You are on page 1of 9

Psychological Trauma:

Theory, Research, Practice, and Policy


In the public domain 2020, Vol. 12, No. 7, 756 –764
ISSN: 1942-9681 http://dx.doi.org/10.1037/tra0000567

Changes in Physiological Reactivity in Response to the Trauma Memory


During Prolonged Exposure and Virtual Reality Exposure Therapy for
Posttraumatic Stress Disorder

Andrea C. Katz Aaron M. Norr and Benjamin Buck


VA Puget Sound Healthcare System, Seattle Division, Seattle, VA Puget Sound Healthcare System, Seattle Division, Seattle,
Washington Washington, and University of Washington School of Medicine

Emily Fantelli, Amanda Edwards-Stewart, Barbara O. Rothbaum


Patricia Koenen-Woods, Kimberlee Zetocha, Emory University School of Medicine
Derek J. Smolenski, and Kevin Holloway
National Center for Telehealth and Technology,
Tacoma, Washington

JoAnn Difede Albert Rizzo


Cornell University University of Southern California

Nancy Skopp, Matt Mishkind, and Gregory Gahm Greg M. Reger


National Center for Telehealth and Technology, VA Puget Sound Healthcare System, Seattle Division, Seattle,
Tacoma, Washington Washington, and University of Washington School of Medicine

Frank Andrasik
The University of Memphis

Objective: A key symptom of posttraumatic stress disorder (PTSD) is hyperreactivity to trauma-relevant


stimuli. Though physiological arousal is reliably elevated in PTSD, the question remains whether this
arousal responds to treatment. Virtual reality (VR) has been posited to increase emotional engagement
during prolonged exposure therapy (PE) for PTSD by augmenting imaginal exposures with trauma-
relevant sensory information. However, the comparative effects of VR exposure therapy (VRE) have
received limited empirical inquiry. Method: Ninety active-duty soldiers with combat-related PTSD

This article was published Online First April 27, 2020. wood, New Jersey. Kimberlee Zetocha is now at the St. Cloud VA Health
X Andrea C. Katz, V A Puget Sound Healthcare System, Seattle Divi- System, St. Cloud, Minnesota. Matt Mishkind is now at the Departments of
sion, Seattle, Washington; Aaron M. Norr and Benjamin Buck, VA Puget Psychiatry and Family Medicine, Helen and Arthur Johnson Depression
Sound Healthcare System, Seattle Division, and Department of Psychiatry Center, School of Medicine, University of Colorado Anschutz Medical
and Behavioral Sciences, University of Washington School of Medicine; Campus.
Emily Fantelli, Amanda Edwards-Stewart, Patricia Koenen-Woods, Kim- The authors would like to disclose the following: JoAnn Difede serves
berlee Zetocha, Derek J. Smolenski, and Kevin Holloway, National Center as a consultant for Pear Therapeutics, Inc. and BehaVR, LLC; Barbara O.
for Telehealth and Technology, Tacoma, Washington; X Barbara O. Rothbaum is a co-owner of Virtually Better, Inc.; and Matt Mishkind is a
Rothbaum, Department of Psychiatry, Emory University School of Medi- paid advisory board member of Meta Pro. This research was supported by
cine; JoAnn Difede, Weill Medical College, Cornell University; Albert
Grant W81XWH-08-2– 0015 from the U.S. Army Medical Research and
Rizzo, Institute for Creative Technologies, University of Southern Califor-
Materiel Command Military Operational Medicine Research Program. This
nia; Nancy Skopp, Matt Mishkind, and Gregory Gahm, National Center for
material is the result of work supported with resources and the use of
Telehealth and Technology; Greg M. Reger, VA Puget Sound Healthcare
facilities at VA Puget Sound Health Care System. The authors would like
System, Seattle Division, and Department of Psychiatry and Behavioral
Sciences, University of Washington School of Medicine; Frank Andrasik, to acknowledge all who contributed to the study including two Army
Department of Psychology, The University of Memphis. Departments of Psychology. The opinions or assertions contained herein
Aaron M. Norr is now at the VISN 20 Northwest Network Mental Illness are the private views of the authors and are not to be construed as official
Research, Education and Clinical Center (MIRECC), Seattle, Washington. or reflecting the views of the Department of Army, the Department of
Amanda Edwards-Stewart, Derek J. Smolenski, and Nancy Skopp are now Defense, or the Department of Veterans Affairs.
at the Psychological Health Center of Excellence, Tacoma, Washington. Correspondence concerning this article should be addressed to Greg M.
Patricia Koenen-Woods is now at the Rockland Psychiatric Center, Or- Reger, VA Puget Sound Healthcare System, Seattle Division, 9600 Vet-
angeburg, New York, and Advanced Psychological Specialists, Ridge- erans Drive, A-116, Tacoma, WA 98493. E-mail: greg.reger@va.gov

756
CHANGES IN GSR DURING VRE VERSUS PE FOR PTSD 757

participating in a randomized-controlled trial to receive PE, VRE, or a waitlist-control (WL) condition


had their physiological reactivity, indexed by galvanic skin response (GSR), to their trauma memories
assessed at pre-, mid-, and posttreatment. Results: Although both VRE and PE conditions showed
reduced GSR reactivity to trauma memories from pre- to posttreatment, only the VRE group differed
significantly from WL. Across the sample, reductions in GSR were significantly correlated with
reductions in self-reported PTSD and anxiety symptoms. Conclusions: This was the first study com-
paring effects of VRE and PE on psychophysiological variables. Given previous research finding limited
differences between VRE and PE in PTSD symptom reduction, these findings lend support to the
rationale for including VR in exposure therapy protocols while raising important questions about the
potential benefits of VRE.

Clinical Impact Statement


This study is among the first to examine how physiological processes change throughout PTSD
treatment and the first to compare standard exposure therapy to therapy augmented with virtual
reality (VR) in active-duty soldiers with PTSD. Results showed that soldiers in VR therapy had
smaller physical reactions to trauma memories compared to those who did not receive treatment,
whereas those who got standard treatment did not. These findings provide insight into possible
mechanisms of PTSD treatment, point to potential objective indicators of early treatment response in
active-duty soldiers, and suggest that VR treatment might lead to earlier symptom reduction.

Keywords: posttraumatic stress disorder, virtual reality, prolonged exposure, psychophysiology, military
personnel

Supplemental materials: http://dx.doi.org/10.1037/tra0000567.supp

A core feature of posttraumatic stress disorder (PTSD) is ele- PE is grounded in emotional processing theory (EPT), which
vated arousal in response to trauma-related stimuli (American postulates that trauma memories are represented in a neural net-
Psychiatric Association, 2013). In addition to subjective experi- work comprising feared stimuli, behavioral and psychological
ences of hyperarousal, evidence from psychophysiological studies avoidance of feared stimuli, and the meaning of this formed
has repeatedly shown that people with PTSD demonstrate height- association (Foa & Kozak, 1986). Although avoidance of feared
ened physiological reactivity to threatening and trauma-related stimuli can be evolutionarily adaptive, in the context of PTSD, the
cues (Grillon, 2002; Orr, Metzger, & Pitman, 2002; Pitman et al., fear network generalizes to nonthreatening stimuli. This resultant
2012). In particular, elevated heart rate (HR) and galvanic skin avoidance makes these generalized fear responses difficult to
response (GSR; i.e., changes in sweat gland activity corresponding change without treatment (Foa & Kozak, 1986). According to
to emotional arousal) have evidenced robust associations with EPT, effective treatment must repeatedly activate this maladaptive
PTSD (Pole, 2007) and are specific to fear-based disorders (i.e., fear network to modify it. To do this, patients must engage emo-
PTSD and phobias), rather than generalized anxiety and depression tionally with traumatic memories during “imaginal exposures.”
(O’Kearney & Parry, 2014). Specifically, PTSD was associated Findings supporting EPT have shown that early, strong emotional
with slower GSR habituation to startling sounds, heightened rest- engagement (EE) during imaginal exposure predicts better out-
ing HR, and larger HR responses to both standardized (e.g., comes when EE is defined as subjective distress (Jaycox, Foa, &
generic sounds of explosions) and idiosyncratic (i.e., individual- Morral, 1998) or physiological response (Wangelin & Tuerk,
specific) trauma cues. A meta-analysis found that idiosyncratic 2015).
trauma cue studies had the largest effect sizes (compared to gen- A variant of PE, virtual reality exposure therapy (VRE), has
erally threatening stimuli) across the widest number of psycho- repeatedly demonstrated efficacy for treating PTSD (Carl et al.,
physiological measures (Pole, 2007). 2019; Difede et al., 2014; Reger et al., 2011, 2016; Rizzo et al.,
In hopes of identifying physiological markers of PTSD that 2011; Rothbaum, Hodges, Ready, Graap, & Alarcon, 2001; Roth-
may be more objective and/or treatment-sensitive than patient baum et al., 2014). VRE is hypothesized to increase EE by en-
self-report, researchers are increasingly focusing on whether hancing imaginal exposure with trauma-relevant sensory stimuli
(and how) physiological correlates of PTSD change with treat- (Robison-Andrew et al., 2014). This increased EE should, theo-
ment (Pitman et al., 2012). For example, Robison-Andrew et al. retically, further increase activation of the fear structure and yield
(2014) found that reductions in the fear-potentiated startle eye- larger reductions in fear responses to nondangerous trauma stimuli
blink response distinguished responders to prolonged exposure over time. However, few studies have examined this question
(PE; Foa, Hembree, & Rothbaum, 2007) from nonresponders. empirically, and none have directly compared changes in psycho-
Others found that combat veterans with PTSD who completed physiology between VRE and PE. Case studies on US Navy
PE displayed reduced HR and GSR reactivity to a script-driven corpsmen found reduced GSR to distressing deployment memories
imagery task from pre- to posttreatment (Wangelin & Tuerk, from pre- to posttreatment and improved recovery during rest
2015). periods following recall (Wood et al., 2007, 2008). The only
758 KATZ ET AL.

controlled trial addressing this topic compared various psycho- provided written informed consent prior to participation and the
physiological responses (GSR, startle, and HR) of 50 combat local Institutional Review Board reviewed and approved the study.
veterans randomly assigned to VRE with D-cycloserine enhance- This article reports on GSR collected during assessments con-
ment, VRE with alprazolam, or VRE with placebo. Veterans ducted as part of the study. Although the physiological assess-
across groups showed significant reductions in startle and GSR ments were in place at study launch, it became evident that the
(but not HR) responses to standardized VR combat scenarios from quality of the physiological data initially being collected was
pre- to posttreatment (Norrholm et al., 2016). As all participants inadequate, and a higher quality, research-grade physiological
received VRE, it was not possible to compare these results to an assessment package was subsequently acquired (Biopac MP150;
existing psychotherapy standard of care. Biopac Systems Inc., 2002). This study consequently reports on
In the current study, we investigated how physiological reactiv- the subsample of soldiers (N ⫽ 90) with valid physiological data
ity—as indexed by GSR—to trauma memories changed over the collected with the improved equipment. Of the total sample, 82
course of a randomized controlled trial (RCT) comparing PE, completed the midtreatment assessment, and 73 completed the
VRE, and a waitlist control. Though multiple markers of physio- posttreatment assessment. Analyses for this study retained all
logical arousal have been examined in the context of PTSD, we randomized participants who had baseline psychophysiological
focused on GSR given its robust associations with PTSD symp- data.
toms, the research highlighting its response to treatment, and its Soldiers were predominantly male (94.444%). Consistent with
minimally invasive nature. The specific aims were: to examine the sample for the parent study (Reger et al., 2016), no statistically
how GSR changes over the course of exposure-based treatment for significant differences were found between groups in the distribu-
PTSD; to test whether VRE reduces physiological reactivity to tion of demographic variables or baseline outcome measures (see
trauma cues over and above the effects of PE; and to explore how Table 1) for this study’s subsample.
changes in psychophysiological correlates of PTSD relate to
changes in PTSD symptoms. Because habituation to a feared
Materials
stimulus is the rationale for exposure-based treatment, we expected
that GSR reactivity to idiosyncratic trauma memories would de- Clinician Administered PTSD Scale (CAPS). The CAPS
crease for both active treatment groups (i.e., VRE and PE) over (Blake et al., 1995) is a structured clinical interview that assesses
time; that is, that the magnitude of the GSR to trauma memories the frequency and intensity of PTSD symptoms according to
would diminish over the course of treatment. Though we expected criteria in the Diagnostic and Statistical Manual of Mental
changes in GSR reactivity in both exposure therapy groups com- Disorders-Fourth Edition (DSM–IV; American Psychiatric Asso-
pared to the control group, we expected greater GSR reductions in ciation, 2000). Frequency and intensity are separately coded 0 to 4.
the VRE group than the PE group, as VRE is posited to increase This study used the ‘last week’ reference at all assessments. CAPS
EE and promote further reductions in fear responses during treat- assessors were clinical psychologists trained to diagnostic compe-
ment. We did not generate specific hypotheses about the relation- tency (␬ ⬎ .800 agreement with trauma expert; Reger et al., 2016)
ship between psychophysiology and self-reported symptoms, as who were blind to treatment group assignment. Participants were
this aim was exploratory. instructed not to disclose their treatment condition to assessors.
The CAPS has demonstrated excellent convergent and discrimi-
Method nant validity, diagnostic sensitivity, sensitivity to symptom
change, and reliability (Weathers, Keane, & Davidson, 2001).
Participants Consistent with previous findings, the intensity and frequency
scales demonstrated adequate internal consistency (␣ ⫽ .869 and
Sample data were drawn from an RCT evaluating the compar- .870, respectively).
ative efficacy of PE and VRE among active-duty U.S. soldiers Beck Anxiety Inventory (BAI). The BAI (Beck & Steer,
with combat-related PTSD in a minimal-attention waitlist- 1990) is a 21-item self-report measure of symptoms of anxiety
controlled design (Reger et al., 2016). Participants (N ⫽ 162) symptoms in the prior week. Items range from 0 (not at all) to 3

Table 1
Sample Demographics and Baseline Characteristics by Treatment Group

Characteristic PE VRE WL p

Number of males (%) 29 (95) 27 (94) 29 (94) .292


Mean Age (SD) 31.258 (6.633) 29.931 (6.923) 30.833 (6.968) .807
Mean CAPS weekly (SD) 77.516 (17.605) 84.551 (16.302) 82.600 (17.350) .261
Mean BAI (SD) 20.936 (8.752) 24.310 (10.351) 24.467 (11.557) .321
Mean BDI (SD) 26.226 (11.026) 28.345 (9.469) 29.233 (10.699) .514
Mean GSR rise (SD) 1.2113 (1.084) 1.296 (1.384) 1.069 (1.969) .747
Mean GSR recovery (SD) 0.117 (1.140) 0.173 (0.705) 0.471 (1.077) .338
Note. PE ⫽ prolonged exposure; VRE ⫽ virtual reality exposure; WL ⫽ waitlist control; CAPS ⫽ Clinician
Administered PTSD Scale; BAI ⫽ Beck Anxiety Inventory; BDI ⫽ Beck Depression Inventory-II; GSR ⫽
galvanic skin response. p values indicate statistically significant differences based on one-way analysis of
variance or ␹2 test.
CHANGES IN GSR DURING VRE VERSUS PE FOR PTSD 759

(severely) and reflect subjective somatic or panic-related symptoms. mon reactions to traumatic events, breathing retraining, repeated
The BAI has been well-validated in clinical samples (Bardhoshi, imaginal exposure to the index trauma memory, cognitive and
Duncan, & Erford, 2016). Consistent with previous findings, the scale emotional processing of material and emotions that emerge during
demonstrated excellent internal consistency (␣ ⫽ .888) within the imaginal exposure, and in vivo exposure to safe but anxiogenic
study sample. circumstances that are avoided (Foa et al., 2007).
Beck Depression Inventory-II (BDI). The BDI-II (Beck, Virtual reality exposure. VRE was identical to PE, except
Steer, & Brown, 1996) is a 21-item self-report measure of depres- imaginal exposure was augmented with VR. VRE participants
sion severity over the past two weeks. Items are rated 0 (least wore a head-mounted display and were placed in a trauma-relevant
severe) to 3 (most severe) and reflect the core symptoms of virtual Iraq/Afghanistan environment (Rizzo et al., 2010). As
depression (e.g., anhedonia, depressed mood, sleep disturbance), participants revisited the memory, therapists customized the envi-
as well as overlapping constructs (e.g., hopelessness, helpless- ronment with relevant stimuli (e.g., sights, sounds, events) to
ness). The BDI-II has been well-validated in clinical samples match the memory. Participants navigated the environment using a
(Erford, Johnson, & Bardoshi, 2016). Consistent with previous joystick, and manufactured smells (e.g., gun powder, body odor)
findings, the scale demonstrated excellent internal consistency could be triggered when relevant. During treatment, participants
(␣ ⫽ .898) within the study sample. sat or stood on a platform with bass shaker speakers such that low
Biopac MP-150 and AcqKnowledge. Psychophysiological frequency sounds (e.g., engines, explosions) were also experienced
data were collected with the Biopac MP150 modular data acqui- as vibrations (see Rizzo et al., 2010, 2011 for details).
sition system and analyzed with the corresponding AcqKnowledge Waitlist control. Participants randomized to the WL condition
4.1 software (Biopac Systems Inc., U.S.A.). Two disposable completed all assessments over the course of five weeks without
Biopac EL507 Ag/AgCl electrodes, pregelled with isotonic gel receiving treatment. After completing all study procedures, they
(0.5% chloride salt) to stabilize measurements and reduce motion were offered participation in PE or VRE.
artifact, were adhered to participants’ most distal phalange (i.e., Assessments. Assessments were conducted at pretreatment,
the fleshy part of the fingertip) on the inside surfaces of the third midpoint (i.e., after 5 PE or VRE sessions or 2.5 weeks of WL),
and fourth fingers on their nondominant hand. Our data collection and posttreatment (i.e., after 10 PE or VRE sessions or 5 weeks of
and analysis procedures were consistent with established guide- WL). Assessments included a CAPS, self-report measures, and a
lines (Boucsein et al., 2012; Dawson, Schell, & Filion, 2016). physiological activation paradigm.
Activation paradigm. After completing the CAPS and self-
report measures, participants underwent a 15-min activation par-
Procedure
adigm to examine their physiological stress response (i.e., GSR) to
Enrollment. Participants were active-duty soldiers with trauma cues. The paradigm comprised three 5-min phases. During
PTSD stemming from military deployment to Iraq or Afghanistan. phase one (baseline), participants sat still, without talking, to
Participants were either referred by clinicians at an Army medical habituate to the psychophysiology equipment. During phase two
center or self-referred after encountering recruitment materials. (trauma recall), participants thought about their index trauma.
Inclusion criteria were: 1) a nonsexual assault index event that met They were asked to recall the memory in vivid sensory and
the diagnostic criterion for trauma according to the CAPS; 2) the emotional detail and cue the assessor once the memory was clear
index trauma occurred at least 3 months prior to study enrollment; so that they would begin recording GSR for this phase. During
and 3) the trauma occurred in an environment similar to those phase three (rest), participants relaxed using pleasant imagery.
available in the Virtual Iraq/Afghanistan software. Soldiers were Physiological data collection and processing. GSR was as-
excluded if they had: 1) initiated a new psychotropic medication or sessed with research-grade physiological assessment equipment
dosage change in the previous 30 days; 2) a history of organic and analyzed with a corresponding software package. GSR data
mental disorder, schizophrenia, other psychotic disorder, or bipolar were collected with a gain setting of 5 ␮Ʊ/V and a low-pass filter
disorder; 3) been hospitalized for suicidal risk or self-harm in past of 1.0 Hz and DC high-pass filter. Data were collected at a
6 months; 4) an ongoing threatening situation (e.g., domestic sampling frequency of 1000 Hz. Skin impedances were under 10
violence); 5) current drug or alcohol dependence; 6) a history of k⍀. After data acquisition, raw GSR data were then averaged
seizures; 7) prior PE treatment; 8) ongoing psychotherapy for across 1-min intervals for each 5-min phase of the stress test,
PTSD; 9) a physical condition interfering with the ability to use a yielding 15 average GSR measurements.
VR head-mounted display or gaming joystick; or 10) a history of
a loss of consciousness for longer than 15 min since entering
Data Analysis
active-duty military service.
Treatment conditions. Participants were randomly assigned Physiological reactivity was operationalized in two ways: the
to PE, VRE, or a minimal-attention waitlist control (WL). The two change in GSR (in microsimiens) from baseline to the trauma
active treatments (PE and VRE) involved 10 weekly or twice- recall (GSR-rise) and the change in GSR from trauma recall to rest
weekly (to accommodate soldiers’ military schedules; Reger et al., (GSR-recovery). GSR-rise was calculated by subtracting the av-
2016) 90- to 120-min psychotherapy sessions. Both PE and VRE erage GSR during baseline from the average GSR during trauma
were delivered in accordance with the PE manual (Foa et al., recall (larger numbers indicating greater GSR activity during
2007), with the exception of VRE participants’ use of the virtual trauma recall relative to baseline). GSR-recovery was calculated
reality system during exposure to the trauma memory. by subtracting the average GSR during recall from the average
Prolonged exposure. PE is a trauma-focused, evidence-based GSR during rest (negative values indicating a drop in GSR from
psychotherapy that involves psychoeducation on PTSD and com- recall to rest).
760 KATZ ET AL.

Mixed-effects regression models were used to examine the relationships. Specifically, positive correlations were found be-
changes in GSR-rise and recovery over the course of treatment. In tween change in CAPS-weekly score and change in GSR-rise from
response to attrition throughout the study, maximum likelihood Time 1 to Time 3, r ⫽ .243, p ⫽ .049 as well as change in general
estimates were used to account for missing data. Omnibus tests anxiety symptoms (BAITime 3 ⫺ BAITime 1) and change in GSR-
examining the overall effects of assessment (i.e., time), treatment rise from Time 1 to Time 3 (r ⫽ .272, p ⫽ .028; see Figures 2 and
group, and assessment-by-treatment interaction were conducted, 3 in online supplemental materials). No significant relationship
and specific hypotheses were tested with linear contrasts. The first was observed between change in GSR-rise and change in depres-
contrast used WL as a reference group to test whether the slopes of sive symptoms.
GSR reactivity of the two exposure treatments (i.e., PE and VRE)
each differed from that of WL. A second contrast compared the Discussion
slopes of GSR reactivity over time in VRE and PE. Two models
were run to examine the effects of treatment on GSR-rise and Our findings show that physiological reactivity to trauma-
GSR-recovery. Each model included linear effects of assessment, related cues, as measured by GSR-rise, reduced over time among
treatment group, and treatment group-by-assessment interaction as individuals in trauma-focused treatments as well as those assigned
fixed effects; intercept was included as a random effect. Given the to a waitlist control condition. These changes were most evident in
association between GSR and ambient temperature (Boucsein et the VRE group, as individuals in VRE experienced a significantly
al., 2012; Dawson et al., 2016), room temperature was included as greater decrease over time compared to those in WL. The visual
a covariate. Pearson correlation coefficients were used to examine presentation of the slopes for each group over time were in the
the relationship between changes in psychophysiology and expected direction, and favor VRE, though the difference in slopes
changes in PTSD symptoms. between VRE and PE was not statistically significant. Interest-
ingly, PE participants’ trajectories did not differ significantly from
the WL group, either. Further, across the entire sample, changes in
Results
physiological reactivity evidenced small-to-moderate correlations
Results from the ombinus mixed-effects regression model with changes in general anxiety and PTSD symptoms. Our find-
(MRM; see Table 2) for GSR-rise indicated an overall downward ings did not show any effects of time or treatment condition on the
trend of assessment, meaning that across groups, the magnitude of second index of physiological reactivity, GSR-recovery.
GSR reactivity decreased over time F(1, 151.9) ⫽ 24.501, p ⬍ The lack of significant difference between VRE and PE in
.001. Results from the omnibus test also showed a significant changes in physiological reactivity is consistent with the primary
treatment-assessment interaction F(2, 152.3) ⫽ 3.898, p ⫽ .022. RCT results demonstrating no statistically significant difference in
Examination of the fixed and random effects (see Table 3) re- clinician-assessed PTSD symptoms between the two treatments at
vealed that the significant treatment-assessment interaction was posttreatment (Reger et al., 2016). However, the visual differences
driven by the VRE group. Specifically, the negative slope of the in the slopes between VRE and PE could offer preliminary support
GSR-rise over time for the VRE group was significantly steeper for the underlying rationale of VRE: that it presents more emo-
than that of the WL (b ⫽ ⫺.713, 95% CI [⫺1.220, ⫺.206]). tionally engaging representations of the trauma memory, thus
Although the negative slope of the GSR-rise over time for the PE yielding greater activation of the fear structure, in turn causing
group was also steeper than that of the WL, this difference was not greater reductions in physiological and emotional reactions to
statistically significant (b ⫽ ⫺.401, 95% CI [⫺.891, .090]). Ad- trauma (Robison-Andrew et al., 2014). It is possible the current
ditionally, the difference in slopes between VRE and PE was not study was simply underpowered to detect these differences. It
statistically significant (b ⫽ .312, 95% CI [⫺.193, .817]; See may be that differences between PE and VRE in psychophysi-
Figure 1). Results from MRMs examining GSR-recovery over ology only emerge at follow-up. It could also be that VRE
time between treatment groups did not yield any statistically might yield earlier treatment response, even if its overall effi-
significant effects (all ps ⬎ .350). cacy is equal to PE.
Pearson correlations examining the relationship between change in In conjunction with prior work (Reger et al., 2016), our results
PTSD symptoms (e.g., CAPS-weeklyTime 3 ⫺ CAPS-weeklyTime 1) present a more nuanced picture regarding outcomes when com-
and changes in physiological reactivity (i.e., GSR-riseTime 3 ⫺ paring PE to VRE. Although VRE appears to enhance reductions
GSR-riseTime 1) over the course of treatment showed positive in physiological responses to trauma relative to WL (while PE did

Table 2
Results of Omnibus Mixed-Effects Regression Models for GSR Rise and Recovery Over Time

Model Variable Numerator df Denominator df F p

Rise Assessment 1 152.254 24.762 <.001


Treatment 2 196.472 1.157 .317
Treatment ⴛ Assessment 2 152.174 3.931 .022
Recovery Assessment 1 162.294 .105 .746
Treatment 2 208.586 .375 .687
Treatment ⫻ Assessment 2 162.240 .544 .581
Note. df ⫽ degrees of freedom; GSR ⫽ galvanic skin response. Parameters in boldface indicate significant
fixed effects.
CHANGES IN GSR DURING VRE VERSUS PE FOR PTSD 761

Table 3
Estimates of Fixed and Random Effects for GSR Rise and Recovery Over Time by
Treatment Group

Model Variable B LCI UCI

Rise Intercept ⴚ fixed .870 .357 1.382


Assessment ⫺.142 ⫺.490 .205
Treatment (VRE v. WL) .533 ⫺.196 1.262
Treatment (PE v. WL) .410 ⫺.298 1.118
Treatment (PE v. VRE) ⫺.125 ⫺.837 .588
Treatment (VRE v. WL) ⴛ Assessment ⴚ.714 ⴚ1.220 ⴚ.208
Treatment (PE v. WL) Assessment ⫺.401 ⫺.892 .088
Treatment (PE v. VRE) Assessment .313 ⫺.192 .817
Intercept ⴚ random .502 .251 .998
Recovery Intercept ⴚ fixed .453 .013 .893
Assessment .027 ⫺.302 .355
Treatment (VRE v. WL) ⫺.210 ⫺.837 .417
Treatment (PE v. WL) ⫺.252 ⫺.857 .354
Treatment (PE v. VRE) ⫺.042 ⫺.652 .567
Treatment (VRE v. WL) Assessment ⫺.118 ⫺.595 .359
Treatment (PE v. WL) Assessment .133 ⫺.330 .597
Treatment (PE v. VRE) Assessment .251 ⫺.226 .728
Intercept ⫺ random .083 .006 1.158
Note. PE ⫽ prolonged exposure; VRE ⫽ virtual reality exposure; WL ⫽ waitlist control; GSR ⫽ galvanic skin
response. Parameters in boldface indicate significant fixed or random effects.

not), and changes in these processes are associated with symptom in superior treatment response (Parsons & Rizzo, 2008). If future
changes across all participants, these differences did not result in studies with larger samples find VRE superior to PE vis a vis
a clear advantage for VRE over PE in symptom reduction. These psychophysiology, this would lend support to the idea that VRE
findings are consistent with previous work showing that treatment works by increasing EE; however, at this time, the mechanism of
group differences in VRE trials are more pronounced for psycho- VRE (as compared to PE) is not clear. It is possible that EE is a
physiological, rather than clinical, measures (Maples-Keller et al., critical mechanism, but that subjective units of distress do not fully
2017; Rothbaum et al., 2014). Data from this trial also showed no capture this construct. Alternatively, it is possible that an unrelated
difference between VRE and PE in relation to average or peak and/or unmeasured mechanism is at play. It is, of course, also
subjective distress during exposures (Reger et al., 2019), which possible that these differences could be attributable to random
was originally proposed as the method by which VRE could result error, or that despite differences in point estimates for GSR, no

Figure 1. Changes in physiological reactivity throughout treatment interval by treatment condition. Results
showed that the slopes of VRE and WL groups over the course of treatment differed significantly. PE ⫽
prolonged exposure; VRE ⫽ virtual reality exposure; WL ⫽ waitlist control; GSR ⫽ galvanic skin response.
762 KATZ ET AL.

measurable benefits accrue to VRE in terms of symptom reduction Because VRE did not generate increased subjective distress
and improvement. during exposure or symptom change relative to PE (Reger et al.,
Identifying mechanisms of change for PE remains an important 2016, 2019), these results invite continued research pertaining to
goal of continued research, with one leading mechanism being broader questions about mechanisms underlying PE and VRE.
between-session habituation of fear responses to trauma-related Physiological responding is often considered an objective marker
cues (Cooper, Clifton, & Feeny, 2017). However, some studies of emotional arousal, and it appears there are divergent effects of
have shown that symptom change is not dependent on within- or VRE on subjective and objective treatment responses. The nature
between-session habituation to trauma-related cues (Bluett, Zoell- of psychophysiological changes in response to treatment clearly
ner, & Feeny, 2014). Our results suggest that while decreases in warrant further investigation. For example, it is possible that
physiological reactivity correlated somewhat with PTSD symptom psychophysiological changes represent early indicators of treat-
decreases, factors not fully captured by our physiological measure ment response, as studies have demonstrated that reductions in
may contribute to treatment benefit. One obvious and previously physiological response to trauma cues differentiate treatment re-
demonstrated factor is subjective discomfort during session (Reger sponders from nonresponders (Griffin, Resick, & Galovski, 2012;
et al., 2019). It is possible that a disconnect exists between phys- Robison-Andrew et al., 2014). One might also assess whether
iologically measured and subjectively experienced reactions to change in physiological responding is a necessary or sufficient
trauma, or that physiological reactivity is merely an indicator of condition of treatment benefits from PE. Given the heterogeneity
treatment change rather than a central mechanism. Additional of PTSD presentations (Galatzer-Levy & Bryant, 2013), factor
factors may further account for treatment changes, such as cogni- analytic studies might examine whether posttreatment physiologic
tive changes (Foa et al., 2004), out-of-session homework comple- changes are associated with particular PTSD symptoms. If repli-
tion (Keller, Zoellner, & Feeny, 2010), therapeutic alliance cated, these results could also have important clinical implications
(McLaughlin, Keller, Feeny, Youngstrom, & Zoellner, 2014), or for providers choosing a treatment approach. Future research could
even therapist effects (Laska, Smith, Wislocki, Minami, & examine whether different PE paradigms might better suit sub-
Wampold, 2013). groups of individuals with PTSD, such as those with primarily
It was somewhat surprising that no effects were observed for the hypervigilance, avoidance, or cognitive symptoms.
Overall, these results provide some support for hypothesized
GSR recovery variable, given findings from case studies of faster
mechanisms of VRE: participants in VRE had greater reductions in
GSR recovery from trauma cues post-VRE (Wood et al., 2007,
GSR over time relative to waitlist, and these changes were mod-
2008). It is possible that the rest phase was too brief to allow for
estly correlated with changes in anxiety and PTSD symptoms.
adequate recovery from the trauma recall. Another possibility is
With replication, this could provide support for proposed mecha-
that the instructions to use pleasant imagery to recover actually
nisms of benefits of VRE, though important questions remain
stimulated the sympathetic nervous system—rather than the para-
about whether VRE has clinical benefits above and beyond stan-
sympathetic nervous system—which perhaps confounded the ac-
dard PE.
tual recovery.
This study is not without limitations. First, we encountered a References
relatively high dropout rate, with only approximately 60% of
individuals in the VRE and PE groups completing all 10 sessions American Psychiatric Association. (2000). Diagnostic and Statistical Man-
of psychotherapy (Reger et al., 2016). Although the dropout rates ual of Mental Disorders (4th ed., text rev.). Washington, DC: Author.
American Psychiatric Association. (2013). Diagnostic and Statistical Man-
did not differ by treatment group (Reger et al., 2016) and missing
ual of Mental Disorders (5th ed.). Washington, DC: Author.
data were handled in a manner consistent with best practices (i.e., Bardhoshi, G., Duncan, K., & Erford, B. T. (2016). Psychometric meta-
maximum likelihood estimation; Graham, 2009), attrition may analysis of the English version of the Beck Anxiety Inventory. Journal
have adversely affected the results. Second, the smaller subsample of Counseling & Development, 94, 356 –373. http://dx.doi.org/10.1002/
attenuated statistical power. Third, participants were all active- jcad.12090
duty soldiers, and most were White males, which limits general- Beck, A. T., & Steer, R. A. (1990). Manual for the Beck Anxiety Inventory.
izability of our results. It has been shown that healthy military San Antonio, TX: Psychological Corporation.
personnel and civilians differ in their physiological responses to Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the Beck
Depression Inventory-II. San Antonio, TX: Psychological Corporation.
stressful scenarios (Parsons et al., 2009), though physiological
Blake, D. D., Weathers, F. W., Nagy, L. M., Kaloupek, D. G., Gusman,
stress reactions have not been compared in military and civilian F. D., Charney, D. S., & Keane, T. M. (1995). The development of a
populations with PTSD. Given the differential scores for clinically clinician-administered PTSD scale. Journal of Traumatic Stress, 8,
significant PTSD symptoms (McDonald & Calhoun, 2010), it is 75–90. http://dx.doi.org/10.1002/jts.2490080106
possible that even disproportionate GSRs seen in PTSD might Bluett, E. J., Zoellner, L. A., & Feeny, N. C. (2014). Does change in
differ between soldiers and the general population. Additionally, distress matter? Mechanisms of change in prolonged exposure for PTSD.
the waitlist period was five weeks, and completion of treatment Journal of Behavior Therapy and Experimental Psychiatry, 45, 97–104.
usually took longer. We elected to accept this limitation rather than http://dx.doi.org/10.1016/j.jbtep.2013.09.003
Boucsein, W., Fowles, D. C., Grimnes, S., Ben-Shakhar, G., Roth, W. T.,
ask soldiers with combat-related PTSD to wait longer for treat-
Dawson, M. E., . . . the Society for Psychophysiological Research Ad
ment. Finally, the trauma recall phase relied solely on soldiers’
Hoc Committee on Electrodermal Measures. (2012). Publication recom-
imaginations (without verbal recounting or audio elements). mendations for electrodermal measurements. Psychophysiology, 49,
Though this was an intentional methodological distinction between 1017–1034. http://dx.doi.org/10.1111/j.1469-8986.2012.01384.x
the assessment sessions and exposure treatment sessions, it may Carl, E., Stein, A. T., Levihn-Coon, A., Pogue, J. R., Rothbaum, B.,
have allowed for experiential avoidance. Emmelkamp, P., . . . Powers, M. B. (2019). Virtual reality exposure
CHANGES IN GSR DURING VRE VERSUS PE FOR PTSD 763

therapy for anxiety and related disorders: A meta-analysis of randomized McDonald, S. D., & Calhoun, P. S. (2010). The diagnostic accuracy of the
controlled trials. Journal of Anxiety Disorders, 61, 27–36. http://dx.doi PTSD checklist: A critical review. Clinical Psychology Review, 30,
.org/10.1016/j.janxdis.2018.08.003 976 –987. http://dx.doi.org/10.1016/j.cpr.2010.06.012
Cooper, A. A., Clifton, E. G., & Feeny, N. C. (2017). An empirical review McLaughlin, A. A., Keller, S. M., Feeny, N. C., Youngstrom, E. A., &
of potential mediators and mechanisms of prolonged exposure therapy. Zoellner, L. A. (2014). Patterns of therapeutic alliance: Rupture-repair
Clinical Psychology Review, 56, 106 –121. http://dx.doi.org/10.1016/j episodes in prolonged exposure for posttraumatic stress disorder. Jour-
.cpr.2017.07.003 nal of Consulting and Clinical Psychology, 82, 112–121. http://dx.doi
Dawson, M., Schell, A., & Filion, D. (2016). The electrodermal system. In .org/10.1037/a0034696
J. Cacioppo, L. Tassinary, & G. Berntson (Eds.), Handbook of psycho- Norrholm, S. D., Jovanovic, T., Gerardi, M., Breazeale, K. G., Price, M.,
physiology (pp. 217–243). Cambridge, England: Cambridge University Davis, M., . . . Rothbaum, B. O. (2016). Baseline psychophysiological
Press. http://dx.doi.org/10.1017/9781107415782.010 and cortisol reactivity as a predictor of PTSD treatment outcome in
Difede, J., Cukor, J., Wyka, K., Olden, M., Hoffman, H., Lee, F. S., & virtual reality exposure therapy. Behaviour Research and Therapy, 82,
Altemus, M. (2014). D-cycloserine augmentation of exposure therapy 28 –37. http://dx.doi.org/10.1016/j.brat.2016.05.002
for post-traumatic stress disorder: A pilot randomized clinical trial. O’Kearney, R., & Parry, L. (2014). Comparative physiological reactivity
Neuropsychopharmacology, 39, 1052–1058. http://dx.doi.org/10.1038/ during script-driven recall in depression and posttraumatic stress disor-
npp.2013.317 der. Journal of Abnormal Psychology, 123, 523–532. http://dx.doi.org/
Erford, B. T., Johnson, E., & Bardoshi, G. (2016). Meta-analysis of the 10.1037/a0037326
English version of the Beck Depression Inventory–2nd ed. Measurement Orr, S. P., Metzger, L. J., & Pitman, R. K. (2002). Psychophysiology of
and Evaluation in Counseling and Development, 49, 3–33. http://dx.doi post-traumatic stress disorder. Psychiatric Clinics of North America, 25,
.org/10.1177/0748175615596783 271–293. http://dx.doi.org/10.1016/S0193-953X(01)00007-7
Foa, E. B., Hembree, E. A., & Rothbaum, B. O. (2007). Prolonged Parsons, T. D., Courtney, C., Cosand, L., Iyer, A., Rizzo, A. A., & Oie, K.
exposure therapy for PTSD: Emotional processing of traumatic experi- (2009). Assessment of psychophysiological differences of west point
ences, therapist guide. New York, NY: Oxford University Press, Inc. cadets and civilian controls immersed within a virtual environment. In
http://dx.doi.org/10.1093/med:psych/9780195308501.001.0001 D. D. Schmorrow, I. V. Estabrooke & M. Grootjen (Eds.), International
Foa, E. B., & Kozak, M. J. (1986). Emotional processing of fear: Exposure conference on foundations of augmented cognition (pp. 514 –523).
to corrective information. Psychological Bulletin, 99, 20 –35. http://dx Springer, Berlin, Heidelberg.
.doi.org/10.1037/0033-2909.99.1.20 Parsons, T. D., & Rizzo, A. A. (2008). Affective outcomes of virtual reality
Foa, E. B., & Rauch, S. A. (2004). Cognitive changes during prolonged exposure therapy for anxiety and specific phobias: A meta-analysis.
exposure versus prolonged exposure plus cognitive restructuring in Journal of Behavior Therapy and Experimental Psychiatry, 39, 250 –
female assault survivors with posttraumatic stress disorder. Journal of 261. http://dx.doi.org/10.1016/j.jbtep.2007.07.007
Consulting and Clinical Psychology, 72, 879 – 884. http://dx.doi.org/10 Pitman, R. K., Rasmusson, A. M., Koenen, K. C., Shin, L. M., Orr, S. P.,
.1037/0022-006X.72.5.879 Gilbertson, M. W., . . . Liberzon, I. (2012). Biological studies of
Galatzer-Levy, I. R., & Bryant, R. A. (2013). 636,120 ways to have post-traumatic stress disorder. Nature Reviews Neuroscience, 13, 769 –
posttraumatic stress disorder. Perspectives on Psychological Science, 8, 787. http://dx.doi.org/10.1038/nrn3339
651– 662. http://dx.doi.org/10.1177/1745691613504115 Pole, N. (2007). The psychophysiology of posttraumatic stress disorder: A
Graham, J. W. (2009). Missing data analysis: Making it work in the real meta-analysis. Psychological Bulletin, 133, 725–746. http://dx.doi.org/
world. Annual Review of Psychology, 60, 549 –576. http://dx.doi.org/10 10.1037/0033-2909.133.5.725
.1146/annurev.psych.58.110405.085530 Reger, G. M., Holloway, K. M., Candy, C., Rothbaum, B. O., Difede, J.,
Griffin, M. G., Resick, P. A., & Galovski, T. E. (2012). Does physiologic Rizzo, A. A., & Gahm, G. A. (2011). Effectiveness of virtual reality
response to loud tones change following cognitive-behavioral treatment exposure therapy for active duty soldiers in a military mental health
for posttraumatic stress disorder? Journal of Traumatic Stress, 25, clinic. Journal of Traumatic Stress, 24, 93–96. http://dx.doi.org/10
25–32. http://dx.doi.org/10.1002/jts.21667 .1002/jts.20574
Grillon, C. (2002). Startle reactivity and anxiety disorders: Aversive con- Reger, G. M., Koenen-Woods, P., Zetocha, K., Smolenski, D. J., Holloway,
ditioning, context, and neurobiology. Biological Psychiatry, 52, 958 – K. M., . . . Gahm, G. A. (2016). Randomized controlled trial of pro-
975. http://dx.doi.org/10.1016/S0006-3223(02)01665-7 longed exposure using imaginal exposure vs. virtual reality exposure in
Jaycox, L. H., Foa, E. B., & Morral, A. R. (1998). Influence of emotional active duty soldiers with deployment-related posttraumatic stress disor-
engagement and habituation on exposure therapy for PTSD. Journal of der (PTSD). Journal of Consulting and Clinical Psychology, 84, 946 –
Consulting and Clinical Psychology, 66, 185–192. http://dx.doi.org/10 959. http://dx.doi.org/10.1037/ccp0000134
.1037/0022-006X.66.1.185 Reger, G. M., Smolenski, D., Norr, A., Katz, A., Buck, B., & Rothbaum,
Keller, S. M., Zoellner, L. A., & Feeny, N. C. (2010). Understanding B. O. (2019). Does virtual reality increase emotional engagement during
factors associated with early therapeutic alliance in PTSD treatment: exposure for PTSD? Subjective distress during prolonged and virtual
Adherence, childhood sexual abuse history, and social support. Journal reality exposure therapy. Journal of Anxiety Disorders, 61, 75– 81.
of Consulting and Clinical Psychology, 78, 974 –979. http://dx.doi.org/ http://dx.doi.org/10.1016/j.janxdis.2018.06.001
10.1037/a0020758 Rizzo, A. S., Difede, J., Rothbaum, B. O., Reger, G., Spitalnick, J., Cukor,
Laska, K. M., Smith, T. L., Wislocki, A. P., Minami, T., & Wampold, B. E. J., & McLay, R. (2010). Development and early evaluation of the Virtual
(2013). Uniformity of evidence-based treatments in practice? Therapist Iraq/Afghanistan exposure therapy system for combat-related PTSD.
effects in the delivery of cognitive processing therapy for PTSD. Journal Annals of the New York Academy of Sciences, 1208, 114 –125. http://dx
of Counseling Psychology, 60, 31– 41. http://dx.doi.org/10.1037/ .doi.org/10.1111/j.1749-6632.2010.05755.x
a0031294 Rizzo, A., Reger, G., Perlman, K., Rothbaum, B., Difede, J., McLay, R.,
Maples-Keller, J. L., Price, M., Jovanovic, T., Norrholm, S. D., Odenat, L., . . . Sharkey, P. M. (2011). Virtual reality posttraumatic stress disorder
Post, L., . . . Rothbaum, B. O. (2017). Targeting memory reconsolidation (PTSD) exposure therapy results with active duty OIF/OEF service
to prevent the return of fear in patients with fear of flying. Depression members. International Journal on Disability and Human Development,
and Anxiety, 34, 610 – 620. http://dx.doi.org/10.1002/da.22626 10, 301–308. http://dx.doi.org/10.1515/IJDHD.2011.060
764 KATZ ET AL.

Robison-Andrew, E. J., Duval, E. R., Nelson, C. B., Echiverri-Cohen, A., Weathers, F. W., Keane, T. M., & Davidson, J. R. (2001). Clinician-
Giardino, N., Defever, A., . . . Rauch, S. A. (2014). Changes in trauma- administered PTSD scale: A review of the first ten years of research.
potentiated startle with treatment of posttraumatic stress disorder in Depression and Anxiety, 13, 132–156. http://dx.doi.org/10.1002/da.1029
combat Veterans. Journal of Anxiety Disorders, 28, 358 –362. http://dx Wood, D. P., Murphy, J., Center, K., McLay, R., Reeves, D., Pyne, J., . . .
.doi.org/10.1016/j.janxdis.2014.04.002 Wiederhold, B. K. (2007). Combat-related post-traumatic stress disor-
Rothbaum, B. O., Hodges, L. F., Ready, D., Graap, K., & Alarcon, R. D. der: A case report using virtual reality exposure therapy with physio-
(2001). Virtual reality exposure therapy for Vietnam veterans with logical monitoring. CyberPsychology & Behavior, 10, 309 –315. http://
posttraumatic stress disorder. The Journal of Clinical Psychiatry, 62, dx.doi.org/10.1089/cpb.2006.9951
617– 622. http://dx.doi.org/10.4088/JCP.v62n0808 Wood, D. P., Murphy, J. A., Center, K. B., Russ, C., McLay, R. N., Reeves,
Rothbaum, B. O., Price, M., Jovanovic, T., Norrholm, S. D., Gerardi, M.,
D., . . . Wiederhold, B. K. (2008). Combat related post traumatic stress
Dunlop, B., . . . Ressler, K. J. (2014). A randomized, double-blind
disorder: A multiple case report using virtual reality graded exposure
evaluation of D-cycloserine or alprazolam combined with virtual reality
therapy with physiological monitoring. Studies in Health Technology
exposure therapy for posttraumatic stress disorder in Iraq and Afghan-
and Informatics, 132, 556 –561.
istan War veterans. The American Journal of Psychiatry, 171, 640 – 648.
http://dx.doi.org/10.1176/appi.ajp.2014.13121625
Wangelin, B. C., & Tuerk, P. W. (2015). Taking the pulse of prolonged
exposure therapy: Physiological reactivity to trauma imagery as an Received September 13, 2019
objective measure of treatment response. Depression and Anxiety, 32, Revision received January 31, 2020
927–934. http://dx.doi.org/10.1002/da.22449 Accepted February 13, 2020 䡲

You might also like