults who survived an avalanche had TRN 16 years later.35 Given the high rates of TRN and their noted persistence over time, specifically addressing this disorder in close temporal proximity to the traumatic experience has the potential to improve not only nightmare-associated distress but overall sleep quality. Another reason to address nightmares is their significant association with suicide,36 with previous studies reporting that nightmares are an independent risk factor for suicide,37 and are associated with a fivefold increase in suicidality.38 Suicide is a pressing health concern in military personnel and veterans39; however, we are not aware of any studies regarding nightmares and suicide in this population.40
Insomnia was found overwhelmingly in those with NDO (86%; mean ISI score = 19.5) compared to those without NDO (54.2%; mean ISI score = 15.2). In military personnel with TRN, insomnia symptoms and self-reported SE were worse than in those with NDO, consistent with the concept that TRN are a more severe disorder.41 These findings are analogous to those reported by Gehrman et al., who evaluated sleep dairies of Vietnam veterans and found PTSD severity and nightmare-related distress correlated with longer self-reported sleep latency, increased awakenings, decreased total sleep time, and greater frequency of nightmares.42
On PSG, the NDO cohort had increased SOL and REM sleep latency. There are a number of small studies that have reported variable PSG characteristics in nightmare sufferers.43–49 REM-specific abnormalities to include absence of early REM periods, increased REM sleep latency, and increased REM periods, as well as NREM changes to include reduced slow wave sleep, have been reported.44,46 Other studies note more global changes to include increased sleep latency, sleep fragmentation, nocturnal awakenings, and periodic limb movements.41,43,45,46 The most consistent PSG finding is an increase in overnight awakenings.49,50 Interestingly, when comparing PSG variables in patients with NDO versus TRN, those with TRN had improved SE, decreased WASO, and increased stage N3 sleep. These findings are distinctly different than their subjective reports and paradoxically suggest that patients with TRN had improved sleep quality in the laboratory.
Specifically regarding the increased stage N3 sleep in patients with TRN, a possible explanation for this unexpected phenomenon is that patients with TRN are chronically sleep deprived in their habitual sleeping environment from their nightmares and hyperarousal, resulting in fragmented poor-quality sleep. When these patients sleep in the monitored sleep laboratory environment that they may perceive as safe, this decreases their usual state of nocturnal hyperarousal, which could translate to more efficient sleep.51 Further support for this theory is that patients with nightmares rarely have this parasomnia captured during an attended in-laboratory PSG. In terms of the evaluation of NDO and TRN, future studies should consider home PSG with video monitoring where these parasomnias are more likely to occur.
Similar to insomnia, the rate of comorbid TBI, as well as anxiety, depression, and PTSD, was higher in the NDO group than the non-NDO group. However, when comparing military personnel with NDO versus TRN, it became apparent that the association with comorbid disorders was primarily in those with TRN. Previous research has established that nightmares, but not necessarily TRN, are associated with anxiety and depression.52 In a recent study, Ulmer et al. evaluated veterans of Iraq/Afghanistan and reported findings similar to ours.53 In their cohort of veterans, 10.1% had TRN, which increased to 54.1% in veterans with comorbid mental health diagnoses. Although TRN are frequently present in PTSD, these findings suggest that TRN occur in a number of behavioral medicine disorders and not exclusively PTSD.
Not unexpectedly, the diagnostic rate of OSA was similar between the NDO and non-NDO groups at 73.4% and 74.6%, respectively. Yet, the TRN group had higher rates of OSA, particularly mild OSA, when compared to the NDO group. A potential reason for this is that the TRN group had higher rates of PTSD and TBI, and in multiple previous studies these disorders have been associated with OSA.20,54 Yet, although the exact etiology of this finding is unknown, it suggests that patients with TRN have a more complex condition. Specifically, they can manifest multiple sleep (ie, nightmares, OSA, insomnia) and comorbid disorders (ie, PTSD, depression, anxiety, TBI) and may represent a phenotype particularly vulnerable to the untoward effects of trauma.55
Our study has limitations that merit discussion. We assessed the prevalence of nightmares in a sample of military personnel referred for the evaluation of sleep disturbances. This does not necessarily represent the overall prevalence of nightmares in the military population. Although we used validated questionnaires to determine NDO, clinical interviews and nightmare diaries that could have further elucidated the etiology and clinical manifestations of their disorder were not performed. The diagnoses of depression, anxiety, PTSD, and TBI were ascertained from the EMR and may not represent definitive diagnoses. The assessment of medications that could affect sleep was limited; a more detailed evaluation, specifically including selective serotonin and norepinephrine reuptake inhibitors, would help determine potential effects on NDO and associated PSG variables. A strength of our study is that it provides both subjective and objective characteristics of NDO in a relatively large cohort of military personnel. Yet, to further the understanding of the effect of nightmares on sleep, future studies should ask whether or not patients had a nightmare during the PSG.
In conclusion, we found a high prevalence of nightmares in military personnel with sleep disturbances. Nightmares are likely underrecognized by clinicians and underreported by military personnel. Subjectively, patients with NDO had worse sleep than those without nightmares, which was further exacerbated in those with TRN. However, it is unknown if this was from their nightmares or their increased insomnia symptoms, comorbid disorders, or a combination of these illnesses. Interestingly, military personnel with TRN appeared to have better sleep quality on PSG than those with NDO only, noting nightmares are rarely reported in the laboratory environment. With the relationship between nightmares and suicidality, this is an unmet aspect of the care of veterans that requires integration into clinical practice and further research. As treatment of nightmares can improve sleep and potentially mitigate suicide risk, it is imperative to evaluate for the presence of nightmares in all trauma survivors with sleep disturbances.
Work for this study was performed at Wilford Hall Ambulatory Surgical Center, JBSA-Lackland, TX. All authors have seen and approved the manuscript. The authors report no conflicts of interest. The opinions and assertions in this manuscript are those of the authors and do not represent those of the Department of the Air Force, Department of the Army, Department of Defense, or the United States government.