Nightmares and insomnia within the acute aftermath of trauma prospectively predict suicidal ideation

Abstract

Study Objectives:

Patients with acute trauma are vulnerable to suicidality following hospitalization. Research suggests nightmares and insomnia may interact to potentiate suicidality, possibly due to nightmares worsening co-occurring insomnia. Nightmares and insomnia are common stress reactions to acute trauma and might compound suicide risk within acutely traumatized patients. We tested the prospective relationship between nightmares and insomnia immediately after trauma on future suicidal ideation (SI).

Methods:

Patients hospitalized in Detroit, Michigan following traumatic injury (mean_age = 39.53 ± standard deviation 14.31 years, 67.0% male, 67.0% Black) completed surveys at 3 posttrauma timepoints: 1 week (time 1 [T1]; n = 88), 1 month (time 2 [T2]; n = 61), and 2 months (time 3 [T3]; n = 59).

Results:

Patients with clinically significant nightmares and comorbid insomnia symptoms at T2 reported the highest rates of SI at T3 (42.9%), whereas patients with insomnia alone (8.0%) or neither sleep disturbance (6.7%) had the lowest SI rates (Ps < .05). We observed an interaction effect wherein insomnia symptoms at T2 predicted increased SI at T3, but only among patients with comorbid nightmares at T2. This interaction remained after accounting for acute stress symptoms at T2. Post hoc analyses showed nighttime awakenings and total wake time at T2 predicted increased SI at T3 with nightmares also moderating this prospective effect.

Conclusions:

These novel results suggest clinically significant nightmares strengthen the association between insomnia and suicidality after trauma. As nearly half of acute trauma patients with nightmares and insomnia experience SI 2 months after trauma, early interventions that target both may curb SI rates.

Citation:

Reffi AN, Kalmbach DA, Cheng P, et al. Nightmares and insomnia within the acute aftermath of trauma prospectively predict suicidal ideation. J Clin Sleep Med. 2025;21(9):1519–1527.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Nightmares and insomnia are common within the acute aftermath of trauma and may explain the inflated risk for suicidality among patients with acute trauma. We evaluated the prospective association between nightmares and insomnia 1 month after trauma and suicidal ideation 2 months after trauma.

Study Impact: Clinically significant nightmares after trauma may worsen the effect of insomnia symptoms on future suicidal ideation. Nightmares may prolong nocturnal wakefulness and contribute to feelings of hopelessness and entrapment in poor sleepers during early trauma recovery. These novel findings suggest that comorbid nightmares and insomnia may increase suicidality risk after trauma, highlighting the potential for early intervention.

INTRODUCTION

Suicide is the tenth leading cause of death within the United States,¹ devastating families and costing society upwards of $500 billion annually.² Certain populations are especially vulnerable to suicide, including patients with acute trauma.^3,4 Compared to the general population, patients hospitalized following trauma face significantly increased risk for suicide attempts or death by suicide after discharge (odds ratio = 4.31), even when accounting for preexisting depression (odds ratio = 3.65).⁴ This highlights a critical need to identify modifiable risk factors for suicide among patients with acute trauma that could be targeted for early intervention.

Emerging research has identified sleep disturbance as a robust risk factor for suicidality across myriad populations.^5–11 More precisely, nightmares and insomnia are both associated with suicidality above and beyond preexisting psychopathology, with specific effects found for suicidal ideation (SI),^12–15 suicide attempts,^16–18 and death by suicide.^19,20 Notably, up to 70% of individuals with nightmares report co-occurring insomnia,²¹ and several studies highlight the deleterious nature of this comorbidity.^22,23 For instance, in a prospective study of patients with schizophrenia, those with comorbid nightmares and insomnia at baseline faced higher risk of suicide attempts 8 years later (odds ratio = 11.10) compared to patients without any sleep disturbance.²⁴ These data suggest comorbid nightmares and insomnia are robust markers for suicidality, possibly due in part to nightmares worsening co-occurring insomnia.^25–28 Given nightmares and insomnia are common, modifiable consequences of acute trauma,^29,30 this comorbid presentation could signal an early risk for suicide within acutely traumatized populations.

The current study therefore sought to characterize the prospective relationship between nightmares and insomnia and subsequent SI within the immediate aftermath of trauma among patients hospitalized following traumatic injury. First, we evaluated whether clinically significant nightmares and insomnia symptoms 1 month posttrauma predicted future SI rates 2 months posttrauma. Specifically, we compared SI rates (2 months posttrauma) across patients reporting clinically significant nightmares and comorbid insomnia symptoms, insomnia alone, nightmares alone, or neither. Second, we tested whether clinically significant nightmares 1 month posttrauma moderated the prospective effect of comorbid insomnia symptoms on future SI 2 months posttrauma while accounting for important covariates, including acute stress disorder (ASD) symptoms within the first month of traumatic injury. Indeed, patients with acute trauma experience a constellation of stress reactions that often include nightmares and insomnia.²⁹ As such, we accounted for ASD to isolate the unique associations between co-occurring nightmares and insomnia and future SI.

METHODS

Participants and procedure

Data were collected as part of a study on Understanding Nightmares after Traumatic Events in Detroit.³¹ We recruited participants following traumatic injury from the surgical general practice and intensive care units within the Level I Trauma Center at Henry Ford Hospital in midtown Detroit, Michigan, between April 2022 and January 2023. Inclusion criteria were as follows: ≥ 18 years old, fluent in English, and no current traumatic brain injury. Regarding exclusion, patients were not approached if they were intubated or sedated, actively recovering from anesthesia, or experiencing delirium. This study was approved by Henry Ford Health’s Institutional Review Board.

Understanding Nightmares after Traumatic Events in Detroit recruited trauma-exposed patients during hospitalization (time 1 [T1]) and prospectively followed them to 1-month (time 2 [T2]) and 2-month (time 3 [T3]) posttrauma follow-up. Study personnel recruited and obtained informed consent from eligible patients within 1 week of hospital admission. Eighty-eight interested patients completed the T1 survey using Qualtrics (Qualtrics, Provo, UT) while hospitalized. After discharge, participants completed follow-up Qualtrics surveys via email at T2 (n = 61) and T3 (n = 59). Study personnel made reminder phone calls to participants who did not complete their follow-up surveys within ∼48 hours of receipt. After T3, participants were compensated for each survey they completed and received a list of regional mental health resources. Figure 1 displays the study flow and measures.

Figure 1. Study flow and timing of assessments.

ISI = insomnia severity index, M = mean, NDI = nightmare disorder index, NWAK = number of awakenings, PCL-5 = posttraumatic stress disorder checklist for DSM-5, PHQ-9 = 9-item patient health questionnaire, SASS = Whole Week Self-Assessment of Sleep Survey, SI = suicidal ideation, SOL = sleep onset latency, T1 = time 1 (M_{days after trauma} = 5.23 ± 4.42), T2 = time 2 (M_{days after trauma} = 24.28 ± 7.53), T3 = time 3 (M_{days after trauma} = 47.81 ± 17.94), WASO = wake after sleep onset.

Measures

Acute insomnia symptoms

Participants completed the 7-item insomnia severity index at T1 and T2 to indicate their pre- and posttrauma acute insomnia disorder symptoms over the past 2 weeks.³² Items 1–3 of the insomnia severity index capture nighttime symptoms of insomnia (difficulty falling asleep, difficulty staying asleep, and problems waking up too early) and items 4–7 capture the disorder’s daytime consequences (eg, dissatisfaction with sleep, worry about sleep problem, etc.). We computed both a sum score, with greater scores indicating more severe acute insomnia symptoms, and a dichotomized variable using a cut-off score of ≥ 11 to group participants by clinically significant insomnia symptoms.³²

Clinically significant nightmares

Participants completed the 5-item nightmare disorder index (NDI) at T1 and T2 to indicate their pre- and posttrauma nightmare disorder symptoms.³³ The NDI is a validated screener aligned with the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5) criteria for nightmare disorder that queries for the presence of nightmares (disturbing, extended, well-remembered dreams) more than 0 times per week (item 1), awakenings from nightmares, after which the individual rapidly becomes alert (item 2), and nightmare-related distress (item 3) and functional impairment (item 4). The NDI also assesses nightmare disorder symptom chronicity (item 5). We computed a dichotomous variable to indicate clinically significant nightmare disorder symptoms if participants reported experiencing any nightmares over the past month (at T1) or past 2 weeks (at T2), waking and quickly becoming alert at least “sometimes,” and experiencing distress and functional impairment due to nightmares at least “somewhat.” ³³

Nocturnal wakefulness

Participants completed the 9-item Whole Week Self-Assessment of Sleep Survey at T1 and T2 to report on various sleep parameters over the past week.³⁴ For descriptive purposes, we report data from the Whole Week Self-Assessment of Sleep Survey on participants’ nocturnal wakefulness over the past week, specifically their number of nighttime awakenings (NWAK), sleep onset latency, and wake after sleep onset. Total wake time (TWT) was operationalized as the sum of their sleep onset latency and wake after sleep onset. These variables were also used for post hoc analyses (as described below).

SI

We measured SI at T1, T2, and T3 using item 9 from the 9-item patient health questionnaire (“Thoughts that you would be better off dead or of hurting yourself in some way”).³⁵ Participants indicated how often they experienced SI from 0 (not at all) to 3 (nearly every day). We computed both a sum score, with greater scores indicating more SI, and a dichotomized variable ≥ 1 to indicate any SI over the past 2 weeks.

Depressive symptoms

Participants completed the 9-item patient health questionnaire at T1, T2, and T3 to indicate their depressive symptoms over the past 2 weeks.³⁵ We computed a dichotomized variable using a cut-off score of ≥ 10 to group participants by provisional major depressive disorder status at T1 and T2 to be used as a covariate.

ASD symptoms

Participants completed the Posttraumatic Stress Disorder Checklist for DSM-5 (PCL-5) at T2 to indicate their ASD symptoms within the first month after their traumatic injury that precipitated their hospitalization. The PCL-5 is a 20-item questionnaire that assesses posttraumatic stress disorder symptoms in relation to a traumatic event using a 5-point scale ranging from 0 (not at all) to 4 (extremely).³⁶ In this context, however, we used the PCL-5 as an assessment of ASD symptoms because participants completed this assessment prior to the 1-month cutoff required for posttraumatic stress disorder. We computed a dichotomous variable using a cutoff score of ≥ 33 to indicate provisional ASD status at T2 to be used as a covariate.³⁷

Data analysis

We removed 5 cases from our inferential analyses to minimize their influence on our SI outcome (2 were attempted suicides, and 3 were patients hospitalized for a trauma during which someone died) (resulting N at T1 = 83). To test our first hypothesis, we carried out chi-square analyses to assess whether rates of SI 2 months after trauma varied by type of sleep disturbance 1 month after trauma (clinically significant symptoms of: insomnia only/nightmares only/both/neither). To test our second hypothesis, we performed a moderation analysis testing whether clinically significant nightmares 1 month after trauma moderated the prospective effect of insomnia symptoms on future SI severity 2 months after trauma (T2 nightmares × T2 insomnia symptoms) (see Figure 2). We adjusted for days elapsed since trauma exposure, pretrauma sleep (T1 insomnia symptoms, T1 nightmares, T1 sleep medications), pre- and posttrauma SI (T1 and T2) and depression (T1 and T2), and co-occurring ASD symptoms (T2). We inspected collinearity statistics and did not find evidence of multicollinearity, as all variables had bivariate correlations < .80, tolerance values > .1, and variance inflation factor values < 10. All predictor variables were z-scored, thus β denotes standardized beta coefficients. We used the PROCESS Macro for SPSS (IBM, Armonk, New York, USA) (model 1) using 5,000 bootstrap samples to produce 95% confidence intervals (CIs).

Figure 2. Interaction between clinically significant nightmares and comorbid insomnia symptoms 1 month after trauma on future suicidal ideation 2 months after trauma.

Patients with acute trauma were recruited during hospitalization (T1) and prospectively followed to 1-month (T2) and 2-month (T3) posttrauma. Model adjusts for days elapsed since trauma, T1 insomnia disorder symptoms, T1 nightmare disorder symptoms, T1 and T2 suicidal ideation, T1 and T2 depression, T2 acute stress disorder, and T1 sleep medications. Insomnia symptoms = insomnia severity index sum score, Nightmares = nightmare disorder index categorical score, suicidal ideation = patient health questionnaire item 9 sum score, T1 = time 1, T2 = time 2, T3 = time 3.

RESULTS

Sample characteristics

Timing of posttrauma assessments

Patients completed the T1 survey within a week of trauma exposure (n = 88; M_{days after trauma} = 5.23 ± 4.42), T2 within 1 month of trauma exposure (n = 61; M_{days after trauma} = 24.28 ± 7.53), and T3 within 2 months of trauma exposure (n = 59; M_{days after trauma} = 47.81 ± 17.94).

Demographics

Most participants reported being male (67.0%, M_age = 39.53) and Black or African American (67.0%) (see Table 1 for complete demographic data). t tests and chi-square analyses revealed that neither participant demographics nor clinical symptoms at T1 (including nightmare severity) were significantly correlated with study attrition at either T2 (P > .05) or T3 follow-up (P > .05).

Table 1.

Demographics (n = 88).

Variables	n (%)
Age (M ± SD)	39.5 (14.3)
Sex (male)	59 (67.0%)
Race*
Black or African American	59 (67.0%)
White	29 (33.0%)
Other	4 (4.4%)
Hispanic/Latino(a)	3 (3.4%)
Highest level of education
Some college credit, no degree	27 (30.7%)
High school diploma/GED	32 (36.4%)
Some high school, no diploma	13 (14.8%)
Other	16 (18.1%)
Working full-time	36 (40.9%)
Annual household income
≤ 20,000	42 (47.7%)
≤ 50,000	22 (25.0%)
≤ 75,000	16 (18.2%)
≤ 100,000	1 (1.1%)
> 100,000	7 (8.0%)
Relationship status
Single	50 (56.8%)
Never married, in relationship	18 (20.5%)
Married	8 (9.1%)
Separated or divorced	12 (13.6%)
Ever diagnosed with a sleep disorder	16 (18.2%)
Currently taking any medications	43 (48.9%)
Currently taking antidepressants	6 (6.8%)
Currently taking sleep medications*	15 (17%)
Over the counter	7 (8.0%)
Prescription	11 (12.5%)

^*Participants could select multiple responses. Sleep disorder history and medication usage were based on self-report. See Table S3 in the supplemental material for all reported medications. GED = General Educational Development, M = mean, SD = standard deviation.

Trauma types that precipitated hospitalization

Patients were most often admitted after a motor vehicle collision (MVC) (42.0%), followed by an assault with a weapon (30.7%), fall (9.1%), physical assault (6.8%), self-injury (2.3%), and sexual assault (1.1%) (see Table S1 in the supplemental material).

Sleep disturbance rates 1 month posttrauma

Approximately 1 month posttrauma exposure, 54.1% of patients screened positive for clinically significant insomnia symptoms only (n = 33/61). Ten patients (16.4%) endorsed clinically significant symptoms of comorbid nightmares and insomnia, whereas no patients endorsed clinically significant symptoms of nightmare disorder without insomnia. Eighteen patients (29.5%) denied clinically significant symptoms of both nightmares and insomnia 1 month posttrauma (see Table S2 in the supplemental material). These rates of clinically significant sleep disturbance 1 month posttrauma were not significantly related to self-reported diagnoses of a sleep disorder (P = .226), including obstructive sleep apnea specifically (P = .509), or current antidepressant use (P = .313) reported during hospitalization.

Regarding how long participants had been bothered by nightmares within the first month of traumatic injury, most participants (61.1%) reported new-onset nightmares, such that 27.8% indicated having been bothered by nightmares for < 1 week, and 33.3% for < 1 month. The remaining participants reported a nightmare history that predated their most recent trauma exposure, with 16.7% indicating being bothered by nightmares for 1–6 months, and 22.2% indicating > 12 months.

Risk for SI 2 months after trauma exposure

Sleep disturbance 1 month posttrauma and future SI 2 months posttrauma

Prospective chi-square analysis indicated that SI rates 2 months posttrauma significantly differed across the 3 sleep disturbance groups (χ² = 6.70, P = .035) (see Table 2).

Table 2.

Rates of suicidal ideation 2 months posttrauma by sleep disturbance 1 month posttrauma.

	Suicidal Ideation 2 Months Posttrauma
Sleep Disturbance 1 Month Posttrauma	No SI	SI	% SI	RR
No nightmares/insomnia	14	1	6.7%	—
Insomnia only	23	2	8.0%	—
Comorbid nightmares and insomnia	4	3	42.9%	6.40

Rates of SI 2 months posttrauma were significantly higher among patients with clinically significant symptoms of comorbid nightmares and insomnia 1 month after trauma compared to patients with insomnia symptoms only or neither (Ps < .05). No patients reported clinically significant symptoms of nightmare disorder only. Insomnia = clinically significant insomnia symptoms (insomnia severity index score ≥ 11), Nightmares = clinically significant nightmare disorder symptoms (nightmare disorder index categorical score), RR = relative risk (compared to the no nightmares/insomnia group), SI = suicidal ideation, PHQ-9 item 9 dichotomized (0 = no SI/1 = SI).

As displayed in Figure 3, patients with clinically significant nightmares and comorbid insomnia symptoms at T2 went on to report the highest rates of SI at T3 (42.9%), whereas SI rates at T3 were lowest among the group without any clinically significant sleep disturbance (6.7%, relative risk = 6.40, P < .05). Patients with clinically significant nightmares and comorbid insomnia symptoms also reported significantly higher rates of SI at T3 than patients with clinically significant insomnia symptoms only (8.0%, P < .05). Rates of SI at T3 did not significantly differ between patients with clinically significant insomnia symptoms only and those with neither sleep disturbance (P > .05).

Figure 3. Rates of SI 2 months posttrauma by type of sleep disturbance 1 month posttrauma.

SI rates significantly differed across groups (χ² = 6.70, P < .05). No patients reported clinically significant symptoms of nightmare disorder only. *P < .05, n.s. = P > .05. Comorbid nightmares & Insomnia = patients screened positive for clinically significant symptoms of both nightmare and insomnia disorder, Insomnia Only = patients screened positive for clinically significant insomnia symptoms only, No Nightmares/Insomnia = patients denied clinically significant symptoms of both nightmare disorder and insomnia disorder, n.s. = not significant, SI = suicidal ideation.

Acute stress symptoms 1 month posttrauma and future SI 2 months posttrauma

Given that ASD often includes sleep disturbances and may itself be predictive of long-term health outcomes, we conducted an exploratory analysis to examine the extent to which ASD symptoms at T2 predict SI at T3. One quarter of patients with ASD at T2 went on to report SI at T3 (25.0%), whereas 0.0% of patients without ASD at T2 reported SI at T3 (P < .05).

Do clinically significant nightmares moderate the suicidogenic effects of insomnia?

Next, we tested whether clinically significant nightmares moderate the effect of comorbid insomnia symptoms on future SI. Our moderation analysis revealed that nightmares at T2 significantly strengthened the effect of insomnia symptoms at T2 on SI at T3, β = 0.53, standard error (SE) = 0.10, 95% CI [0.33, 0.73], P < .001. For patients with clinically significant nightmares at T2, comorbid insomnia symptoms prospectively predicted more SI at T3, β = 1.40, SE = 0.26, 95% CI [0.86, 1.93], P < .001. However, for patients without clinically significant nightmares at T2, the relationship between T2 insomnia symptoms and SI at T3 was nonsignificant, β = −0.02, SE = 0.08, 95% CI [−0.18, 0.15], P = .833. The overall model accounted for 72% of the variance in SI at T3 (R² = .72, P < .001), indicating a large effect size.

Post hoc analyses: interaction effect between clinically significant nightmares and comorbid nocturnal wakefulness

Based on our results, we conducted post hoc analyses to explore whether clinically significant nightmares moderated the relationship between concurrent nocturnal wakefulness and future SI (see Figure S1 in the supplemental material). We operationalized nocturnal wakefulness separately as NWAK and TWT.³⁸ Using these variables, we performed the following 2 post hoc models:

1. T2 NWAK as a predictor of SI at T3, with clinically significant nightmares at T2 as the moderator variable (T2 nightmares × T2 NWAK); and

2. T2 TWT as a predictor of SI at T3, with clinically significant nightmares at T2 as the moderator variable (T2 nightmares × T2 TWT).

Consistent with our main analyses, the presence of clinically significant nightmares at T2 significantly strengthened the effects of NWAK and TWT at T2 on SI at T3: nightmares × NWAK, β = 0.73, SE = 0.11, 95% CI [0.50, 0.97], P < .001; nightmares × TWT: β = 0.63, SE = 0.09, 95% CI [0.45, 0.82], P < .001.

Specifically, for patients with clinically significant nightmares at T2, more nighttime awakenings at T2 prospectively predicted more SI at T3, β = 2.05, SE = 0.30, 95% CI [1.43, 2.67], P < .001. Similarly, for patients with clinically significant nightmares at T2, longer nocturnal wakefulness at T2 prospectively predicted more SI at T3, β = 1.69, SE = 0.23, 95% CI [1.21, 2.16], P < .001. Conversely, these relationships were nonsignificant for patients without clinically significant nightmares at T2 (Ps > .05).

DISCUSSION

Patients hospitalized following acute trauma face a disproportionate risk for suicide attempts and death by suicide after discharge,^3,4 suggesting an urgent need to identify modifiable risk factors amenable to early intervention. Comorbid nightmares and insomnia commonly manifest within the acute aftermath of trauma^29,30,39 and are markers of suicidal thoughts and behaviors.^22–24,40 The current study recruited patients from an urban Level I Trauma Center and found novel evidence that insomnia symptoms within the first month after trauma prospectively predicted increased SI severity 2 months after trauma. Critically, however, insomnia predicted future SI only for patients whose insomnia symptoms co-occurred with clinically significant nightmares. In other words, poorly sleeping patients were at highest risk for SI during trauma recovery if they also experienced frequent, distressing nightmares that were severe enough to disrupt sleep continuity and impair functioning. These findings are the first to highlight the synergistic effects of comorbid nightmares and insomnia shortly after trauma on future risk for suicidality.

Comorbid nightmares and insomnia symptoms after trauma forecast SI

In our sample, patients with acute trauma with clinically significant nightmares and comorbid insomnia symptoms 1 month posttrauma were over 6 times more likely to report SI 2 months posttrauma (42.9%) compared to patients with clinically significant insomnia symptoms only (8.0%) or without either of these sleep disturbances (6.7%) (no patients endorsed clinically significant nightmares only). These findings build on previous reports that comorbid nightmare and insomnia symptoms confer susceptibility to suicidality^22–24 and extend this effect to an acutely traumatized population, suggesting this comorbid presentation during the immediate aftermath of trauma indicates early risk for suicide.⁴¹

Clinically significant nightmares after trauma exacerbate the impact of insomnia symptoms on SI

Our study also provides first evidence of a synergistic effect of comorbid nightmares and insomnia on SI, such that the relationship between insomnia symptoms after trauma and future SI depended on whether clinically significant nightmares were also present within the first month after trauma. Notably, this effect was independent of pretrauma nightmares and insomnia symptoms, as well as pretrauma and concurrent depressive symptoms and SI, suggesting this comorbidity after trauma confers a robust risk for suicidality that is distinct from preexisting and co-occurring vulnerabilities. Importantly, the interaction between comorbid nightmares and insomnia symptoms within the first month posttrauma was also robust to concurrent symptoms of ASD. This finding is meaningful because nightmares and insomnia are a common expression of ASD, thus our results suggest that the suicidogenic effect of posttrauma nightmares and comorbid insomnia symptoms was not a mere proxy for ASD symptoms but rather may signal a unique risk for SI. The suicidogenic effect of this comorbid presentation may hinge on clinically significant nightmares because they are a source of psychological pain during the night when one is awake, socially isolated, and behaviorally dysregulated.^8,40

Comorbid nightmares and insomnia after trauma may exploit the mind after midnight

Emerging work suggests the mechanism(s) undergirding the sleep-suicidality link may not necessarily be any particular sleep disorder per se, but rather the nocturnal wakefulness that follows. This hypothesis—termed the Mind after Midnight—proffers that nocturnal wakefulness itself is a risk factor for suicidality due to diminished cognition and executive function during the biological night.⁹ Thus, being awake at night (“when reason sleeps”)¹¹ may increase suicide risk because the usual processes that inhibit suicidal behaviors are offline (eg, executive function, affect regulation, reward processing).^8,10,11 In this view, insomnia is the kindling that creates a vulnerable state where the mind is primed for nocturnal rumination and distress, while nightmares are the match that reinforce a sense of defeat onto an already fatigued, disinhibited mind prone to suicidal thinking. Indeed, among our patients with clinically significant nightmares 1 month after trauma, future SI was downstream of patients’ reports of both (1) more frequent nighttime awakenings 1 month after trauma and (2) greater total wake time 1 month after trauma. Review of acute trauma patients’ dream reports align with this sense of hopelessness, defeat, and entrapment due to nightmares, especially those that are trauma-related (see Dream Reports in the supplemental material).

Clinical implications

Trauma survivors who develop clinically significant nightmares and comorbid insomnia symptoms within the acute recovery phase may benefit from early, targeted sleep treatment. This comports with accumulating evidence that sleep interventions reduce suicide risk, as patients who utilize sleep medicine appointments are less likely to attempt suicide,⁴² and both cognitive behavioral therapy for insomnia and nightmares reduce SI.^43–45 Given that nearly half of our patients with comorbid nightmares and insomnia experienced SI 2 months after trauma (42.9%), early interventions that target both types of sleep disturbance may help reduce SI rates (eg, cognitive behavioral therapy for insomnia and nightmares⁴⁶). Importantly, the ideal approaches to treating acute sleep disturbance early after trauma exposure are still under investigation and warrant caution.⁴¹ For instance, some evidence points to the use of benzodiazepines immediately after trauma as being potentially harmful.⁴⁷

Limitations, strengths, and future directions

The present findings should be considered within our study’s limitations. Chief among them is our single-item measure of SI as our primary outcome. Although SI predicts future suicide attempts⁴⁸ and death by suicide,⁴⁹ future studies can build on this work by including more comprehensive assessments of suicide risk (eg, suicide planning, attempts). Other measurement-related limitations to our study include the exclusive use of self-report measures, including screeners for both nightmare disorder and insomnia disorder.

Although the ISI, NDI, and PCL-5 align with the DSM-5’s core criteria for insomnia disorder, nightmare disorder, and ASD, respectively, they are nonetheless screeners that cannot determine formal diagnoses. For instance, the NDI and insomnia severity index only assessed acute nightmare and insomnia symptoms in our patients (over the past 2 weeks), yet the DSM-5 requires symptom duration for at least 1 month (for nightmare disorder) and 3 months (for insomnia disorder). Future studies that seek to build on our work are encouraged to perform clinical interviews that can adequately assess symptom duration, rule-out differential diagnoses, and more thoroughly account for the use of concomitant medications (eg, beta-blockers) or recreational drugs (eg, cannabis) that may affect nightmare production. Additionally, our study shows that patients’ perceptions of longer nocturnal wakefulness predict future SI (based on retrospective self-report). However, patient perceptions and objective assessments of sleep are often inconsistent. Thus, it would be interesting for future investigators to test whether objective sleep disturbances in insomnia predict future SI.

Our models also utilized listwise deletion, which reduces sample size and may introduce potential bias that together could affect the representativeness of the results. Although our attrition rates at T2 or T3 were not significantly associated with participant characteristics or nightmare severity at T1, it is still possible that there were variables that contributed to participant dropout that we did not capture (eg, housing instability). Studies that utilize more intensive longitudinal designs (eg, ecological momentary assessments) would be better suited to handle missing data and capture the temporal variability of suicidality, especially with small samples such as ours (T1 n = 88, T2 n = 61, and T3 n = 59). Attrition in acute trauma samples is common, and retention of participants becomes especially challenging within urban populations characterized by predominantly low socioeconomic status. Future studies may improve retention in similar populations by requiring prospective participants to demonstrate engagement before enrollment, such as by completing a specified percentage of sleep diaries. Additionally, our observational design precludes causal inferences about the relationship between posttraumatic sleep and SI. Future studies can test this causal relationship by randomly assigning patients with trauma to a sleep-focused intervention in the immediate aftermath of trauma exposure.

Nonetheless, our prospective study has several notable strengths, including the relatively proximal follow-ups. That is, by capturing SI within approximately 1 month of acute sleep disturbance, our findings suggest sleep disturbance immediately following trauma is an early warning sign of short-term suicide risk. This is clinically relevant to the identification of at-risk patients toward whom timely suicide prevention efforts can be targeted.⁶ By contrast, most longitudinal studies on sleep and suicide utilize longer-term follow-ups (median = 10.5 months⁶) that are more conducive to examining predictors of delayed risk for suicide.⁶ Additionally, many previous studies in this literature assessed sleep at 1 timepoint only,⁶ whereas our repeated assessments enabled us to covary for pretrauma symptoms of clinically significant nightmares and insomnia, nighttime awakenings, and nocturnal wakefulness. In doing so, our findings highlight the impact of acute sleep disturbance within the immediate aftermath of trauma on proximal SI severity, above and beyond basal sleep disturbance. Finally, this study is the first to demonstrate prospective links between clinically significant nightmares and SI using a validated screener aligned with DSM-5 criteria for nightmare disorder, the NDI, whereas previous studies used a single item about nightmares adapted from other questionnaires.³³ Given the NDI was predictive of SI above and beyond ASD in our study, the NDI may help acute care providers discern clinically significant nightmares that warrant intervention from those that may reflect a transient stress response. More research is needed to test this proposition, however.

CONCLUSIONS

This study is the first to highlight the synergistic effects of nightmares and insomnia on future suicide risk in the acute aftermath of trauma. Clinically significant nightmares 1 month after trauma strengthened the prospective relationship between comorbid insomnia symptoms and future SI 2 months after trauma. Patients who developed clinically significant nightmares 1 month after trauma also reported more concurrent nighttime awakenings and nocturnal wakefulness that may, in part, help explain their inflated SI risk. Treating nightmares and insomnia within the immediate aftermath of trauma using sleep-focused cognitive behavioral therapies might mitigate risk for suicide.

DISCLOSURE STATEMENT

All authors have seen and approved this manuscript. This research was conducted at Henry Ford Health System, Detroit, MI. Funding for A.N.R. was provided from the National Institute of Mental Health (K23MH134032). The authors report no conflicts of interest. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

ABBREVIATIONS

ASD

acute stress disorder

DSM

Diagnostic and Statistical Manual of Mental Disorders

M

mean

NDI

nightmare disorder index

NWAK

number of nighttime awakenings

SE

standard error

SI

suicidal ideation

T1

time 1

T3

time 2

T3

time 3

TWT

total wake time