Hindawi
Behavioural Neurology
Volume 2018, Article ID 7940832, 10 pages
https://doi.org/10.1155/2018/7940832
Research Article
Sleep Quality and Emotion Regulation Interact to Predict
Anxiety in Veterans with PTSD
Janna Mantua,1,2 Steven M. Helms,3,4 Kris B. Weymann,3,4 Vincent F. Capaldi II,2
and Miranda M. Lim 3,4
1
Neuroscience & Behavior Program, University of Massachusetts, Amherst, Amherst, MA 01003, USA
Behavioral Biology Branch, Walter Reed Army Institute of Research, Silver Spring, MD 20910, USA
3
VA Portland Health Care System, Portland, OR 97239, USA
4
Oregon Health & Science University, Portland, OR 97239, USA
2
Correspondence should be addressed to Miranda M. Lim; lmir@ohsu.edu
Received 9 February 2018; Revised 23 April 2018; Accepted 13 May 2018; Published 5 June 2018
Academic Editor: Laura Piccardi
Copyright © 2018 Janna Mantua et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Posttraumatic stress disorder (PTSD) is a debilitating and common consequence of military service. PTSD is associated with
increased incidence of mood disturbances (e.g., anxiety). Additionally, veterans with PTSD often have poor-quality sleep and
poor emotion regulation ability. We sought to assess whether such sleep and emotion regulation deficits contribute to mood
disturbances. In 144 veterans, using a double moderation model, we tested the relationship between PTSD and anxiety and
examined whether sleep quality and emotion regulation interact to moderate this relationship. We found that PTSD predicts
higher anxiety in veterans with poor and average sleep quality who utilize maladaptive emotion regulation strategies. However,
there was no relationship between PTSD and anxiety in individuals with good sleep quality, regardless of emotion regulation.
Similarly, there was no relationship between PTSD and anxiety in individuals with better emotion regulation, regardless of sleep
quality. Results were unchanged when controlling for history of traumatic brain injury (TBI), despite the fact that those with
both PTSD and TBI had the poorest emotion regulation overall. Taken together, these results suggest that good-quality sleep
may be protective against poor emotion regulation in veterans with PTSD. Sleep may therefore be a target for therapeutic
intervention in veterans with PTSD and heightened anxiety.
1. Introduction
Posttraumatic stress disorder (PTSD), which can occur after
a traumatic event, is characterized by negative alterations
in cognition, mood, and arousal/reactivity [1]. PTSD is
prevalent in post-9/11 veteran populations; roughly 1 in
7 deployed veterans and 1 in 10 nondeployed veterans carry
a diagnosis of PTSD [2]. Additionally, PTSD can have lasting
negative consequences and significant disability. Compared
to age- and gender-matched veterans without PTSD, veterans with PTSD have a significantly poorer functioning
in a number of tested domains. For example, veterans with
PTSD are less likely to be working, more likely to have
physical limitations, and more likely to have committed
violence in the past several years [3]. Therefore, identifying
contributing factors to impairment in PTSD is critical to
designing effective therapies.
Given the prevailing physiological hyperarousal present
in PTSD, it is perhaps not surprising that sleep issues are
common in this population. In fact, insomnia is the most
commonly reported symptom of PTSD [4]. Yet, sleep issues
are not simply the product of hyperarousal, as they also seem
to be causal in exacerbating PTSD symptomatology. For
instance, in PTSD populations, treating insomnia (via cognitive behavioral therapy) reduces PTSD symptomatology
[5] and treating of obstructive sleep apnea (OSA), which
is also common in PTSD, reduces symptoms as well [6].
Taken together, these results suggest that sleep quality is
an important and modifiable factor that can influence
PTSD severity.
2
Individuals with PTSD also suffer from impaired emotion regulation, which involves the voluntary or involuntary
process of modulating one’s own emotional experience [7].
The process of emotion regulation is implemented using a
number of well-characterized cognitive techniques or strategies. These strategies are not all equally valuable. That is,
some emotion regulation strategies tend to be helpful or
adaptive, while others are not. For instance, “expressive suppression” (trying to hold back or hide emotions) is predictive
of poor emotional outcomes (e.g., anxiety), while “cognitive
reappraisal” (reframing emotional situations to make them
less emotional) is predictive of healthy emotional functioning
[8]. Individuals with PTSD tend to utilize more expressive
suppression than cognitive reappraisal [9], and the overuse
of expressive suppression may exacerbate poor psychological
functioning in this population.
Importantly, the type of strategy used for emotion regulation seems to be moderated by sleep quality. Vantieghem and
colleagues recently showed that individuals with sleep issues
(i.e., insomnia) suppress emotions more frequently and are
less likely to use cognitive reappraisal than those without
insomnia [10]. A separate study showed that poorer sleep
quality (e.g., higher scores on the Pittsburgh Sleep Quality
Index) predicted less cognitive reappraisal use and more
expressive suppression use [10]. These studies suggest that
there is a potential interaction between sleep quality and
emotion regulation strategy use, such that sleep quality influences emotion regulation or vice versa.
The current study therefore is aimed at examining the
potential relationship between sleep quality and emotion
regulation in veterans with and without PTSD. We specifically focus on a core feature of PTSD, that is, anxiety symptomatology, and examine how sleep quality and emotion
regulation interact to moderate this relationship. Using a
cohort of veterans with and without PTSD, we hypothesized
the following: (1) Individuals with PTSD would have higher
anxiety than veterans without PTSD. (2) Individuals with
PTSD would have poorer sleep quality than veterans without
PTSD. (3) Individuals with PTSD would utilize higher levels
of expressive suppression and lower levels of cognitive reappraisal than veterans without PTSD. (4) Sleep quality and
emotion regulation strategy use would interact to moderate
the link between PTSD and anxiety. Furthermore, given that
PTSD is often comorbid with TBI in the veteran population,
we also examined the effects of TBI on these hypotheses as a
contributing factor.
2. Materials and Methods
2.1. Sample. In total, 144 veterans from the Veterans Affairs
(VA) Portland Healthcare System participated under a VA
IRB approved protocol (MIRB number 3641, PI: Lim).
Twenty-four individuals met criteria for PTSD on the basis
of self-reported symptoms (cluster criteria in the PTSD
Checklist 5 or PCL-5, total score > 33 [11]) (“PTSD only”
group). Twenty-three individuals carried a diagnosis of TBI
confirmed in the medical record (“TBI only” group). Eleven
veterans met criteria for PTSD and carried a diagnosis of
TBI (“PTSD + TBI” group). Eighty-six veterans without a
Behavioural Neurology
history of trauma exposure (neither PTSD nor TBI) served
as matched controls (“control” group).
2.2. Questionnaires. The main analyses were conducted
using several self-reported questionnaires: (1) Emotion
Regulation Questionnaire (ERQ) [12], (2) National Institutes of Health Patient-Reported Outcomes Measurement
Information System (NIH PROMIS) anxiety items [13],
and (3) Insomnia Severity Index (ISI) [14]. We also included
two questionnaires for descriptive purposes: (1) Patient
Health Questionnaire-9 (PHQ-9) [15] and (2) the Functional
Outcomes of Sleep Questionnaire-10 (FOSQ-10) [16].
2.2.1. ERQ. The ERQ is a 10-item scale that measures an
individual’s tendency to use specific emotion regulation
strategies [12]. Two emotion regulation strategies in particular (expressive suppression and cognitive reappraisal) are
assessed by this questionnaire. Individual items are 7-point
Likert scales: 1 = “strongly disagree”, 4 = “neutral”, and
7 = “strongly agree”. A higher number indicates that the
participant utilizes the given emotion regulation strategy
more frequently.
2.2.2. NIH PROMIS. The NIH PROMIS battery is a multidomain test with a range of outcomes pertaining to physical,
mental, and social wellbeing. Participants were asked to rate
to what extent four items about anxiety symptoms described
them (e.g., “In the past 7 days, my worries overwhelmed
me.”). Items are 5-point Likert scales: 5 = “never”, 4 = “rarely”,
3 = “sometimes”, 2 = “often”, and 1 = “always”. A lower number indicates higher dysfunction.
2.2.3. ISI. The ISI is a 7-item measure assessing insomnia
severity (i.e., difficulty initiating and staying asleep). Individual items are 5-point Likert scales: 0 = “none”, 1 = “mild”,
2 = “moderate”, 3 = “severe”, and 4 = “very severe”. A higher
score indicates higher dysfunction.
2.2.4. FOSQ-10. The FOSQ-10 is a 10-item measure assessing
whether an individual has poor quality of life due to poor
sleep quality. Individual items are 4-point Likert scales:
1 = “yes, extreme difficulty”, 2 = “yes, moderate difficulty”,
3 = “yes, a little difficulty”, and 4 = “no difficulty”; however,
half of the items are a 5-point Likert scale that includes
a rating of 0 = “I do not do this activity for other reasons”.
The survey has 5 subscales: (1) activity level, (2) vigilance,
(3) intimacy and sexual relationships, (4) general productivity, and (5) social outcomes. A lower score indicates
higher dysfunction.
2.2.5. PHQ-9. The PHQ-9 is a 10-item measure assessing
the presence of depressive symptoms. Individual items,
which enquire about how frequently participants experience the presented symptoms, are a 4-point Likert scale:
0 = “not at all”, 1 = “several days”, 2 = “more than half the
days”, and 3 = “nearly every day.” A higher score indicates
higher dysfunction.
2.3. PTSD Diagnosis. Individuals met criteria for PTSD on
the basis of symptoms reported in the Posttraumatic Stress
Disorder Checklist (PCL-5) [11], using a combination of
Behavioural Neurology
cluster criteria, and total score > 33 [11]. The PTSD Checklist
for DSM-5 (PCL-5) is a 20-item measure used for the screening and provisional diagnosis of PTSD, as well as assessing
symptom severity. Individual items are 5-point Likert scales:
0 = “not at all”, 1 = “a little bit”, 2 = “moderately”, 3 = “quite a
bit”, and 4 = “extremely”. In the current study, PTSD status
was used as a categorical variable (1 = presence of PTSD,
0 = no PTSD).
2.4. TBI Diagnosis. Two trained researchers reviewed medical
records from the Portland VA electronic medical record
system to determine whether veterans had a history of TBI.
Individuals with self-reported TBI that was not confirmed
in the medical record were excluded from the sample.
2.5. Protocol. Consecutive veterans who entered the Portland
VA Sleep Clinic for a clinical diagnostic sleep study were
recruited to participate as part of a larger study (MIRB
number 3641, factors that affect adherence to treatment for
sleep apnea, PI: Lim). All participants were in the lab for
overnight sleep assessments and were asked by a trained
research coordinator whether they would like to participate.
Veterans who wished to participate completed all questionnaires either in person, via mail, or by phone, between the
time of consent and 12 months later.
2.6. Statistical Analyses. IBM SPSS Statistics 23 (Armonk,
NY) was used to conduct statistical analyses. Analyses were
conducted in several phases. First, descriptive statistical tests
were conducted to determine whether groups differed at
cross section in meaningful ways. One-way ANOVA tests
were used to determine whether groups differed for demographic factors (e.g., age), objective sleep characteristics
(e.g., TST and sleep staging), and self-reported functioning
(e.g., via the FOSQ-10 and PHQ-9). If a significant difference
between groups was detected, post hoc Tukey tests were used
to determine which groups differed.
Next, 2 × 2 ANOVA tests (TBI versus non-TBI, PTSD
versus non-PTSD) were used to assess whether groups
differed for the outcome measures of interest. For this
set of analyses, a 2 × 2 ANOVA was utilized rather than
a one-way ANOVA so that we could examine whether
PTSD and TBI interacted to synergistically affect outcomes
of interest. Outcome measures were cognitive reappraisal
(higher = more use of that strategy), expressive suppression
(higher = more use of that strategy), ISI scores (sleep quality;
higher = more sleep problems), and NIH PROMIS anxiety
scores (anxiety; lower = more anxiety). In addition, an
emotion regulation ratio was calculated, which consists of
cognitive reappraisal divided by expressive suppression
(higher = more cognitive reappraisal use, lower = more
expressive suppression use), so both strategies could be
assessed in one continuous outcome measure.
Lastly, given our a priori hypotheses about sleep quality
interacting with emotion regulation to affect anxiety symptomatology in individuals with PTSD, moderation analyses
were conducted to determine whether sleep (ISI scores) and
emotion regulation (emotion regulation ratio) both moderate
the relationship between PTSD (yes/no diagnosis) and
3
M
X
W
Y
Figure 1: Conceptual model of the relationship between PTSD
status, sleep quality, emotion regulation strategy use, and
anxiety symptoms. X = PTSD status; M = sleep quality (ISI
scores); W = emotion regulation strategy use (emotion regulation
ratio); Y = anxiety symptoms (PROMIS anxiety items).
anxiety (NIH PROMIS scores) (see Figure 1 for schematic).
Analyses were conducted with and without TBI as a covariate
to determine whether the presence of TBI impacts the
relationship between PTSD, sleep, emotion regulation, and
anxiety. A moderation model was chosen over a mediation
model due to our a priori focus on the interaction between
factors [17].
Moderation analyses, discussed directly above, were
conducted using the SPSS PROCESS macro (Model 3).
PROCESS calculates statistical significance using 95% confidence intervals to determine the effect of the predictor on
the outcome measure. PROCESS also calculates the interaction effects of the moderators by testing the predictive effect
of each factor at different levels (e.g., a low emotion regulation ratio and good sleep quality or a high emotion regulation ratio with poor sleep quality). Each predictor level was
determined by values plus or minus 1 standard deviation
from the centered mean. In order to eliminate both statistical power limitations and concerns about statistical
assumption violations, 5000 bootstrap iterations were used
within the macro (at least 2000 bootstrap iterations are
recommended [18]).
Given the potential bias in our sample, which was
referred for evaluation of sleep disorders, polysomnography(PSG-) derived measures of sleep were assessed for group differences using ANOVA tests and results are reported below.
3. Results and Discussion
3.1. Preliminary Analyses. Demographic information is listed
in Table 1. Notably, consistent with the demographics of veterans, the sample was predominantly male. There were no
significant group differences for age, sex, or BMI between
groups, although the mean age of the PTSD + TBI group
was notably lower than that of the other groups. There were
no group differences between objective PSG-derived sleep
metrics (total sleep time (TST), sleep efficiency, sleep staging,
and apnea-hypopnea index (AHI)).
There were several mental health/quality-of-life questionnaires for which groups differed. For example, there
was a significant difference between groups for the NIH
PROMIS anxiety items (Figure 2(a)). Consistent with our
hypotheses, post hoc tests showed that the control group
had significantly lower anxiety symptomatology than the
PTSD-only group (mean difference = 2.95, p = 0 002) and
the PTSD + TBI group (mean difference = 5.05, p < 0 001).
The TBI-only group had significantly lower anxiety than
the PTSD + TBI group (mean difference = 3.46, p = 0 03).
4
Behavioural Neurology
Table 1: Demographics (mean ± standard deviation). BMI = body mass image; PHQ-9 = Patient Health Questionnaire version 9; FOSQ10 = Functional Outcomes of Sleep Questionnaire version 10; NIH PROMIS = National Institutes of Health Patient-Reported Outcomes
Measurement Information System anxiety items; TST = total sleep time; SE = sleep efficiency; NREM1 (%) = percent of the night in nonREM sleep stage 1; NREM2 (%) = percent of the night in non-REM sleep stage 2; SWS (%) = percent of the night in slow wave sleep; REM
(%) = percent of the night in rapid eye movement sleep; AHI = apnea/hypopnea index.
Demographics
Age
BMI
Gender (% male)
Mental health and quality of life
NIH PROMIS
PHQ-9
FOSQ-10
ISI
Objective PSG-derived sleep metrics
TST (min)
SE (%)
NREM1 (%)
NREM2 (%)
SWS (%)
REM (%)
AHI
Controls
TBI only
PTSD only
PTSD + TBI
F value
p value
59.1 ± 13.3
33.0 ± 6.8
89.4
56.2 ± 15.3
33.0 ± 7.5
90
55.7 ± 13.5
32.2 ± 5.4
80.1
48.8 ± 21.3
29.3 ± 5.0
100
1.6
1.0
2.7
0.13
0.42
0.45
16.0 ± 3.1
6.6 ± 4.8
15.9 ± 3.1
12.0 ± 6.1
14.4 ± 4.3
8.7 ± 4.7
14.9 ± 3.7
13.7 ± 5.04
13.0 ± 3.6
13.3 ± 5.6
10.9 ± 3.0
17.2 ± 5.9
10.9 ± 4.0
14.0 ± 7.1
12.3 ± 4.3
18.5 ± 5.4
10.6
15.0
15.6
6.8
Purchase answer to see full
attachment