Effects of threat context and cardiac sensitivity on fear responding to a 35% CO2
challenge: A test of the context-sensitivity panic vulnerability model
Michael J. Telcha,*, Jasper A.J. Smitsb, Matt Browna, Mandy Dementa, Mark B. Powersc, Hanjoo Leed,
aDepartment of Psychology, Laboratory for the Study of Anxiety Disorders, The University of Texas at Austin, 1 University Station, Mail Code A8000, Austin, TX 78712, USA
bSouthern Methodist University, Dallas, TX, USA
cUniversity of Pennsylvania, PA, USA
dThe University of Wisconsin at Milwaukee, WI, USA
a r t i c l e i n f o
Received 15 April 2009
Received in revised form
23 March 2010
Accepted 23 March 2010
Theories of panic
Context-sensitivity theory of panic
a b s t r a c t
The present study tested several predictions of a context-sensitivity panic vulnerability model empha-
sizing the interaction between threat context and threat sensitivities. Participants without a history of
panic (N ¼47) completed both global and domain-specific panic relevant sensitivity measures and were
then randomized to undergo a 35% CO2inhalation challenge in the presence or absence of a cardiac
defibrillator (threat context). As predicted by the model, cardiac sensitivity (but not trait anxiety or
anxiety sensitivity) potentiated the effects of the presence of the defibrillator on CO2fear responding.
Moreover, as predicted by the model, the observed potentiation effects of cardiac sensitivity on CO2fear
responding were mediated by participants’ threat appraisals connected to the presence of the defibril-
lator. Theoretical and clinical implications are discussed.
? 2010 Elsevier Ltd. All rights reserved.
The past two decades have witnessed a proliferation of psycho-
logical theories of panic disorder (Barlow, 1988; Beck, Emery, &
Greenberg, 1985; Bouton, Mineka, & Barlow, 2001; Chambless &
Gracely, 1989; Clark, 1986; McNally, 1990; Wolpe & Rowan, 1988).
Fear of anxiety sensations appears as a common thread throughout
manyof thesemodels. Evidencesupporting the link betweenfearof
anxiety sensations and panic comes from several lines of research
including (a) descriptive studies showing elevations in anxiety
sensitivity (AS) among panic disorder patients (Taylor, Koch, &
McNally, 1992; Telch, Jacquin, Smits, & Powers, 2003); (b) labora-
tory studies demonstrating heightened subjective fear and panic in
response to biological challenges such as CO2inhalation among
Gasperini, 1995); (c) studies showing that patients undergoing
cognitive behavioral treatment for panic display significant reduc-
et al., 1993); (d) evidence suggesting that panic disorder symptom
sensations (Smits, Powers, Cho, & Telch, 2004); and (e) prospective
studies suggesting that those scoring high on measures of anxiety
sensitivity have a significant increased risk for developing subse-
1997,1999), panic symptoms (Grant, Beck, & Davila, 2007), or other
anxiety disorders (Schmidt & Zvolensky, 2007; Schmidt, Zvolensky,
& Maner, 2006).
Although the evidence linking anxiety sensitivity to panic is
substantial, some negative findings have emerged (e.g., Koszycki &
Bradwejn, 2001; Struzik, Vermani, Duffin, & Katzman, 2004) and
even among the positive findings, anxiety sensitivity explains only
a small proportion of the variance in panic attack occurrence. These
data highlight the importance of identifying additional causal
factors implicated in panic. Research examining contextual factors
during biological challenge has provided important data on the
psychology of panic. For example, Rapee, Mattick, and Murrell
(1986) observed greater anxious responding to 50% CO2/air chal-
lenge among panic disorder participants who received an explana-
tion for their symptoms compared to panic disorder participants
who did not receive such explanation. Similarly, panic patients
undergoing a CO2challenge without a safe person present respon-
ded with greater fear, and a greater number of catastrophic
* Corresponding author.
E-mail address: Telch@Austin.utexas.edu (M.J. Telch).
Contents lists available at ScienceDirect
Journal of Behavior Therapy and
journal homepage: www.elsevier.com/locate/jbtep
0005-7916/$ e see front matter ? 2010 Elsevier Ltd. All rights reserved.
J. Behav. Ther. & Exp. Psychiat. 41 (2010) 365e372
cognitions than panic patients who did have a safe person present
(Carter, Hollon, Carson, & Shelton, 1995). In addition to perceived
safety, perceived control has also shown to be predictive of
panicking in response to CO2. Panic disorder patients who were led
to believe that they could decrease the concentration of CO2during
relative to patients who were not offered this option (Sanderson,
Rapee, & Barlow, 1989). However, an attempt to replicate this
finding was not successful (Welkowitz, Papp, Martinez, Browne, &
Gorman, 1999). Finally, there is some evidence that attentional
focus to internal cues moderated the relationship between panic
Taken together, results suggest that psychological variables such
as perceived safety, perceived control, and perceived predictability
influence fearful responding to biological challenge among panic
these variablesfunction ascontextual riskfactors forheightened fear
responding to challenge or whether they become panicogenic only
after the emergence of panic disorder. Experiments manipulating
these potential threat-enhancing contextual factors in non-clinical
participants provide some evidence to suggest that they increase
emotional responding to challenge in participants with no history of
panic disorder. For example, the effects of perceived control on non-
clinical subjects’ emotional response to a 450 mg. Caffeine challenge
was examined through an experimental manipulation in which
a caffeine antidote pill placebo was made available to half of non-
clinicalparticipantswithinstructionsthattheycould ingest the pillif
their caffeine-induced sensations became too uncomfortable (Telch,
Silverman, & Schmidt, 1996). Results showed a significant interac-
tion between participants’ anxiety sensitivity and their perceived
control assignment. Specifically, those high in anxiety sensitivity
showed higher fear when they underwent the caffeine challenge
without the placebo caffeine antidote (no perceived control);
whereas low anxiety sensitive participants were not affected by the
perceived control manipulation.
Additional support for the interactive effects of contextual
factors and dispositional factors comes from a study manipulating
offset control during 20% CO2challenge (Zvolensky, Eifert, & Lejuez,
2001). As predicted, only those high in anxiety sensitivity showed
heightened fear in the no offset control condition. Finally, using an
instructional set manipulation, Telch, Harrington, Smits, and
Powers (2005) provided non-clinical participants scoring high or
low in anxiety sensitivity with instructions to expect either arousal
or relaxation during a single 35% CO2inhalation. Consistent with
the hypothesis that unexpected arousal would be perceived as
more threatening, those receiving relaxation instructions were
significantly more likely to panic in response to CO2relative to
those receiving arousal instructions. However, consistent with
earlier findings, this effect was observed only for those high in
Taken together these findings suggest that person and
contextual factors may interact to influence challenge-induced
vulnerability model to account for the often-observed interaction
between anxiety sensitivity and contextual factors in laboratory
studies of panic provocation. This formulation proposed that
contextual factors, which increase either the anticipated threat of
the challenge or the perceived threat of the consequent reactions
(somatic, cognitive, or affective) during the challenge will result in
greater fear responding. Unlike other psychological formulations
of panic that emphasize the threatening misinterpretation of
bodily sensations (Clark, 1986) or the enduring tendency to
perceive anxiety as threatening (McNally, 1990), the current
formulation places central importance on the interaction of
context and dispositional tendencies (sensitivities) in predicting
panic. It should be emphasized that this model is not a theoretical
formulation of panic disorder but rather an explanatory model for
the propensity to experience panic in situations that pose no
objective threat to the individual (i.e., false alarms). Given the high
prevalence of false alarms across the full spectrum of anxiety
disorders as well as the general population, the proposed panic
vulnerability model has applicability across a broad range of
anxiety disorders and non-clinical cases in which panic reactions
The model makes several specific predictions. First, it proposes
that certain dispositional tendencies potentiate the likelihood of
experiencing panic in certain contexts. The use of the term
potentiate here is meant to imply that the combined effects of the
specific sensitivityand the specific context are greater than the sum
of the individual contributions of the presence of the sensitivity
alone or the context alone. Context is defined here as a stimulus
that may be internal as in the case of a somatic cue (e.g., chest
tightness), threatening thought (I’m going to lose control), or
experienced emotion (e.g., anger); or external (e.g., being in
a densely crowded place with a small exit). The threat potentiating
dispositional sensitivities can be quite broad as in the case of trait
anxiety or anxiety sensitivity or may be more narrow or domain-
specific as in the case of cardiac or respiratory sensitivity.
The model also proposes that the strength of the panic poten-
tiation brought about by a specific dispositional sensitivity will be
directly related to the conceptual match between the dispositional
variable under investigation and the threat-relevant context
(context-sensitivity matching hypothesis). Indirect support for this
hypothesis comes from biological challenge studies showing that
more narrow-band sensitivities such as physical concerns (Carter,
Suchday, & Gore, 2001; Zinbarg et al, 2001) or suffocation
concerns (Eke & McNally, 1996; McNally & Eke, 1996; Shipherd,
Beck, & Ohtake, 1996) outperform anxiety sensitivity in predicting
fear response to provocations that elicit a strong physical/respira-
tory reaction i.e., CO2inhalation and voluntary hyperventilation.
A final prediction of the model is that dispositional tendencies
such as anxiety sensitivity or rejection sensitivity potentiate the
panicogenic effects of certain internal or external contextual cues
by increasing the likelihood that an objectively non-threatening
context will be perceived as threatening. For example, those dis-
playing high dissociation sensitivity are more likely to perceive the
effects of hyperventilation or marijuana ingestion to be threatening
and consequently panic. Similarly, possessing high vestibular
sensitivity may contribute to panic when confronting a high plat-
form due to an exaggerated concern of losing one’s balance.
The aim of the presentexperiment was to perform a preliminary
test of the aforementioned context-sensitivity panic vulnerability
theory by experimentally examining the singular and interactive
effects of a putative threat-enhancing contextual cue (i.e., presence
or absence of a cardiac defibrillator) and several putative threat-
enhancing dispositional variables (i.e., trait anxiety, anxiety sensi-
tivity, respiratory sensitivity and cardiac sensitivity) among young
adults undergoing a 35% CO2challenge.
Our rationale for using a cardiac defibrillator to manipulate
threat context was two-fold. First, cardiac concerns are extremely
common among panic patients (Sheehan, 1983), hence it seemed
appropriate to use a contextual manipulation that was directly
related to the threat-relevant concerns of panic attack sufferers.
Second, because the presence of a cardiac defibrillator is not
inherently threatening, any fear-enhancing effects are more easily
attributable to how participants appraise its presence. This was an
important feature because one of the major assumptions of the
proposed theory is that dispositional factors such as trait anxiety
and cardiac sensitivity potentiate the panicogenic effects of
M.J. Telch et al. / J. Behav. Ther. & Exp. Psychiat. 41 (2010) 365e372
contextual cues by increasing the likelihood that the cue will be
perceived as threatening.
Three specific predictions from the theory were tested. The
potentiation hypothesis was tested by examining whether the
interaction of threat context (presence or absence of the defibril-
lator) by cardiac sensitivity predicted significant variance in
subjective fear and probability of panic after controlling for the
main effects of both threat context and sensitivity. The sensitivity-
context matching hypothesis was tested by examining the relative
potentiation strength of each of the dispositional sensitivities. It
was hypothesized that the greatest potentiationwould be observed
for cardiac sensitivity because of its closer conceptual match with
the manipulated threat context (i.e., presence of the defibrillator).
Lastly, we tested the threat appraisal mediation hypothesis by
examining whether any observed potentiation effects were medi-
ated by the perceived threat associated with the presence of the
Participants (N¼47) were 22 male and 25 female undergrad-
uate students enrolled in an introduction to psychology course at
the University of Texas at Austin. They ranged in age from 18 to 27
with a mean of 19.03 (SD¼1.45). Participants were selected from
a large sample (N¼1940) undergoing screening, and received
course credit for their participation. The following exclusion
criteria were used: (a) history of panic attacks; (b) history of
medical conditions that could be aggravated by inhalation of CO2
cardiac failure), respiratory disorders (e.g., asthma, lung fibrosis),
high blood pressure, epilepsy, stroke or seizures; (c) use of
psychotropic medication during the past two weeks; and (d)
pregnant or lactating. Of the 63 participants who volunteered to
participate in the experiment, 13 were excluded because of
a history of panic attacks, and 3 were excluded because of
a history of asthma.
Participants were randomly assigned to receive a single vital
capacity inhalation of 35% CO2and 65% O2under one of two threat
context conditions (i.e., presence or absence of a cardiac defibril-
lator). Measures of trait anxiety, anxiety sensitivity, and depres-
sion were included in thedesign as continuous putative
fear-enhancing dispositional variables. Indices of fear responding
were collected immediately before and after completing the CO2
1.3.1. Anxiety sensitivity
The Anxiety Sensitivity Index-Revised (ASI-R; Taylor & Cox,
1998) consists of 36 items and was constructed as an extension of
the original ASI (Peterson & Reiss, 1987). Respondents are pre-
sented with statements expressing concerns about possible nega-
tive consequences of anxiety such as “It scares me when my hear
beats rapidly.” For each statement, respondents rate each item on
a Likert scale ranging from very little (0) to very much (4). The ASI
total score is computed by summing responses across the 36 items.
Initial factor analytic research on the ASI-R using a clinical sample
(Taylor & Cox, 1998) yielded 4 subscales: (a) Fear of respiratory
symptoms,(b) Fearof publiclyobservable anxietyreactions, (c) Fear
of cardiovascular symptoms, and (d) Fear of cognitive dyscontrol.
However, more recent factor analytic findings from a large cross-
cultural non-clinical sample (Zvolensky et al., 2003) revealed only
two dimensions: (a) Fear of somatic symptoms; and (b) Social/
1.3.2. Cardiac and respiratory sensitivity
The existing factor analytic studies of the ASI-R did not
encompass a subscale that distinctively taps cardiac sensitivity.1
This was essential for the present study given the nature of our
threat context manipulation (presence or absence of cardiac
defibrillator) aimed at examining the context-specificity hypoth-
esis. Thus,we newly constructed
subscales of the ASI-R based on data obtained from an indepen-
dent sample of undergraduate students (N¼1350). First, two of
the authors (MJT and JAJS) independently chose 9 ASI-R items
tapping cardiac or respiratory concerns on the basis of their face
validity (i.e., #1e4, and 8 for respiratory concerns, and 7, 22, 24, 27
for cardiac concerns). Inter-rater reliability between the two raters
for this initial classification was 100%. The first half of the data
were subjected to exploratory factor analysis based on maximum
likelihood extraction and oblique rotation (delta ¼0). This resulted
in a two-factor structure explaining 59.2% of the variance. Item #7
and #8 were excluded due to failing to exhibit differential factor
loadings. Thus, the final items included three items for the Cardiac
sensitivity subscale (i.e., #22, 24, 27) and four items for the
Respiratory subscale (#1, 2, 3, 4). Another exploratory factor
analysis using the 7 final items revealed that the two factors
explained 62.7% of the variance with a clear bifurcation of factor
loadings between the two factors. For the cross-validation of the
two-factor substructure, the other half of the data was subjected
to confirmatory factor analysis using Analysis of Moment Struc-
tures, Version 4.0 (AMOS; Arbuckle & Wothke, 1999). Results
demonstrated a good fit of the model: GFI¼0.94, NFI¼0.95,
IFI¼0.95, and CFI¼0.95. Internal consistency (Cronbach Alpha)
was 0.87 and 0.85 for the Respiratory and Cardiac subscales,
respectively. The Respiratory and Cardiac subscales displayed
a correlation of 0.73 and 0.70 with the total scores of the ASI-R,
respectively. The correlation between the cardiac and respiratory
subscales was 0.35.
1.3.3. Trait anxiety
Trait anxiety was assessed using the State Trait Anxiety Inven-
tory-Trait subscale (Spielberger, Gorsuch, & Lushene, 1970). The
subscale is composed of 20-items designed to assess trait anxiety.
The items are scored on a four-point Likert scale ranging from Not
At All (1) to Very Much So (4). Both scales of the STAI have shown
adequate psychometric properties (Knight, Waal-Manning, &
1.4. Emotional response to CO2inhalation
1.4.1. The Acute Panic Inventory (API; Liebowitz, Gorman, Fyer,
Dillon, & Klein, 1984)
The API is a widely used self-report instrument for assessing
physical and affective reactions to biological challenges. Respon-
1We considered using the Cardiac Anxiety Questionnaire (CAQ) e a well-
established measure of heart-focused anxiety (Eifert et al., 2000) to index cardiac
sensitivity. This measure was developed on a clinical population of cardiac patients
who had undergone angiography; it has been studied extensively with this clinical
populations (Marker, Carmin, & Ownby, 2008; Zvolensky, Eifert, Feldner, & Leen-
Feldner, 2003). However, as noted by Eifert et al. (2000) less than 3% of college
students show elevations on the CAQ, which precluded its use in the current study.
M.J. Telch et al. / J. Behav. Ther. & Exp. Psychiat. 41 (2010) 365e372
lightheadedness, going crazy, losing control) on a 0 (none) to 3
(extreme) Likert scale. Peak fear during the challenge is assessed
using a 0 (not disturbed at all) to 100 (the worst imaginable
experience) scale. The API also includes a question to assess the
occurrence of panic (yes/no) in response to the challenge.
1.5. Appraisal of danger and safety associated with the presence of
1.5.1. The Defibrillator Appraisal Questionnaire (DAQ)
The DAQ is a 4-item author-constructed scale designed to index
participants’ perceived dangerand perceived safetyassociated with
the presence of the defibrillator. Participants rated the following
items on a 5-point (0¼not at all; 4¼completely) Likert scale: (1)
Because of the defibrillator, I felt that I was in a potentially harmful
or dangerous situation; (2) I felt safe knowing that should I need
assistance, there was an available method of medical aid; (3) The
presence of the defibrillator made me question the safety of the
challenge; and (4) The presence of the defibrillator reassured me
that nothing bad was going to happen to me. Scores on items 1 and
3 were averaged to provide an indexof perceived danger associated
with the presence of the defibrillator. Similarly, items 2 and 4 were
averaged to provide an index of perceived safety associated with
the presence of the defibrillator. Examination of the scale used in
the current study revealed a coefficient alpha of 0.95 and 0.69 for
the two subscales, respectively. Only participants randomized to
the defibrillator condition completed this scale.
1.6. Procedures common to both threat context conditions
Following informed consent, participants completed a 9-item
medical history questionnaire designed to identify participants
with medical or medication contraindications to inhalation of CO2
and panic attack history. Participants then completed the ASI-R and
the STAI-Tand were randomlyassigned to either the Defibrillatoror
No Defibrillator (i.e., Control). All participants then watched a 2-
min video clip demonstrating the inhalation procedure along with
the following CO2challenge instructions:
“For this experiment, you will be administered a carbon dioxide
inhalation challenge. Inhaling carbon dioxide is a widely used
research procedure for studying anxiety. The gas mixture that
you will be inhaling consists of 35% carbon dioxide and 65%
oxygen. Both of these components are in the air that we breathe.
However, during the inhalation procedure you are likely to
experience some unusual or odd sensations, such as faintness
The procedure will consist of several steps. First, we will have
you complete several forms to indicate how you are feeling right
now. Next, we will ask you to exhale fully so that all the air is out
of your lungs and have you breathe in a full, complete breath of
the gas mixture. While breathing in the gas it is important that
you hold your nose to insure that only the proper mixture of gas
is inhaled. You will hold the gas mixture in your lungs for 5 full
secs., after which you will exhale the gas. Next, you will
complete a form asking about your reactions to the inhalation of
the gas. Let’s have you watch an actual example of a person
undergoing the CO2inhalation procedure.”
Following instructions and demonstration, the procedure was
practiced with the experimenter using regular room air. Partici-
pants then completed baseline questionnaires.
1.6.1. Defibrillator threat context condition
cart was positioned 2 m directly in front of the participant and the
Experimenter provided the following comment: “Oh, I forgot one
thing; we need this defibrillator here in the room just in case of an
emergency.” Participants then completed the CO2 challenge.
Following inhalation, they completed the API and the DAQ.
1.6.2. Control threat context condition
Participants in the control condition underwent the challenge
without a defibrillator present, after which they completed the API.
2.1. Pre-manipulation equivalence of the groups
Scores on the baseline measures of ASI, STAI, and BDI were
within the range of a non-clinical population (see Table 1). There
were no significant differences on any of the measures at baseline,
indicating that randomization had resulted in comparable groups.
2.2. Test of the potentiation hypothesis
A series of hierarchical regression analyses were conducted to
examine the interactive effects between threat context and dispo-
sitional variables in predicting peak fear and panic occurrence in
response to CO2inhalation. Support for the potentiation hypothesis
is demonstrated if the disposition by threat context interaction
significantly predicts CO2fear/panic after controlling for the main
effects of both the sensitivity and threat context variable. Separate
hierarchical regressionanalyses wereperformed foreach of the five
dispositional sensitivity variables (i.e., cardiac sensitivity, respira-
tory sensitivity, anxiety sensitivity, trait anxiety, and depression).
All continuous predictors were centered to reduce potential mul-
ticollinearity (see Aiken & West,1991). Primary outcome measures
were CO2peak fear (0e100) and CO2panic (yes or no).
In Step 1 of the hierarchical regression analysis, Cardiac sensi-
tivity and Threat context (i.e., defibrillator condition) combined
accounted for 25.0% of the variance in CO2Fear, R2¼0.25, F (2,
44)¼7.32, p<0.005. In Step 2, the interaction term of Cardiac
sensitivity by Threat context explained an additional 11.8% of the
variance (R2change¼0.12, F (1, 43)¼8.03, p<0.01). Once the
interaction term was entered into the model (b¼0.41, t¼2.83,
p<0.01), neither Cardiac sensitivity nor Threat context signifi-
cantly predicted CO2fear. Follow-up simple slope analyses revealed
that Cardiac sensitivity predicted CO2 fear in the defibrillator
Baseline measures of trait anxiety and anxiety sensitivity.
Note STAI-T¼State-Trait Anxiety Inventory-Trait Subscale Score; BDI¼Beck
Depression Inventory; ASI-R Total¼Anxiety Sensitivity Index-Revised Total Score;
ty¼Respiratory sensitivity subscale.
M.J. Telch et al. / J. Behav. Ther. & Exp. Psychiat. 41 (2010) 365e372
condition (b ¼7.97, t¼6.39, p<0.01) but not in the control
condition (b¼2.13, t¼1.52, p>0.10; See Fig. 1).
Identical hierarchical regression analyses were performed
separately foreach of the other sensitivity variables (i.e., respiratory
sensitivity, anxiety sensitivity, trait anxiety, and depression). None
of these dispositional variables interacted significantly with threat
context in predicting CO2fear (See Fig. 2).
A series of hierarchical logistic regression analyses were con-
ducted on the binary outcome measure of CO2panic similar to the
hierarchical linear regression analyses performed for CO2 fear.
Cardiac sensitivity and Threat context were entered into Step 1 and
Cardiac sensitivity by Threat context interaction was entered into
Step 2. The overall model was significant, ?2 Log likelihood¼36.10,
c2¼15.04, p<0.005. Nagelkerke R2, which is a measure of strength
of association similar to R2in ordinary least squares regression
(Nagelkerke, 1991), increased from 0.15 to 0.41 after entering the
interaction of Cardiac sensitivity by Threat context. The Threat
Wald X2(1)¼3.81, p¼0.05; Odds ratio¼2.37 (95% CI¼1.0, 5.63).
However, once the interaction term was entered into the model,
Cardiac sensitivity alone was not a significant predictor, Wald X2
threat context alone was still significantly predictive, Wald X2(1)¼
5.14, p<0.05; Odds ratio¼8.72 (95% CI¼1.34, 56.73). Follow-up
simple effects analyses were conducted to assess the effects of
cardiac sensitivity on CO2Panic for participants in the defibrillator
vs. control condition. Consistent with prediction, Cardiac sensitivity
(1)¼4.81, p<0.05, Odds ratio¼2.35 (95% CI¼1.10, 5.05).
2.3. Test of the context-sensitivity matching hypothesis
To review, the Context-Sensitivity Matching Hypothesis asserts
that the strength of the panic potentiation brought about by
a specific dispositional sensitivity will be directly related to the
conceptual match between the sensitivity variable under investi-
gation and the threat-relevant context. We tested this formulation
by examining the magnitude of potentiation (i.e., R2change for the
interaction term) across the five dispositional sensitivity variables.
demonstrated if cardiac sensitivity leads to a significantly greater
fear potentiation effect (i.e., greater R2change for the sensitivi-
ty?context interaction) relative to the sensitivity?context inter-
action terms for the less context-relevant dispositional sensitivities
(e.g., trait anxiety). A series of five hierarchical regression analyses
(one for each sensitivity variable) were performed. In each analysis,
the effects of threat context (dummycoded 0 or 1) and the centered
sensitivity variable were entered in Step 1. The threat con-
text?sensitivity interaction term was entered in Step 2. Peak fear
in response to the CO2 challenge served as the outcome in all
Hypothesis, the cardiac sensitivity?threat context interaction
term was the only sensitivity?threat context interaction term to
significantly contribute to the variance in peak fear (see Fig. 2).
The approach for testing the context-sensitivity matching
hypothesis for CO2panic was identical to that used for CO2fear. A
series of five hierarchical logistic regression analyses (one for each
sensitivity variable) were performed. Fig. 3 presents the results of
these analyses. Consistent with prediction, of the five threat con-
text?cardiac sensitivity interaction was significant Wald X2(1)¼
3.81, p¼0.05; Odds ratio¼2.37 (95% CI¼1.0, 5.63). Consistent
with the context-sensitivity matching hypothesis, the magnitude of
the potentiation effect (i.e., 26% of the variance in CO2panic was
accounted for by the interaction of cardiac sensitivity and threat
context) was markedly higher than that observed for the other less
conceptually matched sensitivities (see Fig. 3).
Follow-up simple effects analyses comparing the percentage of
those reporting CO2panic for the defibrillator vs. no defibrillator
conditions as a function of cardiac sensitivity (high vs. low) illus-
trates the marked potentiation effect of cardiac sensitivity on CO2
panic in the presence of the defibrillator (see Fig. 4). Specifically,
58% of the high cardiac anxiety sensitive participants reported
panic during the CO2inhalation in the presence of the defibrillator
relative to only 10% in its absence [Wald X2(1)¼4.81, p<0.05,
Odds ratio¼2.35 (95% CI¼1.10, 5.05)].
Fig. 1. The interactive effect of threat context and cardiac sensitivity on CO2fear.
Fig. 2. Proportion of variance explained in CO2Fear from a series of hierarchical regression analyses.
M.J. Telch et al. / J. Behav. Ther. & Exp. Psychiat. 41 (2010) 365e372
2.4. Appraisal mediation hypothesis
The appraisal mediation hypothesis was tested according to the
analytic steps outlined by Baron and Kenny (1986). However,
because data on the putative mediator e DAQ-Perceived Danger
were collected only for participants randomized to the defibrillator
condition, our test for mediation focused on whether the rela-
tionship between cardiac sensitivity and CO2fear/panic would be
mediated by specific threat appraisals associated with the presence
of the defibrillator. In Step1, the cardiac sensitivityexplained 64% of
the variance in CO2fear, R2¼0.640, F (1, 23)¼40.85, p<0.001.
Likewise, the probability of CO2Panic was also significantly asso-
ciated with Cardiac sensitivity, Wald X2(1)¼4.81, p<0.05,
OR¼1.46 (95% CI, 1.10, 5.05). These data confirm that the first
condition for mediation was met.
In Step 2, we tested the association between Cardiac Sensitivity
and the proposed mediator (i.e., the a path). Cardiac sensitivity was
R2¼0.40, F (1, 23)¼15.62, p<0.005. These data confirm that the
second condition for mediation was met.
In Step 3, we tested the relationship between the proposed
mediatorandfearful responding toCO2. DAQ-Perceived Danger was
significantly associated with CO2Fear, R2¼0.316, F (1, 23)¼10.61,
p<0.005. However, the relationship with CO2 Panic was only
marginally significant, Wald X2(1)¼3.29, p¼0.07 OR¼1.49 (95%
CI, 0.97, 2.29). These data confirm that the thirdnecessarycondition
for mediation was met for CO2Fear but not for CO2Panic. Accord-
ingly, we performed subsequent analyses only for CO2fear.
In Step 4, we tested the relationship between Cardiac Sensitivity
and CO2Fear aftercontrolling for the proposed mediator. According
to Baron and Kenny (1986), evidence for full mediation exists when
the relationship between the IV and DV is no longersignificant after
controlling for the effects of the mediator, whereas evidence for
partial mediation exists when the relationship between IV and DV
is significantly attenuated (but still significant) after controlling for
mediator effects. The strength of the relationship between Cardiac
Sensitivity and CO2Fear was reduced considerably after controlling
for DAQ-Perceived Danger Scores. This reduction was consistent
with partial mediation, since the association between Cardiac
Sensitivityand CO2Fear was still significant after controlling for the
mediator, b¼0.74, t¼4.51, p¼<0.001.
We employed the distribution of products test (MacKinnon,
Fairchild, & Fritz, 2007; MacKinnon, Lockwood, Hoffman, West, &
Sheets, 2002) to test the significance of the mediated pathway.
The distribution of products test involves multiplying the regres-
sion coefficients of the two segments of the mediated pathway (i.e.,
a?b) and calculating the 95% confidence interval (CI) for this
product (95% CI). Because the 95% CI (0.057, 0.527) did not include
0, we can conclude that the mediated pathway was statistically
significant (MacKinnon et al., 2004).
Finally, we calculated the proportion mediated (PM; Shrout &
Bolger, 2002) as an index of the effect size of the mediated
pathway. PM is the proportion of the total effect of the independent
variable on the dependent variable (i.e., the c path) mediated by the
mediator and is calculated by the formula PM ¼(a?b)/c. The
proportion of the effect of cardiac sensitivity on CO2Fear mediated
by perceived danger was 40.95%.
The central aim of the present study was to test predictions
derived from the context-sensitivity vulnerability model of panic
that posits a central role for the interaction of threat-enhancing
context and threat-enhancing dispositional sensitivities in the
pathogenicity of panic. Our approach was to have non-clinical
participants with varying levels of trait anxiety, anxiety sensitivity
(two commonly-studied dispositional sensitivity measures in the
experimental study of panic), respiratory sensitivity and cardiac
sensitivity (more domain-specific sensitivity variables) undergo
Fig. 3. Proportion of variance explained in CO2panic from a series of hierarchical logistic regression analyses.
Low Cardiac Sensitivity
High Cardiac Sensitivity
% Participants Reporting Panic
Fig. 4. The interactive effect of threat context and cardiac sensitivity on CO2panic.
M.J. Telch et al. / J. Behav. Ther. & Exp. Psychiat. 41 (2010) 365e372
a 35% CO2challenge in the presence or absence of a threat-relevant
contextual cue (i.e., cardiac defibrillator).
The prediction that threat-relevant sensitivity variables would
potentiate the anxiogenic effects of threat-relevant contexts was
supported by the significant interaction observed between the
experimentally manipulated threat context (defibrillator present or
absent) and cardiac sensitivity. Participants scoring high in cardiac
sensitivity showed significantly higher subjective fear and proba-
bility of panic in response to CO2inhalation in the presence of the
cardiac defibrillator. In contrast, high cardiac sensitivity in the
absence of the defibrillator was not significantly associated with
higher subjective fear or probability of panic in response to CO2
challenge. These findings are consistent with previous biological
challenge investigations demonstrating a significant interaction
between anxiety sensitivity and manipulated threat contexts such
as the presence or absence of a caffeine antidote (Telch et al.,1996)
or instructional sets leading to the expectation of either challenge-
induced arousal or relaxation (Telch et al., 2005). Taken together,
these data argue against an additive model in which dispositional
and contextual factors each only independently contribute to the
prediction of fearful responding to challenge. Additional support
for this contention comes from analyses from the current experi-
ment demonstrating that neither cardiac sensitivity nor threat
context (defibrillator condition) predicted unique variance in fear
responding after statistically controlling for the effects of the
Examination of the other sensitivity variables provides some
evidence supporting the specificity of cardiac sensitivity as
a potentiator of fear responding to CO2in the presence of the
defibrillator. Specifically, cardiac sensitivity showed a significant
interaction with the defibrillator condition, while respiratory
sensitivity and anxiety sensitivity did not. These findings provide
preliminary support for the proposed formulation asserting that
the conceptual linkage between a threat-relevant sensitivity and
context will affect the magnitude of the potentiation effect. Our
previous attempt at testing this sensitivity-context matching
hypothesis using a dissociation challenge paradigm failed to
demonstrate specificity in that both dissociation sensitivity and
anxiety sensitivity potentiated fear responding to the challenge
equally (see Leonard, Telch, & Owen, 2000). Perhaps limitations of
our measure of dissociation sensitivity or the modest potency of
our dissociation challenge contributed tothe failure todemonstrate
specificity of the sensitivity variable.
Our findings may provide some clues as to the potential
mechanisms through which the presence of the cardiac defibril-
lator exerted its effect on the relationship between cardiac sensi-
tivity and fearful responding to CO2. Our mediational analyses
revealed that cardiac sensitivity was a strong predictor of partici-
pants’ ratings of perceived danger associated with the defibrillator.
In turn, perceived danger associated with the defibrillator signifi-
cantly predicted fearful responding to the CO2challenge. Finally,
the effects of the defibrillator’s presence on CO2subjective fear was
significantlyattenuated aftercontrolling for the effects of perceived
danger associated with the defibrillator. This finding provides
evidence in support of the third prediction derived from the theory,
namely asserting that dispositional factors such as cardiac sensi-
tivity potentiate the panicogenic effects of contextual cues by
increasing the likelihood that the context will be perceived as
It should be noted that for purposes of testing the proposed
model we have conceptualized both cardiac sensitivity and anxiety
sensitivity as continuous rather than categorical constructs.
However, there is some evidence (e.g., Bernstein et al., 2006, 2007;
Schmidt, Kotov, Lerew, Joiner, & Ialongo, 2005) suggesting that
anxiety sensitivity may be taxonic. To date, tests of the taxonic
hypothesis for AS have been mixed (see Broman-Fulks et al., 2008).
Conclusions regarding the applicability of the proposed context-
sensitivity model to sensitivities that are taxonic in nature await
Several limitations of the study deserve comment. First,
although the study of those who have not yet developed naturally
occurring false alarm panic reactions has the advantage of ruling
out that the putative vulnerability factor under investigation is
simply a consequence or concomitant of pre-existing panic, our
data only speak directly to predicting fear responding to a labora-
tory challenge. Prospective risk studies and randomized prevention
trials are needed to determine how the interaction of contextual
factors and dispositional sensitivities contribute to the onset of
naturally occurring panic reactions across the full spectrum of
anxiety disorders. Second, a more powerful test of the context-
sensitivity matching hypothesis would have been to also include
a non-cardiac threat context condition as part of the experimental
manipulation in order to demonstrate that elevated cardiac sensi-
tivity potentiates CO2 fear responding in the presence of the
cardiac-relevant contextual cue (i.e., defibrillator) but not in the
presence of a safety cue unrelated to cardiac concerns (e.g.,
epinephrine pen). Third, statistical power to detect modest effects
was limited due to our relatively small sample size.
Some clinical implications of our findings deserve mention.
First, our data suggest that there may be some therapeutic justifi-
cation for employing a domain-specific approach in the assessment
of threat-relevant sensitivities. Global dispositional measures such
as trait anxiety and anxiety sensitivity may lack sufficient speci-
ficity to aid the clinician in individually tailoring treatment strat-
egies. In contrast, the inclusion of assessment strategies that assist
the clinician in identifying patients’ specific threat-relevant sensi-
tivities such as cardiac sensitivity, respiratory sensitivity, rejection
sensitivity, heat sensitivity, fatigue sensitivity, to name a few e may
assist clinicians to more effectively tailor intervention strategies to
match peoples’ sensitivity profiles. A second treatment implication
of our findings comes from our preliminary demonstration that the
appraisal of threat connected to the presence of the defibrillator
accounted for the observed potentiation effect of cardiac sensitivity
on fear responding to CO2. These data speak to the potential
importance of including intervention elements that target patients’
negatively biased appraisals.
Taken together, our findings lend preliminary support for the
proposed context-sensitivity panic vulnerability model. Future
laboratory and prospective risk studies investigating additional
contextual factors and their interaction with both global and
domain-specific dispositional sensitivities may yield greater
understanding of false alarm fear reactions and their prevention.
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting inter-
actions. Thousand Oaks: Sage.
Arbuckle, J. L., & Wothke, W. (1999). Amos 4.0 user's guide. Chicago, IL: SPSS.
Barlow, D. H. (1988). Anxiety and its disorders: The nature and treatment of anxiety
and panic (1st ed.). New York: Guilford Press.
Baron, R. M., & Kenny, D. A. (1986). The moderatoremediator distinction in social
psychological research: conceptual, strategic, and statistical considerations.
Journal of Personality and Social Psychology, 51, 1173e1182.
Beck, A. T., Emery, G., & Greenberg, R. L. (1985). Anxiety disorders and phobias. A
cognitive perspective. USA: Harper Collins.
Bernstein, A., Zvolensky, M. J., Kotov, R., Arrindell, W. A., Taylor, S., Sandin, B., et al.
(2006). Taxonicity of anxiety sensitivity: a multi-national analysis. Journal of
Anxiety Disorders, 20, 1e22.
Bernstein, A., Zvolensky, M. J., Norton, P. J., Schmidt, N. B., Taylor, S., Forsyth, J. P.,
et al. (2007). Taxometric and factor analytic models of anxiety sensitivity:
integrating approaches to latent structural research. Psychological Assessment,
Bouton, M. E., Mineka, S., & Barlow, D. H. (2001). A modern learning theory
perspective on the etiology of panic disorder. Psychology Review, 108, 4e32.
M.J. Telch et al. / J. Behav. Ther. & Exp. Psychiat. 41 (2010) 365e372
Broman-Fulks, J. J., Green, B. A., Olatunji, B. O., Berman, M. E., Arnau, R. C., Download full-text
Deacon, B. J., et al. (2008). The latent structure of anxiety sensitivity e revisited.
Assessment, 15, 188e203.
Carter, M. M., Hollon, S. D., Carson, R., & Shelton, R. C. (1995). Effects of a safe person
on induced distress following a biological challenge in panic disorder with
agoraphobia. Journal of Abnormal Psychology, 104, 156e163.
Carter, M. M., Suchday, S., & Gore, K. L. (2001). The utility of the ASI factors in
predicting response to voluntary hyperventilation among nonclinical partici-
pants. Journal of Anxiety Disorders, 15(3), 217e230.
Chambless, D. L., & Gracely, E. J. (1989). Fear of fear and the anxiety disorders.
Cognitive Therapy & Research, 13, 9e20.
Clark, D. M. (1986). A cognitive approach to panic. Behavior Research and Therapy,
Eifert, G. H., Thompson, R. N., Zvolensky, M. J., Edwards, K., Haddad, J. H.,
Frazer, N. L., et al. (2000). The cardiac anxiety questionnaire: development and
preliminary validity. Behaviour Research and Therapy, 38, 1039e1053.
Eke, M., & McNally, R. J. (1996). Anxiety sensitivity, suffocation fear, trait anxiety,
and breath- holding duration as predictors of response to carbon dioxide
challenge. Behaviour Research and Therapy, 34, 603e607.
Gorman, J. M., Papp, L. A., Coplan, J. D., Martinez, J. M., Lennon, S., Goetz, R. R., et al.
(1994). Anxiogenic effects of CO2and hyperventilation in patients with panic.
American Journal of Psychiatry, 151, 547e553.
Grant, D. M., Beck, J. G., & Davila, J. (2007). Does anxiety sensitivity predict symp-
toms of panic, depression, and social anxiety? Behaviour Research and Therapy,
Knight, R. G., Waal-Manning, H. J., & Spears, G. F. (1983). Some norms and reliability
data for the State-Trait Anxiety Inventory and the Zung Self-Rating Depression
scale. British Journal of Clinical Psychology, 22, 245e249.
Koszycki, D., & Bradwejn, J. (2001). Anxiety sensitivity does not predict fearful
responding to 35% carbon dioxide in patients with panic disorder. Psychiatry
Research, 101, 137e143.
Leonard, K. N., Telch, M. J., & Owen, K. K. (2000). Fear response to dissociation
challenge. Anxiety, Stress & Coping: An International Journal, 13, 355e369.
Liebowitz, M. R., Gorman, J. M., Fyer, A. J., Dillon, D. J., & Klein, D. F. (1984). Effects of
naloxone on patients with panic attacks. American Journal of Psychiatry, 141,
MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. Annual
Review of Psychology, 58, 593e614.
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A
comparison of methods to test mediation and other intervening variable effects.
Psychological Methods, 7(1), 83e104.
MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the
indirect effect: Distribution of the product and resampling methods. Multi-
variate Behavioral Research, 39(1), 99e128.
Maller, R. G., & Reiss, S. (1992). Anxiety sensitivity in 1984 and panic attacks in 1987.
Journal of Anxiety Disorders, 6, 241e247.
Marker, C. D., Carmin, C. N., & Ownby, R. L. (2008). Cardiac anxiety in people with
and without coronary atherosclerosis. Depression and Anxiety, 25, 824e831.
McNally, R. J. (1990). Psychological approaches to panic disorder: a review.
Psychological Bulletin, 108, 403e419.
McNally, R. J., & Eke, M. (1996). Anxiety sensitivity, suffocation fear, and breath-
holding duration as predictors of response to carbon dioxide challenge. Journal
of Abnormal Psychology, 105, 146e149.
Nagelkerke, N. J. D. (1991). A note on a general definition of the coefficient of
determination. Biometrika, 78, 691e692.
Penava, S. J., Otto, M. W., Maki, K. M., & Pollack, M. H. (1998). Rate of improvement
during cognitive-behavioral group treatment for panic disorder. Behaviour
Research and Therapy, 36, 665e673.
Perna, G., Barbini, B., Cocchi, S., Bertani, & Gasperini, M. (1995). 35% CO2challenge in
panic and mood disorders. Journal of Affective Disorders, 33, 189e194.
Peterson, R. A., & Reiss, S. (1987). Anxiety sensitivity index. Palos Heights, IL: Inter-
national Diagnostic Systems.
Rapee, R., Mattick, R., & Murrell, E. (1986). Cognitive mediation in the affective
component of spontaneous panic attacks. Journal of Behavior Therapy & Exper-
imental Psychiatry, 17, 245e253.
Sanderson, W. C., Rapee, R. M., & Barlow, D. H. (1989). The influence of an illusion of
control on panic attacks induced via inhalation of 5.5% carbon dioxide-enriched
air. Archives of General Psychiatry, 46, 157e162.
Schmidt, N. B., Kotov, R., Lerew, D. R., Joiner, T. E., & Ialongo, N. S. (2005). Evaluating
latent discontinuity in cognitive vulnerability to panic: a taxometric investi-
gation. Cognitive Therapy and Research, 29, 673e690.
Schmidt, N. B., Lerew, D. R., & Jackson, R. J. (1997). The role of anxiety sensi-
tivity in the pathogenesis of panic: prospective evaluation of spontaneous
panic attacks during acute stress. Journal of Abnormal Psychology, 106,
Schmidt, N. B., Lerew, D. R., & Jackson, R. J. (1999). Prospective evaluation of anxiety
sensitivity in the pathogenesis of panic: replication and extension. Journal of
Abnormal Psychology, 108, 532e537.
Schmidt, N. B., & Trakowski, J. (1999). Attentional focus and fearful responding in
patients with panic disorder during a 35% CO2challenge. Behavior Therapy, 30,
Schmidt, N. B., & Zvolensky, M. J. (2007). Anxiety sensitivity and CO-sub-2 challenge
reactivity as unique and interactive prospective predictors of anxiety pathology.
Depression and Anxiety, 24, 527e536.
Schmidt, N. B., Zvolensky, M. J., & Maner, J. K. (2006). Anxiety sensitivity:
prospective prediction of panic attacks and axis I pathology. Journal of Psychi-
atric Research, 40, 691e699.
Sheehan, D. (1983). The anxiety disease. New York: Scribner.
Shipherd, J. C., Beck, J. G., & Ohtake, P. J. (1996). Anxiety sensitivity, suffocation fear,
trait anxiety, and breath-holding duration as predictors of response to carbon
dioxide challenge. Behaviour Research and Therapy, 34(8), 603e607.
Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental
studies: new procedures and recommendations. Psychological Methods, 7(4),
Smits, J. A. J., Powers, M. B., Cho, Y., & Telch, M. J. (2004). Mechanism of change in
cognitive-behavioral treatment of panic disorder: evidence for the fear of fear
mediational hypothesis. Journal of Consulting and Clinical Psychology, 72,
Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). STAI manual for the State-
Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press.
Struzik, L., Vermani, M., Duffin, J., & Katzman, M. A. (2004). Anxiety sensitivity as
a predictor of panic attacks. Psychiatry Research, 129, 273e278.
Taylor, S., & Cox, B. J. (1998). An expanded anxiety sensitivity index: evidence for
a hierarchic structure in a clinical sample. Journal of Anxiety Disorders, 12,
Taylor, S., Koch, W. J., & McNally, R. J. (1992). How does anxiety sensitivity vary
across the anxiety disorders? Journal of Anxiety Disorders, 6, 249e259.
Telch, M. J. Psychological model of panic provocation. In J. Abelson (Chair).
Cognitive and pharmacological factors in laboratory panic. Symposium con-
ducted at the meeting of the American Psychiatric Association, May 1995,
Miami Beach, FL.
Telch, M. J., Harrington, P., Smits, J. A. J., & Powers, M. B. (2005). Anxiety sensitivity
and arousal expectedness in moderating anxious responding to 35% carbon
dioxide inhalation. Unpublished manuscript.
Telch, M. J., Jacquin, K., Smits, J. A. J., & Powers, M. B. (2003). Emotional responding
to hyperventilation as a predictor of agoraphobia status among individuals
suffering from panic disorder. Journal of Behavior Therapy & Experimental
Psychiatry, 34, 161e170.
Telch, M. J., Lucas, J. A., Schmidt, N. B., Hanna, H. H., Jaimez, T. L., & Lucas, R. A.
(1993). Group cognitive-behavioral treatment of panic disorder. Behaviour
Research and Therapy, 31, 279e287.
Telch, M. J., Silverman, A., & Schmidt, N. B. (1996). Effects of anxiety sensitivity and
perceived control on emotional responding to caffeine challenge. Journal of
Anxiety Disorders, 10, 21e35.
Welkowitz, L. A., Papp, L., Martinez, J., Browne, S., & Gorman, J. M. (1999).
Instructional set and physiological response to CO2inhalation. American Journal
of Psychiatry, 156, 745e748.
Wolpe, J., & Rowan, V. C. (1988). Panic disorder: a product of classical conditioning.
Behaviour Research and Therapy, 26, 441e450.
Zinbarg, R. E., Brown, T. A., & Barlow, D. H. (2001). Anxiety sensitivity, panic, and
depressed mood: a reanalysis teasing apart the contributions of the two levels
in the hierarchical structure of the Anxiety Sensitivity Index. Journal of
Abnormal Psychology, 110, 372e377.
Zvolensky, M. J., Arrindell, W. A., Taylor, S., Bouvard, M., Cox, B. J., Stewart, S. H., et al.
(2003). Anxiety sensitivity in six countries. Behaviour Research and Therapy, 41,
Zvolensky, M. J., Eifert, G. H., Feldner, M. T., & Leen-Feldner, E. (2003). Heart-focused
anxiety and chest pain in postangiography medical patients. Journal of
Behavioral Medicine, 26(3), 197e209.
Zvolensky, M. J., Eifert, G. H., & Lejuez, C. W. (2001). Offset control during recurrent
20% carbon dioxide-enriched air induction: relation to individual difference
variables. Emotion, 1, 148e165.
M.J. Telch et al. / J. Behav. Ther. & Exp. Psychiat. 41 (2010) 365e372