Content uploaded by Sherry H Stewart
Author content
All content in this area was uploaded by Sherry H Stewart
Content may be subject to copyright.
Differences in High and Low Anxiety Sensitive
Women’s Responses to a Laboratory-Based Cold
Pressor Task
Lindsay S. Uman
1
, Sherry H. Stewart
1,2
, Margo C. Watt
1,3
and
Amber Johnston
4
1
Department of Psychology, Dalhousie University, Halifax, NS, Canada;
2
Department of
Psychiatry, Dalhousie University, Halifax, NS, Canada;
3
Department of Psychology, St.
Francis Xavier University, Antigonish, NS, Canada;
4
Department of Psychology, University
of Guelph, Guelph, ON, Canada
Abstract. Anxiety sensitivity (AS) has been linked to a variety of disabling chronic health conditions,
including pain-related conditions. A recent study has found that healthy women with high AS
reported significantly higher levels of sensory and affective pain on an experimental cold pressor task
compared to women with low AS. However, this study found no differences between AS groups for
pain tolerance or pain threshold. In the present study, which was designed to replicate and extend
these findings, 90 undergraduate university women were selected for inclusion in 1 of 2 AS groups
(high or low) based on their screening scores on a 16-item measure of AS. Participants were tested
individually on a lab-based cold pressor task using a variety of self-report and observer-measured
variables. Data analyses revealed that, as expected, the high AS participants reported significantly
more fear in response to the cold pressor on a relevant item of the McGill Pain Questionnaire – Short
Form (SF-MPQ) than did the low AS participants. Also as expected, the high AS participants
reported more pain in response to the cold pressor on the Present Pain Index (PPI) of the SF-MPQ
than did the low AS participants. High AS participants did not differ from low AS participants on
other aspects of the cold pressor response (e.g. pain threshold, pain tolerance, pain recovery). These
results support the role of pain-related fear as a mediating variable between AS and increased
perceived pain intensity. Key words: anxiety sensitivity; pain; fear; health; adult; laboratory; non-
clinical; acute
Received February 13, 2006; Accepted May 22, 2006
Correspondence address: Lindsay S. Uman, Department of Psychology, Dalhousie University, Life
Sciences Centre, 1355 Oxford Street, Halifax, Nova Scotia, Canada, B3H 4J1. Tel: +902 425 9133;
Fax: +902 494 6585. E-mail: luman@dal.ca
Anxiety sensitivity (AS) is the fear of anxiety-
related sensations and thoughts (e.g. increased
heart rate, decreased concentration, sweating;
Reiss & McNally, 1985) based on the belief
that they will have negative or harmful
consequences. AS involves the tendency to
misinterpret these sensations and thoughts,
which can lead to avoidance behavior. For
example, a person with high AS may fear that
an increased heart rate signifies an impending
heart attack, which leads him/her to avoid
situations leading to increases in heart rate.
In recent years, researchers have explored
the role of AS in the maintenance and
exacerbation of chronic health conditions,
such as gastrointestinal dysfunction, asthma,
and chronic pain (Asmundson, Wright, &
Hadjistavropoulos, 2000). For example,
Asmundon, Norton, and Veloso (1999) com-
pared patients with recurrent headaches who
were categorized as low, medium, or high in
AS. Headache patients with high AS reported
significantly greater depression, trait anxiety,
pain-related escape/avoidance, and pain-related
# 2006 Taylor & Francis ISSN 1650-6073
DOI 10.1080/16506070600898512
Cognitive Behaviour Therapy Vol 35, No 4, pp. 189–197, 2006
fear compared to patients with low or
medium AS levels. In another study, high
AS patients with chronic back pain reported
greater pain-related fear and displayed
greater avoidance than low AS chronic back
pain patients (Asmundson & Norton, 1995).
Previous research also suggests that AS may
be a better predictor of pain-related avoidance
than is pain severity. A recent study used
structural equation modeling to examine the
role of AS in fear and avoidance behavior in
patients with recurrent headaches (Norton &
Asmundon, 2004). AS showed a direct sig-
nificant loading on fear of pain. Interestingly,
this loading was even greater than the loading
of pain severity on fear of pain. An earlier
structural equation modeling study
(Asmundson & Taylor, 1996) found compar-
able results with patients who experienced
chronic musculoskeletal pain: AS directly
exacerbated fear of pain, even after control-
ling for the effects of pain severity, and
indirectly promoted pain-related escape and
avoidance via its influence on fear of pain.
Some studies examining the role of AS in
pain experience and behavior have made use
of laboratory-based tasks that induce acute
pain, such as the cold pressor. Behavioral
indices of experimental cold pressor pain
generally assess pain threshold (the moment
participants begin to feel pain), pain tolerance
(the moment participants remove their arm
from the water), and pain recovery (how long
it takes the pain to subside once their arm is
removed from the water). Theoretically, one
might hypothesize that high AS participants
should experience lower pain tolerance
because of their greater tendency for pain
avoidance (Asmundson et al., 1999).
However, studies have been mixed as to
whether high AS individuals are more sensi-
tive to pain on such behavioral indices of
response to cold pressor pain (see review by
Keogh et al., this issue).
There is some evidence to support the role of
AS in some aspects of response to acute or
experimental pain among clinical populations.
One study using participants meeting DSM-IV
criteria for panic disorder, found that both
diagnostic status and AS significantly predicted
pain and anxiety during an induced pain task
using a cold pressor (Schmidt & Cook, 1999).
However, AS appeared to have an indirect
association with pain via its contribution to
anxiety. In other words, AS influenced pain by
way of its impact on anxiety, which in turn
contributed to greater subjective pain experi-
ences.
Few studies have examined the role of AS
in acute or experimental pain, using non-
clinical samples. Keogh and Mansoor (2001)
investigated the effect of AS and coping on
pain responses in healthy women participating
in a cold pressor task. The women were
categorized into high and low AS groups
and within each group they were randomly
instructed to either avoid or attend to the pain
sensations. As expected, collapsed across pain
coping instruction conditions, the high AS
women reported greater levels of sensory and
affective pain. Of equal importance was the
finding that there were no differences between
the AS groups for measures of pain threshold
or pain tolerance.
Gender has also been shown to play a role
in the relationship between AS and pain.
Keogh and Birkby (1999) found that gender
exerted a main effect on pain experiences
whereby females exhibited lower pain thresh-
old and tolerance than males. Females with
high AS also reported more sensory pain than
females with low AS, but no effect of AS on
sensory pain was found among males. In a
more recent study, Keogh, Hamid, Hamid,
and Ellery (2004) investigated the effect of AS,
gender and negative interpretative biases in
the perceptions of acute chest pain in patients
without a history of cardiac problems, using a
battery of self-report measures. As with the
Keogh and Birkby (1999) study, they found
that AS was positively related to pain for
women but not for men. In addition, negative
interpretative biases played a mediating role
in the relationship between AS and pain for
women, but this mediating effect was not
apparent for the men. These results suggest
that the positive relationship between AS and
pain may be stronger among, or unique to,
females compared to males.
The objective of the present study was to
help elucidate the relationship between AS
level and experimental pain in a non-clinical
sample of undergraduate students. Given
prior findings that the relationships between
AS and pain may be stronger in women (e.g.
Keogh & Birkby, 1999; Keogh et al., 2004),
we focused exclusively on women in the
present study (see Keogh & Mansoor, 2001).
190 Uman, Stewart, Watt and Johnston COGNITIVE BEHAVIOUR THERAPY
To expand on the study by Keogh and
Mansoor (2001), we devised the present study
to: (i) allow coping responses to vary naturally
(rather than instructing participants to attend
to or avoid the pain sensations); (ii) include
additional behavioral and subjective indices of
responses to the cold pressor (we added pain
recovery time, and measures of total subjec-
tive pain experienced); and (iii) examine the
mediating role of the SF-MPQ fear item in
explaining the expected relationships of AS to
increased perceived pain and increased pain
avoidance (i.e. lower pain tolerance). The
following 4 hypotheses were tested: (i) the
high AS group would display lower pain
tolerance (i.e. greater pain avoidance) than
the low AS group, but no AS group differ-
ences were expected for the other behavioral
indices (i.e. pain threshold and recovery); (ii)
the high AS group would experience greater
negative affective responses (particularly fear)
in response to the cold pressor task than the
low AS group; (iii) the high AS group would
experience greater perceived pain in response
to the cold pressor than the low AS group on
the Present Pain Index (PPI) and Visual
Analogue Scale (VAS) ratings of the SF-
MPQ; and (iv) the greater perceived pain and
lower pain tolerance experienced/displayed by
the high AS group would be mediated by their
greater fear of the cold pressor pain relative to
low AS controls.
Method
Participants
Participants included 90 female undergradu-
ate students from 2 universities in the
Canadian province of Nova Scotia, who were
recruited based on their Anxiety Sensitivity
Index (ASI) scores obtained during in-class
screenings. Participants were selected for the
present study if their ASI score was above
(High AS group; n540) or below (Low AS
group; n550) 1 standard deviation (SD) from
the non-clinical women’s norm (M517.9,
SD58.7) (Maller & Reiss, 1992). The mean
(and SD) ASI screening scores were 34.20
(7.87) and 9.12 (4.06) for the high and low AS
groups, respectively. There were no significant
differences between High and Low AS groups
on any of the assessed demographic variables.
Mean (and SD) ages for the High and Low AS
groups were 19.30 (3.08) and 18.84 (1.83)
years, respectively, t (88)50.88, p50.38. Mean
(and SD) education levels for the High and
Low AS groups were 1.03 (0.16) and 1.12
(0.48) years of university, respectively, t
(88)521.31, p50.19 (t-value corrected for
violations of assumptions of homogeneity of
variance). Mean (and SD) family income
levels (on a 1–8 scale) for the High and Low
AS groups were 5.18 (2.25) and 5.68 (2.00),
respectively, t (88)521.12, p50.26. Finally,
94% of those in the High AS group and 92%
of those in the Low AS group were Caucasian,
x
2
50.10, p50.76.
Exclusion criteria included participants
with chronic health conditions (e.g. heart
conditions) and participants taking medica-
tion to manage anxiety or pain (i.e. present
analgesic and/or narcotic use). Such partici-
pants were excluded due to concerns that
chronic health conditions might preclude
exposure to the cold pressor, and that use of
these medications might alter pain responses.
Students participated voluntarily and did not
receive any form of compensation for their
participation. Ethical approval was obtained
from the Research Ethics Boards for both
universities.
Measures
Anxiety Sensitivity Index (ASI; Peterson &
Reiss, 1992). The ASI is a 16-item self-report
questionnaire designed to tap the amount of
fear one experiences in response to anxiety-
related symptoms and sensations. For each
item, participants are instructed to indicate
their degree of agreement on a Likert scale
ranging from very little (0) to very much (4).
All items are summed for a total score ranging
between 0 and 64. The ASI has excellent
psychometric properties including good test-
retest reliability, and construct and criterion
validity (Peterson & Reiss, 1992; Reiss,
Peterson, Gursky, & McNally, 1985).
McGill Pain Questionnaire- Short Form (SF-
MPQ; Melzack, 1987). The SF-MPQ is a
self-report measure comprised of 4 subscales
assessing the subjective experience of pain.
The first subscale is a 10-cm visual analogue
scale (VAS) with anchors of ‘‘no pain’’ and
‘‘worst possible pain’’. The second subscale,
called the Present Pain Index (PPI), requires
participants to rate the severity of pain they
are experiencing on a Likert scale ranging
VOL 35, NO 4, 2006 Anxiety sensitivity and cold pressor pain 191
from no pain (0), mild (1), discomforting (2),
distressing (3), terrible (4), to excruciating (5).
The remaining 2 subscales make up the Pain
Rating Index (PRI) which is comprised of 15
words. The first PRI subscale has 11 words
describing sensory qualities of pain (e.g.
shooting) and the second PRI subscale has 4
words describing affective qualities of pain
(e.g. fearful). Participants are asked to rate the
degree to which each of these words reflects
the pain they are experiencing on a scale
ranging from none (1) to severe (4). Melzack
(1987) has shown that the SF-MPQ has good
internal consistency, and good convergent
validity with the long form MPQ. The SF-
MPQ PRI also has good factorial validity as it
is best represented by a 2-factor solution
(Wright, Asmundson, & McCreary, 2001)
consistent with the structure of the original
PRI scale (Melzack, 1987). The dependent
measures from the SF-MPQ used in this study
were pre- to post-cold pressor scores on the
VAS, PPI ratings, and the sensory and
affective PRI subscale scores. We also sepa-
rately examined the ‘‘fear’’ item from the
affective subscale. This item was chosen a
priori because it has the highest loading on the
affective SF-MPQ factor (loading50.80;
Wright, Asmundson, & McCreary, 2001)
and because ‘‘fear’’ responses to the cold
pressor also have a strong theoretical rele-
vance to AS.
Cold-Pressor Task
Following the procedures of previous pain
induction studies (e.g. Keogh & Birkby, 1999;
Keogh & Mansoor, 2001), the cold pressor
task was used as a safe method to experimen-
tally induce pain. Participants were asked to
wash their non-dominant arm in soap and
water. They were then asked to submerge this
arm up to the elbow in a basin of warm water
(36–38
˚
C) for 2 minutes to help ensure that all
participants’ arms were equivalent in tem-
perature at baseline. The 2 minutes was timed
using a stop-watch and participants were
instructed to remove their arm once the time
had lapsed. Participants were then asked to
place the same arm in the cold pressor, a tank
filled with ice-cold water (2
˚
C). Participants
were asked to try to keep their arm in the
water for as long as they could, but were told
that they could remove it at any time if they
could no longer tolerate the pain. Although
they were not explicitly told, a maximum of 2
minutes was set for the total amount of time
they could keep their arm in the water, since
after this period the arm will generally start to
go numb (Wolff, 1984). Each participant was
asked to verbally indicate their pain threshold
(the moment they began to feel pain), pain
tolerance (the moment they removed their
arm from the water due to intolerable pain),
and their pain recovery (how long it took for
the pain to subside once their arm was
removed from the water).
Procedure
Female undergraduates were recruited from
the student mass testing pools at Dalhousie
and St Francis Xavier Universities in Nova
Scotia over a 2-year period. Participants were
recruited from among those participating in
another study (see Watt et al., this issue). A
total of 186 women were approached about
participating and 130 indicated their will-
ingness to be contacted to participate in the
cold pressor study. Of the 130 willing indivi-
duals, we were able to reach and schedule 90
for the present study.
Upon arrival, the study was explained to
each participant, and written informed con-
sent was obtained. Participants were tested
individually in a small room at each univer-
sity, and the study was administered by a
trained honors or graduate student in psy-
chology. After completing informed consent,
participants began the cold pressor task.
Immediately upon removing their arm from
the warm water, participants were asked to
complete the first (baseline) VAS subscale of
the SF-MPQ (Melzack, 1987). Participants
were then asked to place the same arm in the
cold pressor. The experimenter timed and
recorded pain tolerance, threshold, and recov-
ery for each participant. Larger values on
these 3 indices represent greater pain thresh-
old, greater pain tolerance, and slower pain
recovery times, respectively. Pain tolerance
was conceptualized as a measure of pain
avoidance given that pain avoidance should
theoretically represent the inverse of pain
tolerance (i.e. pain avoidance5pain intoler-
ance). Immediately upon removing their arms
from the cold pressor, participants were asked
to complete the second VAS scale and the
remaining scales of the SF-MPQ, starting with
the PRI and ending with the PPI (Melzack,
192 Uman, Stewart, Watt and Johnston COGNITIVE BEHAVIOUR THERAPY
1987). Finally, participants were fully
debriefed.
Results
No outliers were identified and the data
for each group approximated a normal distri-
bution for each dependent measure. Tests of
homogeneity of variance were conducted
for all analyses and corrections were made
when appropriate. There were no signifi-
cant AS group differences in baseline pain
ratings between the High AS (M50.15,
SD50.42) and Low AS (M5 0.10, SD50.23)
groups on the VAS1, t (88)50.70, p50.489,
d50.15.
1
A mixed model 26 2 (AS Group 6 Time
(pre- vs post-cold pressor)) Analysis of
Variance (ANOVA) conducted on the VAS
pain ratings revealed a significant main effect
of time with the VAS-2 (post-cold pressor
rating; M56.95, SD51.87), reflecting signifi-
cantly more pain than the VAS-1 (pre-cold
pressor rating; M50.12, SD50.34), F (1,
88)51175.63, pv0.001. This main effect
established the efficacy of the cold pressor in
inducing experiences of pain in the present
study. However, there was no significant main
effect of AS group (HAS: M53.61, SD50.96;
LAS: M53.48, SD50.96), F (1, 88)50.42,
n.s., and the hypothesized AS group 6 Time
interaction was not significant, F (1,88)50.18,
n.s. Thus, perceived pain increased similarly
from baseline due to the cold pressor, for both
high and low AS participants, according to
their ratings on the VAS.
All other hypotheses were tested using a set
of independent-samples (AS group) t-tests.
One-tailed tests were conducted because
directional predictions had been made a
priori. The t-test on the PPI revealed the
hypothesized significant difference between
the 2 AS groups (t (88)51.91, pv0.05,
d50.41), with the high AS participants
reporting more present pain in response to
the cold pressor than the low AS participants
(Figure 1).
The set of t-tests on the behavioral indices
of response to the cold pressor revealed
no significant group differences for pain
threshold (t (88)520.47, n.s., d50.10), pain
tolerance (t (88)521.01, n.s., d50.22), or pain
recovery (t
(88)520.66, n.s., d50.14). The
associated means and SDs (all in seconds)
were as follows: pain threshold (HAS:
M511.93, SD519.03; LAS: M513.49,
SD512.74), pain tolerance (HAS: M549.63,
SD533.37; LAS: M556.63, SD533.20), and
pain recovery (HAS: M562.22, SD544.06,
LAS: M568.50, SD545.84).
For the t-tests on the SF-MPQ sensory and
affective subscale scores, contrary to predic-
tion, there was no significant difference
between AS groups for the sensory (HAS:
M515.30, SD55.52,LAS:M514.68, SD5
16.64; t (88)50.47, n.s., d50.10) or the affective
(HAS: M52.23, SD52.30,LAS:M51.63,
SD52.50; t (88)51.14, n.s., d50.24) subscales.
However, as expected, significant group differ-
ences were found on the ‘‘fear’’ item of the SF-
MPQ, with the High AS group experiencing
more fear than the Low AS group in response to
the cold pressor pain experience, t (74.15)52.24,
pv0.05, d50.52 (t-value adjusted for violation
of the assumption of homogeneity of variances)
(Figure 2).
Overall, the results indicated AS group
differences in subjective pain experience (on
the PPI) as well as on the degree of fear in
response to the cold pressor task, with high AS
participants showing greater responses in each
case. We wished to test whether the greater
levels of subjective pain reported by the high AS
participants on the PPI might be ‘‘mediated’’ by
their greater fear in response to the sensations
induced by the cold pressor. We thus followed
steps outlined by Baron and Kenny (1986) for
testing mediation. Specifically, we first tested
whether there was a significant relationship
between the independent variable (AS group)
and the dependent variable (PPI scores). This
significant relationship is indicated in Figure 3
Figure. 1. Present Pain Index (PPI) scores on the McGill Pain
Questionnaire – Short Form in response to cold pressor for
HAS and LAS groups (bars represent standard errors).
1
Effect sizes (Cohen’s d) of 0.2 are small, 0.5 medium, and
0.8 large (Cohen, 1977).
VOL 35, NO 4, 2006 Anxiety sensitivity and cold pressor pain 193
by the path labeled ‘‘a’’ (r50.20, pv0.05).
Next, we tested whether there was a significant
relationship between the independent variable
(AS group) and the potential mediator (fear
rating on the SF-MPQ). This significant
relationship is indicated in Figure 3 by the path
labeled ‘‘b’’ (r50.24, pv0.05). Third, we tested
whether there was a significant relationship
between the potential mediator (fear rating)
and the dependent variable (PPI scores). This
significant relationship is indicated in Figure 3
by the path labeled ‘‘c’’ (r50.61, pv0.001).
Finally, we examined whether the significant
relationship between the independent variable
(AS group) and the dependent variable (PPI
scores) persisted after controlling for the
potential mediator (fear rating). As shown in
parentheses in Figure 3, the path labeled ‘‘a’’ is
no longer statistically significant (r
p
50.07, n.s.)
after controlling for paths ‘‘b’’ and ‘‘c’’. In
contrast, the relationship between the potential
mediator (fear ratings) and the dependent
measure (PPI scores) remained significant after
controlling for the independent variable (AS
group) (r
p
50.59, pv0.001; as illustrated in
parentheses in Figure 3 path ‘‘c’’). This pattern
of results suggests that fear acted as a mediating
variable in explaining the relationship between
AS and heightened pain ratings as assessed by
the PPI (Figure 3). We could not test the
hypothesized mediating role of fear in explain-
ing the expected relation between AS group and
pain avoidance because there was no significant
relationship between AS group and our mea-
sure of pain avoidance in this study (i.e. lower
pain tolerance).
Discussion
The results of this study examining the
relationship between AS and pain provide
partial support for 3 of the 4 hypotheses. As
expected, there were no significant differences
between the high and low AS groups for pain
threshold or recovery. Keogh and Mansoor
(2001) similarly found no significant differ-
ences between their high and low AS groups
for pain threshold. Our findings extend those
of Keogh and Mansoor (2001) by showing
that the lack of AS group differences on
Figure. 2. Fear ratings on the McGill Pain Questionnaire –
Short Form in response to the cold pressor for HAS and LAS
groups (bars represent standard errors).
Figure. 3. The mediational role of fear of pain (SF-MPQ fear rating) in explaining the relationship of anxiety sensitivity (AS
group: HAS vs LAS) to subjective pain ratings (SF-MPQ PPI). Notes: AS group based on scores on the Anxiety Sensitivity
Index. SF-MPQ5McGill Pain Questionnaire – Short Form. PPI5Present Pain Index. **pv0.001; *pv0.05 (1-tailed tests).
r5correlation coefficient; r
p
5partial correlation coefficient.
194 Uman, Stewart, Watt and Johnston COGNITIVE BEHAVIOUR THERAPY
behavioral responses to the cold pressor
extend to pain recovery time.
Contrary to expectations, there was no
significant difference between groups on pain
tolerance. We conceptualized pain tolerance
to reflect a form of avoidance whereby
participants high in AS would remove their
arm from the cold water sooner (i.e. avoiding
the pain and/or the associated anxiety-related
sensations). However, despite this theoretical
basis, studies have generally not found differ-
ences in pain tolerance as a function of AS
level (e.g. Keogh & Mansoor, 2001). This may
imply that the avoidance characteristic of
participants with high AS may be limited to
situations in which the bodily symptoms are
unexplained or unexpected. Alternatively,
sample selection bias may account for the
lack of AS group differences on pain toler-
ance, whereby those high AS participants less
willing to tolerate pain, may have been less
willing to volunteer for the present cold
pressor study. Another possibility is that the
measure of lower pain tolerance on the cold
pressor is not a valid index of pain avoidance.
Nevertheless, our results do provide further
support for the notion that although there
may be differences in how people with high
AS and low AS conceptualize the experience
of pain (see below), there is no objective
evidence that they are experiencing the pain
differently across several behavioral indices of
response to acute experimental pain.
We found partial support for our second
hypothesis that the high AS group would
experience greater negative affective reac-
tions (particularly fear) in response to a cold
pressor task than the low AS group. Although
there were no significant differences between
high and low AS groups on the affective or
sensory subscales of the SF-MPQ, the high AS
group did endorse significantly greater fear
ratings compared to their low AS counterparts.
The failure to observe AS group differences in
sensory and affective pain is inconsistent with
Keogh and Mansoor (2001). However, the
observed AS group difference in fear ratings on
the SF-MPQ is consistent with other studies in
the literature supporting a relationship
between AS and fear of pain (e.g. Norton &
Asmundson, 2004). Individuals with high AS
could experience more fear in response to
experimental pain because the pain sensations
may produce physiological sensations similar
to anxiety symptoms, or because the situation
(pain induction) may produce stress which
triggers anxiety symptoms.
Finally, we found partial support for the
hypothesis that the high AS group would
experience greater perceived pain in response
to the cold pressor than the low AS group, at
least as measured by the PPI. This finding
complements the aforementioned result indi-
cating that the high AS group also attributed
more fear to the painful experience. Thus, high
AS appears to color the way individuals
conceptualize the more subjective components
of pain. However, we only found support for
this hypothesis on the PPI and not the VAS
ratings of the SF-MPQ. It is not clear why this
distinction emerged since both scales presum-
ably tap present pain intensity. At least 3
explanations are possible. First, although both
the PPI and VAS require participants to
subjectively rate their pain on a continuum,
the PPI assigns affective labels to each item (e.g.
‘‘excruciating’’), whereas the VAS uses numer-
ical anchors only. It may be that the use of such
affective labels allows the PPI to tap into the
anxiety-related affective components of pain
better than the VAS, such that the PPI is more
sensitive to AS group differences in pain
experience. Second, the VAS and PPI were
administered at somewhat different times. The
VAS was always administered first (immedi-
ately after the participant removed her hand
from the cold pressor) and the PPI was always
administered last in the series of SF-MPQ
items. Thus, it could be that, for some reason
(e.g. slight numbness wearing off), AS group
differences in subjective pain experience only
emerged several seconds after the participants
removed their arms from the pain stimulus.
Third, the VAS was always completed before,
and the PPI was always completed after, the 15
PRI subjective experience items. Thus, it is also
possible that AS group differences in subjective
pain experience only emerge after participants
have reflected on their affective (e.g. fear)
responses to the pain induction procedure.
These various possibilities are deserving of
further investigation.
Similar to the findings of Schmidt and
Cook (1999) in panic disorder patients, we
also found support for the hypothesized
mediating role of fear in explaining the
relationship between AS and pain intensity
on the PPI in our sample of non-clinical
VOL 35, NO 4, 2006 Anxiety sensitivity and cold pressor pain 195
women. Once the relationships between AS and
fear, and between fear and pain intensity, were
controlled, the relationship between AS and
pain experience on the PPI was no longer
significant. Although the fear rating in this
model is based only on 1 item of the SF-MPQ,
it introduces an interesting additional compo-
nent to traditional pain models. The amended
Vlaeyen-Linton fear-avoidance model of chro-
nic pain conceptualizes the appraisal and
experience of pain as feeding into pain cata-
strophizing (with AS feeding into catastrophiz-
ing), which in turn leads to pain-related fear
and associated avoidance, depression, and
disability (Norton & Asmundson, 2004). We
propose that perhaps pain-related fear should
be depicted as a mediating variable in this
model, whereby the relationship between AS
and pain is accounted for by the increased fear
experienced by high AS individuals when they
encounter a painful stimulus. Further research
will be needed to determine if the mediating role
of fear holds in further investigations of AS and
pain, and to further explicate the mechanisms
involved.
There are several limitations to the present
study. First, the MPQ-SF fear item was
chosen as the main measure of fear because
it provided a state measure of fear in response
to the cold pressor (as opposed to trait fear).
Although relying on only 1 item is an
important limitation, it is not uncommon to
use a single measure (e.g. VAS) to assess state
anxiety during an experimental manipulation
(e.g. Schmidt & Cook, 1999). Secondly, low
AS participants may not represent the most
appropriate control group for comparisons
with high AS participants because they may
be a distinct population with unique beha-
vioral patterns (Shostak & Peterson, 1990). In
addition, failure to observe AS group differ-
ences on some variables (e.g. sensory and
affective pain; see Keogh & Mansoor, 2001)
may have been due to the fact that all of the
present participants were recruited from
another study in which a random half of the
high and low AS participants had undergone
cognitive behavior therapy (CBT) to reduce
AS, prior to taking part in the present study
(see Watt et al., this issue). This may have
served to dampen AS group differences to the
cold pressor in the present study. Given this
concern, we checked the magnitude and
pattern of effects on all cold pressor variables
among only those who had received the
control intervention. There was no evidence
that the magnitude of the effects increased
when those who had received CBT were
eliminated; however, this analysis was admit-
tedly underpowered. Another difference
between our study and that of Keogh and
Mansoor (2001) is that they instructed half of
their participants to focus on the pain and the
other half to avoid the pain, while we
provided no specific coping instructions. It is
possible that allowing individuals to use their
own naturalistic coping strategies serves to
minimize AS group differences in overall
sensory and affective responses to pain.
Further examination of this issue appears
warranted to explain the circumstances under
which AS group differences emerge in overall
sensory and affective responses to pain. It is
also important to acknowledge that we made
an assumption that pain tolerance has real-
word meaning (i.e. not tolerating pain in an
experimental task could reflect a broader
tendency to avoid pain in real-life situations).
This assumption needs to be tested empirically
in future research by examining the relations
of lower pain thresholds on the cold pressor to
established measures of pain avoidance.
Finally, given that our sample was female
and predominantly Caucasian, the general-
izability of our findings is unclear.
In terms of clinical implications, our findings
suggest that AS may be an important construct
to consider in the assessment and treatment of
pain patients. Furthermore, for patients high in
AS, it is likely that cognitive-behavioral inter-
ventions specifically designed to help reduce or
manage AS (see Watt et al., this issue, for an
example of such an approach), may indirectly
lead to better management of chronic and
acute pain. Our findings suggest that a focus on
managing fear of pain among high AS
individuals may be particularly helpful within
such interventions. Finally, prevention and
early interventions targeting high AS indivi-
duals may also have important implications
for reducing the development of problems
associated with chronic pain.
Acknowledgements
The authors acknowledge Kerry MacSwain’s
assistance with data collection. This research
project was funded by an operating grant to
196 Uman, Stewart, Watt and Johnston COGNITIVE BEHAVIOUR THERAPY
M. Watt and S. Stewart from the Nova Scotia
Health Research Foundation (NSHRF). S.
Stewart is funded by an Investigator Award
from the Canadian Institutes of Health
Research and by a Killam Professorship from
the Dalhousie University Faculty of Science.
L. Uman is supported by a graduate student-
ship from NSHRF.
References
Asmundson, G. J. G., & Norton (1995). Anxiety
sensitivity in patients with physically
Unexplained chronic back pain: A preliminary
report. Behaviour Research and Therapy, 33,
771–777.
Asmundson, G. J. G., & Taylor, S. (1996). Role of
anxiety sensitivity in pain-related fear and
avoidance. Journal of Behavioral Medicine, 19,
573–582.
Asmundson, G. J. G., Norton, P. J., & Veloso, F.
(1999). Anxiety sensitivity and fear of pain with
recurring headaches. Behaviour Research and
Therapy, 37, 703–713.
Asmundson, G. J. G., Wright, K. D., &
Hadjistavropoulos, H. D. (2000). Anxiety
sensitivity and disabling chronic health condi-
tions: State of the art and future directions.
Scandinavian Journal of Behaviour Therapy, 29,
100–117.
Baron, R. M., & Kenny, D. A. (1986). The
moderator-mediator variable distinction in
social psychological research: Conceptual, stra-
tegic and statistical considerations. Journal of
Personality and Social Psychology, 51, 1173–
1182.
Cohen, J. (1977). Statistical Power Analysis for the
Behavioral Sciences (2nd edn). New York:
Academic Press.
Keogh, E., & Birkby, J. (1999). The effect of
anxiety sensitivity and gender on the experience
of pain. Cognition Emotion, 13, 813–829.
Keogh, E., & Mansoor, L. (2001). Investigating the
effects of anxiety sensitivity and coping on the
perception of cold pressor pain in healthy
women. European Journal of Pain, 5, 11–25.
Keogh, E., Hamid, R., Hamid, S., & Ellery, D.
(2004). Investigating the effect of anxiety
sensitivity, gender and negative interpretative
bias on the perception of chest pain. Pain, 111,
209–217.
Maller, R. G., & Reiss, S. (1992). Anxiety
sensitivity in 1984 and panic attacks in 1987.
Journal of Anxiety Disorders, 6, 241–247.
Melzack, R. (1987). The Short-Form McGill Pain
Questionnaire. Pain, 30, 191–197.
Norton, P. J., & Asmundson, G. J. G. (2004).
Anxiety sensitivity, fear, and avoidance
behaviour in headache pain. Pain, 111,
218–223.
Reiss, S., & McNally, R. J. (1985). The expectancy
model of fear. In S. Reiss & R. R. Bootzin
(Eds), Theoretical issues in behavior therapy
(pp. 107–121). New York: Academic Press.
Reiss, S., Peterson, R. A., Gursky, D. M., &
McNally, R. J. (1985). Anxiety sensitivity,
anxiety frequency and the prediction of fear-
fulness. Behaviour Research and Therapy, 24,
1–8.
Peterson, R. A., & Reiss, S. (1992). Anxiety
Sensitivity Index Revised Test Manual.
Worthington, OH: IDS Publishing
Corporation.
Schmidt, N. B., & Cook, J. H. (1999). Effects of
anxiety sensitivity on anxiety and pain during a
cold pressor challenge in patients with panic
disorder. Behaviour Research and Therapy, 35,
313–323.
Shostak, B. A., & Peterson, R. A. (1990). Effects of
anxiety sensitivity on emotional response to a
stress task. Behaviour Research and Therapy, 28,
513–521.
Wolff, B. B. (1984). Methods of testing pain
mechanisms in normal man. In P. D. Wall, &
R. Melzack (Eds), Textbook of Pain (pp.186–
194). Edinburgh: Churchill Livingstone.
Wright, K. D., Asmundson, G. J. G., & McCreary,
D. R. (2001). Factorial validity of the Short-
Form McGill Pain Questionnaire (SF-MPQ).
European Journal of Pain, 5, 279–284.
VOL 35, NO 4, 2006 Anxiety sensitivity and cold pressor pain 197