Insomnia Symptoms Predict the Development of Posttraumatic Stress Symptoms Following an Experimental Trauma

Summary

Insomnia symptoms prior to traumatic event exposure predict the development of posttraumatic stress symptoms (PTSD symptoms). However, potential mechanisms underlying the association between insomnia and risk for PTSD symptoms have not been prospectively tested. The current study used the trauma film paradigm to test whether insomnia symptoms prior to analogue trauma exposure predict subsequent analogue PTSD symptoms, and potential mediators of this relationship, among an at-risk sample of 108 participants. Results indicated that, after covarying for negative affectivity, insomnia symptoms in the two weeks prior to analogue trauma exposure significantly predicted increased PTSD symptoms three days and one week post-exposure. Moreover, distress immediately after exposure and posttraumatic avoidance mediated the association between insomnia symptoms and PTSD symptoms one week after exposure. Effect sizes were small. The current study uses an analogue trauma and PTSD symptoms to model clinical symptoms, includes an additional intervention prior to analogue trauma, and lacks a control film. Findings suggest increased reactivity to trauma exposure and subsequent reminders, and attempts to suppress trauma memories may be mechanisms in the association between insomnia symptoms and risk for PTSD symptoms.

Insomnia symptoms are common among those with posttraumatic stress disorder (PTSD), a disorder in which re-experiencing of traumatic memories, avoidance of trauma-related cues, negative alterations of cognition and mood, and hyperarousal symptoms are experienced for at least a month following trauma exposure (American Psychiatric Association, 2013). Insomnia symptoms among those with PTSD have been linked to negative outcomes (Germain, Buysse, & Nofzinger, 2008; Milanak et al., 2019), and insomnia may be an etiological factor in PTSD (for reviews, see Germain et al., 2008). Insomnia pre-trauma prospectively predicts PTSD in injured patients and Iraq combat veterans (Bryant, Creamer, O’Donnell, Silove, & McFarlane, 2010; Gehrman et al., 2013; Koffel, Polusny, Arbisi, & Erbes, 2013). Further, posttrauma insomnia predicts PTSD among combat veterans and motor vehicle accident survivors (Koren, Arnon, Lavie, & Klein, 2002; Wright et al., 2011). Healthy individuals allowed to sleep after viewing an analogue trauma film (i.e., a film with traumatic content chosen to provoke intrusive memories; Holmes & Bourne, 2008), compared to those who did not, experienced fewer intrusions (Kleim, Wysokowsky, Schmid, Seifritz, & Rasch, 2016). Taken together, insomnia may be a risk factor for PTSD.

Despite the growing body of literature indicating that insomnia symptoms predict the development of PTSD, the mechanism of this association is unclear. Sleep problems may lead to increased emotional reactivity to stressors (Minkel et al., 2012; Short, Allan, & Schmidt, in press) and difficulties down-regulating negative emotional responding (Mauss, Troy, & LeBourgeois, 2013). Thus, those with insomnia may experience increased subjective and physiological negative emotional reactivity during trauma (i.e., peritraumatic distress). Heightened peritraumatic distress (i.e., high levels of distress during exposure), is in turn a well-established risk factor for PTSD (Birmes et al., 2005).

Furthermore, insomnia symptoms may interfere with natural recovery posttrauma. For example, intrusive memories are common in the aftermath of trauma exposure. Yet, individuals with insomnia may be more distressed by these intrusions and attempt to suppress them (Harvey, 2003; Harvey, Tang, & Browning, 2005). Unfortunately, attempts to suppress intrusions or avoid trauma cues lead to paradoxical rebounds in PTSD symptoms (Clohessy & Ehlers, 1999; Foa, Hembree, & Rothbaum, 2007; van den Hout & Engelhard, 2004). In sum, insomnia symptoms may heighten distress, and motivate suppression or avoidance of trauma reminders, increasing risk for PTSD.

Prior research demonstrates that insomnia increases risk for PTSD, and suggests peritraumatic distress and posttrauma suppression and avoidance may be mediators. However, no prior work has prospectively examined mediators of the association between insomnia and PTSD symptoms. As such, the current study utilized an analogue trauma design (Holmes & Bourne, 2008) to measure peritrauma (during exposure) and posttrauma factors that mediate associations between insomnia and PTSD. We hypothesized that: 1) self-reported insomnia symptoms will significantly predict increased analogue PTSD symptoms after analogue trauma (i.e., accident film), 2) insomnia will predict increased distress immediately after the analogue trauma, avoidance of trauma reminders, and attempts to avoid/suppress intrusive memories, 3) the effect of insomnia symptoms on Week 1 PTSD symptoms will be mediated by increased peritrauma distress and posttrauma avoidance and suppression in a temporal mediation model, and 4) an alternate mediator model (with depression as mediator) will not be significant. We hypothesized these effects would persist above and beyond trait negative affectivity, a shared risk factor for both insomnia and PTSD symptoms (Vujanovic et al., 2013).

Method

Participants

The sample (N=108) were undergraduates recruited through the university’s research pool to participate in a clinical trial examining effects of an intervention to analogue PTSD symptoms (Boffa & Schmidt, 2019). Power was determined for the larger clinical trial. Interested undergraduates participated in a brief screening survey for the psychology department to determine their eligibility for various studies, including a measure of a risk factor for PTSD (anxiety sensitivity), for the current study. Participants were invited to participate if they demonstrated elevations on a risk factor for PTSD, anxiety sensitivity (Olatunji & Fan, 2015), or a fear of anxious arousal (i.e., anxiety sensitivity cognitive concerns > 9 on the Anxiety Sensitivity Index; Taylor et al., 2007). The majority were women (77.8%) and identified as White (81.5%), with an additional 9.3% African American, 6.5% Asian/Pacific Islander, and 2.8% ‘Other’ (e.g., biracial). On average participants were 19.03 years old (SD=2.06).

Measures

Insomnia Severity Index (ISI).

The ISI is a seven-item self-report questionnaire that assesses severity of insomnia during the prior two-week period (Morin, Belleville, Bélanger, & Ivers, 2011). The ISI has excellent psychometric properties (Morin et al., 2011). The ISI was administered at baseline, and demonstrated good internal consistency (α=.82).

Patient Health Questionnaire – 8 item version (PHQ-8).

The PHQ-8 (Kroenke & Spitzer, 2002) is a self-report measure of depression symptoms (American Psychiatric Association, 2013). The PHQ-8 omits reference to suicidal ideation, and was administered at baseline to be used as an alternate mediator. The third item inquiring about sleep disturbance was omitted to reduce redundancy with the ISI. The PHQ-8 has excellent psychometric properties (Kroenke & Spitzer, 2002). The seven items used in the present investigation displayed good internal consistency (α=.80).

Positive and Negative Affect Schedule – Negative Affect subscale (PANAS-NA).

The PANAS (Watson & Clark, 1994) is a 10-item self-report measure of trait negative affect, and was administered at the baseline time point. The PANAS-NA has good psychometric properties (Watson & Clark, 1994), and had good internal consistency in the present study (α=.83).

Life Events Checklist (LEC).

The LEC is a self-report measure assessing 16 potentially traumatic events and was administered at Baseline to characterize prior trauma exposure for the sample. The LEC has demonstrated convergent validity but may not assess whether or not the event reaches the criteria for a DSM-5 traumatic event (Weathers et al., 2013).

PTSD Checklist – 5 (PCL-5).

The PCL-5 is a 20-item self-report measure of PTSD symptoms and was given at Baseline to measure PTSD symptoms from prior trauma exposure. The PCL-5 has demonstrated convergent and divergent validity, test-retest reliability, and internal consistency in previous research and demonstrated excellent internal consistency in the current study (α=.94; Blevins, Weathers, Davis, Witte, & Domino, 2015; Weathers, Litz, Herman, Huska, & Keane, 1993).

Accident Fear Questionnaire – Phobic Avoidance subscale (AFQ-PA).

The AFQ-PA is a 10-item subscale of the AFQ that indexes how likely individuals are to avoid situations involving motor vehicles (e.g., driving as a passenger, driving yourself, driving in certain weather conditions, crossing streets alone) following an accident on a 9-point scale. The AFQ-PA has demonstrated sound psychometric properties (Asmundson, Cox, Larsen, Frombach, & Norton, 1999). In the present investigation the AFQ-PA was administered at baseline, and again following the trauma film to assess whether the trauma film was distressing. Internal consistency at baseline and post-film was good (α’s=.79 and .89, respectively).

Impact of Event Scale- Revised (IES-R).

The IES-R (Weiss, 2007) is a 22-item questionnaire assessing past-week PTSD symptoms. The IES-R includes a total score and subscales reflecting intrusions, avoidance, and hyperarousal. The IES-R has demonstrated good psychometric properties, and may be a sensitive measure of PTSD symptoms in individuals experiencing mild symptoms (Creamer, Bell, & Failla, 2003), thus has been used in analogue trauma studies (Olatunji & Fan, 2015). The IES-R was administered one, three, and seven days after exposure. Instructions on the IES-R were modified such that participants were instructed to indicate their symptoms related to the accident film. All items referencing sleep disturbance (Items 2, 15, 20) were removed. Internal consistency was good for one-day, three-day, and week-one total (α’s=.92, .92, and .90, respectively), re-experiencing (all α’s=.87), and avoidance scores (α’s=.81, .85, and .80), and acceptable for hyperarousal scores (α’s=.78, .71, and .75).

State Anxiety.

Participants were prompted, “Please rate how anxious you feel right now” on a 10-point Likert-style scale. Participants were administered this question before and after the trauma film, and after the thought monitoring period.

Response to Intrusions Questionnaire (RIQ).

The RIQ (Clohessy & Ehlers, 1999) is a 19-item measure assessing strategies for responding to and coping with distressing intrusive thoughts. In the present study, only the six-item thought suppression scale was utilized. RIQ items were administered at the week-one follow-up appointment (e.g., I try to push them out of my mind, I drink alcohol or smoke, I think about something else). Participants were asked to rate how often they used strategies for dealing with unwanted memories of the trauma film on a 4-point scale. The RIQ has good psychometric properties (Clohessy & Ehlers, 1999), and internal consistency was excellent in the present sample (α=.90).

Procedure

Participants provided written informed consent upon arrival to the lab. Next they completed self-report measures (i.e., ISI, PHQ-8, PANAS-NA [trait], LEC, PCL-5) before a randomly assigned intervention (Cognitive Anxiety Sensitivity Treatment [CAST]) or physical health control (Figure 1). For further details see Boffa and Schmidt (2019), which discusses the positive impact of CAST in reducing PTSD symptoms in this sample. Following the intervention, participants watched the 10-minute accident film described below. After the film, participants completed the State Anxiety measure and AFQ-PA before participating in a ten-minute thought monitoring period, not relevant to the current study. After ten-minutes, participants provided an additional measure of State Anxiety. Finally, participants were given instructions for responding to mobile surveys, and their follow-up appointment was scheduled. A three-minute Qualtrics survey including the IES-R was sent via text message to participants at 7:00 p.m. one (Day 1) and three days (Day 3) following their baseline appointment. Seven days after their baseline appointment (Week 1), they returned to the lab to complete self-report measures, and were awarded class credit for their participation. All procedures were approved by the Florida State University’s Institutional Review Board (Protocol 2018.25400).

Figure 1.

Depiction of study design.

Trauma Film Paradigm

Participants were shown a 10-minute, continuous film clip with 20 scenes of motor vehicle accidents with high probability of serious injury or death. This film has been used in prior studies (Olatunji & Fan, 2015); the film and its permitted use were obtained through personal correspondence with the authors.

Data Analytic Plan

First, multiple regression analyses were used to test whether insomnia symptoms significantly predicted analogue PTSD symptoms and symptom clusters. Insomnia, negative affectivity, and condition were entered as predictors of total PTSD symptoms and symptom clusters. Second, four additional multiple regression analyses were used to determine whether insomnia symptoms significantly predicted potential mechanisms accounting for increased PTSD symptoms: increased peritraumatic distress (measured by a 1-item scale immediately after the analogue trauma and after a delay), and posttrauma avoidance (IES-R) and suppression (RIQ). Again, analyses covaried for negative affectivity and treatment condition. Finally, we conducted a temporal multiple mediation model using Preacher and Hayes’ Process macro for SPSS (Preacher & Hayes, 2004). Bias-corrected 95% confidence intervals were constructed using 5000 bootstrap sampling procedures. Insomnia symptoms were entered as a predictor, along with peritrauma distress, condition, and negative affectivity as predictors. Peritrauma delayed distress, Day 3 avoidance, and Week 1 suppression were entered as mediators. Week 1 PTSD symptoms, excluding the avoidance symptom cluster and sleep-related items, to prevent overlap with the predictors, were the dependent variable. To determine whether our mediation model was specific to our proposed mediators vs. other potential mediators (i.e., depressive symptoms), we conducted a rival mediational model. Insomnia symptoms and negative affectivity were entered as predictors, depressive symptoms were entered as a mediator, predicting Week 1 PTSD symptoms (without the avoidance cluster or sleep items).

In regards to missing data, some data were missing at the individual time points, including day one (17.6%), day three (12.0%), and week one (3.7%). Participants were deleted listwise in individual analyses for which they had missing data. Participants with missing data were not significantly different from those with complete data in terms of insomnia symptoms or reactivity (distress) to the trauma film for those missing day one (ps>.087), day three (ps>.121), or week one (ps>.479).

Results

Descriptive Statistics.

Means, standard deviations, and zero-order correlations for all variables are presented in Table 1. Insomnia symptoms were above the clinical cut off of 10 (Morin et al., 2011). Nearly all participants (98.1%) reported negative life events on the LEC, which involved directly experiencing, witnessing, or learning about potentially traumatic events that may not necessarily meet criteria for a traumatic event. The majority (62.0%) reported directly experiencing a motor vehicle accident, a natural disaster (61.1%), an uncomfortable sexual experience (37.0%), physical assault (25.9%), other serious accident (22.2%), witnessing a sudden or violent death (21.3%), sexual assault (15.7%), life-threatening illness or injury (8.3%), a fire or explosion (6.5%), and assault with a weapon (4.6%). Mean PTSD symptoms at Baseline were 26.81 (SD = 17.77), below the clinical cut-off of 33 on the PCL-5 (Blevins et al., 2015). Mean analogue PTSD symptoms at Day 1 after viewing the trauma film were higher than in previous research (Olatunji & Fan, 2015), potentially because we recruited an at-risk sample. Correlations were in the expected directions, with significant positive correlations between most study variables.

Table 1.

Means, standard deviations, and zero-order correlations for all variables of interest.

	1	2	3	4	5	6	7	8	9	10
1. Insomnia	-	-	-	-	-	-	-	-	-	-
2. NA	.25*	-	-	-	-	-	-	-	-	-
3. Pre-film Anxiety	.16	.43***	-	-	-	-	-	-	-	-
4. Post-film Anxiety	.21*	.30**	.62***	-	-	-	-	-	-	-
5. Post-film Anxiety Recovery	.25*	.30**	.65***	.69***	-	-	-	-	-	-
6. Day 1 PTSD symptoms	.26*	.24*	.27*	.66***	.44***	-	-	-	-	-
7. Day 3 PTSD symptoms	.31*	.23*	.21*	.48***	.36**	.69***	-	-	-	-
8. Day 3 Avoidance	.27*	.22*	.28**	.41***	.32**	.62***	.90***	-	-	-
9. Week 1 PTSD symptoms	.29**	.28**	.22*	.49***	.43***	.72***	.65***	.60***	-	-
10. Week 1 Suppression	.29**	.23*	.15	.31**	.29**	.54***	.51***	.50***	.55***	-
M (SD)	12.35 (5.47)	26.86 (7.40)	4.50 (1.85)	5.95 (2.06)	4.56 (2.00)	18.19 (12.45)	9.62 (10.41)	4.15 (5.00)	7.84 (8.23)	14.55 (5.05)

Note: NA=negative affectivity; PTSD symptoms=posttraumatic stress symptoms; M=mean; SD=standard deviation. All PTSD symptoms totals are calculated excluding sleep items.

^*=p<.05

^**=p<.01

^***=p<.00

Manipulation Check.

We conducted two t tests, which suggested state anxiety significantly increased from immediately prior (M=4.47, SD=1.85) to after (M=5.95, SD=2.06) the film (t(101)=7.26, p<.001). Second, accident fear also increased from immediately prior (M=32.15, SD=13.21) to after (M=41.82, SD=17.67) the film (t (102)=5.55, p<.001).

Direct Effects on IES-R.

At Day 1, insomnia symptoms did not significantly predict total PTSD symptoms or any symptom clusters (ps>.057, Table 2). At Day 3, insomnia symptoms significantly predicted PTSD symptoms (B=.50, p=.014), intrusions (B=.17, p=.037), avoidance (B=.20, p=.042), and hyperarousal (B=.14, p=.013). At Week 1, insomnia symptoms significantly predicted total PTSD (B=.39, p=.009), intrusions, (B=.15, p=.029), and hyperarousal (B=.12, p=.005), but not avoidance (B=.12, p=.087). Effect sizes were small.

Table 2.

Multiple regression models of the direct effects of insomnia on PTSD symptoms, covarying for negative affectivity and treatment condition.

	B	t	p	sr2
Day 1 PTSD symptoms
Total	.48	1.93	.057	.04
Intrusions	.18	1.84	.069	.04
Avoidance	.19	1.67	.098	.03
Hyperarousal	.11	1.52	.133	.03
Day 3 PTSD symptoms
Total	.50	2.50	.014	.06
Intrusions	.17	2.12	.037	.05
Avoidance	.20	2.06	.042	.04
Hyperarousal	.14	2.53	.013	.07
Week 1 PTSD symptoms
Total	.39	2.66	.009	.06
Intrusions	.15	2.21	.029	.04
Avoidance	.12	1.73	.087	.03
Hyperarousal	.12	2.89	.005	.08

Note. PTSD = Posttraumatic stress disorder.

Mediation Model.

The entire model was significant and accounted for 41.8% of the variance in Week 1 PTSD symptoms (F (7, 75)=7.71, p<.001, Figure 2). In terms of the a path, insomnia symptoms significantly predicted peritraumatic distress, Day 3 avoidance, and Week 1 intrusive thought suppression. The total indirect effect of insomnia on Week 1 PTSD was significant (B=.18, 95% CI[.05, .39]) via peritraumatic distress (B=.08, 95% CI [.01, .21]), Day 3 avoidance (B=.08, 95% CI [.01, .22]), but not Week 1 intrusive thought suppression (B=.02, 95% CI [−.01, .10]).

Figure 2.

Multiple mediation model.

Note: Condition, trait negative affectivity, and peritrauma anxiety were included as covariates. a=a path from independent variable to mediator, b=b path from mediator to outcome, c=c path (a*b), B=unstandardized regression coefficient, PTSD symptoms=posttraumatic stress symptoms (excluding the avoidance cluster and sleep-related symptoms). *=p<.05, **=p<.01, ***=p<.001

Alternate Models.

Results indicated a significant a path from insomnia to depressive symptoms (B=.23, SE=.07, p=.001), but not a significant b path from depressive symptoms to analogue PTSD symptoms (B=.04, SE=.15, p=.787). The indirect effect was also not significant (B=.01, 95% CI [−.06, .11]).

Discussion

Results from the current study support a growing line of research suggesting that insomnia symptoms increase risk for developing PTSD (Bryant et al., 2010; Gehrman et al., 2013; Koffel et al., 2013). Specifically, insomnia symptoms prior to undergoing the trauma film paradigm significantly predicted PTSD symptoms across each symptom cluster three days and one week following exposure. We also identified potential mechanisms of this association: peritrauma anxiety, intrusive memory suppression, and avoidance.

Consistent with prior studies, insomnia symptoms were a predictor of analogue PTSD symptoms, across all symptom clusters, even when covarying for other risk factors for PTSD, such as negative affectivity (Bryant et al., 2010; Gehrman et al., 2013; Koffel et al., 2013; Mellman, Pigeon, Nowell, & Nolan, 2007). Inconsistent with hypothesis, insomnia symptoms were not associated with increased PTSD symptoms the first day after trauma. This is potentially because it is normative to have PTSD symptoms in the immediate aftermath of trauma (Ehlers, Hackmann, & Michael, 2004), which typically resolve in days or weeks. With the analogue trauma design, PTSD symptoms on Day 1 are typically the highest (Olatunji & Fan, 2015), drastically reducing at each subsequent time point. Thus, although unexpected, insomnia symptoms may only predict ongoing PTSD symptoms, indicative of pathological outcomes, rather than acute PTSD symptoms, which are normative. Effects of insomnia on PTSD symptoms are fairly consistent at Day 3 and Week 1 across symptom clusters.

We also investigated mechanisms accounting for the insomnia-PTSD association. First, considering evidence that poor sleep is associated with increased negative emotional reactivity to stressors (Minkel et al., 2012), we examined whether insomnia symptoms predicted increased peritraumatic distress. Insomnia symptoms predicted ongoing distress 10 minutes after exposure. Thus, insomnia symptoms may predict difficulties recovering from a stressful experience. This is consistent with research noting that those with poor sleep quality have difficulty downregulating negative emotions (Mauss et al., 2013), as well as neurobiological research finding that brain areas responsible for adaptive emotion regulation functions such as cognitive reappraisal are negatively impacted by poor sleep (Klumpp et al., 2017; Yoo, Gujar, Hu, Jolesz, & Walker, 2007). In turn, the indirect effect of insomnia on later PTSD symptoms via peritraumatic distress was significant, indicating that prolonged distress in response to trauma may be a mechanism between insomnia symptoms and PTSD symptoms. The pathway between insomnia and PTSD symptoms via increased distress in response to trauma has been proposed in prior literature (e.g., Bryant et al., 2010), but, to our knowledge, this is the first study to test it.

There was also support for the role of insomnia symptoms in avoidance, and avoidance as a mediator of the association between insomnia and PTSD symptoms. This is consistent with prior research linking insomnia symptoms with avoidance of anxiety-related stimuli (Short & Schmidt, 2017), and theoretical models of PTSD emphasizing the critical role of avoidance in PTSD’s etiology (Foa, Huppert, & Cahill, 2006). Future research should attempt to replicate this novel finding that insomnia symptoms may exacerbate avoidance, and, in turn, PTSD symptoms.

Finally, insomnia symptoms were associated with increased suppression of intrusive memories after analogue trauma exposure. To our knowledge, this is the first study indicating that those with insomnia are prone to not only attempt to suppress pre-sleep intrusions (Harvey, 2003), but also trauma-related intrusions. This may reflect a growing line of research indicating that sleep is crucial for the processing of emotional memories (Goldstein & Walker, 2014). Sleep problems could interfere with the processing of a traumatic experience, resulting in increased intrusive memories and desire to suppress these. However, intrusion suppression did not mediate the impact of insomnia on PTSD symptoms. This may be due to including several mediators in analyses, preventing the ability to detect the role of suppression in the insomnia-PTSD relationship. However, in contrast with prior research, the association between intrusive memory suppression and PTSD symptoms was not observed in the current study (Clohessy & Ehlers, 1999; van den Hout & Engelhard, 2004). This may be due to the present analogue trauma design, in which intrusions may not be as distressing as in directly experienced trauma exposure.

Rival mediator analyses indicated that an alternate mediator, depressive symptoms, was not significant, supporting the specificity of peritrauma distress and avoidance as mediators of the insomnia-PTSD association. Overall, findings are consistent with a model in which insomnia symptoms increase reactivity to trauma as well as trauma reminders, and may lead to suppression or avoidance, exacerbating PTSD symptoms.

The current study has clinical implications. Reducing insomnia prior to trauma exposure could prevent the development of PTSD. This may be relevant for at-risk populations, such as military personnel and first responders. There are many brief and effective treatments for insomnia, including Cognitive Behavioral Therapy for Insomnia (CBT-I), as well as efficacious computerized treatments for insomnia (Espie et al., 2012). Thus, these treatments may be feasible to disseminate to such at-risk populations.

Findings from the current study must be considered in the context of its limitations and opportunities for future research. First, although the trauma film paradigm is well-established (Holmes & Bourne, 2008), it is not a Criterion A stressor (American Psychiatric Association, 2013). As such, results may not generalize to individuals experiencing a traumatic event. Second, we did not assess other types of sleep disturbance, such as nightmares, that may confer risk for PTSD (van Liempt, van Zuiden, Westenberg, Super, & Vermetten, 2013), or other potential mechanisms for the insomnia-PTSD association, such as hyperarousal, cognitive functioning or impairment in fear and safety learning. There was also no data available regarding the night’s sleep prior to or following analogue trauma exposure. Future research should incorporate other types of sleep disturbance (including the nights prior and following exposure), and other potential mechanisms to determine specificity of these findings. Third, we relied on self-report measures for the constructs of interest. Future research should incorporate multimodal measures of sleep (e.g., actigraphy), and anxious arousal (e.g., skin conductance) for a more complete assessment of these associations. Fourth, the effect size of insomnia on subsequent analogue PTSD was small. However, considering that treatments for insomnia are brief, portable, and efficacious, insomnia may be clinically relevant (Espie et al., 2012). Fifth, the intervention some received prior to the analogue trauma, CAST, significantly reduces insomnia symptoms (Short, Allan, Raines, & Schmidt, 2015), as well as PTSD (Short et al., 2017). Future research should examine these effects in the absence of an intervention. Sixth, although most mediation analyses were temporal mediation models, the model including intrusive memory suppression was not. Future research should measure intrusive memory suppression prior to PTSD symptom to conduct temporal models. Seventh, participants were not randomized to a trauma film vs. a control film, thus it cannot be determined whether the analogue symptoms observed were specific to trauma film exposure, and meditational models are not causal. Future research should include a control film. Eighth, the sample was undergraduates, and results may not generalize to all at-risk for PTSD symptoms. Future research should utilize representative samples. Ninth, we did not conduct a clinical interview and cannot determine if any members of the sample already met criteria for PTSD, so future research should formally screen for PTSD in similar designs.

The current study adds to a growing body of literature suggesting insomnia symptoms are an important risk factor in the development of PTSD and is the first to prospectively test potential mechanisms of this relationship. Findings identify increased peritrauma distress and posttrauma avoidance as mechanisms in the insomnia-PTSD relationship. These results support theoretical models proposing problems with emotional responding to stressors as a key factor explaining the insomnia-PTSD risk relationship.