a psychometric and clincial investigation of anxiety...
TRANSCRIPT
A PSYCHOMETRIC AND CLINCIAL INVESTIGATION OF ANXIETY SENSITIVITY IN ANXIETY DISORDERS
Kerry Ann Armstrong, B.SocSc (Psych) (Hons)
Dissertation Submitted to the Queensland University of Technology for the Degree of Doctor of Philosophy School of Psychology and Counselling
Faculty of Health
2004
ii
iii
ACKNOWLEDGEMENTS
There have been many people who have provided their support, assistance
and understanding whilst I completed this degree. I would like to take this
opportunity to say thank you to everyone who helped me along the way.
First and foremost I would like to say thank you to my principal supervisor
Dr Nigar Khawaja, whom without her supervision, support and guidance this
dissertation would not be possible. Nigar is perhaps the most generous and
professional person that I have had the pleasure of working with and I have learned a
great deal from her, as both a student and from her supervising me for registration as
a Psychologist.
I would also like to thank Professor Tian Oei from the University of
Queensland, not only for being an associate supervisor but also for allowing me to
collect data from his CBT clinic. Tian was always available for a supervision
meeting when I needed one and was helpful and kind in his encouragement in all
aspects of the dissertation. Also, a thank you goes to Ms Sue Fell from the Toowong
CBT clinic for all the support and encouragement she offered to me during the
course of the PhD, as well as all of her assistance in the collection of data.
A big thank you to Professor Mary Sheehan, for not only being an associate
supervisor, but for all the assistance and support she showed me when unforseen
circumstances arose.
I would like to thank the late Professor Larry Evans for his assistance in
collecting data from his clinic. Larry would always take the time to ask me how I
iv
was coming along in the dissertation and was very generous in helping me in my
data collection efforts. I would also like to thank Dr Stephen Cox for all his
assistance in his role as Statistics Advisor in the Confirmatory Factor Analysis
section of this PhD.
Further, I don’t think an acknowledgements page would be complete without
a note made regarding the ‘behind the scenes’ team that offer their encouragement
and support on a daily basis.
First, I would like to thank my fiancé Dan for always taking an interest in
what I do, for listening to me when I talk, and for understanding that a dissertation is
not a Monday to Friday, 9 to 5 job.
Thank you also to Kim Johnston who, although she has her own PhD to
complete, has always taken the time to talk to me about my dissertation, help me sort
out my thinking and even enter data. Kim was always available, at any time of day
or night, to help in anyway she could.
I would like to thank all the staff from the Centre for Accident Research and
Road Safety – Queensland (CARRS-Q) for their assistance, encouragement, and
understanding over the years. In particular I would like to thank Cynthia Schonfeld,
who has supported me in every way since day one. I would also like to thank
Dianne Jensen for being ‘Aunty Di’ and for showing me how to format documents
properly.
Finally, I would like to thank the Post Grad Community, in particular Eve
Dwyer and Trish Obst for reading the final draft; as well as Jenny Summerville,
v
Collette Roos, Jane Shakespeare-Finch, Katina Damoulius, and Leith Harding for
being the supportive people that they are and always being available for anything.
vi
vii
CERTIFICATION OF ORIGINAL WORK
I, Kerry Ann Armstrong, certify that to the best of my knowledge this dissertation,
which is submitted in partial fulfilment of the requirements of Doctor of Philosophy
degree undertaken at the Queensland University of Technology, is my own work,
except as acknowledged in the text. The material contained within has not been
submitted, in whole or in part, for a degree at this or any other university.
Publications arising from this dissertation are specified.
Signed……………………………………………………………………………….. Kerry Ann Armstrong School of Psychology and Counselling Faculty of Health Queensland University of Technology February, 2004
viii
ix
PARTS OF THE THESIS SUBMITTED FOR PUBLICATION OR PRESENTED AT CONFERENCES
Papers Submitted for Publication Armstrong, K. A., Khawaja, N. G., & Oei, T. P. S. (Submitted). Confirmatory Factor Analysis and Psychometric Properties of the Anxiety Sensitivity Index – Revised in Australian Clinical and Normative Populations. European Journal of Psychological Assessment. Papers Presented at Conferences International Armstrong, K. A., & Khawaja, N. G. (2003). An investigation of Anxiety Sensitivity: Confirmatory factor analysis and psychometric properties of the 21- item Anxiety Sensitivity Index. Poster Presented at the 37th Annual Convention for the Association for Advancement of Behavior Therapy, Boston, MA. National Armstrong, K. A., & Khawaja, N. G. (2003). Anxiety Sensitivity and Differences in Diagnostic Groups. Paper presented as part of a symposium entitled ‘Questionnaires: Are they of any use?’ at the Australian Association for Cognitive Behavioural Therapy, Adelaide, 42, South Australia. Armstrong, K. A., & Khawaja. N. G. (2002). An investigation of the Anxiety Sensitivity Index – Revised (ASI-R) in a clinical population. Australian Journal of Psychology, 54 (Supplement), (Combined Abstracts for the Australian Psychology Conference), Gold Coast, 10, Queensland. Armstrong, K. A., & Khawaja. N. G. (2002). An investigation of Anxiety Sensitivity: Confirmatory factor analysis and psychometric properties of the Anxiety Index – Revised (ASI-R). Australian Association for Cognitive and Behavioural Therapy, 25th National Conference, Brisbane, 41, Queensland.
x
xi
ABSTRACT Anxiety sensitivity is a cognitive, individual difference variable that is differentiated
by an individual’s fear of anxiety sensations and centred on the belief that such
sensations result in harmful consequences. In order to test anxiety sensitivity, Reiss,
Peterson, Gursky, and McNally (1986) developed the Anxiety Sensitivity Index
(ASI). However, one contentious issue in the area concerns the factor analytic
structure of anxiety sensitivity and this has important consequences for the construct.
Numerous investigations have been conducted using the ASI, and the results have
varied appreciably with some researchers arguing for a unidimensional construct.
However the general consensus now is that anxiety sensitivity is multidimensional.
It has been argued that the repeated attempts to clarify the dimensionality of anxiety
sensitivity, using the 16- item ASI, is problematic because the scale was never
designed to measure a multidimensional construct in the first instance. Thus, the
objective of the dissertation was to critically examine the anxiety sensitivity
construct by using an expanded, multidimensional measure of anxiety sensitivity
referred to as the Anxiety Sensitivity Index – Revised ([ASI-R] Taylor & Cox, 1998)
and establish the psychometric properties of the measure by conducting a series of
empirical investigations to assess the clinical utility of the measure.
A series of three empirical investigations are presented in the current
dissertation. The first investigation aimed to critically examine the factor structure
and psychometric properties of the ASI-R. Confirmatory factor analysis using a
clinical sample of adults revealed that the ASI-R could be improved substantially
through the removal of 15 problematic items in order to account for the most robust
xii
dimensions of anxiety sensitivity. The modified measure was re-named the 21- item
Anxiety Sensitivity Index (21- item ASI) and re-analysed with a large sample of
nonclinical adults, revealing configural and metric invariance across groups.
Further, comparisons with other alternative models that also include comparisons
with previous published ASI models indicated the 21- item ASI to be the best fitting
model for both groups. There was also evidence of internal consistency, test-retest
reliability, and construct va lidity for both samples. The aim of the second
investigation was to critically examine differences between and within various
anxiety classifications, a mood disorder classification, and a nonclinical control
sample, with respect to both general and specific dimensions of anxiety sensitivity as
identified by the 21- item ASI. In most instances, the results revealed that the
differences between and within the diagnostic groups were consistent with
theoretical expectations. Finally, the third investigation aimed to examine
differences within each diagnostic category before and after cognitive behavioural
therapy in order to provide a further test of validity for the revised 21- item ASI. The
results revealed significant differences within all but one diagnostic group on the pre
and post-treatment scores, using the global and specific dimensions of the 21-item
ASI.
The strengths, theoretical contribution, limitations, and directions for future
research are discussed. It is concluded that the overall findings relating to the series
of empirical investigations presented in the current dissertation make a significant
and valid theoretical contribution to the field of anxiety sensitivity in particular, and
anxiety research in general, by enhancing our understand ing of anxiety sensitivity
xiii
and how the 21- item ASI can be used to improve therapeutic interventions in clinical
practice.
xiv
xv
TABLE OF CONTENTS
Acknowledgements iii Certification of Original Work vii Parts of Dissertation Submitted for Publication and
Presented at Conferences ix Abstract xi Table of Contents xv List of Tables xxi List of Figures xxiii Introduction xxv Chapter 1: Theoretical approaches to the fear of anxiety: Introduction to
expectancy theory and anxiety sensitivity
1.1 Introduction 3 1.2 Theoretical Background 5 1.3 Expectancy Theory 8 1.4 Chapter Summary 10
Chapter 2: Review of the literature on anxiety sensitivity 2.1 Introduction 15
2.2 Introduction to the Anxiety Sensitivity Index 15 2.3 Anxiety Sensitivity and Trait Anxiety 18 2.4 Factor Analytic Investigations of the ASI 20 2.5 The Anxiety Sensitivity Index – Revised 40 2.6 Predictive Validity of Anxiety Sensitivity 48 2.7 Anxiety Sensitivity and Nonclinical Populations 51
2.8 Anxiety Sensitivity and Clinical Populations 54 2.8.1 Panic Disorde r with or without Agoraphobia 54 2.8.2 Posttraumatic Stress Disorder 58 2.8.3 Social Phobia 60 2.8.4 Obsessive-Compulsive Disorder and Generalised Anxiety Disorder 65 2.8.5 Specific Phobia 66 2.8.6 Major Depressive Disorder 68
2.9 Anxiety Sensitivity and Cognitive Behavioural Therapy 73 2.9.1 Anxiety Sensitivity and Combined Cognitive Behavioural and Pharmacological Therapy 77
2.10 Rationale for Study One 80 2.11 Rationale for Study Two 83 2.12 Rationale for Study Three 84 2.13 Chapter Summary 85
xvi
Chapter 3: Method
3.1 Participants 89 3.1.2 Participants from the clinical setting 89 3.1.3 Participants from the nonclinical setting 90
3.2 Design 90 3.3 Materials/Measures 90
3.3.1 Anxiety Sensitivity Index (ASI) and Anxiety Sensitivity Index – Revised (ASI-R) 91 3.3.2 Beck Anxiety Inventory 92 3.3.3 Catastrophic Cognitions Questionnaire (Modified) 93 3.3.4 Fear Questionnaire (FQ) 95 3.3.5 Zung Self-Rating Depression Scale (Zung – SDS) 95 3.3.6 COPE Questionnaire (COPE) 96 3.3.7 Stress Subscale of the 21- item Depression Anxiety Stress Scale (DASS-Stress) 96 3.3.8 Generalised Self-Efficacy Scale (SES) 97
3.4 Procedure 97 3.4.1 Participants from the clinical setting 97
3.4.2 Participants from the nonclinical setting 100 Chapter 4: Confirmatory factor analysis and psychometric properties of the Anxiety
Sensitivity Index-Revised in Aust ralian clinical and nonclinical populations
4.1 Introduction 105 4.2 Method 109
4.2.1 Participants 109 4.2.2 Measure 109 4.2.3 Procedure 110 4.2.4 Statistical Analysis 110
4.3 Results 111 4.3.1 Clinical and nonclinical groups CFA 111 4.3.2 Model modifications of Taylor and Cox’s (1998) Model 114 4.3.3 Test of Configural Invariance of Modified Model 116 4.3.4 Test of Metric of Modified Model 116 4.3.5 Comparisons of Alternative Models 121 4.3.6 Comparisons of the 21- item ASI and competing ASI models 123 4.3.7 Internal consistency and test-retest reliability of the 21-item ASI 127 4.3.8 Concurrent Validity of the 21- item ASI 128 4.3.9 Discriminant Validity of the 21-item ASI 130
4.4 Discussion 133 4.5 Chapter Summary 137
xvii
Chapter 5: Differences in anxiety sensitivity between and within diagnostic and nonclinical groups using the 21-item Anxiety Sensitivity Index
5.1 Introduction 141 5.2 Method 144
5.2.1 Participants 144 5.2.2 Design 147
5.2.3 Measures 148 5.2.4 Procedure 148
5.3 Results 148 5.3.1 Data Screening and Cleaning 148 5.3.2 Preliminary Analyses 151 5.3.3 Tests of Hypotheses 153
5.3.3.1 Analysis of differences between individuals with a primary diagnosis of panic disorder, GAD, PTSD, depression, and nonclinical controls on the total score of the 21- item ASI 153
5.3.3.2 Analysis of differences between individuals with a primary diagnosis of panic disorder, GAD, PTSD, depression, and nonclinical controls on the specific dimensions of the 21-item ASI 155
5.3.3.3 Analysis of differences for the panic disorder diagnostic group on the first- order dimensions of anxiety sensitivity, as assessed by the 21- item ASI 164
5.3.3.4 Analysis of differences for the GAD diagnostic group on the first-order dimensions of anxiety sensitivity, as assessed by the 21- item ASI 166
5.3.3.5 Analysis of differences for the PTSD diagnostic group on the first-order dimensions of anxiety sensitivity, as assessed by the 21- item ASI 167
5.3.3.6 Analysis of differences for the depression diagnostic group on the first-order dimensions of anxiety sensitivity, as assessed by the 21- item ASI 169
5.3.3.7 Analysis of differences for the nonclinical control group on the first-order dimensions of anxiety sensitivity, as assessed by the 21-item ASI 171
5.4 Discussion 173 5.5 Chapter Summary 183
xviii
Chapter 6: Examination of differences within diagnostic categories after cognitive behavioural therapy using the 21-item Anxiety Sensitivity Index
6.1 Introduction 189 6.2 Method 191
6.2.1 Participants 191 6.2.2 Measures 194
6.2.3 Design 194 6.2.4 Treatment 194
6.3 Results 195 6.3.1 Data Screening and Cleaning 195 6.3.2 Reliability Analyses of Scale 197 6.3.3 Preliminary Analyses 197 6.3.4 Tests of Hypotheses 200
6.3.4.1 Analysis of differences for each diagnostic group before and after CBT for overall level of anxiety sensitivity 200
6.3.4.2 Analysis of differences for the panic disorder group before and after CBT for the first-order dimensions of anxiety sensitivity, as assessed by the 21-item ASI 201
6.3.4.3 Analysis of differences for the GAD group before and after CBT for the first-order dimensions of anxiety sensitivity, as assessed by the 21- item ASI 203
6.3.4.4 Analysis of differences for the PTSD group before and after CBT for the first-order dimensions of anxiety sensitivity, as assessed by the 21- item ASI 205
6.3.4.5 Analysis of differences for the depression group before and after CBT for the first- order dimensions of anxiety sensitivity, as assessed by the 21- item ASI 206
6.3.4.6 Analysis of differences on the BAI, CCQ-M, FQ and Zung – SDS after CBT 208
6.4 Discussion 209 6.5 Chapter Summary 213
Chapter 7: General discussion
7.1 Introduction 219 7.2 Measurement of Anxiety Sensitivity 219 7.3 Clinical Utility 221 7.4 Therapy 222 7.5 Theoretical Contribution 222
xix
7.6 Limitations of the Research 224 7.7 Future Directions 226 7.8 Conclusion 229
References 231 Appendix A: Anxiety Sensitivity Index (ASI) 249
Appendix B: Anxiety Sensitivity Index – Revised (ASI-R) 253
Appendix C: 21- item Anxiety Sensitivity Index (21- item ASI) 257
Appendix D: Questionnaire Booklet for clinical and nonclinical Participants’ 261
Appendix E: List of Contents of Cognitive Behavioural Therapy Manuals for Anxiety Disorders at the Toowong Private Hospital 277
Appendix F: List of Contents of Cognitive Behavioural Therapy Manuals for Depression at the Toowong Private
Hospital 283 Appendix G: Telephone Screening Sheet for Clinical Participants
Recruited from Advertisements in Local Media 291
Appendix H: Non-published handbook for Information about Anxiety Disorders prepared by Dr Nigar Khawaja and Kerry Ann Armstrong 297
Appendix I: Instructions for participants from First Year Subject Pool 327
xx
xxi
LIST OF TABLES
Table 2.1 Factor analytic studies of the Anxiety Sensitivity Index 21 Table 2.2 Factor analytic studies of the Anxiety Sensitivity Index-
Revised 41
Table 4.1 Clinical group CFA of three hypothesised ASI-R models 113
Table 4.2 Nonclinical group CFA of three hypothesised ASI-R models 114
Table 4.3 Nonclinical group CFA of the 21-item ASI Hierarchical model 116
Table 4.4 Metric Invariance CFA statistics between the clinical and nonclinical group datasets using the 21- item ASI 117
Table 4.5 Clinical groups CFA of the unifactorial, orthogonal 4- factor model and 3-factor hierarchical models of the 21- item ASI 121
Table 4.6 Nonclinical groups CFA of the unifactorial, orthogonal 4-factor model and 3-factor hierarchical models of the 21- item ASI 123
Table 4.7 Clinical groups CFA of alternative hypothesised ASI models 125
Table 4.8 Nonclinical groups CFA of alternative hypothesised ASI models 126
Table 4.9 Internal Consistency and Test-Retest Reliability of the 21-item ASI for the Clinical and Nonclinical Groups 127
Table 4.10 Pearson Correlations among the 21- item ASI, 16- item ASI, BAI, CCQ-M, FQ, Zung-SDS, and DASS-Stress Scale for the clinical group 129
Table 4.11 Pearson Correlations among the 21- item ASI, 16- item ASI, BAI, CCQ-M, FQ, Zung-SDS, and DASS-Stress Scale for the nonclinical group 130
Table 4.12 Pearson Correlations among the 21- item ASI, Self-Efficacy, and COPE questionnaires for the clinical group 131
Table 4.13 Pearson Correlations among the 21- item ASI, Self-Efficacy, and COPE questionnaires for the non clinical group 132
xxii
Table 4.14 Between Groups Differences of the 21- item ASI and first-order dimensions: Means, Standard Deviations and Univariate Results for the Clinical and Nonclinical groups 133
Table 5.1 Means, Standard Deviations and Age Range for diagnostic groups 146
Table 5.2 Sample size of Demographic Variables for the diagnostic groups 147
Table 5.3 Means and Standard Deviations for the Diagnostic and Nonclinical Control Groups on the 21-item ASI dimensions 152
Table 6.1 Means, Standard Deviations and Age Range for d iagnostic groups 192
Table 6.2 Sample size of Demographic Variables for the diagnostic groups 193
Table 6.3 Internal Consistency of the 21- item ASI for the Clinical Group Pre and Post-Cognitive-Behavioural Therapy 197
Table 6.4 Means and Standard Deviations for the Diagnostic Groups before and after Cognitive-Behavioural Therapy 199
xxiii
LIST OF FIGURES
Figure 4.1 Clinical group factor loadings for the items of the 21- item ASI with a hierarchical model of four first-order dimensions and a general higher-order dimension 119 Figure 4.2 Nonclinical group factor loadings for the items of the 21-item ASI with a hierarchical model of four first-order dimensions and a general higher-order dimension 120 Figure 5.1 Mean differences and standard error between panic disorder,
GAD, PTSD, depression, and nonclinical control groups and the 21-item ASI total score 155
Figure 5.2 Mean differences and standard error between panic disorder, GAD, PTSD, depression, and nonclinical control groups and the 21-item ‘Fear of Respiratory Symptoms’ dimension 158
Figure 5.3 Mean differences and standard error between panic disorder, GAD, PTSD, depression, and nonclinical control groups and the 21-item ‘Fear of Publicly Observable Symptoms’ dimension 160
Figure 5.4 Mean differences and standard error between panic disorder, GAD, PTSD, depression, and nonclinical control groups and the 21-item ‘Fear of Cardiovascular/Stroke Symptoms’ dimension 162
Figure 5.5 Mean differences and standard error between panic disorder, GAD, PTSD, depression, and nonclinical control groups and the 21-item ‘Fear of Cognitive Dyscontrol’ dimension 163
Figure 5.6 Mean differences and standard error for the panic disorder group and the four first-order dimensions of the 21-item ASI 165
Figure 5.7 Mean differences and standard error for the GAD group and the four first-order dimensions of the 21- item ASI 167
Figure 5.8 Mean differences and standard error for the PTSD group and the four first-order dimensions of the 21- item ASI 169
Figure 5.9 Mean differences and standard error for the depression group and the four first-order dimensions of the 21- item ASI 171
xxiv
Figure 5.10 Mean differences and standard error for the nonclinical control group and the four first-order dimensions of the 21-item ASI 173
Figure 6.1 Mean differences and standard error on the 21-item ASI total score for the panic disorder, GAD, PTSD and depression diagnostic groups before and after Cognitive- Behavioural Therapy 201
Figure 6.2 Mean differences and standard error for the panic disorder group on the four first order dimensions of the 21- item ASI before and after Cognitive-Behavioural Therapy 203
Figure 6.3 Mean differences and standard error for the GAD group on the four first order dimensions of the 21-item ASI before and after Cognitive-Behavioural Therapy 205
Figure 6.4 Mean differences and standard error for the PTSD group on the four first order dimensions of the 21-item ASI before and after Cognitive-Behavioural Therapy 206
Figure 6.5 Mean differences and standard error for the Depression group on the four first order dimensions of the 21-item ASI before and after Cognitive-Behavioural Therapy 208
xxv
INTRODUCTION
Since the concept of anxiety sensitivity was initially presented as part of
Reiss and McNally’s (1985) interactional model of expectancy theory, it has
received considerable attention in the extant literature. According to Reiss’s
expectancy theory (Reiss & McNally, 1985), anxiety sensitivity is a cognitive,
individual difference variable that is characterised by the individual’s fear of anxiety
related sensations and based on the belief that such sensations result in harmful
consequences.
In order to test anxiety sensitivity, Reiss, Peterson, Gursky, and McNally
(1986) developed the Anxiety Sensitivity Index (ASI). Several investigations, using
the ASI, have revealed that anxiety sensitivity is capable of not only distinguishing
panic disorder from other anxiety disorders (see Apfledorf, Shear, Leon, & Portera,
1994); but also of predicting who will respond fearfully in panic provocation
procedures (see Rapee, Brown, Antony, & Barlow, 1992) as well as who will
develop panic attacks as part of prospective research (see Schmidt, Lerew, &
Jackson, 1997). Thus the construct validity of anxiety sensitivity, using the ASI, has
been well documented. The construct of anxiety sensitivity has also been linked to
other clinical problems such as major depression (Otto, Pollack, Fava, Uccello &
Rosenbaum, 1995; Taylor, Koch, Woody & McLean, 1996).
The factor analytic structure of anxiety sensitivity has important
consequences not only for the construct but also for the role it plays in anxiety and
non-anxiety related pathology. Numerous investigations have been conducted using
the ASI, and the results have varied appreciably with some researchers arguing for a
xxvi
unidimensional structure (Reiss et al., 1986; Taylor, Koch, McNally, & Crockett,
1992); and others arguing for a multidimensional construct (Blais et al., 2001;
Peterson & Heilbronner, 1987; Schmidt & Joiner, 2002; Stewart, Taylor & Baker,
1997; Taylor, 1996; Telch, Shermis & Lucas, 1989; Zinbarg, Barlow, & Brown,
1997). According to Zinbarg, Mohlman, and Hong (1999), there is sufficient,
convergent evidence from several factor analytic examinations to argue that the ASI
contains three first-order dimensions that load onto a single, second-order or general
anxiety sensitivity dimension.
It is interesting that the broad consensus of the factor structure of the ASI is
one in which the construct is viewed as multidimensional because the ASI was
originally developed to measure a unitary, anxiety sensitivity construct (Reiss et al.,
1986). Given that investigations of the first-order dimensions of the ASI have
validated the importance of the multidimensional perspective of anxiety sensitivity
(see Zinbarg et al., 1999 for review), Taylor and Cox (1998) developed the 36-item
Anxiety Sensitivity Index – Revised (ASI-R), to provide a more comprehensive
measure of the first-order anxiety sensitivity dimensions. While the 36-item ASI-R
retains the same instructions and response format as the 16-item ASI, it contains a
broader selection of items, and hence dimensions, for assessing the explicit domains
of anxiety sensitivity. Thus, the main objective of the current dissertation is to
empirically investigate not only the factor structure of the ASI-R but also the
reliability and validity of the measure with both clinical and nonclinical populations
because it is contended that the numerous limitations of the 16-item ASI are of
concern as investigator’s may not be confident when making any theoretical claims
xxvii
based on the measure due to argument that the reliability and validity of the measure
has been compromised.
This dissertation is organised into four sections. The first section consists of
three chapters. The aim of chapter one was to provide a review of the theoretical
conceptualisations of the ‘fear of anxiety’ construct and describe in detail Reiss and
McNally’s (1985) expectancy model of fear, in which the construct of anxiety
sensitivity is embedded. Chapter two provides a literature review of the research
that has been conducted in the area, highlights the limitations in the current literature
and concludes with a sound rationale and argument for the dissertation. The third
chapter provides a detailed account of the participants, measures and procedures
employed. The second section consists of three empirical investigations that aims to
provide a psychometric and clinical investigation of anxiety sensitivity (see chapters
four to six). The third section provides a general discussion of the entire dissertation
(see chapter seven) and highlights the strengths, theoretical contribution, limitations,
and directions for future research. The references and appendices are presented in
the fourth and final section.
Anxiety Sensitivity 1
Chapter One: Theoretical approaches to the fear of anxiety: Introduction to
expectancy theory and anxiety sensitivity
1.1 Introduction 3
1.2 Theoretical Background 5
1.3 Expectancy Theory 8
1.4 Chapter Summary 10
Anxiety Sensitivity 2
Anxiety Sensitivity 3
CHAPTER ONE
THEORETICAL APPROACHES TO THE FEAR OF ANXIETY:
INTRODUCTION TO EXPECTANCY THEORY AND ANXIETY
SENSITIVITY
1.1 Introduction
While some degree of anxiety is natural and normal, past research has
demonstrated that anxiety disorders represent one of the most significant mental
health concerns worldwide (Lewinsohn, Gotlib, Lewinsohn, Seeley, & Allen,
1998). The notion that many individuals with anxiety or anxiety disorders are
afraid of experiencing anxiety has stimulated a great deal of research into
clarifying the role ‘fear of anxiety’ plays in the development and maintenance of
anxiety and anxiety disorders. One construct that has occupied an important
place in research is the concept of anxiety sensitivity. Anxiety sensitivity is
defined as a cognitive individual difference variable that is delineated by an
individual’s fear of anxiety related sensations and is based on the belief that such
sensations result in harmful consequences (Reiss, 1991; Reiss & Havercamp,
1996; Reiss & McNally, 1985). For example, while some individuals are fearful
of flying, insects, or of performing in public, they may also be fearful of the
sensations or symptoms that are associated with anxiety or panic; symptoms
which these stimuli might generate.
Due to the fear of anxiety symptoms, individuals who exhibit elevated
anxiety sensitivity are hypothesised to respond fearfully to sensations of
involuntary autonomic arousal, which, in turn, magnifies the symptoms and the
experience of anxiety. For example, an individual who exhibits high anxiety
Anxiety Sensitivity 4
sensitivity is fearful that shortness of breath may lead to suffocation or that a
rapid heart beat indicates an impending heart attack. Conversely, an individual
who exhibits low anxiety sensitivity is likely to regard such sensations as
unpleasant but not life threatening (McNally, 1999; Reiss & McNally, 1985).
This ‘fear of anxiety’ sensations might help to perpetuate a vicious cycle,
whereby pre-existing beliefs about physiological arousal (or autonomic change)
may predispose an individual to respond fearfully, which could lead to an
increase in the experience of anxiety (McNally, 1994; Taylor, Koch & McNally,
1992).
Anxiety sensitivity is distinguishable from other cognitive theories of
anxiety because it is regarded as a dispositional construct. It is similar to, but
distinct from, trait anxiety and can conceivably be attributed to individual
differences in verbal, observational or direct learning experiences of illness-
related behaviour (McNally, 1999). For example, anxiety sensitivity denotes a
specific propensity to respond fearfully to the sensations of anxiety themselves,
whereas trait anxiety denotes a general propensity to respond fearfully to a wide
range of stressors in general. It should be noted that both are regarded as stable
individual difference variables (McNally, 2002). Therefore, not only do
individuals differ in their propensity to experience anxiety symptoms (i.e., trait
anxiety), they also differ in their propensity to respond fearfully to them (i.e.,
anxiety sensitivity). While anxiety sensitivity is regarded as a dispositional
construct, it should not be confused with the concept of anticipatory anxiety,
which is an occurrent construct and denotes an escalation in distress at the
prospect of imminent challenge or threat (McNally, 2002). Finally, Reiss (1991)
has argued that anxiety sensitivity is a fundamental fear, and therefore distinct
Anxiety Sensitivity 5
from common fears. That is, the motive for avoiding any stimulus that has the
potential to invoke the symptoms of anxiety can arguably be due to a fear of
anxiety rather than a fear of the stimulus itself.
Whilst the construct of anxiety sensitivity is not new, it is preceded by a
great deal of research and various theoretical models have been offered to
explain the phenomenon of fear of anxiety including psychoanalytic,
interoceptive learning and cognitive theories. The purpose of this introductory
chapter is to review the conceptualisations of the ‘fear of anxiety’ construct from
several different theoretical perspectives. The review will culminate in a
discussion of Reiss and McNally’s (1985) expectancy model of fear, in which
the construct of anxiety sensitivity is embedded.
1.2 Theoretical Background
The importance of the fear of anxiety has been recognised by many
researchers from different theoretical orientations. Both Freud (1926/1995) and
Fenichel (1945) reported that many individuals with anxiety or phobias acquire a
‘fear of anxiety’ and become fearful of recurrent experiences. According to the
psychoanalytic view, anxiety is a disturbing symptom that results from
intrapsychic conflict that could be focused either internally or externally
(Wolman & Stricker, 1994). Conversely, according to the existential view,
Frankl (1975) posited that anticipatory anxiety was the fearful expectation that
the anxiety would reoccur and described a vicious circle that occurs when an
individual’s expectations of a fearful symptom of anxiety (e.g., excessive
sweating) produces or precipitates the symptom, therefore reinforcing the
individual’s anticipatory anxiety.
Anxiety Sensitivity 6
Following the psychoanalytic and existential views of anxiety, the
behaviourists’ view of learning generated a different explanation to account for
the relationship between anxiety disorders and the fear of anxiety. For example,
Eysenck’s (1968) theory of the incubation of anxiety/fear response proposes that
when an individual is highly anxious or fearful, short exposures to a conditioned
stimulus will lead to the incubation of anxiety without the unconditioned
stimulus through a positive feedback cycle in which the conditioned response
(such as anxiety and fear) provides reinforcement of the conditioned stimulus
only. This is seen as a departure from traditional theories of learning which
propose that the absence of an unconditioned stimulus would lead to a reduction
in the conditioned response. Eysenck’s (1982) hypothesis also included an
individual differences factor in order to account for the personality differences
among individuals. This is significant as Eysenck (1982) recognised the
importance of differences among individuals in the predisposition to become
conditioned by a fear of anxiety.
Goldstein and Chambless (1978) proposed an alternative concept to the
fear of anxiety. Using the term ‘fear of fear’, Goldstein and Chambless (1978)
conceptualised the fear of anxiety as a consequence of the experience of panic
attacks through the process of interoceptive conditioning. According to these
researchers, individuals who have experienced one or more panic attacks become
increasingly vigilant to their internal bodily sensations. Interpretation of these
bodily sensations as a sign of another impending panic attack has the potential to
lead to such an experience. As such, it is this physiological arousal that becomes
the conditioned stimuli for the conditioned response of a panic attack (Goldstein
& Chambless, 1978). While interoceptive conditioning is regarded as an
Anxiety Sensitivity 7
important form of learning (Razran, 1961), many researchers at the time
considered the fear of anxiety as a secondary consequence of panic attacks
because it required the experience of panic attack in order for it to develop.
Thus, it was unable to explain the acquisition of fear of anxiety in individuals
who had never experienced a panic attack.
In contrast to the views held by behaviourists, cognitive theorists have
proposed an alternative explanation of the fear of anxiety by arguing that many
individuals with anxiety misinterpret their bodily symptoms as catastrophic and
therefore believe that these sensations are an indication of physical or social
doom. Consequently, cognitive theorists have postulated that individuals with
anxiety disorders appear to be particularly sensitive to a systematic bias in
cognitive processing due to the cognitive misinterpretation of the physical or
psychosocial experience of anxiety as dangerous (Beck & Emery, 1979; Beck,
Emery, & Greenberg, 1985; Clark, 1986; Khawaja & Oei, 1992, 1998). Thus,
the cognitive paradigm of anxiety disorders postulates that it is not the events per
se, but rather the individual’s overactive or catastrophising cognitive processing
of harmless physiological perturbations or psychosocial consequences which, in
turn, may become misinterpreted as threatening or dangerous (Beck et al., 1985;
Khawaja & Oei, 1992, 1998; McNally, 1990). For example, Beck et al. (1985)
and Clark (1986) have suggested that panic attacks are the result of an
individual’s catastrophic misinterpretations of benign physiological agitations.
The physiological agitations or sensations which are primarily misinterpreted,
involve characteristic non-pathological anxiety such as heart palpitations,
shortness of breath, and dizziness. Catastrophic misinterpretations involve
perceiving the physiological sensations to be much more harmful than they
Anxiety Sensitivity 8
would be in actuality. According to Clark (1986), such misinterpretations can
lead to the development of more anxiety that has the potential to increase an
individual’s level of arousal and consequently increase the physiological
sensations which were the focus of alarm in the first instance. Therefore, the
misinterpretation of physiological sensations as harmful or dangerous may lead
to a cycle whereby faulty interpretations lead to an increase in the experience of
anxiety and consequently may result in an anxiety disorder such as panic (Beck
et al., 1985; Clark, 1986; Khawaja & Oei, 1992, 1998). This process has the
potential to contribute to the belief that some physiological sensations result in
harmful consequences. Taylor (1995) has argued that this vicious cycle is driven
by the fear of anxiety-related sensations, or anxiety sensitivity. However, while
the strength of the cognitive models is in its description of panic experiences,
Cox (1996) has argued that it appears weaker in distinguishing a quantifiable
individual difference variable that predisposes some individuals to make
catastrophic misinterpretations and the construct of anxiety sensitivity may
provide this missing component. It is this construct that has received a great deal
of attention and research into the fear of anxiety phenomenon.
1.3 Expectancy Theory
The concept of anxiety sensitivity was initially presented as part of Reiss
and McNally’s (1985) interaction model of expectancy theory. According to
Reiss and McNally (1985), expectancy theory is teneted on three significant
suppositions; (1) almost all individuals are motivated to evade stimuli that
provoke expectations of danger, (2) anxiety sensitivity is a stable, individual
difference variable along which difference can be measured, and (3) individuals
learn that particular stimuli in the environment have the potential to make them
Anxiety Sensitivity 9
anxious. The central tenet of Reiss and McNally’s (1985) expectancy model of
fear allows for general estimates in which increases or decreases in one variable
reflect increases or decreases in another variable which can be expressed by the
following formula:
Fb = Ed + (Ea x Sa) where Fb (fear behaviour) denotes a motivation to avoid a feared stimulus; Ed
(danger expectancy) denotes a motivation to avoid a stimulus that may cause
harm; Ea (anxiety expectancy) denotes a tendency of a fearful stimulus to induce
an expectation of becoming anxious; and Sa (anxiety sensitivity) is the degree to
which an individual is fearful of becoming anxious.
Reiss and McNally’s (1985) expectancy theory proposes that fear
behaviour (Fb) is the function of two fear components; a danger component and
an anxiety component. The first component, danger expectancy (Ed), is one in
which individuals learn that a specific external situation or object is a dependable
signal of danger in the immediate environment. The other anxiety components,
anxiety expectancy (Ea) and anxiety sensitivity (Sa) are a combination in which
an individual is averse to becoming anxious because he or she learns that a given
stimulus is a reliable indication of anxiety arousal. Thus, the total hypothetical
tendency to avoid a specific stimulus is the computation of the motivation
associated with the avoidance of the danger component and the motivation
associated with the avoidance of the anxiety component (Reiss & McNally,
1985). The significance of Reiss and McNally’s (1985) expectancy model is that
it relies upon two categorically different explanations for avoiding a specific
stimulus; (1) individuals who view a particular stimulus as dangerous are
motivated to avoid that stimulus; and (2) individuals who view a stimulus as
Anxiety Sensitivity 10
anxiety arousing are also likely to avoid that stimulus. As such, Reiss and
McNally’s (1985) expectancy model hypothesises that the propensity to perceive
anxiety as threatening (i.e., anxiety sensitivity) is a dispositional construct that
functions when a situationally based anxiety expectation is activated. Therefore,
the theory predicts that fear behaviour is influenced by the interaction of the
expectation of anxiety and an individual’s propensity to perceive fear as
threatening. It is this construct of anxiety sensitivity that provides the context for
the current dissertation.
1.4 Chapter Summary
The aim of the current chapter was to provide an outline of the theoretical
approaches to the fear of anxiety sensitivity and provide an overview of
expectancy theory as well as the concept of anxiety sensitivity. The concept of
anxiety sensitivity is defined as a cognitive, individual difference variable that is
delineated by an individual’s fear of anxiety related sensations and based on the
belief that such sensations result in harmful consequences (Reiss, 1991; Reiss &
Havercamp, 1996; Reiss & McNally, 1985). It was contended that the construct
of anxiety sensitivity was conceptualised largely on the ‘fear of fear’ or ‘fear of
anxiety’ theories first proposed by Goldstein and Chambless (1978). Finally, it
was proposed that due to the fear of anxiety symptoms, individuals who exhibit
elevated anxiety sensitivity are hypothesised to respond fearfully to sensations of
involuntary autonomic arousal, which in turn, magnify the symptoms and the
experience of anxiety.
The literature review to follow begins with an introduction to the
empirical literature regarding the anxiety sensitivity construct as well as the most
common measure used in adult populations, the Anxiety Sensitivity Index (ASI
Anxiety Sensitivity 11
– Reiss, Peterson, Gursky, & McNally, 1986). Construct validation that outlines
the debates in the field and prospective uses of the ASI will also be discussed.
Studies showing support for both the unidimensional and multidimensional
models of anxiety sensitivity will be discussed and the problems and issues
surrounding the various methods of analysis and theoretical limitations of this
measure will be outlined. Following this, the most fundamental question
regarding the optimal internal structure and dimensionality of the ASI will be
explored extensively. The expanded version of the ASI – the Anxiety Sensitivity
Index – Revised (Taylor & Cox, 1998) will be introduced and explored in detail.
Finally the cross-sectional applications of the ASI, experimental manipulation
and clinical applications will be discussed. A concluding argument and the
rationale of the dissertation will be provided.
Anxiety Sensitivity 12
Anxiety Sensitivity 13
Chapter Two: Review of the literature on anxiety sensitivity
2.1 Introduction 15
2.2 Introduction to the Anxiety Sensitivity Index 15
2.3 Anxiety Sensitivity and Trait Anxiety 18
2.4 Factor Analytic Investigations of the ASI 20
2.5 The Anxiety Sensitivity Index – Revised 40
2.6 Predictive Validity of Anxiety Sensitivity 48
2.7 Anxiety Sensitivity and Nonclinical Populations 51
2.8 Anxiety Sensitivity and Clinical Populations 54
2.8.1 Panic Disorder with or without Agoraphobia 54
2.8.2 Posttraumatic Stress Disorder 58
2.8.3 Social Phobia 60
2.8.4 Obsessive-Compulsive Disorder and
Generalised Anxiety Disorder 65
2.8.5 Specific Phobia 66
2.8.6 Major Depressive Disorder 68
2.9 Anxiety Sensitivity and Cognitive Behavioural Therapy 73
2.9.1 Anxiety Sensitivity and Combined Cognitive
Behavioural and Pharmacological Therapy 77
2.10 Rationale for Study One 80
2.11 Rationale for Study Two 83
2.12 Rationale for Study Three 84
2.13 Chapter Summary 85
Anxiety Sensitivity 14
Anxiety Sensitivity 15
CHAPTER TWO
REVIEW OF THE LITERATURE ON ANXIETY SENSITIVITY
2.1 Introduction
The previous chapter provided an overview of the theoretical approaches
to the fear of anxiety as well as a detailed account of the theory in which the
construct of anxiety sensitivity is embedded. Anxiety sensitivity was defined as
a cognitive, individual difference variable that is delineated by an individual’s
fear of anxiety related sensations and based on the belief that such sensations
result in harmful consequences (Reiss, 1991; Reiss & Havercamp, 1996; Reiss &
McNally, 1985). Due to the fear of anxiety sensations, individuals who exhibit
elevated anxiety sensitivity are hypothesised to respond fearfully to sensations of
involuntary autonomic arousal, which in turn, magnify the symptoms and
experience of anxiety. These ‘fear of anxiety’ sensations may lead to a vicious
cycle, whereby pre-existing beliefs about physiological arousal (or autonomic
change) may predispose an individual to respond fearfully, which can lead to an
increase in the experience of anxiety (McNally, 1994; Taylor, Koch, & McNally,
1992).
2.2 Introduction to the Anxiety Sensitivity Index
In order to test the anxiety sensitivity construct, Reiss and colleagues
(Reiss, Peterson, Gursky, & McNally, 1986) developed the Reiss-Epstein-Gursky
Anxiety Sensitivity Index (ASI), which is most cited measure for this construct to
date. (The ASI is presented in Appendix A). Based on their Expectancy Model
of Fear, Reiss et al. (1986) proposed that the concept of anxiety sensitivity
warranted investigation because of its theoretical role in the development of
Anxiety Sensitivity 16
anxiety disorders and presented an initial set of three studies which were designed
to empirically validate the ASI.
In the first study designed to evaluate the psychometric properties of the
ASI, Reiss et al. (1986) recruited two independent samples of university students
enrolled in an introductory psychology course and administered the ASI in a
group format. The ASI is a self-report measure containing 16-items relating to
fear of anxiety symptoms (e.g., ‘It scares me when I feel faint’; ‘It scares me
when my heart beats rapidly’; ‘When I notice that my heart is beating rapidly, I
worry that I might have a heart attack’). The ASI instructs participants to rate
each item on a five-point scale that ranges from (0) ‘very little’ to (4) ‘very
much’ with an individual’s total ASI score the sum of the responses to the 16-
items. Principal component factor analysis for both samples revealed that the
ASI contained a single anxiety sensitivity factor and that the items were highly
inter-related. Evidence of test-retest reliability over a two-week period was also
reported with Pearson product-moment correlations ranging from r = .71 to .74
for the men and women respectively, and r = .75 for the total sample. Reiss et
al. (1986) concluded this first study by arguing that the findings were evidence
of adequate psychometric properties of the ASI.
The second study reported by Reiss et al. (1986) was designed to
evaluate the criterion validity of the ASI testing the scale’s association with
psychopathological conditions such as agoraphobia and other anxiety disorder
classifications. Based on the clinical observation of Goldstein and Chambless
(1978) and Mathews, Gelder, and Johnston (1981) in which agoraphobia was
particularly associated with a ‘fear of fear’ as well as the notion that anxiety
sensitivity is one of two components of the ‘fear of fear’, Reiss et al. (1986)
Anxiety Sensitivity 17
hypothesised that ASI scores would be more elevated for individuals with a
diagnosis of agoraphobia than for those without such a disorder. Further, Reiss
et al. (1986) hypothesised that ASI scores would be elevated in those individuals
with a diagnosis of an anxiety disorder other than agoraphobia, than for
university students. Participants included 31 men and 32 women who were
patients of one of two clinics. Eleven participants were diagnosed with
agoraphobia according to the Diagnostic and Statistical Manual of Mental
Disorders – Third Edition (DSM-III; American Psychiatric Association [APA],
1980) with the remaining 52 individuals diagnosed with an anxiety disorder
other than agoraphobia. Scores of the university students from their first
investigation were also utilised. Results indicated that while individuals with a
diagnosis of agoraphobia reported significantly elevated ASI scores when
compared to individuals with an anxiety disorder other than agoraphobia,
individuals with an anxiety disorder other than agoraphobia also reported
significantly elevated ASI scores when compared to the university student group.
While the results provided evidence that the ASI possesses criterion validity, the
finding that anxiety sensitivity is especially associated with agoraphobia should
be regarded as preliminary because of the small size of the agoraphobic sample.
Reiss et al’s. (1986) final study was designed to validate the distinction
between anxiety and anxiety sensitivity. Reiss et al. (1986) argued that this
distinction was necessary in order to address the concern that the ASI is a measure
of anxiety sensitivity rather than just anxiety. Using stepwise regression analysis,
Reiss et al. (1986) reported that the ASI was capable of explaining up to 23.7% of
the additional variance of the Fear Survey Schedule – II (FSS-II), which is a
measure of subjective feelings of fear, over that which was explained by the
Anxiety Sensitivity 18
Taylor Manifest Anxiety Scale (TMAS) and the Anxiety Frequency Checklist
(AFC) using the university student sample from the first investigation. Thus, it
was argued that this result provided evidence to validate the distinction between
anxiety sensitivity and anxiety symptoms.
While the Reiss et al. (1986) study was important in understanding the
concept of anxiety sensitivity, one central controversy to emerge involved the
classification of the construct. The research and findings related to this
conceptual argument are discussed in the following section.
2.3 Anxiety Sensitivity and Trait Anxiety
Following the first series of studies into the construct of anxiety
sensitivity, one contentious issue to emerge was whether anxiety sensitivity was
distinguishable from trait anxiety. As defined earlier, anxiety sensitivity is a
specific tendency to respond fearfully to sensations associated with anxiety. The
argument that anxiety sensitivity is little more than trait anxiety has been
conceptually refuted by McNally (1989) who argued that anxiety symptoms
should not induce fear in individuals with trait anxiety who do not have
concurrent high anxiety sensitivity since anxiety sensitivity refers to fears of
anxiety symptoms that are based on the belief that these symptoms have harmful
consequences. Thus, McNally’s (1989) argument was based on the hypothesis
that anxiety symptoms are not intrinsically aversive stimuli, therefore only those
individuals with high anxiety sensitivity should experience fear of such
symptoms.
Lilienfeld, Jacob, and Turner (1989) challenged the conceptual and
empirical distinction between anxiety sensitivity and trait anxiety by arguing that
the results related to anxiety sensitivity could be explained by “the more
Anxiety Sensitivity 19
parsimonious hypothesis that the ASI simply measures trait anxiety” (p. 101).
There is some evidence in this claim in that several studies using the ASI and
measures of trait anxiety have reported moderately high correlations. For
example, Reiss et al. (1986) reported correlations on two samples of participants
of .43 and .50 between the ASI and the TMAS. Shostak and Peterson (1990)
reported a correlation of .36 between the ASI and the State Trait Anxiety
Inventory – Trait (STAI-T), while McNally (1989) and McNally and Lorenz
(1987) reported similar correlations of r = .55 and r = .46 respectively.
It is important to note that while Lilienfeld and colleagues (1989) have
taken these moderately high correlations as evidence that the construct of anxiety
sensitivity may be contaminated by trait anxiety, there is also sufficient empirical
evidence to support the claim that anxiety sensitivity and trait anxiety are related
but distinct constructs. For example, Reiss (1991) examined 11 datasets in
which correlations among state anxiety, trait anxiety, and anxiety sensitivity
were calculated. All 11 studies reported that correlations between trait anxiety
and anxiety sensitivity were only moderate, with the r squares ranging from zero
to 36% variance overlap. Similar results were reported by Taylor, Koch, and
Crockett (1991) who reviewed a number of studies and found a median overlap
of 21% variance between the ASI and measures of trait anxiety. However, while
trait anxiety is a higher-order construct that represents a propensity to respond
fearfully to stressors in general (McNally, 1999), results from several programs
of research have established that anxiety sensitivity is a lower-order construct
that represents a specific propensity to respond fearfully to one’s own anxiety
symptoms (Maller, 1988; McNally, 1989; McNally & Lorenz, 1987; Rapee &
Medoro, 1994; Reiss, 1991; Schmidt, Lerew, & Jackson, 1997; Taylor, 1996).
Anxiety Sensitivity 20
As a result of the numerous investigations into the classification of anxiety
sensitivity, a general consensus now exists within the available literature that
suggests anxiety sensitivity represents a genuine individual difference variable,
which is not only conceptually but also empirically distinct from trait anxiety.
While the above research appears to provide support for the usefulness of
the anxiety sensitivity construct as measured by the ASI, it is important to note
that one of the most contentious issues in the area of this construct concerns the
factor analytic structure of anxiety sensitivity. According to Reiss and
McNally’s (1985) Expectancy Theory, anxiety sensitivity is conceptualised as
unidimensional. Accordingly, Taylor, Koch, McNally, and Crockett (1992) have
argued that the ASI is comprised of a unidimensional factor structure. While
some research has been conducted that supports this view of anxiety sensitivity,
others have found support for a multidimensional structure, with all
investigations using varying sample sizes, factor analytic techniques as well as
varying rationales as to how factors are to be extracted. The following sections
presents the empirical research to date regarding the internal structure, items and
psychometric properties of the original ASI as well as the relatively new Anxiety
Sensitivity Index – Revised (ASI-R) measure developed by Taylor and Cox
(1998).
2.4 Factor Analytic Investigations of the ASI
The majority of factor analytic investigations of the ASI have employed
the method of exploratory factor analysis, with a limited number using
confirmatory factor analysis. This section will go through each investigation of
anxiety sensitivity in a chronological order. A table detailing each study in brief
is provided (see Table 2.1).
Anxiety Sensitivity 21
Table 2.1. Factor analytic studies of the Anxiety Sensitivity Index Author (s)
Sample Method of Factor Extraction
Method of Factor Rotation
Number of Factor (s)
Reiss et al. (1986) Two independent samples of university students (N = 49 and N = 98)
PCA None 1
Peterson and Heilbronner (1987)
122 self described ‘tense, anxious, or nervous’ university students
PCA Oblique and orthogonal 1
Telch, Shermis, and Lucas (1989)
840 university students PAF and PCA Varimax 4
Wardle, Ahmad, and Hayward (1990)
160 individuals with a diagnosis of agoraphobia and 120 nonclinical controls
PCA Varimax 4
Anxiety Sensitivity 22
Taylor, Koch, and Crockett (1991)
142 spider phobic university students and 93 individuals with a diagnosis of anxiety, depression or other disorder
PCA Oblique 1
Taylor et al. (1992)
142 spider phobic university students and 327 individuals with a diagnosis of an anxiety or stress-related disorder
CFA Oblique and orthogonal 4 (as previously identified
by Telch et al., 1989)
Cox, Parker, and Swinson (1996)
365 university students and 216 individuals with a diagnosis of panic disorder
CFA Orthogonal and oblique 4 (as previously identified by Wardle et al., 1990)
Taylor, Koch, Wood, and McLean (1996)
52 individuals with a diagnosis of panic disorder, 46 individuals with a diagnosis of major depression, and 37 individuals with a dual diagnosis of both
PCA Oblique 3
Zinbarg, Barlow, and Brown (1997)
432 individuals with a diagnosis of an anxiety or depressive disorder
PFA and CFA Oblique 3
Anxiety Sensitivity 23
Blais et al. (2001) 340 individuals with panic disorder, social phobia, or major depression and 50 university students
PCA Orthogonal and Oblique 2 (using 11 of the original
16 items of the ASI)
Schmidt and Joiner (2002)
233 individuals from the general community and 809 university students
PCA and PAF and CFA
Orthogonal and Oblique 2 (using 10 of the original
16 items of the ASI)
Note. PCA = Principal components analysis; PAF = Principal axis factor analysis; CFA = Confirmatory factor analysis.
Anxiety Sensitivity 24
In the original study of the ASI, Reiss et al. (1986) employed principal
components analysis (PCA) and reported a unidimensional solution, with 13 of
the 16-items loading .4 or more on the first factor and a high degree of inter-
relation between scale items. They concluded this study by arguing that the
results provided evidence that the ASI measures a coherent factor. However, this
investigation is flawed because they offered no rationale as to how many factors
were to be extracted as well as the very small sample size employed.
Peterson and Heilbronner (1987) investigated the factor analytic structure
of the ASI using a sample of 122 self-described ‘tense, anxious, or nervous’
university students who volunteered to take part in a relaxation study. Using
PCA, four factors were extracted using Kaiser’s criterion of eigenvalues ≥ 1 rule.
These four factors were subjected to an oblique simple structure rotation, in
which two factors associated with an a priori ‘fear of consequences’ and two
factors with an a priori ‘fear of physical sensations’ emerged. Using a specified
two-factor structure to further evaluate the a priori factors, an orthogonal rotation
was conducted which resulted in 11 of the 16-items loading on the first factor
and five items loading onto the second factor. Three items cross-loaded
substantially onto the first factor. Factor loadings for each analysis were not
provided and a rational was not offered for employing both rotational analyses.
As such, it is unclear what Peterson and Heilbronner (1987) were trying to show
as the implications for employing both rotational analyses are contradictory
because the oblique rotational analysis specifies that the dimensions of the ASI
are related whereas the orthogonal rotational analysis specifies that the
dimensions of the ASI are unrelated. Peterson and Heilbronner (1987) argued
that their results provided support for a strong single factor ASI and suggested
Anxiety Sensitivity 25
that only the total scale score should be used in future clinical and research work
due to the oblique rotation resulting in four factors with only a few items in each
factor and the high cross-loadings from the second factor of the two-factor
orthogonal rotation. The results of Peterson and Heilbronner (1987) study are
not definitive as they are vague as to how many factors should be retained as a
result of their conflicting rotational analyses and they concluded that the ASI was
best viewed as unidimensional when the unidimensional model was not tested.
In order to test the hypothesis that the ASI measured a unitary personality
variable, Telch, Shermis, and Lucas (1989) administered the ASI to 840
university students. In the first of two separate factor analyses of the ASI, Telch
et al. (1989) employed an interated principal axis factor (PAF) analysis with a
varimax rotation and Kaiser’s criterion of eigenvalues ≥ 1 as the factor extraction
rule. The procedure resulted in one-factor solution with all but three items
loading onto the principal factor. However, as the proportion of variance
explained by the unidimensional structure was relatively low (25%), PCA
followed by varimax rotation and Kaiser’s criterion of eigenvalues ≥ 1 was
performed in order to maximise the variance explained in the data. This second
procedure resulted in four factors capable of explaining 53.5% of the total
variance. The four factors obtained were (a) concern about physical sensations,
(b) concern about mental/cognitive incapacitation, (c) concern about loss of
control, and (d) concern about heart/lung failure. Factor three and four only
consisted of two and three items respectively and two items from factor two and
two items from factor four had moderate to high loadings on the first factor.
Telch et al. (1989) concluded by challenging the view in which ASI is regarded
as unidimensional and instead argued that the ASI measures several loosely
Anxiety Sensitivity 26
associated cognitive appraisal domains involved with the anticipated negative
consequences of anxiety. However, it should be noted that this interpretation
may be limited by the fact that Telch et al. (1989) imposed an orthogonal
rotation, in which dimensions are forced to be unrelated to one another. As such,
it would have been interesting, based on their conclusion, to investigate the factor
structure after an oblique rotation was applied and compare it to the orthogonal
solution.
Wardle, Ahmad, and Hayward (1990) assessed the factor structure of the
ASI in 160 individuals with agoraphobia and 120 nonclinical controls. Using
PCA with varimax rotation and Kaiser’s criterion of eigenvalues ≥ 1 rule for each
sample separately resulted in a four-factor solution that was capable of
explaining more than 60% of the total variance. For the agoraphobic sample, the
pattern of factor loadings was consistent with factors reflecting (a) fear of
physical sensations related to cardiovascular health, (b) fear of loss of mental
control, (c) fear of gastrointestinal upset, and (d) concern of publicly observable
anxiety symptoms. The item loadings for the nonclinical sample were regarded
as meaningless and uninterpretable. Examination of a two and three-factor
solution did not improve the meaningfulness of the internal structure of the ASI
and the four-factor solution was retained as it was regarded as a “very clear
factor structure” (Wardle et al., 1990, p. 330), despite the presence of six cross-
loading items in the agoraphobic sample and 11 cross-loading items in the
nonclinical sample. Wardle et al. (1990) concluded their investigation by
arguing that the findings supported a multidimensional view of anxiety
sensitivity without discussing the implications of the cross-loaded items.
Anxiety Sensitivity 27
Taylor, Koch, and Crockett (1991) examined the factor structure of the
ASI using 142 university students who indicated that they were spider phobic
according to the Fear Survey Schedule III (FSS-III) and 93 individuals from a
clinical setting who were diagnosed according to the DSM-III-R (APA, 1987)
criteria for anxiety, depression or other disorder. As Taylor et al. (1991) were
aware of the risk of over-extraction of factors using Kaiser’s criterion, they also
employed Cattell’s scree test as well as Thurstone’s criteria for simple structure
for determining the number of factors to retain in each sample. Using these three
rules of factor extraction in addition to PCA with oblique rotation, Taylor et al.
(1991) concluded that the ASI represented a unidimensional solution that
measures a fear of bodily sensations.
In another examination, Taylor et al. (1992) used confirmatory factor
analysis (CFA) to evaluate competing models previously published in the
available literature. The four models identified for inclusion in their study
included Reiss and McNally’s (1995) unifactorial solution, as well as Peterson
and Heilbronner (1987), Telch et al. (1989) and Wardle et al. (1990) four-factor
solutions. Taylor et al. (1992) examined responses from two samples including
142 spider-phobic university students and 327 individuals from a clinical setting
who met the DSM-III-R (APA, 1987) criteria for an anxiety or stress-related
disorder. For each four-factor model, Taylor et al. (1992) examined both the
oblique and orthogonal forms of each solution using a cut-off of at least .30 to
define a salient factor loading. Taylor et al. (1992) choose to retain the
orthogonal solutions because the oblique solutions for both the clinical and
student samples resulted in highly correlated factors (clinical sample range rs =
.34 to .92; student sample range rs = .03 to .97). They argued that the presence
Anxiety Sensitivity 28
of highly correlated factors indicated that the oblique factors tended to collapse
on one another. However it is not clear why the researchers rejected this solution
in favour of the orthogonal models as it is inconsistent with theoretical view of
anxiety sensitivity. Using only the orthogonal solutions and unifactorial model,
Taylor et al. (1992) identified the Telch et al. (1989) solution as the best-fitting
model for the clinical sample. It is important to note that examination of the chi-
square statistic as well as the Tucker-Lewis and Mulaik Normed Fit Index for all
models revealed that each model was well below the criteria of .8 to .9 for
retaining a specified model. In concluding their investigation, Taylor et al.
(1992) argued that the oblique solution could be seen as part of a hierarchical
factor structure in which all the oblique factors load onto a higher-order factor.
However, because the factors were highly correlated, Taylor et al. (1992) argued
they could best be regarded as measuring a single facet of anxiety sensitivity
because the four-factor model appeared to be the result of constraining the
factors to be orthogonal. Thus, these investigators proposed that the ASI
measures a single facet of anxiety sensitivity and is best regarded as unifactorial.
As such, it is unclear what Taylor et al’s. (1992) investigation revealed as their
findings and conclusions are confusing and arguably, not consistent with the
theory in which anxiety sensitivity is based.
In an investigation similar to Taylor et al. (1992), Cox, Parker, and
Swinson (1996) investigated the ASI by employing CFA to the unifactorial
model of Reiss and McNally (1985) and four-factor models of Peterson and
Heilbronner (1987), Telch et al. (1989) and Wardle et al. (1990). Each model
was tested using orthogonal and oblique solutions and using responses from two
samples consisting of 365 university students and 216 individuals who met the
Anxiety Sensitivity 29
DSM-IV (APA, 1994) criteria for panic disorder. Using multiple goodness-of-fit
indicators, Cox et al. (1996) reported the unidimensional model did not meet the
criteria standards for adequacy of fit in either the clinical or student sample.
Conversely, for both the clinical and student samples, the Peterson and
Heilbronner (1987) four-factor oblique model met all criteria for the goodness-
of-fit indices; however the chi-square statistic was significant. The orthogonal
model was not found to be an adequate fit of the data in either sample. The
Telch et al. (1989) four-factor oblique model was found to be a good fit of the
data in the student sample and an adequate fit of the clinical sample. Again, the
orthogonal model was not found to be a good fit of the data in either the clinical
or student sample. The four-factor oblique model based on Wardle et al.’s
(1990) agoraphobic sample met the criteria for the goodness-of-fit indices in the
student sample; however the chi-square statistic was significant. Neither the
orthogonal or oblique models were a good fit of the data for the clinical sample.
Finally, for both the clinical and student samples, the four-factor oblique and
orthogonal models based on Wardle et al’s. (1990) analysis with nonclinical
adults did not meet the criteria for retaining a specified model. Cox et al. (1996)
compared the goodness-of-fit indices for all three models that indicated
acceptable fit indices and found that the four-factor model of Peterson and
Heilbronner (1987) provided the best fit of the data in both the clinical and
student samples.
It is important to note that cut-off parameters for the fit indices Cox et al.
(1996) used were not as stringent as those used today. As such, not one of the
four models tested would have been retained as satisfying today’s goodness-of-fit
criteria. Cox et al’s. (1996) investigation found empirical support for a
Anxiety Sensitivity 30
multidimensional model of the ASI and they argued that there may be some
important domains within the anxiety sensitivity construct. However, it was
noted that one limitation to using the ASI in a multidimensional manner is that
there are too few items on the scale to produce reliable anxiety sensitivity
subscales. As such, Cox et al. (1996) argued that future work in the area of
anxiety sensitivity would be enhanced if the “ASI item pool was expanded to
allow for reliable subscales to be developed” (p. 596).
Taylor, Koch, Woody, and McLean (1996) examined the relationship
between anxiety sensitivity and depression in 52 individuals who met the DSM-
III-R (APA, 1987) criteria for panic disorder, 46 who met the criteria for major
depression and 37 individuals who met the criteria for both. Using the rule of
parallel analysis in order to determine the number of factors to retain, Taylor et
al. (1996) used PCA with oblique rotation and reported a three-factor solution
that accounted for 59% of the total variance. The three factors retained were (a)
fear of publicly observable arousal-related reactions, (b) phrenophobia or fear of
cognitive dyscontrol, and (c) fear of somatic sensations. Although all items had
salient loading of ≥ .40, it is important to note that factor two had only three
salient loadings and two items from factor three significantly cross-loaded onto
factor one. Taylor et al. (1996) concluded that the ASI as well as the construct of
anxiety sensitivity is multidimensional and posited that Lilienfeld et al’s. (1993)
argument, in which anxiety sensitivity is viewed as hierarchically arranged, may
be correct.
In order to test the hypothesis that the ASI is hierarchically arranged,
Zinbarg, Barlow, and Brown (1997) conducted both an exploratory and
confirmatory factor analytic investigation. Examining responses of 432
Anxiety Sensitivity 31
individuals from a clinical setting who met the DSM-III-R (APA, 1987) criteria
for an anxiety or depressive disorder, Zinbarg et al.’s (1997) first study employed
a first-order principal factor analysis and oblimin rotation. Three factors that
accounted for 44% of the total variance were extracted. Rather than employing
the Kaiser’s criterion to determine the number of factors to extract, Zinbarg et al.
(1997) decided to extract the maximum number of factors that would allow
replicability to other multidimensional solutions of the ASI already published in
the available literature. Using coefficients of convergence, Zinbarg et al. (1997)
reported that the three-factor solution converged well with the results of other
investigations, but became unstable when a fourth factor was extracted. The
three factors retained were labelled (a) AS-Physical Concerns, (b) AS-Mental
Incapacitation Concerns, and (c) AS-Social Concerns. Factor one was correlated
r = .44 with factor two and r = .36 with factor three, while the second factor was
correlated r = .33 with the third factor. In order to test the presence of a higher-
order general factor, Zinbarg et al. (1997) factor analysed the first-order
correlation matrix using principal factor extraction combined with Kaiser’s
criterion and scree test and reported a single general order factor. The loadings
of the three first-order factors on the high-order factor were .69, .64, and .52
respectively. Zinbarg et al. (1997) argued that the higher-order factor was indeed
a general factor evidenced by 15 of the 16 ASI items having a loading of .30 or
greater on it.
For their second investigation, Zinbarg et al. (1997) employed CFA in
order to test the hierarchical model of anxiety sensitivity identified in their
exploratory work as well as a unifactorial model and an alternative model with
three factors constrained to be orthogonal. Examining responses from the same
Anxiety Sensitivity 32
432 individuals used in their previous exploratory investigation, Zinbarg et al.
(1997) found that the three-factor hierarchical model provided a better fit of the
data than either the model with three factors constrained to be orthogonal or the
unifactorial model. It is important to note however that examination of the
goodness-of fit parameters for retaining a specified model were not as stringent
as they are today and it is entirely possible that all three models would have been
rejected using current standards. Further, Zinbarg et al. (1997) did not use a
separate sample on which to confirm their exploratory work. Thus, it is unclear
whether the hierarchical model would be invariant in another clinical sample if
the factor structure had not already been explored and fitted to the data.
Zinbarg et al. (1997) concluded their investigation by arguing that their
model is dissimilar to the description offered by Taylor et al. (1992) because the
latter did not perform the appropriate statistical tests of comparing the fit
between oblique multidimensional, unifactorial and orthogonal multidimensional
models. Similarly, they argued that as the estimation of the general factor
saturation exceeded .50, the total score of the ASI provides an interpretable
measure of anxiety sensitivity because it is consistent with the hypothesis of a
single construct that accounts for the majority of the variance in ASI total scores.
Therefore, Zinbarg et al. (1997) interpreted their results as countering the
hypothesis that the ASI contains multiple orthogonal factors and instead is best
viewed as hierarchical containing three first-order factors that load onto a single
second-order factor. The researchers suggested that future investigations should
focus on revising the ASI with the aim of producing higher general factor
saturation and increasing the reliability of the AS-Social Concerns subscale
through the inclusion of additional items.
Anxiety Sensitivity 33
Recently two other exploratory investigations have emerged that have
utilised different methodologies and rationales than the approach taken by
Zinbarg et al. (1997). The first of these was conducted by Blais et al. (2001).
In order to evaluate the ASI through the process of item analysis, Blais et
al. (2001) examined responses from 340 outpatients who met the DSM-III-R
(APA, 1987) criteria for panic disorder, social phobia, or major depression as
well as 50 university students. Initial item-to-scale correlations revealed that
items 1, 5, and 7 had adjustable correlations below the acceptable value of .30.
Follow-up between groups’ tests revealed that items 1, 5, 7, and 8 consistently
failed to discriminate between the groups in a theoretically expected manner,
while item 13 mainly reflected social phobia. Use of PCA with Kaiser’s
criterion, scree test, and root curve criterion for all of the items of the ASI
revealed either a three or four-factor solution. Although the two-factor structures
were similar, the four-factor solution was retained as it produced more primary
loadings and also had fewer hyperplaning items than did the three-factor
solution. Factor one contained 7 items and accounted for 37% of the variance.
Items on this factor were consistent with the assessment of fear of arousing
sensations. Factor two contained only 2 items and accounted for 9% of the
variance. Items on this factor appeared to measure fear related to loss of control.
Both items on factor two were shown to have adjusted item-to-scale correlations
below the cut-off level of .30. Factor three contained 3 items and accounted for
8% of the variance. This factor was considered stable as all three items loaded
exclusively on it with a loading of .70 or greater and was consistent with the
assessment of fear of loss of mental control. Finally, factor four contained only 2
items and accounted for 7% of the variance. Items on this factor appeared to
Anxiety Sensitivity 34
measure concern of gastrointestinal symptoms. Item 13 failed to load
significantly on any of the four factors and item 11 was significantly cross-
loaded on both factors four and one. Orthogonal or oblique rotations contained
essentially identical factor loadings.
Blais et al. (2001) has argued that as factors two and four only contained
two items each, it was possible that these factors would be unstable and would
have a reduced probability of replication in future investigations. As such, a
follow-up study using data from three previously published investigations (Ball,
Otto, Pollack, Uccello, & Rosenbaum, 1995; Eke & McNally, 1996; Schmidt et
al., 1997) was performed using the original 16-item ASI and the modified ASI
with five items eliminated (1, 5, 7, 8, and 13). Blais et al. (2001) referred to this
scale as the 11-item ASI. Results from all three investigations revealed that the
11-item ASI performed as well as the original 16-item ASI. Thus, Blais et al.
(2001) results revealed that through the removal of five problematic items (31%
of the scale); improvements could be made to the original ASI without loss of
construct validity.
Whilst the results from Blais et al. (2001) investigation showed some
promise regarding the 11-item ASI, it is important to note that it is not without
limitations. Firstly, in their examination of the ASI, Blais et al. (2001) retained
11 of the 16 original items that corresponded to one dimension relating to a fear
of somatic sensations of anxiety and another representing fear of loss of mental
control. Whilst these two dimensions are similar to those observed in other
investigations of the original ASI it is important to note that Blais et al’s. (2001)
investigation fails to provide clearer understanding of anxiety sensitivity than has
previously been reported in the literature. The two dimensions retained by Blais
Anxiety Sensitivity 35
et al. (2001) do not add significantly to the existing literature as they are too
general to highlight the specific anxiety sensitivity fears. Similarly, as the
dimension ‘fear of loss of mental control’ contains only three items, this
dimension could benefit from further item expansion, whereas the absence of a
dimension capable of accounting for fears related to publicly observable anxiety
symptoms excludes potentially important information from investigation using
the current scale.
Another limitation in their investigation concerns the proposition that the
methodology was to some extent flawed. In their analyses, Blais et al. (2001)
applied factor analysis to a participant pool comprised of clinical and nonclinical
participants. Whilst they assumed that it was logical to combine these two
groups into one heterogeneous sample to improve the generalisability of results,
the investigation could have been improved by retaining the clinical and
nonclinical participants as two separate testing groups. As their investigation
involved modifications to the original ASI item pool, a more rigorous test would
have been to make item modifications based on one a priori group of participants
and repeat the analysis in the second group of participants in order to test the
configural and metric invariance of their modified ASI model. As studies of the
original ASI suggest that the factor structure is convergent across clinical and
nonclinical populations (see Zinbarg, Mohlman, & Hong, 1999), this would have
been a more stringent test of the 11-item ASI model.
In a similar investigation to the above, Schmidt and Joiner (2002)
evaluated the ASI in a community sample of 233 individuals with no history of
psychiatric illness or spontaneous panic. Initial examination of the corrected
item-total correlations revealed that items 1, 5, and 7 were found to have the
Anxiety Sensitivity 36
fewest correlations greater than .20. Results of PCA and PAF using both
orthogonal and oblique rotations suggested retaining two factors after it was
determined that a four-factor solution would not be stable. As such, Schmidt and
Joiner (2002) argued that items 1, 5, and 7 be removed from the measure.
Additional analyses revealed that through the removal of these three problematic
items, the internal consistency of the scale improved slightly. PCA and PAF of
the revised 13-item ASI revealed that, in general, analyses using orthogonal
rotations were consistent with a three-factor solution consisting of dimensions
relating to ‘fears of mental catastrophe’, ‘cardiopulmonary fears’, and ‘vasovagal
fears’. Conversely, analyses using oblique rotations were consistent with a two-
factor solution as the items from the ‘vasovagal fears’ factor loaded onto the
‘fears of mental catastrophe’ and ‘cardiopulmonary fears’ factors.
Schmidt and Joiner (2002) retained both models and examined them
again through the process of confirmatory factor analysis. For comparison
purposes, the orthogonal and oblique multifactorial models of Peterson and
Heilbronner (1987), Telch et al. (1989), Blais et al. (2001) and four separate
unifactorial solutions were also examined. Using the same population on which
the exploratory analyses were conducted, Schmidt and Joiner (2002) found that a
two-factor oblique model produced the best overall fit of all models tested, by
meeting the various goodness-of-fit cut-off criteria using only 10 items from the
ASI. The first factor, labelled ‘fear of mental catastrophe’ contained 6 items (2,
11, 12, 13, 15, and 16); while the second factor, labelled ‘fear of
cardiopulmonary symptoms’ contained 4 items (6, 8, 9, and 10). While the fit
indices for this model were marginal at best, Schmidt and Joiner (2002) retested
Anxiety Sensitivity 37
this 10-item, two-factor model on another population of 809 undergraduate
university students and again found a marginal fit of the student data.
This investigation is an improvement on the Blais et al. (2001) study as
the methodology was enhanced by employing the hypothesis testing approach of
confirmatory factor analysis and retesting using a separate sample of participants.
The investigation served to highlight the inadequacy of the previously identified
ASI models in the available literature and revealed that the shortfalls of the ASI
could be improved by expanding the item pool in order to construct more reliable
subscales.
In summary, it is important to note that Reiss et al. (1986) originally
developed the ASI to measure the single anxiety sensitivity component of their
interactional model of Expectancy Theory. Thus, the ASI was originally
developed to measure a unidimensional construct. According to Reiss et al.
(1986), anxiety sensitivity was defined as a cognitive, individual difference
variable delineated by an individual’s fear of anxiety related sensations and based
on the belief that such sensations result in harmful consequences. Thus the ASI
was developed in line with this conceptual definition. However, results of
exploratory and confirmatory factor analytic investigations of the ASI reviewed
at this juncture have revealed that most investigators have found that anxiety
sensitivity is in fact comprised of multiple dimensions. There has been some
argument over the exact number and nature of dimensions with some
investigators finding support for as few as two dimensions (Blais et al., 2001;
Schmidt & Joiner, 2002), and others finding support for three (Stewart et al.,
1997; Taylor, 1996; Zinbarg et al., 1997), or four dimensions (Peterson &
Heilbronner, 1987; Telch et al., 1989). Further, while different methods of
Anxiety Sensitivity 38
analysis have been employed, some inappropriately and some without a sound
theoretical rationale Zinbarg et al. (1997) contended that anxiety sensitivity, as
measured by the 16-item ASI, is hierarchically structured, consisting of multiple
first-order dimensions that relate to a fear of social, somatic, or psychological
consequences of anxiety that load onto a single second-order dimension.
However, while some level of agreement is emerging regarding this
interpretation, it is important to note that a consensus has not yet been reached on
this key issue.
While the ASI is important in furthering our understanding of anxiety
sensitivity, the repeated attempts to clarify the dimensionality of anxiety
sensitivity using the 16-item ASI is problematic, as it was never designed to
measure a multidimensional construct. As such, the original ASI contains a
number of flaws that require attention. As an unresolved question concerns the
number and nature of the first-order anxiety sensitivity dimensions, it is
important to note that several investigations have repeatedly revealed that the
ASI contains too few items to reliably identify the major anxiety sensitivity
facets. For example results from several investigations (see Deacon &
Valentiner, 2001; McWilliams & Cox, 2001; Stewart et al., 1997) have revealed
that the ASI contains only two items capable of measuring the domain of ‘fear of
publicly observable anxiety and anxiety-related symptoms’. A further limitation
with the ASI is that it also contains too few items to determine whether the ‘fear
of physical sensations’ dimension consists of a number of dimensions such as
fear of respiratory symptoms, fear of cardiovascular symptoms and fear of
gastrointestinal symptoms. Thus, precisely which physical sensations are feared
is unclear. Examination of the scale reveals that the ASI only contains two items
Anxiety Sensitivity 39
capable of assessing cardiovascular concerns, two items capable of assessing
gastrointestinal concerns and one item capable of assessing respiratory concerns.
Tabachnick and Fidell (2001) have argued that interpretation of factors defined
by only one or two variables is problematic for psychological test measurement
research as it results in poorly defined constructs. As such, it is clear that the
ASI contains too few items for these specific dimensions to be assessed
adequately as the emergence of a factor with only one or two items is
conceptually meaningless. Further, caution should be used in the interpretation
of scores from the various dimensions that have been derived from use with the
original ASI, because of the instability that may result from dimensions that do
not contain an adequate number of items.
A further problem with the ASI is that it contains a number of ambiguous
statements (e.g., ‘unusual body sensations scare me’; ‘it is important for me to
stay in control of my emotions’; ‘it embarrasses me when my stomach growls’;
or ‘it scares me when I am nervous’). Items such as these are problematic
because they do not act as reliable markers for the specific anxiety sensitivity
dimensions as individuals could obtain high scores for a number of different
reasons.
Given that investigations of the first-order dimensions of the ASI have
validated the importance of the multidimensional perspective of anxiety
sensitivity (see Zinbarg et al., 1999 for review), Taylor and Cox (1998)
developed the 36-item Anxiety Sensitivity Index – Revised (ASI-R), in order to
provide a more comprehensive measure of the first-order anxiety sensitivity
dimensions. While the 36-item ASI-R retains the same instructions and response
format as the 16-item ASI, it contains a broader selection of items, and hence
Anxiety Sensitivity 40
dimensions, for assessing the explicit domains of anxiety sensitivity. In addition,
the ASI-R contains 10 items which have been adopted from the original ASI.
These items were included because they were considered reliable markers for
specific anxiety sensitivity dimensions (The ASI-R is presented in Appendix B).
A table detailing the three published investigations of the ASI-R is provided (see
Table 2.2).
2.5 The Anxiety Sensitivity Index – Revised
In their first study designed to examine the internal structure of the ASI-
R, Taylor and Cox (1998) examined responses from 155 individuals attending
outpatient clinics who also met the DSM-IV (APA, 1994) criteria for an anxiety
disorder, trichotillomania, major depression, Tourette’s disorder, kleptomania,
somatoform pain disorder, adjustment disorder, and psychological factors
affecting a medical condition. The ASI-R is a self-report measure containing 36
items, 10 of which are from the 16-item ASI, that relate to a fear of anxiety
symptoms (e.g., ‘smothering sensations scare me’; it is important for me not to
appear nervous in public’; ‘when I feel a pain in my chest, I worry that I’m going
to have a heart attack’; ‘When my thoughts seem to speed up, I worry that I
might be going crazy’). It contains the same instructions and format of the ASI,
with participants rating each item on a five-point scale that ranges from (0) ‘very
little’ to (4) ‘very much’.
Drawing on the domains derived from previous factor analytic
investigations of the ASI, Taylor and Cox (1998) constructed the ASI-R to
measure six a priori dimensions of anxiety sensitivity that related to a fear of
cardiovascular, respiratory, gastrointestinal, publicly observable, dissociative and
neurological anxiety symptoms. Using the rule of parallel analysis in order to
Anxiety Sensitivity 41
Table 2.2. Factor analytic studies of the Anxiety Sensitivity Index-Revised Author (s)
Sample Method of Factor Extraction
Method of Factor Rotation
Number of Factor (s)
Taylor and Cox (1998)
155 outpatients PCA and PAF Oblique 4
Zvolensky et al. (2003) Nonclinical participants from six countries, including Canada (N = 478), France (N = 701), Mexico (N = 418), Netherlands (N = 536), Spain (N = 480), and the United States (N = 173)
PFA Oblique 2
Deacon, Abramowitz, Woods, and Tolin (2003)
Two independent samples of university students ( N = 558 and N = 444)
PCA and PAF Oblique 4
Note. PCA = Principal components analysis; PAF = Principal axis factor analysis.
Anxiety Sensitivity 42
determine the number of factors to retain at both the mean and 95th percentile
eigenvalue, Taylor and Cox (1998) used both PCA and PAF with oblique
rotation and reported a four-factor solution that was just below the cut-off values
of the mean and 95th percentiles. The four-factor solution accounted for 60% of
the total variance for PCA and 55.1% of variance for PAF. The four factors
retained were (a) fear of respiratory symptoms, (b) fear of publicly observable
anxiety reactions, (c) fear of cardiovascular symptoms, and (d) fear of cognitive
dyscontrol. The pattern of salient loadings was very similar across both PCA and
PAF with only a small number of hyperplane items.
The matrix of correlations for the four first-order factors computed from
the PCA solution was analysed again via PCA in order to identify the presence of
second-order factors. Similarly, the matrix of correlations for the four first-order
factors computed from the PAF solution was also analysed again via PAF in
order to identify any second-order factors. For each analysis, only one
eigenvalue was greater than 1, which led to a single second-order factor being
extracted. This second-order or general factor accounted for 52% of the variance
in the PCA solution and 44.9% of the variance in the PAF solution. As such, the
results indicated the presence of a hierarchical solution in which four first-order
dimensions loaded on a single second-order or general anxiety sensitivity
dimension. In order to account for the individual variance of the first and
second-order factors as well as to compute the loadings of each item on the
second-order factor and residualised loadings of each item on the first-order
factors, a Schmidt and Leiman transformation and application of Jensen and
Weng’s formulae were conducted. This process revealed that the pattern of
loadings of the first-order factors were similar to those obtained by PCA and
Anxiety Sensitivity 43
PAF with oblique rotation and that the second-order factor accounted for
substantially more variance (40.8% for PCA) than the four first-order factors
(6.7%, 5.7%, 6.6%, and 4.6% for PCA respectively). As a result, Taylor and
Cox (1998) argued that these findings suggest “anxiety sensitivity arises largely
from a general anxiety sensitivity factor, with more modest contributions from
the four specific factors” (p. 475).
In their second study, correlations between the ASI-R first and second-
order factors and the original ASI, Beck Anxiety Inventory (BAI), Beck
Depression Inventory (BDI) and medication status were investigated (Taylor &
Cox, 1998). There was evidence of large correlations (rs ≥ .60) with the ASI,
which was predictable given that the ASI-R contains 10 of the 16-items from the
ASI. The scale’s first and second-order dimensions were also moderate to highly
correlated with the BAI and moderately correlated with the BDI. The ASI-R
factors were uncorrelated with medication status. Thus, there was sufficient
support for the concurrent validity of the ASI-R.
Taylor and Cox’s (1998) third investigation compared the factor scores of
the ASI-R across four groups of participants that included individuals with a
diagnosis of panic disorder and assessed before treatment; individuals with a
diagnosis of panic disorder and assessed after treatment; a group of individuals
classified as ‘other anxiety disorders’ and assessed before treatment; and the
remaining individuals who were classified as ‘non anxiety disorders’ and were
assessed before treatment. Using one way analysis of variance for each factor
score with Newman-Keuls post hoc comparisons for significant differences
between groups, Taylor and Cox (1998) reported that the panic disorder group
who were assessed at pre-treatment scored significantly higher on all of the first
Anxiety Sensitivity 44
and second-order dimensions when compared to the other diagnostic groups.
Conversely, a significant difference was not observed between the ‘other anxiety
disorders’ group assessed at pre-treatment when compared to the ‘nonanxiety
disorder’ group who were also assessed at pre-treatment for any of the first or
second-order ASI-R dimensions. Conversely, it is interesting to note that both
the ‘other anxiety disorders’ and ‘nonanxiety disorder’ groups assessed at pre-
treatment scored significantly higher on the ‘fear of respiratory symptoms’
dimension as well as on the second-order or total score dimension when
compared to the panic disorder group assessed at post-treatment.
Since Taylor and Cox (1998), only two investigations have appeared in
the extant literature that attempt to determine the factor structure and
psychometric properties of the expanded ASI-R. The first investigation,
conducted by Zvolensky et al. (2003), sought to determine the factor structure
and internal consistency of the ASI-R using a large, diverse sample of
nonclinical participants from six countries, including Canada, France, Mexico,
the Netherlands, Spain and the United States. PFA with oblique rotation was
applied to each of the six samples as well as a within-group correlation matrix.
Parallel analysis found that a two-factor solution was the most replicable across
all seven datasets. Zvolensky et al. (2003) reported that the two dimensions
retained related to (1) fear of somatic sensations and (2) fear of social-cognitive
concerns. Subsequent investigations revealed that the two dimensions contained
an acceptable level of internal consistency; the two dimensions were moderate to
strongly correlated in each dataset, denoting the presence of a hierarchical
structure; and factor comparability coefficients indicated that the different
versions of both dimensions were, in most cases, consistent with one another
Anxiety Sensitivity 45
across datasets. Zvolensky et al. (2003) concluded their investigation by
proposing that although their investigation offered important insight into the
nature of some aspects of anxiety sensitivity, it was an early first attempt to
examine the construct using the ASI-R in a large, diverse sample and as such,
should be regarded in this context.
Deacon, Abramowitz, Woods, and Tolin (2003) set out to replicate
Taylor and Cox’s (1998) results and provide an adequate evaluation of the ASI-
R’s psychometric properties. Using both PCA and PAF with oblique rotation on
a large sample of undergraduate university students, these researchers reported
that parallel analysis using the mean and 95th percentile eigenvalues revealed
four dimensions relating to (1) beliefs about the harmful consequences of
somatic sensations, (2) fear of publicly observable anxiety reactions, (3) fear of
cognitive dyscontrol, and (4) fear of somatic sensations without explicit
consequences. The presence of a hierarchical solution was examined using the
same methodology as Taylor and Cox (1998) and again, a single second-order
dimension was extracted indicating that the ASI-R is hierarchically arranged.
Coefficients of congruence indicated that while the dimensions relating to fear of
publicly observable anxiety reactions and cognitive dyscontrol were highly
replicable to the same dimensions obtained by Taylor and Cox (1998); the two
somatic dimensions showed less convergence. In addition, correlations between
the ASI-R dimensions and other related measures were theoretically consistent
with the anxiety sensitivity construct. However, due to differences between the
somatic factors obtained in their study when compared to the results obtained by
Taylor and Cox (1998); Deacon et al. (2003) conducted another investigation
Anxiety Sensitivity 46
using a second, independent sample of undergraduate university students in order
to increase confidence in the reliability of their factor analytic findings.
Employing the same methodology to identify the number of factors to
retain, Deacon et al. (2003) again found that the ASI-R contained a hierarchical
solution of four first-order dimensions that loaded onto a single, second-order
dimension. Essentially, this solution was identical to the solution identified in
their previous investigation with the dimensions assessing cognitive dyscontrol
and publicly observable anxiety symptoms closely replicating those obtained by
Taylor and Cox (1998). Conversely, the two dimensions relating to somatic
sensations consisted of a blend of the two somatic dimensions reported by Taylor
and Cox (1998) and therefore did not replicate their findings. Deacon et al.
(2003) concluded their investigation by arguing that future studies of the ASI-R
using diverse populations were warranted.
Results of the above investigations of the ASI-R reveal that the revised
measure has the potential to open up new and important avenues for further
investigation. Although a great deal of research has been conducted into the
anxiety sensitivity construct, investigators have repeatedly noted the inherent
problems of measuring the construct when exclusively using the 16-item ASI.
Thus, in an effort to present a more comprehensive measure of the anxiety
sensitivity construct, Taylor and Cox (1998) developed the 36-item ASI-R, with
initial research showing promise.
It is important to reiterate that most of the 16-item ASI models evaluated
in the literature have revealed poor fit of the data and, at best, the best-fitting
models have provided only a marginal fit of the data. The 16-item ASI was
constructed to measure what was originally conceptualised as a unitary construct
Anxiety Sensitivity 47
(Reiss et al., 1986) and because the 16-item ASI contains a relatively small
number of items, most of which measure fears of somatic sensations (e.g.,
Stewart et al., 1997), the scale is too abbreviated to adequately measure the major
anxiety sensitivity facets. It has been ascertained that the 16-item ASI contains
too few items to determine whether the ‘fear of somatic sensations’ dimension,
for example, may actually consist of several factors, such as fears of cardiac
symptoms and fears of gastrointestinal symptoms. Further, past research has
repeatedly revealed that the 16-item ASI dimension related to ‘Social Concerns’
tends to be plagued by low levels of internal consistency when compared to the
other dimensions and usually explains the least amount of variance in the overall
solution (Peterson & Plehn, 1999; Stewart et al., 1997, Taylor et al., 1996).
Finally, evidence of a number of ambiguous statements contained in the 16-item
ASI is problematic because such items do not act as reliable markers for the
specific anxiety sensitivity dimensions. Overall, the numerous limitations of the
16-item ASI are of concern as they reduce an investigator’s confidence when
making any theoretical claims about the anxiety sensitivity construct.
In light of the limitations of the 16-item ASI, the 36-item ASI-R
developed by Taylor and Cox (1998) is a promising instrument for measuring
anxiety sensitivity. The research presented at this point, using the ASI-R, has
revealed that the lower-order dimensions demonstrate theoretically consistent
relationships with criterion variables and that the second-order dimension
correlate very well with the 16-item ASI, indicating that both instruments
measure the same construct. Further, given that the content validity of the ASI-R
is an improvement on the 16-item ASI, it is contended that it is a better measure
for use in studies investigating anxiety sensitivity dimensions.
Anxiety Sensitivity 48
Given the important role that anxiety sensitivity plays in anxiety
pathology, the following sections aim to discuss the research and findings related
to how the construct is manifested between and within various anxiety and mood
pathologies as well as normative groups. As a psychological construct, anxiety
sensitivity has become associated with understanding panic pathology in
particular and understanding negative emotional functioning found in other
anxiety and mood psychopathology in general. Further, as the degree to which
treatment, and in particular cognitive-behavioural therapy (CBT), successfully
impacts upon anxiety sensitivity will also be examined and discussed. It is
important to note that in all instances the research presented from this point
forward has been conducted using the 16-item ASI only.
2.6 Predictive Validity of Anxiety Sensitivity
From the time when Reiss and McNally (1985) first proposed the
expectancy model of fear, extensive research has culminated to demonstrate a
relationship between anxiety sensitivity and various anxiety-related phenomena.
Some of the most persuasive evidence of the predictive validity of anxiety
sensitivity and the ASI has come from prospective research. For example, in a
prospective study designed to predict the frequency and intensity of panic attacks
after a 3-year period, Maller and Reiss (1992) administered the ASI to 151
university students in 1984 and retested these same students in 1987 using the
ASI, Panic Attack Questionnaire (PAQ), Anxiety Disorders Interview Schedule-
Revised (ADIS-R) and the State-Trait Anxiety Inventory (STAI). The presence
or absence of an anxiety disorder was evaluated by a ‘blind’ judge using the
DSM-III-R criteria (APA, 1987). Maller and Reiss (1992) found that after a 3-
year period, scores on the ASI were not only predictive of the intensity and
Anxiety Sensitivity 49
prevalence of panic attack, but that individuals with elevated levels of anxiety
sensitivity were approximately five time more likely to have developed an
anxiety disorder (panic disorder being the most common diagnosis) than those
individuals with low anxiety sensitivity scores. The test-retest reliability for the
ASI from 1984 to 1987 was r = .71. This provided evidence that the construct of
anxiety sensitivity was not only stable, but also that anxiety sensitivity is an
individual difference variable that plays an important role in understanding the
development and maintenance of anxiety and anxiety disorders. However, as the
presence of panic disorder and agoraphobia was not measured in 1984, the
results of Maller and Reiss’s (1992) study should to be interpreted with caution,
as it is unclear whether participants had a pre-existing panic disorder or
experienced panic attacks.
In response to Maller and Reiss’s (1992) study, Ehlers (1995) conducted
a one-year prospective investigation of 39 individuals who met the DSM-III-R
(APA, 1987) diagnosis of current panic disorder, panic disorder in remission (n =
17), infrequent panic (n = 46), or simple phobia (n = 22). In addition to this, 45
control participants were also included in the study. Ehlers (1995) results
showed that while elevated levels of anxiety sensitivity did not predict
maintenance or relapse among individuals with panic disorder who were
receiving treatment during the follow-up, it was capable of predicting a poorer
outcome for those individuals who did not receive treatment. Anxiety sensitivity
was also capable of predicting a poorer outcome in the frequency of spontaneous
panic for individuals who experienced infrequent panic attacks and were left
untreated. Untreated individuals with infrequent panic who reported elevated
levels of anxiety sensitivity were more likely to continue experiencing panic
Anxiety Sensitivity 50
attacks compared to individuals with infrequent panic and lower levels of anxiety
sensitivity. Individuals with a diagnosis of simple phobia, or the control
participants who experienced their first panic attack during follow-up, showed
elevated anxiety sensitivity at the initial assessment stage when compared to
individuals who did not experience panic attacks. Finally, Ehlers (1995) argued
that even though multiple regression analyses employed a liberal significance
level of p < .10, anxiety sensitivity demonstrated incremental validity in
predicting the continuation of panic disorder for the clinical sample as well as
panic attacks during follow-up for the total sample beyond what could be
attributable to previous panic attack frequency and other predictor variables. As
such, Ehlers’ (1995) study suggests that elevated levels of anxiety sensitivity are
related to poorer outcomes for untreated individuals experiencing anxiety or an
anxiety disorder.
In another prospective study, Schmidt et al. (1997) studied a large (N =
1,172) nonclinical population of first year cadets at the United States Air Force
Academy over a highly stressful five-week period and found that after
controlling for individuals’ history, elevated levels of anxiety sensitivity
predicted the development of anxiety and depressive symptomatology,
spontaneous panic attacks as well as functional impairment and disability created
by anxiety. Schmidt et al. (1997) argued that the findings provided persuasive
evidence for anxiety sensitivity as a risk factor in the development of panic
attacks and other anxiety symptoms.
Finally, in a recent prospective study of panic symptoms, panic attacks,
panic disorder and anxiety sensitivity among 178 individuals who were
undergraduate university students, Plehn and Peterson (2002) found that after
Anxiety Sensitivity 51
controlling for history of panic symptoms, the ASI was the strongest predictor of
the occurrence of panic symptoms and panic attacks after a period of 11 years.
As a result of the prospective research conducted, there is sufficient
evidence to demonstrate that anxiety sensitivity is a dispositional construct that
plays an important role in the development of subsequent anxiety symptoms.
Further, it is a valuable identifier of individuals at risk of panic attacks or an
anxiety disorder. As such, individuals with elevated anxiety sensitivity are more
likely to have past histories of panic attacks and are more likely to react
anxiously to physiological arousal.
2.7 Anxiety Sensitivity and Nonclinical Populations
In addition to prospective research, numerous investigations have also
been conducted in order to predict and understand panic experiences in
nonclinical populations. For example, in one study designed to determine
whether elevated levels of anxiety sensitivity are a result of panic attack
experiences or whether elevated levels of anxiety sensitivity exist independently
of a history of unpredictable panic attacks, Donnell and McNally (1990)
administered the ASI and PAQ to 389 university students and found that
individuals with elevated levels of anxiety sensitivity more frequently reported a
greater personal and family history of panic than did individuals in the other sub-
groups. However, while most individuals with elevated levels of anxiety
sensitivity reported never experiencing an unpredictable panic attack, this same
finding was also observed for individuals who reported low levels of anxiety
sensitivity. As a result, Donnell and McNally (1990) argued that the findings
from their investigation supported the notion of anxiety sensitivity as being a
Anxiety Sensitivity 52
vulnerability factor in the experience of panic and not simply a result of panic
experiences.
In a similar study examining the relationship of anxiety sensitivity to the
occurrence of nonclinical panic attacks, Cox, Endler, Norton, and Swinson
(1991) classified 265 university students into high, medium or low anxiety
sensitivity groups and classified as nonclinical panickers according to DSM-III-R
(APA, 1997) criteria. Fifty percent of individuals with elevated anxiety
sensitivity reported experiencing both cued and non-cued panic attacks in the
past year. Additionally, individuals with elevated anxiety sensitivity were more
likely to anticipate un-cued panic attacks in the future when compared to
individuals who reported lower levels of anxiety sensitivity. Similar results have
also been reported by Asmundson and Norton (1993) who found that among 461
university students, individuals with elevated levels of anxiety sensitivity were
more likely to experience non-cued panic attacks as well as greater cued panic
attacks when compared to individuals with medium or low levels of anxiety
sensitivity. Brown and Deagle (1992) examined the prevalence and nature of
panic in 171 university students using the ADIS-R and reported that individuals
with panic endorsed significantly higher levels of anxiety sensitivity, trait
anxiety, depression, cognitions, and self-focused attention than did individuals
without panic. As such, the results of Cox et al. (1991), Asmundson and Norton
(1993), and Brown and Deagle (1992) reveal that anxiety sensitivity is frequently
associated with panic attacks in nonclinical individuals.
In addition to those studies that have only utilised nonclinical
populations, the concept of anxiety sensitivity has also been found to be elevated
in groups of individuals with anxiety disorders when compared to normal
Anxiety Sensitivity 53
controls. For example, Rapee, Ancis and Barlow (1988) administered the ASI to
155 undergraduate university students and 18 individuals with a DSM-III-R
(APA, 1987) diagnosis of panic disorder with or without agoraphobia and found
that individuals with panic disorder scored significantly higher overall anxiety
sensitivity when compared to individuals with nonclinical panic attacks and those
who had never experienced a panic attack before. Wardle, Ahmad, and Hayward
(1990) examined anxiety sensitivity in 160 individuals with agoraphobia and 120
nonclinical control participants and reported that the agoraphobic group scored
significantly elevated levels of anxiety sensitivity when compared to normative
controls. Similarly, Reiss et al. (1986) found that individuals with a diagnosis of
agoraphobia or an anxiety disorder other than agoraphobia displayed a
significantly higher level of anxiety sensitivity when compared to a sample of
university students. Taylor, Koch, and McNally (1992) found that with the
exception of simple phobia, anxiety sensitivity was elevated in individuals with
panic disorder, post-traumatic stress disorder (PTSD), generalised anxiety
disorder (GAD), obsessive-compulsive disorder (OCD) and social phobia when
compared to a normative control sample. Thus, it appears that anxiety sensitivity
is able to differentiate between individuals with an anxiety disorder and
nonclinical controls in a theoretically expected manner.
Finally, representative investigations on the ASI for normative
populations have also been conducted. Results from several investigations (Cox
et al., 1991; Lilienfeld, 1997; Peterson & Heilbronner, 1987; Reiss et al., 1986;
Stewart et al., 1997; Telch et al., 1989) have demonstrated that university
student’s mean ASI scores are usually one to two standard deviations lower when
compared to clinical populations; while studies utilising results from the general
Anxiety Sensitivity 54
community tend to be even lower (see Asmundson & Stein, 1994; Wardle et al.,
1990). However, what is not yet known is how individuals from the general
community respond with reference to the first-order dimensions of either the
original ASI or 36-item ASI-R and as yet, there is no theoretical foundation on
which to base hypotheses.
2.8 Anxiety Sensitivity and Clinical Populations
The prediction that fear of anxiety related sensations may serve as a
psychological risk factor in the development of anxiety disorders in general and
panic attacks in particular is a central doctrine in anxiety sensitivity theory. As
such, considerable research has investigated anxiety sensitivity among the
various anxiety disorder and psychopathology classifications. Additionally, as
elevated levels of anxiety sensitivity are more likely to lead to a vicious cycle of
anxiety amplification that has the potential to culminate in a panic attack, there
has also been a great deal of research generated that is interested in whether
anxiety sensitivity is capable of discriminating panic disorder from other anxiety
disorders.
2.8.1 Panic disorder with or without Agoraphobia. According to the
DSM-IV-TR (APA, 2000), panic disorder is depicted by the occurrence of
repeated, unexpected panic attacks that are accompanied by concern about
having another panic attack, the possible repercussions or consequences of the
panic attacks, or considerable behavioural change related to the attacks. A panic
attack is defined as:
“a discrete period in which there is the sudden onset of intense
apprehension, fearfulness, or terror, often associated with feelings of
impending doom. During these attacks, symptoms such as shortness of
Anxiety Sensitivity 55
breath, palpitations, chest pain or discomfort, choking or smothering
sensations and fear of ‘going crazy’ or losing control are present” (DSM-
IV-TR, APA, 2000, p. 429).
As individuals with panic disorder exhibit characteristic traits or
attributions about the implications of experiencing a panic attack, some
individuals fear that the attacks signify the presence of a life-threatening illness
such as cardiac disease, seizure or respiratory disorder whereas others may fear
that the panic attacks are a sign that they are going crazy or losing control. In
addition to the concern about the experience and implications of panic attacks,
individuals with panic disorder also report continuous or sporadic feelings of
anxiety that are not related to any specific situation or event. Conversely, others
develop excessive apprehension about the outcome of everyday activities and
encounters, especially those related to health. For example, individuals with
panic disorder are more likely to anticipate a catastrophic outcome from a mild
physical symptom, such as having a headache and interpreting it as evidence for
the presence of a brain tumour or hypertensive crisis. Finally, panic disorder is
distinguishable from other anxiety disorders because the panic attacks are not
situationally bound or situationally predisposed (APA, 2000).
In their investigation that introduced the 16-item ASI, Reiss et al. (1986)
found that individuals with a DSM-III (APA, 1980) diagnosis of agoraphobia
displayed significantly higher anxiety sensitivity scores when compared to a
mixed group of individuals with another anxiety disorder, such as OCD, social
phobia, and simple phobia. Conversely, in an investigation that examined how
anxiety sensitivity varied across the DSM-III-R (APA, 1987) anxiety disorders,
Taylor et al., (1992) administered the ASI to 313 individuals who were receiving
Anxiety Sensitivity 56
CBT for an anxiety disorder and found that with the exception of posttraumatic
stress disorder (PTSD), individuals with panic disorder scored significantly
higher on the ASI when compared to individuals with another anxiety disorder.
When they compared panic disorder group scores from those of PTSD at the
mean item level, Taylor et al. (1992) found that individuals with panic disorder
reported significantly greater mean scores on items relating to a fear of cardiac,
respiratory, and intestinal symptoms as well as the general experience of anxiety.
Conversely, when comparing panic disorder from those other anxiety disorders
combined, Taylor et al. (1992) found that individuals with panic disorder scored
significantly greater mean scores on all but two items relating to embarrassment
or perceptions of others. As such, it appears that based on the results of this
investigation, there is very little difference between each diagnostic groups
concerns relating to publicly observable anxiety symptoms.
In a similar study of the ASI, Sandin, Chorot, and McNally (1996) tested
the validity of Spanish version of the ASI in a sample of 126 individuals with a
DSM-III-R (APA, 1987) anxiety disorder diagnosis as well as 30 normative
controls. Sandin et al. (1996) reported that individuals with panic disorder
scored significantly higher overall anxiety sensitivity compared to individuals
with another anxiety disorder and to normal controls. In addition, while
individuals with social phobia and GAD scored significantly greater anxiety
sensitivity compared to normal controls, there was no significant difference
between normal controls, individuals with simple phobia and OCD. Mean level
item analysis between the panic disorder group scores and other anxiety disorder
groups combined found that the mean score of individuals with panic disorders
was significantly greater on 10 out of the 16-items. These findings are similar to
Anxiety Sensitivity 57
those of Taylor et al. (1992) with the exception that the Sandin et al. (1996) study
did not include a PTSD diagnostic group. As such, a comparison between panic
disorder and PTSD diagnostic groups to determine whether mean ASI scores for
both groups would be similar could not be performed.
Using a sample of 143 individuals who met the DSM-III-R (APA, 1987)
criteria for panic disorder with or without agoraphobia or a different anxiety
disorder, Apfeldorf, Shear, Leon, and Portera (1994) reported that the individuals
in the panic disorder group scored significantly elevated levels of anxiety
sensitivity when compared to individuals in the mixed anxiety disorder group
and concluded that the ASI was capable of distinguishing between individuals
with panic disorder and individuals with other anxiety disorders.
While it is evident that individuals with panic disorder endorse higher
overall levels of anxiety sensitivity when compared to individuals with another
anxiety disorder, it is interesting to find that mean ASI scores for individuals
with PTSD are comparable with those observed in panic disorder. Taylor et al.
(1992) hypothesised that the similarity in mean ASI scores for both diagnostic
groups could be due, in part, to the similarity between panic attacks and
flashbacks. For instance, according to the DSM-IV-TR, a hallmark feature of
PTSD is the persistent re-experiencing of the traumatic event in the form of
thoughts, images or perceptions (flashbacks) with intense psychological distress
and physiological reactivity upon exposure to internal or external cues (APA,
2000). Thus, both panic attacks and flashbacks are similar in that they are both
intense events activated by internal or external fear stimuli and share similar
symptomatology (Taylor et al., 1992). As flashbacks have an elemental course
Anxiety Sensitivity 58
similar to panic attacks, it is proposed that both panic disorder and PTSD have an
underlying common denominator of high arousal.
2.8.2 Posttraumatic Stress Disorder. The DSM-IV-TR characterises
PTSD as exposure to a trauma in which an individual witnesses an event or
events that involve actual or threatened death or serious injury to self or others
and responds with intense fear, helplessness or horror (APA, 2000). Individuals
with PTSD experience recurrent and intrusive recollections of the event,
distressing dreams about the event or even dissociative states, often referred to as
‘flashbacks’ (APA, 2000). Additionally, intense psychological anguish or
physiological reactivity often occurs when the individual is exposed to triggering
events that resemble or symbolise an aspect of the traumatic event (APA, 2000).
These symptoms can include trouble falling or staying asleep, irritability or
outbursts of anger, concentration difficulties, hypervigilance, and exaggerated
startle response.
In a investigation examining the frequency of PTSD in a community
based substance abuse program, Bronin, Norton, Asmundson, Dicurzio and
Pidlubney (2000) examined ASI scores for a total of 91 participants who met the
DSM-IV (APA, 1994) diagnostic criteria for substance abuse or dependence who
were also grouped as having PTSD, possible PTSD, or no PTSD according to the
Modified PTSD Symptom Scale. Bronin et al. (2000) found that individuals with
a diagnosis of PTSD scored significant higher mean scores on the ASI when
compared to those individuals with a ‘possible’ diagnosis of PTSD or without
PTSD.
Lehman and Cheung (2002) investigated the prevalence of anxiety
sensitivity, PTSD, depression and alcohol-related problems in 120 veterans with
Anxiety Sensitivity 59
chronic hepatitis C and found that individuals with a history of PTSD reported
elevated levels of anxiety sensitivity compared to individuals without a history of
PTSD. Conversely, no significant differences were observed between
individuals with a diagnosis of alcohol use disorder and individuals without such
a diagnosis or individuals with a diagnosis of depression and those individuals
without a history of depression.
Lang, Kennedy, and Stein (2002) evaluated the relationship between
anxiety sensitivity and PTSD in 30 women with no history of exposure to serious
trauma, 23 women who never developed PTSD but who had been exposed to
intimate partner violence, and 19 women who had been exposed to intimate
partner violence and were also diagnosed with PTSD according to the Structured
Clinical Interview for DSM-IV Diagnoses (SCID). Lang et al. (2002) reported
that traumatised women diagnosed with PTSD scored significantly higher
anxiety sensitivity compared to traumatised women without PTSD or women
without history of trauma. Similarly, while women with PTSD were found to be
significantly more fearful of the physical, psychological and social consequences
of anxiety symptoms than women without history of trauma, only the physical
and psychological dimensions were capable of differentiating between
traumatised women without or with PTSD. Item analysis revealed that women
with PTSD endorsed items relating to a fear of going crazy, trembling, mental
illness or unusual bodily sensations than traumatised women without PTSD.
Again, it is interesting to note that when differences between groups are
examined at the item level, there appears to be no significant difference between
the classified groups and items measuring publicly observable anxiety symptoms.
Anxiety Sensitivity 60
Fedoroff, Taylor, Asmundson, and Koch (2000) evaluated the importance
of anxiety sensitivity and various other trauma-related beliefs in predicting PTSD
symptom severity and treatment-related changes in 81 individuals who had been
exposed to a motor vehicle accident (MVA). Multiple regression analyses
indicated that after controlling for medication and pain severity, anxiety
sensitivity and pain severity emerged as significant predictors of PTSD
symptoms, while MVA-related beliefs were not significant predictors once
anxiety sensitivity, pain severity and medication status were entered into the
equation. Multiple regression analyses of a subsample of 28 participants who
completed CBT revealed that reductions in anxiety sensitivity and pain severity
were significant predictors of reductions in PTSD symptoms. Fedoroff et al.
(2000) argued that the results suggest that anxiety sensitivity is an important
cognitive risk factor for exacerbating and maintaining PTSD symptoms.
Results of these studies are interesting as they relate back to the DSM-IV-
TR (APA, 2000) description of PTSD in that individuals experience intense
psychological distress and physiological reactivity upon exposure to internal or
external cues. Thus, as flashbacks contribute to the increase in arousal and
psychological distress experienced by individuals with PTSD, this process may
assist in creating not only a fear of anxiety related symptoms, but also elevated
anxiety sensitivity in a process similar to that of panic disorder.
2.8.3 Social Phobia. According to the DSM-IV-TR (APA, 2000), social
phobia is characterised by an apparent and persistent fear of social or
performance circumstances in which embarrassment may possibly occur.
Individuals with social phobia express concern about humiliation or
embarrassment and are afraid that others will judge them to be nervous, anxious,
Anxiety Sensitivity 61
pathetic, ‘crazy’, or stupid when confronted with feared social or performance
situations (APA, 2000). Additionally, there are a number of anxiety symptoms
that individuals with social phobia almost always experience in feared social
situations such as sweating, nausea, gastrointestinal discomfort, diarrhoea,
palpitations, tremors, muscle tension, blushing, confusion and disorientation
(APA, 2000).
Investigations of anxiety sensitivity and diagnostic groups have found
elevated anxiety sensitivity among individuals with social phobia. Although the
nature of studies vary, there is evidence to suggest that individuals with social
phobia tend to obtain elevated ASI scores that approach those obtained in studies
of individuals with panic disorder For example, inn a comparison of anxiety
sensitivity among individuals with social phobia or panic disorder, Hazen,
Walker and Stein (1995) compared 47 individuals who met the DSM-III-R
(APA, 1987) criteria for panic disorder and 47 individuals with a diagnosis of
generalised social phobia on the ASI and found while anxiety sensitivity was
elevated in both diagnostic categories, individuals with social phobia scored
significantly higher on an item relating to the social embarrassment of a growling
stomach, whereas individuals with panic disorder scored significantly higher on
items relating to fear of a rapid heart beat, shortness of breath and a range of
bodily sensations related to anxiety. However, there were no significant
differences between individuals with panic disorder and social phobia for items
evaluating fear of cognitive dyscontrol or subjective experience of anxiety
(Hazen et al., 1995). As such, Hazen et al. (1995) argued that the findings
provide support for the assertion that the fear of somatic symptoms and the
Anxiety Sensitivity 62
consequences of such symptoms are particularly relevant to the phenomenology
of panic disorder.
Conversely, in a examination of the core beliefs in social phobia and
panic disorder among 102 individuals with a DSM-III-R (APA, 1987) diagnosis
of panic disorder only, social phobia only, comorbid social phobia and panic
disorder, or comorbid social phobia and depressive disorder, Ball, Otto, Pollack,
Uccello, and Rosenbaum (1995) found that while there was no significant
difference between anxiety sensitivity and individuals with a diagnosis of panic
disorder, social phobia, comorbid social phobia/panic disorder or comorbid
social phobia/depressive disorder, there was a trend for individuals with social
phobia only to report less anxiety sensitivity than the other diagnostic groups.
Thus, the results of Ball et al. (1995) reveal that a considerable overlap exists
amongst these diagnostic groups. According to Cox, Borger, and Enns (1999), it
is possible for elevated levels of anxiety sensitivity to be observed in individuals
with social phobia given that they may fear publicly observable anxiety
symptoms if they believe that such symptoms result in detrimental consequences
socially. Conversely, individuals with panic disorder would be more likely to
fear symptoms that relate to cardiac and respiratory arousal if they believe that
such symptoms results in detrimental consequences physically (Cox et al., 1999).
Such considerations emphasise the need for dimensional analyses rather than
item analyses in order to elucidate the nature of anxiety sensitivity among
various clinical conditions.
Maidenberg, Chen, Craske, Bohn, and Bystritsky (1996) investigated the
specificity of attentional bias using a computerised version of the Stroop for 30
individuals who met the DSM-III-R (APA, 1987) criteria for panic disorder or
Anxiety Sensitivity 63
social phobia and 15 control participants and found that while individuals with
panic disorder and social phobia scored significantly higher overall ASI scores
compared to normal controls, there was no significant difference between the two
diagnostic groups. Maidenberg et al. (1996) surmised that it would be reasonable
to expect that the emotionality rating for both diagnostic groups would be
comparable for words relating to panic or social threat. However, results
revealed that individuals with social phobia demonstrated specific interference
for words relating to social- threat compared to words relating to panic-threat,
whereas individuals with panic disorder demonstrated specific interference for
words relating to panic-threat compared to words relating to social-threat. As
such, Maidenberg et al. (1996) posited that individuals with panic disorder may
possess a wider range of fears than individuals with social phobia and although
physical sensations are of concern for individuals with social phobia, it would
appear that the central concern remains one of social evaluation and its
consequences, or a concern with the physical sensations of anxiety limited to
social circumstances.
Among 101 individuals with a DSM-III-R (APA, 1987) or DSM-IV
(APA, 1994) diagnosis of social phobia with panic attacks, social phobia without
panic attacks, or social phobia with comorbid panic disorder, Scott, Heimberg,
and Jack (2000) found that individuals with social phobia who also experienced
panic attacks reported elevated levels of anxiety sensitivity when compared to
individuals with social phobia but who did not experience panic attacks. Scott et
al. (2000) also reported that there was no difference in anxiety sensitivity scores
among individuals with a comorbid diagnosis of social phobia and panic disorder
when compared to individuals with social phobia who also experienced panic
Anxiety Sensitivity 64
attacks. What did appear was a different pattern of responding between
individuals with social phobia who did and did not experience panic attacks, with
those who experienced panic attacks reporting to be significantly more concerned
of the physical symptoms of anxiety when compared to individuals who did not
experience panic attacks. There was also a trend for those individuals who
experienced panic attacks to report greater fear regarding the experience of
shakiness as well as a racing heart compared to individuals who did not
experience panic attacks. As such, Scott et al. (2000) have argued that that
anxiety sensitivity is an associated characteristic of panic attacks in anxiety
disorders other than panic disorder and may function differently in individuals
with an additional diagnosis of panic rather than with primary panic disorder.
In a similar study of situational panic among 135 individuals categorised
as experiencing social phobia with comorbid panic disorder, social phobia with
situational panic or non-panicking social phobia according to the ADIS-R, Jack,
Heimberg, and Mennin (1999) found that while there was no significant
difference between individuals with social phobia and situational panic and
individuals with comorbid social phobia and panic disorder; individuals with
social phobia and situational panic reported significantly higher anxiety
sensitivity when compared to individuals with non-panicking social phobia.
According to the results of Scott et al. (2000) and Jack et al. (1999) it appears
that individuals with social phobia, who also experience panic attacks, do not
endorse significantly less anxiety sensitivity than those individuals with a dual
diagnosis of social phobia and panic disorder. Accordingly, it is reasonable to
argue that while individuals with social phobia are most likely fearful of publicly
observable anxiety symptoms and believe that these symptoms have harmful
Anxiety Sensitivity 65
social consequences, individuals with social phobia who also experience
unpredicted or situational panic attacks are fearful not only of publicly
observable anxiety symptoms, but that they are similar to individuals with a
diagnosis of panic disorder in that they are fearful of the physical symptoms of
anxiety as well.
2.8.4 Obsessive-Compulsive Disorder and Generalised Anxiety Disorder.
While there is limited research regarding anxiety sensitivity and obsessive-
compulsive disorder (OCD), it is important to note that some investigators have
found elevated ASI scores in this population (Taylor et al., 1992; Zeitlin &
McNally, 1993), while others have not (Sandin et al., 1996). Similarly, there is a
distinct lack of research regarding anxiety sensitivity and GAD. However, of the
small number of studies that have included GAD as part of their methodology,
there is some evidence to suggest that this diagnostic group also display elevated
levels of anxiety sensitivity when compared to normative samples (Taylor et al.
1992; Sandin et al., 1996). What is not known is whether these diagnostic
groups would endorse distinctive facets of anxiety sensitivity (Cox et al., 1999).
For example, according to the DSM-IV-TR (APA, 2000), GAD is characterised
by disproportionate anxiety and worry (apprehensive expectation) about a
number of events and activities. Individuals with GAD find it very difficult to
control their worry or apprehensive expectations and experience a number of
symptoms that include agitation or restlessness, fatigue, concentration
difficulties, petulance or irritability, muscle tension and disturbed sleep (APA,
2000). The intensity, length, or occurrence of the anxiety and apprehension that
an individual with GAD experiences is regarded to be far out of proportion to the
real probability or impact of the feared event and they find it difficult to keep the
Anxiety Sensitivity 66
troublesome thoughts from interfering with daily activities and events. It is also
very difficult for individuals with GAD to stop worrying (APA, 2000).
Additionally, there are a number of somatic sensations that many individuals
with GAD experience such as sweating, nausea, or diarrhoea and may also
experience trembling, twitching, feelings of shakiness, as well as muscular aches,
soreness and exaggerated startle response (APA, 2000).
Conversely, according to the DSM-IV-TR, OCD is characterised as
involving recurrent obsessions or compulsions that typically last for more than
one hour per day or cause noticeable distress or considerable impairment (APA,
2000). The obsessions observed in OCD are regarded as repetitive, invasive
thoughts, images or impulses that cause noticeable anxiety or distress. To reduce
the distress and anxiety that accompanies an obsession and to neutralise such
thoughts, images or impulses, an individual with OCD will usually engage in
some other thought or action (the compulsion) (APA, 2000). What is noticeable
about these two diagnostic categories is that they both share cognitive elements
or actions related to mental processes. Therefore, according to Cox et al. (1999)
it is possible that individuals with OCD and GAD might be more fearful of a loss
of cognitive control than of a fear of physical or publicly observable anxiety
symptoms. However, this is not yet established. It is possible that this lack of
research stems from the diagnostic criteria of both disorders in that they are not
accompanied by a fear of or list of physical or somatic sensations similar to those
observed in panic disorder, PTSD or social phobia.
2.8.5 Specific Phobia. In contrast, available research to date has not
supported the view that elevated levels of anxiety sensitivity are observed in
individuals with specific phobia. For example, while Lumley and Melamed’s
Anxiety Sensitivity 67
(1992) research of university students with a phobia of blood and Kleinknecht’s
(1988) research of individuals with a history of fainting due to blood/injury fears
reported elevated levels of anxiety sensitivity when compared to normal controls,
other investigations have failed to observe these findings (Sandin et al., 1996;
Taylor et al., 1992). According to the DSM-IV-TR, specific phobia is
characterised as a discernible and persistent fear of clearly discernible, yet
constrained objects or situations (APA, 2000). For an individual with specific
phobia, the focal point of the fear may be expected harm from a particular aspect
of the object or situation and may also involve concern and apprehension
regarding a loss of control, panicking, the experience of anxiety symptoms such
as palpitations or shortness of breath and that fainting may possibly occur when
exposed to the feared stimulus (APA, 2000). Also, while the anxiety is
consistently felt immediately on confrontation with the phobic stimulus, the level
of anxiety or fear usually fluctuates as a function of both the degree of proximity
to the phobic stimulus and the degree to which escape from the phobic stimulus
is limited (APA, 2000).
Taylor et al. (1992) suggested that individuals with specific phobia might
not develop a fear of anxiety symptoms because the fearful stimulus is highly
predictable and bound to specific situations or events. Cox et al. (1999) have
posited that fundamental fears rather than anxiety sensitivity may be more
pertinent for certain types of specific phobias, such as blood/injury phobia and
that is may well be that particular facets of anxiety sensitivity (e.g., fear of
physical sensations) may be pertinent for some types of specific phobia.
However, this is not yet empirically established.
Anxiety Sensitivity 68
2.8.6 Major Depressive Disorder. According to the DSM-IV-TR, major
depressive disorder is described as a clinical course that is distinguished by one
or more major depressive episodes (APA, 2000). In a clinical setting, individuals
experiencing a major depressive episode commonly list tearfulness, irritability,
brooding, obsessive rumination, anxiety, phobias, excessive worry over physical
health and complain of pain as their presenting complaint.
The mood in a major depressive episode is commonly depicted as
depressed, miserable, bleak, or discouraged (APA, 2000). Individuals often
complain about a loss of interest or pleasure in everyday activities as well as a
reduction in appetite (APA, 2000). Psychomotor changes include either agitation
or retardation and the most common sleep disturbance reported by individuals
experiencing a major depressive episode is middle or terminal insomnia (APA,
2000). Additionally, decreased energy, weariness, exhaustion and fatigue are
common and individuals may also report sustained fatigue without physical
exertion (APA, 2000).
The sense of worthlessness, unimportance or guilt linked with a major
depressive episode can be comprised of unrealistic negative evaluations of one’s
worth or guilty preoccupations or cogitations over trivial past failings (APA,
2000). The sense of unimportance or guilt may be delusional in scope, however
holding oneself responsible for being ill and for failing to meet occupational or
interpersonal responsibilities as a result of the depression is very common and
not usually considered sufficient to meet the this criterion (APA, 2000). In
addition, many individuals experiencing a major depressive episode report an
impaired capability to think, concentrate or make decisions and may appear to be
unfocused, easily distracted or may complain of memory problems (APA, 2000).
Anxiety Sensitivity 69
This symptom is very similar to the cognitive symptoms that are experienced by
individuals who meet the criteria for an anxiety disorder. Finally, individuals
may have thoughts of death, suicidal ideation or may have already engaged in
suicide attempts (APA, 2000).
According to the DSM-IV-TR, dysthymic disorder and major depressive
disorder are differentiated based on severity, chronicity, and persistence (APA,
2000). For major depressive disorder, the depressed mood is required to be
present for most of the day, nearly every day, for a period of at least two weeks,
whereas dysthymic disorder is required to be present for more days than not over
a period of at least two years (APA, 2000). The difference between dysthymic
disorder and major depressive disorder is particularly complicated because the
two disorders share similar symptoms and the differences between them in onset,
duration, persistence, and severity are not easy to evaluate retrospectively (APA,
2000). Typically major depressive disorder consists of one or more distinct
major depressive episodes that can be distinguished from the individual’s normal
functioning, whereas dysthymic disorder is characterised by chronic, though less
severe depressive symptoms that have been present for many years (APA, 2000).
Therefore, as major depressive disorder and dysthymic disorder are characterised
as involving similar cognitive symptoms to those experienced by individuals who
meet the criteria for an anxiety disorder, and that the two diagnostic
classifications share comparable symptoms that are not easy to differentiate.
While there is a lack of research into anxiety sensitivity and specific
phobias, GAD or OCD, investigations have been conducted that have found
anxiety sensitivity appears to be elevated in individuals with major depression.
For example, in their initial study of anxiety sensitivity and major depression for
Anxiety Sensitivity 70
15 individuals who met the DSM-III-R (APA, 1987) diagnostic criteria for major
depression and 26 normative controls, Otto, Pollack, Fava, Uccello, and
Rosenbaum (1995) reported, in the absence of a comorbid anxiety disorder, that
individuals with a diagnosis of major depression reported elevated levels of
anxiety sensitivity when compared to the control group. In their second study,
Otto et al. (1995) examined ASI scores both pre and post-fluoxetine treatment for
144 individuals and found that while pre-treatment scores on the ASI were
elevated, there was no significant difference between individuals with a
diagnosis of depression without comorbidity and those individuals with a
diagnosis of depression with comorbid past or present panic disorder or other
anxiety disorder. However, mean scores did decline from 27.1 to 20 for 86
individuals with depression after eight weeks of fluoxetine treatment.
In an investigation of the relationship between anxiety sensitivity and
depression for 135 individuals with a DSM-III-R (APA, 1987) diagnosis of panic
disorder without major depression, major depression without panic disorder, or
major depression with panic disorder, Taylor et al. (1996) replicated Otto et al.’s
(1995) findings and reported that individuals with a comorbid diagnosis of panic
disorder and major depression reported the highest level of anxiety sensitivity
followed by individuals diagnosed with panic disorder and major depression
respectively.
In an effort to identify the distinct facets of anxiety sensitivity involved in
this elevation, Taylor et al. (1996) used principal components analysis and found
a three-factor solution in which individuals with comorbid panic and depression
displayed significantly greater ‘fear of publicly observable symptoms’ (factor
one) and ‘fear of cognitive dyscontrol’ or ‘phrenophobia’ (factor two) than
Anxiety Sensitivity 71
individuals with a diagnosis of panic disorder alone, which in turn was greater
than that of individuals with a diagnosis of major depression alone. While there
was no significant difference between individuals with comorbid panic and
depression or individuals with a diagnosis of panic disorder alone on the third
factor, ‘fear of bodily sensations’, both groups displayed significantly higher
mean scores than individuals with a diagnosis of major depression alone for this
factor. Taylor et al. (1996) argued that these results were important for the
conceptualisation of anxiety sensitivity as they revealed that not all components
of the construct are related to fears, phobia, or panic.
In an attempt to determine whether anxiety sensitivity acts as a specific
vulnerability factor in the pathogenesis of anxiety and depression, Schmidt,
Lerew, and Joiner (1998) examined a large group of 1401 military recruits
undergoing 5 weeks of basic training at a United States Air Force Academy as
well as 53 individuals who met the DSM-IV (APA, 1994) diagnostic criteria for
panic disorder with or without agoraphobia using a covariance strategy. They
found that for both groups, anxiety sensitivity was capable of accounting for
symptom specificity in regards to anxiety but was not predictive of depression
when changes in anxiety symptoms where accounted for. However, using the
same factors described by Taylor et al. (1996), Schmidt et al. (1998) reported that
a fear of cognitive dyscontrol was likely to account for the association between
anxiety and depression because it was a predictor of both anxiety and depression
symptoms.
Cox, Enns, and Taylor (2001) evaluated the mediating effects of
rumination of the anxiety sensitivity factor ‘fear of cognitive dyscontrol’ in
predicting the severity of a depressed mood in 142 individuals who met the
Anxiety Sensitivity 72
criteria for DSM-IV (APA, 1994) diagnosis of major depression. Results of
regression analyses found that, after controlling for neuroticism, the anxiety
sensitivity facet ‘fear of cognitive dyscontrol’ was a significant predictor of
severity of depression. Additionally, after rumination was statistically
controlled, the anxiety sensitivity facet ‘fear of cognitive dyscontrol’ no longer
predicted variance in the severity of depression. Thus, Cox et al’s. (2001)
investigation revealed that ‘fear of cognitive dyscontrol’ or ‘phrenophobia’ is
mediated by the rumination of symptoms of depression.
The relationship between anxiety sensitivity and depression may be due
to both anxiety and depression sharing similar features. Otto et al. (1995)
conjectured that the cognitive distortions that are characteristic of anxiety may
also be common to both anxiety sensitivity and depression. Taylor et al. (1996)
proposed that the relationship may be due to the anxiety sensitivity component
‘fear of cognitive dyscontrol’ or ‘phrenophobia’. Taylor et al. (1996) has argued
that phrenophobia is a depression-specific form of anxiety sensitivity, unrelated
to anxiety and panic. Conversely, Schmidt et al. (1998) have argued that
phrenophobia is related to both anxiety and depression. They argue that the
cognitive symptoms of phrenophobia such as impairment in concentration,
depersonalisation and derealisation are conceptually consistent with anxiety
sensitivity and have the potential to lead to specific fears (Schmidt et al., 1998).
While research has revealed that anxiety sensitivity is associated with a
number of clinical disorders, it is important to note that this area could benefit
from further critical analysis. The research to date strongly suggests that anxiety
sensitivity, at least as measured by the ASI, is comprised of multiple facets rather
than a single construct. For example, some facets (e.g., fear of cognitive
Anxiety Sensitivity 73
dyscontrol) may be more important for understanding a particular disorder (e.g.,
major depression). However, as has been argued previously, the numerous
limitations of the 16-item ASI are of concern as they reduce an investigator’s
confidence when making any theoretical claims about the anxiety sensitivity
construct. Therefore, further investigations, not only into the differences in the
various aspects of anxiety sensitivity between different disorders, but also within
each disorder itself, and using the expanded ASI-R, are clearly warranted to
broaden our understanding of this construct.
The following section is concerned with the degree to which treatment,
and in particular CBT, can successfully impact upon anxiety sensitivity.
Evidence from several lines of investigations has revealed that CBT is capable of
facilitating a reduction in anxiety sensitivity scores, as measured by the ASI.
2.9 Anxiety Sensitivity and Cognitive-Behavioural Therapy
As part of a larger study designed to investigate the validity of the ASI as
a measure of anxiety sensitivity, McNally and Lorenz (1987) reported that mean
ASI scores for 23 individuals who met the DSM-III (APA, 1980) criteria for
agoraphobia with panic attacks declined from 38.3 (SD = 11.2) to 19.2 (SD =
12.4) following 10-12 weeks of CBT that focused on education about panic
disorder and agoraphobia, cognitive restructuring, training in diaphragmatic
breathing, in vivo exposure and exposure homework assignments. Post-
treatment ASI scores remained stable, 15.2 (SD = 8.6), after a six-month follow-
up period. Although this study was limited in that it did not employ a wait-list
control group, the results do suggest that reductions in anxiety sensitivity can be
obtained with CBT.
Anxiety Sensitivity 74
In an examination of the efficacy of an eight-week group CBT for
individuals with a DSM-III-R (APA, 1987) diagnosis of panic disorder, Telch,
Lucas, Schmidt, Hanna, Jaimez, and Lucas (1993) randomly assigned 67
individuals to either a treatment group that emphasised education and corrective
information, diaphragmatic breathing, cognitive restructuring, and interoceptive
exposure, or to a delayed treatment control group. Mean ASI scores for
individuals who received treatment declined from 33.74 (SD = 11.15) to 13.94
(SD = 8.52) after the eight-weeks of CBT and post-treatment scores remained
stable at six-month follow-up assessment. In contrast, the delayed treatment
control groups mean ASI score was in the mid 30s and remained stable across the
treatment interval. Approximately 64% of individuals in the treatment group and
9% of individuals in the control treatment group displayed full recovery from
panic disorder at the completion of the study.
Penava, Otto, Maki, and Pollack (1997) examined the rate of
improvement during manual driven, group CBT for 37 individuals with a
diagnosis of panic disorder. The treatment program consisted of 12 sessions that
included informational interventions and training in diaphragmatic breathing,
interoceptive exposure, cognitive restructuring, training in progressive muscle
relaxation as well as naturalistic and situational exposure. Penava et al. (1997)
reported that for a subset of 26 individuals, significant decreases in anxiety
sensitivity were observed consistently across the three months of treatment with
a mean change from pre-treatment to post-treatment of 35.0 (SD = 12.5) and 21.5
(SD = 12.9) respectively. The change in ASI scores from pre-treatment to post-
treatment was significantly correlated (r = .62) with individuals’ symptom
improvement.
Anxiety Sensitivity 75
One of the most direct examinations of changes in ASI scores following
CBT has been conducted by Hazen, Walker, and Eldridge (1996). Hazen et al.
(1996) examined the relationship between anxiety sensitivity and treatment
outcome in a sample of 106 patients with a primary DSM-III-R (APA, 1987)
diagnosis of panic disorder, with or without agoraphobia. The active treatment
centred on the use of a self-help manual developed by Clum (1990) and was
delivered in the context of a therapist-directed group, a self-help group, or
individual use. These three treatment conditions were compared with each other
and with a wait-list control group. The results revealed that the pre-treatment
ASI scores for the four groups were in the mid to low 30s, with no significant
differences between groups. The post-treatment scores (M = 35.1) remained
almost two standard deviations above the normative mean for the wait-list
control group; demonstrating that a pre-to-post treatment change in anxiety
sensitivity did not occur for this group. Further, scores for the wait-list group
were highly correlated between the pre and post treatments conditions (r = .74),
indicating that the test-retest reliability for the ASI was acceptable.
In contrast, ASI scores decreased significantly in the three active-
treatment conditions. Post-treatment scores were lowest in the therapist-directed
treatment group (M = 17.8) and were in the normative range. In addition,
reductions in anxiety sensitivity were associated with the degree of improvement
as assessed by the Clinical Global Improvement ratings, with the effect size for
changes in ASI scores being greater in magnitude than those obtained on the Fear
Questionnaire Agoraphobia subscale and the Sheehan Patient-Rating Anxiety
Scale, which were used as other measures of treatment outcome. As a result of
their investigation, Hazen et al. (1996) concluded that not only is anxiety
Anxiety Sensitivity 76
sensitivity responsive to CBT but that the ASI is a sensitive measure of treatment
effects.
Shear, Pilhonis, Cloitre, and Leon (1994) examined the specificity of
CBT for 45 individuals who met the DSM-III-R (1987) criteria for panic disorder
with or without agoraphobia and a non-prescriptive intervention that involved
reflective listening that focused on life problems and stressors in relation to panic
disorder. Twenty-four individuals were randomly assigned to 12 sessions of
CBT that focused on diaphragmatic breathing, muscle relaxation, cognitive
restructuring, and exposure to interoceptive and agoraphobia stimuli, with the
remaining 21 individuals assigned to 12 sessions of non-prescriptive treatment.
Both groups received three-weeks of educational intervention that focused on the
phenomenology of panic disorder. Mean ASI scores for individuals who
received CBT declined from 30.8 (SD = 15.0) to 20.7 (SD = 10.5) after 15 weeks
of treatment, with post-treatment scores declining to 17.9 (SD = 9.6) at six-month
follow-up assessment. Interestingly, mean ASI scores for individuals in the non-
prescriptive treatment group also declined from 31.6 (SD = 11.2) to 21.4 (SD =
13.8) after the 15 weeks of treatment. Six-month follow-up assessment revealed
post-treatment ASI scores declined approximately eight points to 13.7 (SD =
11.8). Shear et al. (1994) conducted an additional two-way ANOVA to compare
anxiety sensitivity and panic remission status by treatment group at the three
measurement periods. No significant difference in mean ASI scores was found
between individuals in remission of panic episodes and those individuals not in
remission at any measurement period. Additionally, no significant difference in
mean ASI scores was observed between individuals in remission of panic
episodes in either the CBT or non-prescriptive treatment groups. Treatment, in
Anxiety Sensitivity 77
either modality, led to a significant decrease in anxiety sensitivity that declined
even further at six-month follow-up. These results are interesting as ASI scores
may covary with successful treatment of panic disorder regardless of the
treatment modality. However, this would require further empirical investigation.
2.9.1 Anxiety Sensitivity and Combined Cognitive- Behavioural and
Pharmacological Therapy. While the investigations discussed to this point have
focused on using CBT therapy in isolation, it is interesting to note that other
investigators have chosen to combine CBT as well as pharmacological therapy
with some interesting results. For example, Hegel, Ravaris, and Ahles (1994)
investigated the efficacy of alprazolam taper and CBT for 22 individuals who
met the DSM-III-R (APA, 1987) criteria for panic disorder. The CBT
component consisted of 12 sessions that emphasised education of the
physiological, cognitive and behavioural model of panic, diaphragmatic
breathing retraining, progressive muscle relaxation training, cognitive
restructuring, interoceptive exposure and homework assignments. The
alprazolam taper schedule began in week four of the CBT sessions so that all
individuals would be medication free by week 12. Hegel et al. (1994) reported
that for the 22 individuals who completed the study, significant decreases in
anxiety sensitivity were observed with a mean change from pre-treatment to
post-treatment of 33.68 (SD = 9.41) and 21.68 (SD = 8.76) respectively, with
post-treatment scores remaining consistent at six-month follow-up, mean = 20.23
(SD = 9.30), and 12 month follow-up, mean = 19.32 (SD = 7.13) follow-up.
In a study designed to investigate whether CBT was capable of
facilitating slow-taper alprazolam discontinuation and maintenance of drug
abstinence among 21 individuals who met the DSM-III-R (APA, 1987) criteria
Anxiety Sensitivity 78
for panic disorder with agoraphobia, Spiegel, Bruce, Gregg, and Nuzzarello
(1994) randomly assigned 10 individuals to receive drug maintenance as well as
a slow, flexible drug taper, and 11 individuals to the same medication condition
treatment condition as well as 12 weeks of concurrent, individual CBT. Of the
20 individuals who completed the study, Spiegel et al. (1994) reported that there
was no significant difference between groups in the rate of alprazolam
discontinuation. However, during the six-month follow-up period, 50% of
individual who discontinued alprazolam without CBT relapsed and resumed
alprazolam treatment. Conversely, no individual in medication treatment
condition who also underwent CBT experienced relapse.
In a follow-up of the Spiegel et al. (1994) study; Bruce, Spiegel, Gregg,
and Nuzzarello (1995) conducted forward stepwise regression analyses in order
to examine whether it was possible to identify individuals’ ability to discontinue
alprazolam treatment and remain medication free during a six-month follow-up
period. Once individuals were stabilised on alprazolam, predictor variables that
included the baseline to post-taper change in ASI scores, a diary of withdrawal
and side effects of symptoms severity, stabilised alprazolam dose, total duration
of continuous benzodiazepine use before taper and baseline scores on the
Mobility Inventory for Agoraphobia were measured. Bruce et al. (1995) reported
that the baseline to post-taper changes in ASI scores was the only significant
predictor associated with discontinuation success, with individuals reporting
minor declines in anxiety sensitivity at a greater chance of relapse.
Westra, Stewart, and Conrad (2002) examined the impact of manner of
benzodiazepine use on 10-week group session CBT outcome for 43 individuals
who met the DSM-IV (APA, 1994) criteria for panic disorder with agoraphobia.
Anxiety Sensitivity 79
For the entire sample, a significant decrease in ASI scores was observed with a
mean change from pre-treatment to post-treatment of 43.97 (SD = 8.54) and
26.02 (SD = 12.49) respectively. While group comparisons of participants
classified as unmedicated (n = 10), as needed benzodiazepine users (n = 15) or
regular benzodiazepine users (n = 13) revealed no significant differences among
the three groups at pre-treatment, individuals who used benzodiazepines as
needed scored significantly higher on the ASI when compared to individuals
classified as either unmedicated or regular users of benzodiazepines at the
completion of CBT. There was no significant post-treatment difference between
unmedicated or regular users of benzodiazepines. Thus, individuals who used
benzodiazepines as needed reported less anxiety sensitivity reduction than the
other two groups following CBT. This is important as previous research by
Telch et al. (1993) has reported that anxiety sensitivity scores mediate changes in
symptom status over the course of CBT, whilst elevated anxiety sensitivity at the
completion of CBT has been used to predict relapse to benzodiazepines use at 6-
months post-taper (see Bruce et al., 1995).
In conclusion, previous studies (see Hazen et al., 1996; McNally &
Lorenz, 1987; Penava et al., 1997; Shear et al., 1994; Telch et al., 1993) have
revealed that CBT interventions are efficacious in facilitating reductions in
individuals’ total ASI scores. However, what is not yet known is whether CBT
interventions are successful in reducing the first-order or specific anxiety
sensitivity dimensions. Thus, it is posited that it is important to re-examine
changes in anxiety sensitivity following successful treatment, using the expanded
ASI-R as the content validity is better than the 16-item ASI. Further, as anxiety
sensitivity is regarded as a cognitive construct, evidence of changes in anxiety
Anxiety Sensitivity 80
sensitivity scores following successful completion of CBT provides additional
evidence of construct validity of such an expanded measure.
The following section aims to provide a clear and concise rational that
will inform the reader of the theoretical and empirical importance of this
dissertation. Using arguments constructed from the literature presented in the
current chapter, it is proposed that the current dissertation is an important
addition to the existing literature and makes both a valid theoretical and
empirical contribution in the area of anxiety sensitivity.
2.10 Rationale for Study One
As described in Section 2.2, Reiss and colleagues (1986) developed the
16-item ASI in order to provide an adequate measure of the anxiety sensitivity
construct. This was an important first step towards developing a measure that
was not only psychometrically sound, but was also capable of stimulating
extensive research. Reiss and colleagues (1986) developed the 16-item ASI to
measure a unidimensional construct because according to Reiss and McNally’s
(1985) expectancy theory, anxiety sensitivity is a single, coherent construct
(Section 1.3). While this view has been supported by a number of investigations,
others have found support for a multidimensional structure (see Section 2.4).
Thus, the factor analytic structure of anxiety sensitivity is arguably one of the
most contentious issues in the literature to date, with investigations using varying
sample sizes, factor analytic techniques, and varying rationales as to how factors
are to be extracted. Results from both exploratory and confirmatory factor
analytic investigations of the ASI have frequently revealed that anxiety
sensitivity is comprised of multiple dimensions (see Section 2.4). Further, there
has been some argument over the exact number and nature of dimensions
Anxiety Sensitivity 81
contained within the 16-item ASI, with some investigators arguing for two
dimensions (Blais et al., 2001; Schmidt & Joiner, 2002); and others arguing for
three (Stewart et al., 1997; Taylor, 1996; Zinbarg et al., 1997), or four
dimensions (Peterson & Heilbronner, 1987; Telch et al., 1989).
It is important to note that most of the 16-item ASI models evaluated in
the literature review (Section 2.4) revealed a poor fit of the data, and even the
best-fitting models provided only a marginal fit of the data. There are three main
reasons why the 16-item ASI has failed to perform adequately. Firstly, the 16-
item ASI was constructed to measure what was originally conceptualised as a
unitary construct (Reiss et al., 1986). Secondly, as it contains relatively few
items, of which most measure fear of somatic sensations (e.g., Stewart et al.,
1997). As such, it is proposed that the scale is too abbreviated to adequately
measure the major and specific anxiety sensitivity facets. In addition, there are a
number of ambiguous statements contained in the 16-item ASI, which is
problematic because such items do not act as reliable markers for the specific
anxiety sensitivity dimensions. Overall, the numerous limitations of the 16-item
ASI are of concern as an investigator may not be confident when making any
theoretical claims based on the measure due to its compromised reliability and
validity.
While the ASI was important in furthering the understanding of anxiety
sensitivity, repeated attempts to clarify the dimensionality of anxiety sensitivity
using the 16-item ASI is problematic because, as stated previously, the scale was
never designed to measure a multidimensional construct. In an attempt to
expand the narrow focus of the 16-item ASI, Taylor and Cox (1998) developed
the 36-item Anxiety Sensitivity Index – Revised (ASI-R). Taylor and Cox’s
Anxiety Sensitivity 82
(1998) primary aim was to provide a more comprehensive measure of the first-
order anxiety sensitivity dimensions (see Section 2.5). The 36-item ASI-R
contains a broader selection of items, and hence dimensions, for assessing the
explicit domains of anxiety sensitivity. It retains the same instructions and
response format as the 16-item ASI, and contains 10 items which have been
adopted from the original ASI.
Since the Taylor and Cox (1998) publication, two other investigations
that have appeared that attempt to determine the factor structure and
psychometric properties of the expanded ASI-R (see Section 2.5). When
combined, the results of the three published investigations using the 36-item
ASI-R reveal that the revised measure has the potential to open up new and
important avenues for further investigation. Due to the limitations of the 16-item
ASI, the 36-item ASI-R developed by Taylor and Cox (1998) is a promising
instrument for measuring anxiety sensitivity. Research presented in the literature
review (Section 2.5) has shown that the first-order dimensions of the 36-item
ASI-R demonstrate theoretically consistent relationships with criterion variables
and the second-order dimension correlates very well with the 16-item ASI,
indicating that both instruments measure the same construct. Further, given that
the content validity of the ASI-R is an improvement on the 16-item ASI, it is
arguably a better measure for use in studies investigating anxiety sensitivity
dimensions. As such, it is of interest to critically examine not only the factor
structure of the ASI-R, but also the reliability and validity of the measure with
both normative and clinical populations before its utility can be assessed.
The first investigation of the present research attempted to empirically
evaluate the factor structure of the ASI-R for both clinical and nonclinical
Anxiety Sensitivity 83
samples using confirmatory factor analysis (CFA) on the hypothesised models
identified by Deacon et al. (2003), Taylor and Cox (1998), and Zvolensky et al.
(2003). In the event of failure to identify any of these measurement models, the
first investigation would endeavour to modify the ASI-R in order to identify the
best possible fitting model on the clinical sample and retest using a separate,
nonclinical sample. Any retained model/models would be retested against
alternative ASI-R and ASI models before endorsing model fit. Finally, any
retained model/models would be examined for acceptable internal consistency,
test-retest reliability, and discriminant validity for both a clinical and nonclinical
sample.
2.11 Rationale for Study Two
Previous sections (2.6, 2.7, and 2.8) have revealed that a central doctrine
in anxiety sensitivity theory is the prediction that fear of anxiety sensations may
serve as a psychological risk factor in the development of anxiety disorders in
general and panic attacks in particular. As such, considerable research has been
conducted into investigating anxiety sensitivity amongst the various anxiety
disorder and psychopathology classifications (Sections 2.8.1 to 2.8.6). Of the
impressive amount of research that has been conducted which revealed that
anxiety sensitivity is associated with a number of clinical disorders, it is
important to note that all of the investigations have been conducted using the 16-
item ASI. It has been suggested previously (Section 2.4) that the original ASI is
not capable of measuring the specific domains of anxiety sensitivity as it was
never designed to measure a multidimensional construct. Further, some facets of
anxiety sensitivity (e.g., fear of cognitive dyscontrol) may be more important for
understanding a particular disorder (e.g., major depression). Whilst it is noted
Anxiety Sensitivity 84
that there is some comorbidity in anxiety and mood disorders, it is important to
note that they also display unique features. Therefore, keeping this point in view,
it is of interest to investigate not only the differences between, but also within,
different diagnostic classifications using an expanded measure of anxiety
sensitivity, in order to broaden our understanding of this construct.
As such, the second investigation examined the differences between
various anxiety classifications, a mood disorder classification and a nonclinical
control sample, with respect to both general and specific dimensions of anxiety
sensitivity as identified by the expanded measure retained from the first
investigation. It is proposed that evidence of the expanded measure’s ability to
discriminate between groups in a theoretically expected manner will lead to
enhanced discriminant validity, which in turn, increases confidence in the
construct validity of the expanded anxiety sensitivity measure. Further, the
second investigation also examined the differences within each anxiety and mood
disorder classification as well as the nonclinical controls. This phase of the
research sought to explore the specific dimensions of anxiety sensitivity in order
to determine whether individual diagnostic groups endorsed particular facets of
anxiety sensitivity over others in a theoretically expected manner.
2.12 Rationale for Study Three
Finally, there is the issue of anxiety sensitivity and therapy. In order to
confirm any expanded measure of anxiety sensitivity it is important to investigate
the changes in anxiety sensitivity as a result of therapy. Previous investigations
have reported that anxiety sensitivity is responsive to CBT and that the ASI is
capable of capturing treatment effects (Section 2.9). However, it remains unclear
as to whether CBT interventions are capable of reducing the first-order or
Anxiety Sensitivity 85
specific anxiety sensitivity dimensions. Thus, changes in the general and
specific facets of anxiety sensitivity for each diagnostic classification, as a result
of CBT, were assessed in the third and final investigation. It was proposed that
evidence of reductions in the general and specific facets of anxiety sensitivity
would provide potentially important information relating to the construct validity
of the expanded anxiety sensitivity measure.
2.13 Chapter Summary
In summary the current dissertation attempted to address the
aforementioned issues by conducting a series of empirical investigations as all
previous investigations reviewed in the literature have utilised the 16-item ASI.
As such, it is of interest to re-examine changes in anxiety sensitivity using an
expanded measure.
Anxiety Sensitivity 86
Anxiety Sensitivity 87
Chapter Three: Method
3.1 Participants 89
3.1.2 Participants from the clinical setting 89
3.1.3 Participants from the nonclinical setting 90
3.2 Design 90
3.3 Materials/Measures 90
3.3.1 Anxiety Sensitivity Index (ASI) and Anxiety
Sensitivity Index – Revised (ASI-R) 91
3.3.2 Beck Anxiety Inventory 92
3.3.3 Catastrophic Cognitions Questionnaire (Modified) 93
3.3.4 Fear Questionnaire (FQ) 95
3.3.5 Zung Self-Rating Depression Scale (Zung – SDS) 95
3.3.6 COPE Questionnaire (COPE) 96
3.3.7 Stress Subscale of the 21-item Depression
Anxiety Stress Scale (DASS-Stress) 96
3.3.8 Generalised Self-Efficacy Scale (SES) 97
3.4 Procedure 97
3.4.1 Participants from the clinical setting 97
3.4.2 Participants from the nonclinical setting 100
Anxiety Sensitivity 88
Anxiety Sensitivity 89
CHAPTER THREE
METHOD
3.1 Participants
A total of 683 individuals, recruited from either a clinical or nonclinical
setting, participated in the current study. Each will be discussed separately.
3.1.2 Participants from the clinical setting. The first sample consisted of
248 individuals who were considered as having met the criteria for a principal
diagnosis of one of the DSM-IV anxiety diagnostic categories, major depression
or dysthymia and who also received CBT in a group format. Of this sample, 90
(36.3%) were male and 158 (63.7%) were female. The mean age of the male
participants was 44.79 years (SD = 10.86; range = 18-71 years) and the mean age
of the female participants was 42.23 years (SD = 12.23; range = 19-72 years).
Most (81.9%) were born in Australia and most (91.9%) spoke English as
their primary language. With respect to highest level of education attained,
34.7% had completed high school; 18.2% had completed a TAFE certificate or
associate diploma degree; 16.9% had completed a university undergraduate
degree; 12.5% had completed a university postgraduate degree; and 10.5% had
attained another level of education (e.g., apprenticeship or trade certificate).
Only 3.6% of the data was missing with respect to level of education attained.
All participants (100%) were in receipt of pharmacological treatment at the time
of participation. Finally, the principal DSM-IV diagnoses for the clinical group
included Panic Disorder (n = 97), Generalised Anxiety Disorder (GAD; n = 49),
Posttraumatic Stress Disorder (PTSD; n = 22); Major Depression or Dysthymic
Disorder (Depression; n = 58), Social Phobia (n = 14) and Other Anxiety
(consisting of individuals who met the diagnostic criteria for Obsessive
Anxiety Sensitivity 90
Compulsive Disorder, Specific Phobia or Hypochondrias, [n = 8]). Less than
12% of the participants from the Anxiety Disorder sample had a comorbid
diagnosis of a Depressive Disorder and only 6% of the participants from the
Depressive Disorder sample had a comorbid Anxiety Disorder diagnosis.
Therefore, it is important to note that some comorbidity amongst diagnostic
groups was evident.
3.1.3 Participants from the nonclinical setting. The second sample
consisted of 435 individuals from the community who were recruited from either
a tertiary institution’s internal telephone directory or who were enrolled in a first
year psychology unit. Of this second sample, 113 (26%) were male and 322
(74%) were female. The mean age of the male participants was 26.54 years (SD
= 11.62; range = 17-64 years) and the mean age of the female participants was
24.69 years (SD = 9.80; range = 17-63 years). Most (81.8%) were born in
Australia and most (82.1%) spoke English as their primary language. With
respect to highest level of education attained, 67.8% had completed high school;
8.1% had completed a certificate or associate diploma degree; 16.6% had
completed a university undergraduate degree; 6.2% had completed a university
level postgraduate degree; and 1.4% had attained another level of education (e.g.,
apprenticeship or trade certificate).
3.2 Design
The design of each study is discussed in the appropriate chapter.
3.3 Materials/Measures
A booklet (see Appendix D) was compiled for all participants and
comprised of a consent form, demographic questionnaire and test measures. The
Anxiety Sensitivity 91
questionnaires in the booklets were randomly arranged to control for order
effects.
3.3.1 Anxiety Sensitivity Index (ASI) and Anxiety Sensitivity Index –
Revised (ASI-R). The ASI (Reiss, Peterson, Gurskey & McNally, 1986) and the
ASI-R (Taylor & Cox, 1998) are both instruments for measuring the construct of
anxiety sensitivity. The ASI consists of 16 items, whereas the ASI-R comprises
of 36 items, 10 of which have been adopted from the original version. Each item
of the ASI and ASI-R is rated on a five-point Likert-type scale that ranges from
zero to four, where zero represents very little fear or worry and four represents
very much fear or worry of a particular anxiety symptom (Reiss et al., 1986). All
the items of the ASI and ASI-R were employed in the present study to assess the
participants’ fear of anxiety-related sensations, making a total of 42 items. This
method follows the author’s decision to include all items relating to both
questionnaires so that the relationship between both scales could be compared.
In their initial factor analytic studies, Cox, Taylor, Borger, Fuentes, and
Ross (as cited in Taylor & Cox, 1998) reported that the ASI-R contained six
lower order dimensions that related to a fear of cardiovascular, respiratory,
gastrointestinal, publicly observable, dissociative and neurological symptoms
that all contained an adequate level of internal consistency. However, in the only
published investigation of the ASI-R, Taylor and Cox (1998) reported that the
ASI-R consists of four lower-order anxiety sensitivity dimensions that load onto
a single high-order anxiety sensitivity dimension. The items corresponding to
each scale are as follows (1) Fear of Respiratory Symptoms (item numbers 1, 2,
3, 4, 5, 6, 7, 8, 9, 10, 11, 12); (2) Fear of Publicly Observable Anxiety Symptoms
(item numbers 13, 14, 15, 16, 17, 18, 19); (3) Fear of Cardiovascular Symptoms
Anxiety Sensitivity 92
(item numbers 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30) and (4) Fear of
Cognitive Dyscontrol (item numbers 31, 32, 33, 34, 35, 36). Principal
components factor loadings of the four lower-order anxiety sensitivity
dimensions on the higher-order dimension were reported as 6.7, 5.7, 6.6 and 4.6
respectively (Taylor & Cox, 1998). The higher-order dimension reportedly
accounts for 40.8 percent of the variance in the total ASI-R score (Taylor & Cox,
1998). Taylor and Cox (1998) also reported that the ASI-R contained acceptable
discriminant validity as it was capable of differentiating between diagnostic
categories as well as acceptable concurrent validity with the single higher-order
and four lower-order dimensions correlating with the original ASI, Beck Anxiety
Inventory and Beck Depression Inventory. However, it is important to note that
the internal consistency and test-retest reliability coefficients for this four-factor
model were not reported.
3.3.2 Beck Anxiety Inventory. In order to assess the severity of
participants’ anxiety symptomatology, the Beck Anxiety Inventory (BAI) (Beck,
Epstein, Brown, & Steer, 1988; Beck & Steer, 1993) was employed. The BAI is
a widely used, standardised self-report questionnaire that measures a range of
anxiety symptoms. The BAI consists of 21 items that assess the severity of
anxiety in adults and adolescents over a one-week period, with each question
being weighted to reflect symptom severity. The anxiety symptoms for each
question are rated on a four-point Likert-type scale that ranges from zero to three,
where zero indicates that the symptom has not been present in the past week and
three indicates that the symptom was severely present and the respondent could
barely stand it (Beck & Steer, 1993). A total score is obtained by adding
individual ratings on successive questions. Thus scores can range from zero to
Anxiety Sensitivity 93
63, with a higher total score indicating more severe anxiety symptomatology
(Beck & Steer, 1993). Similarly, Beck and Steer (1993) established cut-off
points for total scores in order to determine the severity of the anxiety symptoms.
Scores of 0-7 reflect 'minimal' anxiety, 8-15 reflects 'mild' anxiety, 16-25 reflects
'moderate' anxiety, and scores of 26-63 reflect 'severe' anxiety. Finally, Beck et
al. (1998), Hewitt and Norton (1993), and Creamer, Foran, and Bell (1995) have
reported that the BAI measures two principal aspects of anxiety, namely
cognitive and somatic symptoms. The ‘Cognitive’ subscale measures the
components of anxiety that are characterised by distorted cognitive functioning
and fearful thoughts, whereas the ‘Somatic’ subscale measures the components
of anxiety that are characterised by symptoms of physiological arousal (Creamer
et al., 1995).
The BAI is internally consistent with a high Cronbach alpha .92 for the
total score and .87, and .85 for the cognitive and somatic subscales respectively
(Beck & Steer, 1993; Hewitt & Norton, 1993). Test-retest reliability after one
week was reported by Beck and Steer (1993), as .75 (p <.001). In addition to
this, the BAI has acceptable content, concurrent, construct, discriminant and
factorial validity (Beck & Steer, 1993).
3.3.3 Catastrophic Cognitions Questionnaire (Modified). In order to
assess the nature of participants’ cognitions as they relate to anxiety, the
Catastrophic Cognitions Questionnaire -Modified (CCQ-M) (Khawaja, Oei, &
Baglioni, 1994) was employed. Based on the cognitive models of anxiety, the
CCQ-M is a valid and reliable self-report inventory, capable of measuring
individuals’ catastrophic cognitions. It is a modified version of the Catastrophic
Cognition Questionnaire (CCQ) (Khawaja & Oei, 1992) which was developed
Anxiety Sensitivity 94
primarily based on Beck et al’s. (1985) theory that postulates anxious individuals
misinterpret their emotional and physiological reaction states as dangerous
(Khawaja & Oei, 1992). The CCQ-M consists of 21 items that measure an
individual’s response to items that assess whether they perceive elements of
unpleasant emotions, physical changes, or thinking difficulty as dangerous. Each
question is weighted to reflect the degree of dangerousness. The degree of
dangerousness for each question is rated on a five-point Likert-type scale that
ranges from one to five, where the lowest level of the scale reflects the absence
of catastrophic misinterpretations and the highest level of the scale indicates the
maximum degree of catastrophic misinterpretations (Khawaja & Oei, 1992). The
items are divided into three subscales consisting of seven items each. The
‘Emotional Catastrophes’ subscale measures the misinterpretation of feelings,
such as anger or agitation, as dangerous; the ‘Physical Catastrophes’ subscale
reflects the misinterpretation of somatic symptoms, such as suffocation or having
a stroke, as dangerous; and the third subscale, ‘Mental Catastrophes’ reflects the
misinterpretation of mental dysfunctioning, such as losing memory or loss of
rational thinking, as dangerous (Khawaja et al., 1994).
The CCQ-M was validated using clinical and nonclinical populations.
The CCQ-M has good internal consistency with both clinical and nonclinical
populations. For the clinical sample, the Cronbach alphas for ‘Emotional
Catastrophes’, ‘Physical Catastrophes’, and ‘Mental Catastrophes’ factors were
.88, .85, and .91 respectively and for the nonclinical sample the Cronbach alphas
for ‘Emotional Catastrophes’; ‘Physical Catastrophes’; and ‘Mental
Catastrophes’ factors were .83, .85, and .89 respectively (Khawaja et al., 1994).
Test-retest reliability over a two-week interval, estimated on individuals with an
Anxiety Sensitivity 95
anxiety disorder, was acceptable. The reliability coefficients for the total scale as
well as ‘Emotional Catastrophes’; ‘Physical Catastrophes’; and ‘Mental
Catastrophes’ factors were reported as .63, .71, .58, and .67 (p > .001)
respectively (Khawaja et al., 1994). Additionally, the CCQ-M possesses
acceptable concurrent validity and differentiates clinical from nonclinical
samples.
3.3.4 Fear Questionnaire (FQ). The FQ (Marks & Mathews, 1979) is a
15-item self-report questionnaire designed to measure how phobic an individual
is in relation to a fearful stimulus as well as to monitor change during treatment.
Items are rated on a nine-point Likert-type scale that ranges from zero to eight,
with zero representing that the fearful stimulus would not be avoided and eight
representing that the fearful stimulus would always be avoided. Thus, scores can
range from zero to 120, with a higher total score indicating greater phobia.
Scores can also be divided into three, five-item subscales capable of measuring
agoraphobia (Ag), blood-injury phobia (B/I), and social phobia (Soc). The test-
retest reliability on a clinical population was reported to be adequate for the total
phobia scale and each of the three subscales (Marks & Mathews, 1979). An
investigation using an Australian sample revealed adequate internal consistency
for the full scale (α = .83 calculated using 15 items) and within each subscale
(Ag = .81, B/I = .71, and Soc = .74) as well as good discriminant validity (Oei,
Moylan & Evans, 1991).
3.3.5 Zung Self-Rating Depression Scale (Zung – SDS). The Zung - SDS
(Zung, 1965) is a 20-item self-report questionnaire designed to assess an
individuals’ depression severity by measuring the frequency of depressive
symptoms. Items are rated on a four-point Likert-type scale ranging from 1 to 4,
Anxiety Sensitivity 96
with reverse scoring for half the items reflecting positive well being. The Zung
SDS also contains four subscales capable of measuring an individual’s mood,
and somatic, psychomotor and cognitive symptoms as they relate to a depressive
episode. Results from several studies have established the Zung - SDS as a
reliable and valid instrument for measuring depressive symptoms (Biggs, Wylie
& Ziegle, 1978; Gabrys & Peters, 1985).
3.3.6 COPE Questionnaire (COPE). The COPE (Carver, Scheier &
Weintraub, 1989) is a 52-item self-report multidimensional coping inventory
designed to measure the different ways in which individuals respond to stress.
Items are rated on a four-point Likert-type scale that ranges from one to four.
Scores can be summed to measure 13 subscales that include active coping,
planning, suppression of competing activities, restraint coping, seeking social
support for instrumental reasons, seeking social support for emotional reasons,
positive interpretation and growth, acceptance, turning to religion, focus on
venting of emotions, denial, behavioural disengagement and mental
disengagement. Overall, the COPE contains adequate reliability and as well as
acceptable concurrent and discriminant validity.
3.3.7 Stress Subscale of the 21-item Depression Anxiety Stress Scale
(DASS-Stress). The stress subscale of the 21-item DASS (Antony, Bieling, Cox,
Enns & Swinson, 1998) contains seven items designed to measure individuals’
levels of chronic, non-specific arousal. The seven items are rated on a four-point
Likert-type scale that ranges from zero to three. The items are designed to assess
the extent to which an individual experienced difficulty relaxing, nervous
arousal, and being easily upset/agitated, irritable/over-reactive and impatient over
the past week. Scores can be summed to measure individual’s stress severity
Anxiety Sensitivity 97
rating. The internal consistency and concurrent validity of the DASS-21was in
the acceptable to excellent range (Antony et al., 1998).
3.3.8 Generalised Self-Efficacy Scale (SES). The generalised scale of the
SES (Sherer, Maddux, Mercandante, Prentice-Dunn, Jacobs & Rogers, 1982).
The generalised self-efficacy subscale is designed to measure the extent to which
an individual perceives their level of personal mastery. The scale is composed of
two subscales, general and social self-efficacy. In this dissertation, only the 17-
item generalised self-efficacy subscale was utilised. Responses to items are rated
on a five-point Likert-type scale that ranges from one to five. Sherer et al. (1992)
has demonstrated that the generalised self-efficacy subscale contains acceptable
internal consistency (Cronbach alpha = .86).
3.4 Procedure
As there were two groups of participants who took part in all of the
studies in this dissertation, the procedure for each will be discussed separately.
3.4.1 Participants from the clinical setting. The participants who formed
the clinical sample were selected from three clinics: The Cognitive-Behavioural
Therapy Unit at the Toowong Private Hospital; The Anxiety Disorders Unit at
the Wesley Private Hospital and the School of Psychology and Counselling
Clinic at the Queensland University of Technology. Each will be discussed in
turn.
The participants who formed the sample from the Cognitive-Behavioural
Therapy Unit at the Toowong Private Hospital were referred to the clinic by a
psychiatrist or general practioner in the Brisbane metropolitan area. Once
referred, participants were placed on a wait-list for the next available program
and a small sample received the ASI and ASI-R two weeks before the
Anxiety Sensitivity 98
commencement of treatment for reliability analyses. This time interval was
chosen because it followed Reiss et al’s. (1986) test-retest interval in their
development and examination of the original ASI. Treatment was manual driven
CBT for Anxiety or Depression (see Appendix E and F for an outline of the
manual) and consisted of five and a half contact hours (minus 30 minutes for
morning tea and one hour for lunch) twice a week for four weeks and there were
approximately 15 participants per treatment session. At the beginning of both
the first and final treatment sessions, participants were provided with the
questionnaire battery and invited to take part in the investigations as well as to
receive feedback on how they scored on the measures both before and after
treatment. Finally, all participants were invited back to the Cognitive-
Behavioural Therapy Unit after a period of three months for an afternoon
refresher program and to monitor their progress. However, as this component of
the program was not a part of the focus of this dissertation, further details
regarding this have not been included. A Professor of Psychology or a
provisionally registered psychological intern from the Clinical Psychology
Masters or PhD programs at the University of Queensland conducted the
therapeutic sessions and a registered Psychologist or Psychiatrist using an
unstructured interview based on the DSM-IV interviewed all participants prior to
participation in the study. All data from this location was collected over a period
of two and a half years.
The participants who formed the sample from the Anxiety Disorders Unit
at the Wesley Private Hospital were referred to the clinic by a general medical
practioner in the Brisbane metropolitan area. Once referred, participants had an
initial consultation with a psychiatrist to determine the nature of the presenting
Anxiety Sensitivity 99
problem, diagnosis and effective treatment plan. At this initial consultation, and
before commencement of any treatment, participants were provided with the
questionnaire battery and invited to take part in the studies. Interested
participants returned the complete questionnaires at the next appointment. All
data from this location was collected over a period of one year.
Finally, the participants who formed the sample from the School of
Psychology and Counselling Clinic at the Queensland University of Technology
volunteered to take part in a study of anxiety from advertisements in local
newspapers. Initial contact from interested participants involved telephoning the
author (who is also a conditionally registered Psychologist with the State of
Queensland and under supervision from the Principal Supervisor) and briefly
describing their problem. This process was utilised in order to provide interested
participants with further information about the study as well as identify those
who were possibly experiencing anxiety and would benefit from participation. A
screening sheet (see Appendix G) was developed by the Principal Supervisor and
author in order to identify individuals who would be suitable for participation in
the current investigations. Once suitable individuals were identified, an
appointment time was made and the questionnaire battery was mailed out with
instructions that it was to be completed prior to the appointment. All
appointments involved an interview which asked the participant about their
presenting problem and involved the administration of a Structured Clinical
Interview for DSM-IV Diagnosis. All interviews were conducted by a
conditionally registered Psychologist or by a fully-registered Clinical
Psychologist. Once each interview was completed and the diagnoses determined,
the interviewer explained to the participant about the nature of the problem,
Anxiety Sensitivity 100
prevalence, and treatment options available. In addition, each participant was
provided with an un-published handbook on the effects and treatment of anxiety
and anxiety disorders that also included a section on places where assistance is
available for the treatment of anxiety disorders in the Brisbane area (see
Appendix H). All data from this location was collected over a period of six
months.
Finally, information on the reliability of DSM-IV (APA, 1994) diagnoses
was available for two of the three clinics (the Cognitive-Behavioural Therapy
Unit at the Toowong Private Hospital and the School of Psychology and
Counselling Clinic at the Queensland University of Technology) and was
conducted on a small sample of the participants (n = 70). In one instance this
involved 100% agreement of an individual’s primary diagnosis between the
interviewer who was a conditionally registered psychologist and a supervising,
registered clinical psychologist for n = 30 participants when using a Structured
Clinical Interview for DSM-IV Diagnosis. The second instance involved 90%
agreement between the referring psychiatrist’s diagnosis and the clinic’s intake
interview and primary diagnosis for n = 40 participants. Reliability of diagnoses
for the participants from the Anxiety Disorders Unit at the Wesley Private
Hospital could not be collected as the Director of this unit passed away
unexpectantly and the clinic closed down.
3.4.2 Participants from the nonclinical setting. The participants who
formed the nonclinical sample were selected from either a first year psychology
unit at the Queensland University of Technology or the Queensland University of
Technology internal telephone directory. The procedure for each group will be
discussed separately.
Anxiety Sensitivity 101
Initially, participants who were recruited from a first year psychology
unit were invited to take part at a scheduled time and date from advertisements
placed on the first year Psychology notice board. Half of the testing sessions
were scheduled in the morning hours and half were scheduled in the afternoon
hours and a small sample of these participants completed the ASI and ASI-R two
weeks later for reliability analyses. Once the first year subject pool was
exhausted, the author visited three introductory Psychology units during teaching
time and verbally invited interested participants to take part at the end of the
class.
As a part of each testing session, all participants received uniform
information about the nature of the study and what they were required to do,
from an overhead projection transparency sheet (see Appendix I). All
participants were informed that there were no right or wrong answers and they
were free to discontinue their participation at any stage. The participants
completed the questionnaires in approximately 30 minutes and once all
questionnaires were returned, the participants were invited to remain for a 10-
minute debriefing session, which explained the current literature surrounding the
project as well as listed a number of support agencies within Brisbane and
surrounding areas available to those individuals who wished to utilise these
services. Additionally, all participants were informed that the QUT Counselling
Services Centre (Carseldine Campus) was available to offer support and
assistance if necessary. All data for this group was collected over a period of six
months.
In order to increase the generalisability of the nonclinical sample, it was
decided to sample beyond the first year subject pool by random recruitment of
Anxiety Sensitivity 102
participants from the Queensland University of Technology internal telephone
directory. Initially, the Queensland University of Technology Human Resource
Department printed and mailed out the names and department of every individual
employed at the University to the author. This list was then limited to those
individuals whose qualifications did not extend to the title of lecturer or above.
This was done in order to control for the effects of education as it is typical that
individuals with the title of lecturer or above have a Masters or PhD degree
which is not reflective of the general community.
For the first random selection, every tenth person was selected and
information about the study as well as the questionnaire battery was posted via
internal mail. Participants were invited to read the information about the study
and if interested, complete the consent form and questionnaire battery and return
both to the author’s internal address in separate packages to ensure
confidentiality. This process was repeated with the exception that now every
fifth person on the list was selected for mail-out. In all, the response rate was
approximately 30%. The un-published handbook on the effects and treatment of
anxiety and anxiety disorders was posted to each participant upon request.
It is important to note that ethical clearance was sought and obtained for
recruitment of both the clinical and nonclinical. Likewise, informed consent was
obtained prior to participation for participants in both the clinical and nonclinical
setting. For all participants, a registered Psychologist was always present or
contactable to provide support if necessary. The following chapter is the first
empirical chapter, which is concerned with investigating the previously identified
models of the ASI-R in order to determine acceptable model fit using both
clinical and nonclinical samples.
Anxiety Sensitivity 103
Chapter Four: Confirmatory factor analysis and psychometric properties of the
Anxiety Sensitivity Index-Revised in Australian clinical and nonclinical
populations
4.1 Introduction 105
4.2 Method 109
4.2.1 Participants 109
4.2.2 Measure 109
4.2.3 Procedure 110
4.2.4 Statistical Analysis 110
4.3 Results 111
4.3.1 Clinical and nonclinical groups CFA 111
4.3.2 Model modifications of Taylor and Cox’s (1998)
Model 114
4.3.3 Test of Configural Invariance of Modified Model 116
4.3.4 Test of Metric of Modified Model 116
4.3.5 Comparisons of Alternative Models 121
4.3.6 Comparisons of the 21-item ASI and competing
ASI models 123
4.3.7 Internal consistency and test-retest reliability of
the 21-item ASI 127
4.3.8 Concurrent Validity of the 21-item ASI 128
4.3.9 Discriminant Validity of the 21-item ASI 130
4.4 Discussion 133
4.5 Chapter Summary 137
Anxiety Sensitivity 104
Anxiety Sensitivity 105
CHAPTER FOUR
CONFIRMATORY FACTOR ANALYSIS AND PSYCHOMETRIC
PROPERTIES OF THE ANXIETY SENSITIVITY INDEX-REVISED IN
AUSTRALIAN CLINICAL AND NONCLINICAL POPULATIONS
4.1 Introduction
As seen in Section 2.2 of the literature review, the 16-item ASI was
developed by Reiss et al. (1986) to measure the anxiety sensitivity construct.
Numerous investigations of the scale have been conducted and the results have
varied considerably with some researchers arguing for a unidimensional structure
with 16 items (Reiss et al., 1986) or 14 items (Taylor et al., 1992); and others
arguing for a multidimensional construct a few as two dimensions (Blais et al.,
2001; Schmidt & Joiner, 2002), and as many as three (Stewart et al., 1997;
Taylor, 1996; Zinbarg et al., 1997), or four dimensions (Peterson & Heilbronner,
1987; Telch et al., 1989). Zinbarg et al. (1999) have argued that there is
sufficient evidence to suggest that anxiety sensitivity is hierarchically
constructed and the 16-item ASI contains three, partially distinct, first-order
dimensions that correspond to: (1) fear of physical sensations, (2) fear of publicly
observable anxiety and anxiety-related symptoms, and (3) fear of mental
incapacitation and that these three first-order dimensions load onto a single,
higher, second-order anxiety sensitivity dimension.
It is interesting that the broad consensus regarding the factor structure of
the ASI is that the construct is viewed as multidimensional, when ASI was
originally developed to measure a unitary, anxiety sensitivity construct (Reiss et
al., 1986). As seen in Section 2.4 of the literature review, the attempts to clarify
Anxiety Sensitivity 106
the dimensionality of anxiety sensitivity using the ASI is problematic because the
scale contains only 16 items; most of which assess fear of physical sensations
(see Stewart et al., 1997). Given that investigations of the first-order dimensions
of the ASI have validated the importance of the multidimensional perspective of
anxiety sensitivity (see Zinbarg et al., 1999 for review), Taylor and Cox (1998)
developed the 36-item Anxiety Sensitivity Index – Revised (ASI-R), in order to
provide a more comprehensive measure of the first-order anxiety sensitivity
dimensions. Results of their investigation found that the ASI-R contained a
hierarchical solution, which corresponded to four first-order dimensions related
to: (1) fear of respiratory symptoms, (2) fear of publicly observable symptoms,
(3) fear of cardiovascular symptoms, and (4) fear of cognitive dyscontrol and all
loaded onto a single, second-order or general anxiety sensitivity dimension.
Taylor and Cox (1998) also reported that the ASI-R contained acceptable
concurrent and discriminant validity; however the reliability coefficients for this
four-factor model were not reported.
Since the Taylor and Cox (1998) publication, only two investigations
have appeared that attempt to determine the factor structure and psychometric
properties of the expanded ASI-R (see Section 2.5 for review). The first was
conducted by Zvolensky et al. (2003). Using a large, diverse sample of
nonclinical participants from six countries, Zvolensky et al. (2003) found two
dimensions relating to: (1) fear of somatic sensations and (2) fear of social-
cognitive concerns to be the most replicable across all seven datasets.
Subsequent investigations revealed that the two dimensions demonstrated
acceptable psychometric properties.
Anxiety Sensitivity 107
The second investigation was conducted by Deacon et al. (2003) who
found four dimensions relating to: (1) beliefs about the harmful consequences of
somatic sensations, (2) fear of publicly observable anxiety reactions, (3) fear of
cognitive dyscontrol, and (4) fear of somatic sensations without explicit
consequences. Further, a single second-order dimension was also reported, again
indicating that the ASI-R was hierarchically structured. Replication of their
study using a separate sample found essentially the same factor structure as
reported above (Deacon et al., 2003). Evidence of acceptable psychometric
properties for both investigations was also reported.
The ASI-R opens up new and potentially important avenues for further
investigation, and although a great deal of research has been conducted into the
anxiety sensitivity construct, investigators have repeatedly noted the inherent
problems of measuring the construct when exclusively using the 16-item ASI.
This issue was addressed by Taylor and Cox (1998), who have argued that the
challenge for future research is to examine the ASI-R in other clinical samples as
well as to advance the understanding of the number and nature of first-order
anxiety sensitivity dimensions. In light of the limitations of the 16-item ASI, the
36-item ASI-R developed by Taylor and Cox (1998) is a promising instrument
for measuring anxiety sensitivity. The research that has been conducted using
the ASI-R has revealed that the lower-order dimensions demonstrated
theoretically consistent relationships with criterion variables and that the second-
order dimension correlated very well with the 16-item ASI, indicating that both
instruments measure the same construct. Further, given that the content validity
of the ASI-R is better than the 16-item ASI, it is proposed that it is an improved
measure for investigating anxiety sensitivity dimensions. Therefore, it is posited
Anxiety Sensitivity 108
that it is important to critically examine not only the factor structure of the ASI-R
but also the reliability and validity of the measure with both clinical and
nonclinical populations before its utility can be assessed.
The purpose of the current investigation was to empirically evaluate the
factor structure of the ASI-R for both a clinical and nonclinical sample by using
confirmatory factor analysis (CFA) on the hypothesised models identified by
Deacon et al. (2003); Taylor and Cox (1998); and Zvolensky et al. (2003). This
technique is advantageous over exploratory factor analysis in that it is based on a
strong theoretical and empirical foundation and tests the adequacy of a
hypothesised model by statistical means. Therefore, CFA is a much more
sophisticated technique to evaluate the underlying measurement model once
initial exploratory work has been done (Gerbing & Hamilton, 1996). As such, the
main hypothesis for the current investigation was that the models identified by
Deacon et al. (2003); Taylor and Cox (1998); and Zvolensky et al. (2003) will be
an adequate fit of both clinical and nonclinical datasets.
In the event of failure to identify either the Deacon et al. (2003); Taylor
and Cox (1998); and Zvolensky et al. (2003) measurement models, the aim of the
current investigation was to modify the ASI-R in order to identify the best
possible fitting model on the clinical sample. As there are no CFA studies of the
ASI-R available in the literature, any adequate hypothesised model would also be
subjected to a repeat CFA in a nonclinical sample as well as a multiple groups
CFA in order to test configural versus metric invariance of the model
respectively. As studies of the original ASI suggest that the factor structure is
convergent across clinical and nonclinical populations (see Zinbarg et al., 1999),
it is also important to test whether this holds true for the ASI-R. As such, the
Anxiety Sensitivity 109
second hypothesis was that any identified model that consists of adequate fit
indices will display both configural and metric invariance using a separate
nonclinical sample.
In order to test that any retained measurement models based on the ASI-R
are the best possible fit of the data, another aim of the investigation was to test
alternative, competing models of the ASI-R and original ASI before endorsing
any model fit. As such the third hypothesis was that any identified model, based
on the ASI-R that consists of adequate fit indices, will be a better fit of the data
than a unifactorial model or a multifactorial orthogonal or hierarchical model
based on a different structure. The fourth hypothesis was any identified model,
based on the ASI-R that consists of adequate fit indices, will also be a better fit
of the data than previously validated ASI models in the available literature.
The final aim was to examine and report the internal consistency, test-
retest reliability, and discriminant validity of any retained model or models as
well as to examine whether a pattern of theoretically consistent relationships
exists with conceptually related variables. Therefore, the final hypothesis was
that any retained model will contain acceptable internal consistency, test-retest
reliability, and discriminant validity.
4.2 Method
4.2.1 Participants
The participants who comprised the clinical and nonclinical samples have
been described in detail elsewhere (see Sections 3.1.2 and 3.1.3).
4.2.2 Measure
The measures used in the present investigation have been described in
detail elsewhere (see Sections 3.3.1 to 3.3.8).
Anxiety Sensitivity 110
4.2.3 Procedure
The procedure for both the clinical and nonclinical control groups has
been described in detail elsewhere (see Sections 3.4.1 and 3.4.2).
4.2.4 Statistical Analysis
For the confirmatory factor analysis, all hypothesised models were
examined with EQS Windows V7.5b (Bentler, 1995) using maximum likelihood
(ML) estimation. EQS employs a range of goodness-of-fit indices to estimate the
adequacy of the proposed model under investigation. Traditionally, a non-
significant chi-square statistic (χ²) or a statistic less than two times the degrees of
freedom is indicative of a good fitting model. Additionally, Hu and Bentler
(1999) recommend using the Satorra-Bentler scaling corrected chi-squared
statistic (SBχ²) if sample size is small or the data departs from normality because
it works well under non-robust conditions. However, since the χ² statistic is
highly sensitive to sample size, it is now accepted practice to employ a
combination of fit indices in conjunction with the chi-square statistic to
determine the adequacy of model fit. In addition to the χ² statistic, the fit of the
CFA models was assessed by the root mean square error of approximation
(RMSEA) which is capable of assessing how well a hypothesised model
reproduces the sample covariance matrix. For the RMSEA, a cut-off value
ranging from 0.05 or lower indicates good model fit and values up to 0.08
represent moderate model fit (Bentler, 1990; Browne & Cudeck, 1993; Hu &
Bentler, 1999). The Non-Normed Fit Index (NNFI) and Comparative Fit Index
(CFI) were also employed in order to compare the hypothesised model with the
null hypothesis. Generally, a cut-off value > .90 for the NNFI and CFI is
considered to be consistent with moderate model fit (Bentler, 1990) and a cutoff
Anxiety Sensitivity 111
value close to 0.95 indicates good model fit (Hu & Bentler, 1999). Akaike’s
information criterion (AIC), in which lower values between competing models
indicate better model fit, was also examined and reported for the purposes of
model comparison. Finally, the Statistical Package for Social Sciences Version
10 (SPSS V10) was employed to examine internal consistency, test-retest
reliability and construct validity for each sample.
4.3 Results Prior to analysis, the clinical and nonclinical datasets were assessed in
order to examine the assumptions of univariate and multivariate normality which
revealed moderate skewness and kurtosis on a number of variables. As
multivariate normality is not always obtainable in psychopathological research,
the decision was made to retain the variables without applying transformations
because skewness reflects logical and valid responses for both populations. The
decision was also made to employ the Sattora-Bentler χ² statistic, which is
utilised in place of the unadjusted χ² as it provides a descending correction for
the level of kurtosis when there is evidence of non-normality (Bentler, 1995).
The Kaiser-Meyer-Olkin measure of sampling adequacy was .937 and .927 for
the clinical and nonclinical datasets respectively.
4.3.1 Clinical and nonclinical groups CFA
The three hypothesised models of Deacon et al. (2003); Taylor and Cox
(1998); and Zvolensky et al. (2003) which had been derived from exploratory
factor analysis were tested separately for both the clinical and nonclinical groups
using CFA. For all hypothesised models, items that had the highest loading on a
dimension were included as an indicator of the dimension under investigation.
The four-factor hierarchical model identified by Deacon et al. (2003) included 12
Anxiety Sensitivity 112
items (items 32, 33, 11, 19, 26, 13, 25, 16, 7, 15, 9, 14) to form the first
dimension ‘beliefs about the harmful consequences of somatic sensations’; 9
items (items 30, 35, 30, 22, 1, 12, 24, 17, 18) to form the second dimension ‘fear
of publicly observable anxiety reactions’; 6 items (items 2, 23, 34, 31, 10, 36) to
form the third dimension ‘fear of cognitive dyscontrol’; and 9 items (items 5, 4,
3, 6, 8, 27, 29, 21, 28) to form the fourth dimension ‘fear of somatic sensations
without explicit consequences’. The four-factor hierarchical model identified by
Taylor and Cox (1998) included 12 items (items 1-12) to form the first
dimension ‘fear of respiratory symptoms’; 7 items (items 13-19) to form the
second dimension ‘fear of publicly observable symptoms’; 11 items (items 20-
30) to form the third dimension ‘fear of cardiac symptoms’; and 6 items (items
31-36) to form the fourth dimension ‘fear of cognitive dyscontrol’. Conversely,
the two-factor hierarchical model identified by Zvolensky et al. (2003) included
19 items (items 7, 22, 21, 24, 27, 23, 3, 1, 4, 8, 2, 10, 6, 26, 28, 20, 5, 30, 25) to
form the first dimension ‘fear of somatic sensations’ and 17 items (items 29, 13,
9, 17, 14, 16, 19, 18, 15, 11, 12, 31, 34, 35, 33, 36, 32) to form the second
dimension ‘fear of social-cognitive concerns’. For each of the hypothesised
models under investigation the variance of the first item for each of the first-
order dimensions was set to 1 and the remaining items were allowed to vary
freely in order to set the scale. The variance from the second-order dimension to
the first-order dimensions was also set to 1.
Overall, confirmatory factor analysis of the three hypothesised
hierarchical models identified by Deacon et al. (2003); Taylor and Cox (1998);
and Zvolensky et al. (2003) failed to provide an adequate fit of the data in either
the clinical group (see Table 4.1) or nonclinical group (see Table 4.2). It can be
Anxiety Sensitivity 113
seen that all of the goodness-of-fit indices are well out of range of the
recommended cutoff criteria for retaining a hypothesised model.
Table 4.1
Clinical group CFA of three hypothesised ASI-R models
Note. *p < .001
Hypothesised Model
χ² SBχ² df Model AIC
NNFI CFI RMSEA
Deacon et al. (2003)four-factor hierarchical model
3898.50*
1511.95*
590
2718.50
.546
.575
.152
Taylor and Cox (1998) four-factor hierarchical model
2170.15*
1722.68*
590
990.15
.781
.795
.105
Zvolensky et al. (2003) two-factor hierarchical model
3226.83*
2562.26*
592
2042.83
.639
.661
.136
Anxiety Sensitivity 114
Table 4.2
Nonclinical group CFA of three hypothesised ASI-R models
Note. *p < .001
Note. *p < .001
Note. *p < .001
4.3.2 Model modifications of Taylor and Cox’s (1998) Model
As the four-factor hierarchical model identified by Taylor and Cox (1998)
resulted in the best available goodness-of-fit indices for all three hypothesised
models under investigation, further analyses were performed using this model as
a basis on which to determine structural modifications. Additionally it was
determined that the clinical group data was the most representative group on
which to base model modifications. Therefore, further analyses of the clinical
group data, using EQS, were performed in an attempt to develop a more robust
and possibly more parsimonious model.
Initially, a unidimensional model was identified in order to test the fit of
the clinical dataset. However, this model failed to converge. As such,
Hypothesised Model
χ² SBχ² df Model AIC
NNFI CFI RMSEA
Deacon et al. (2003) four- factor hierarchical model
4604.23*
1033.73*
590
3427.23
.532
.562
.125
Taylor and Cox (1998) four-factor hierarchical model
2420.99*
1773.57*
590
1240.99
.801
.788
.085
Zvolensky et al. (2003) two-factor hierarchical model
4110.22*
2897.90*
592
2926.22
.592
.616
.117
Anxiety Sensitivity 115
modifications were made on the basis of the four-factor model identified by
Taylor and Cox (1998). Firstly, the covariance between items within each of the
factors was examined by inspecting items with large standardised residuals.
Bentler (1995) has argued “large values of standardised residuals point to the
variables that are not being well explained by the model” (p. 91). For example, it
was found that item 15 ‘I think it would be horrible for me to faint in public’
exhibited a large unmodelled covariance with item 14 ‘I believe it would be
horrible to vomit in public’ and item 5 ‘It scares me when I feel faint’. However,
there was not a large unmodelled covariance between item 14 and 5. This
revealed that item 15 was indicating a possible relationship between itself and
these two items when such a relationship was not hypothesised to exist. As such
the decision was made to remove item 15 from the model. Other items that were
removed from the model because they were resulting in large unmodelled
covariances were items 11, 12, 28 and 29.
The remaining 31 items were further evaluated on the basis of the
Lagrange Multiplier (LM) test. Parameters that were identified as improving
model fit at the multivariate level were examined to determine whether they were
significantly cross-loading onto more than one or more factors. Although some
cross-loading may be important for measuring overall anxiety sensitivity,
significant cross-loadings were removed in order to improve model fit and
develop a more parsimonious measure with a clean factor structure (see
Anderson & Gerbing, 1998). Further, unless a theory suggests items may be
cross-loaded, the presence of cross-loading items could be attributed to a
statistical artifact (Anderson & Gerbing, 1998). Therefore, in order to generate
factors that measured distinct aspects of anxiety sensitivity, any item that resulted
Anxiety Sensitivity 116
in a z score of >1.96 on more than one factor was removed from the solution.
This process resulted in items 5, 6, 7, 8, 9, 10, 20, 25, 26, and 30 being removed,
even though they also loaded onto the original factors. As such a more
parsimonious, 21-item four-factor hierarchical model was identified as the best
possible fit of the clinical data.
4.3.3 Test of Configural Invariance of Modified Model
In order to validate the solution obtained from the clinical dataset and
determine whether the same items from the clinical sample loaded onto the same
dimensions as the nonclinical sample, a separate CFA was performed using the
nonclinical sample data in order to test the configural invariance of the model.
Results revealed that the 21-item four-factor hierarchical model was an
acceptable fit of the nonclinical population data (see Table 4.3).
Table 4.3
Nonclinical group CFA of the 21-item ASI Hierarchical model
χ² SBχ² df Model
AIC
NNFI CFI RMSEA
21-item ASI
Hierarchical
Model
537.32*
403.04*
185
167.32
.922
.932
.066
Note. *p < .001
4.3.4 Test of Metric Invariance of Modified Model
Similarly, a multiple group CFA was performed in order to test the metric
invariance of the model between both the clinical and nonclinical group datasets.
Anxiety Sensitivity 117
The results revealed that the 21-item four-factor hierarchical model was again an
acceptable fit of both the clinical and nonclinical group datasets (see Table 4.4).
Table 4.4
Metric Invariance CFA statistics between the clinical and nonclinical group
datasets using the 21-item ASI
χ² df Model
AIC
NNFI CFI RMSEA
21-item ASI
Hierarchical
Model
1163.25*
391
381.25
.913
.919
.05
Note. *p < .001
Examination of each constraint indicated significantly different factor
loadings existed between the clinical and nonclinical groups on items 2, 16, 23,
24, 32, 33, 34, and 36 as well between the first-order dimensions 1, 3, and 4 and
the second-order or general anxiety sensitivity factor. Specifically items 2, 16,
24, 33, 34, and 36 were stronger indicators for the clinical group when compared
to the nonclinical groups’ factor loadings. Conversely, items 23 and 32 were
stronger indicators for the nonclinical groups’ when compared to the clinical
group. In addition, the loadings on dimensions 1, 3, and 4 were also stronger
indicators of the general anxiety sensitivity dimension for the nonclinical group
when compared to the clinical groups’ loadings. The loadings for the clinical
group are reported in Figure 4.1 and the loadings for the nonclinical group are
reported in Figure 4.2. It can be seen that the 21 items from the ASI-R resulted
in four first-order dimensions comprising of (1) Fear of respiratory symptoms
Anxiety Sensitivity 118
(four items); (2) Fear of publicly observable symptoms (six items); (3) Fear of
cardiovascular/stroke symptoms (five items); and (4) Fear of cognitive
dyscontrol (six items) that loaded onto a single, second-order dimension for both
the clinical and nonclinical samples.
Anxiety Sensitivity 119
Figure 4.1. Clinical group factor loadings for the items of the 21-item ASI with a hierarchical model of four first-order dimensions and a general higher-order dimension.
Publicly Observable
V 4
V 2
V 3 Respiratory
V 1
V 13
V 14
V 16
V 17
V 18
V 19
.60
.59
Cardiovascular/Stroke
V 27
V 24
V 23
V 22
V 21
Cognitive Dyscontrol
V 36
V 35
V 34
V 33
V 32
V 31
.55
.52
.82
.89
.68
.90
.71
.56
.60
.74
.87
.85
.77
.76
.91
.90
.75
.74
.82
.88
.91
.84
.83
Anxiety Sensitivity
Anxiety Sensitivity 120
Figure 4.2. Nonclinical group factor loadings for the items of the 21-item ASI with a hierarchical model of four first-order dimensions and a general higher-order dimension.
Publicly Observable
V 4
V 2
V 3 Respiratory
V 1
V 13
V 14
V 16
V 17
V 18
V 19
.62
.62
Cardiovascular/Stroke
V 27
V 24
V 23
V 22
V 21
Cognitive Dyscontrol
V 36
V 35
V 34
V 33
V 32
V 31
.79
.79
.74
.88
.67
.89
.67
.50
.62
.76
.83
.85
.83
.74
.83
.87
.79
.80
.87
.84
.86
.73
.81
Anxiety Sensitivity
Anxiety Sensitivity 121
For the clinical group, all items loaded very well (86% were above 0.70)
on their respective factors. The four first-order dimensions also loaded well (>
0.50) on the second-order anxiety sensitivity dimension. Similarly, for the
nonclinical group all items loaded very well (81% were above 0.70) on their
respective dimensions and the four first-order dimensions also loaded well (>
0.60) on the second-order anxiety sensitivity dimension. However, it is
important to note that while the number of first-order anxiety sensitivity
dimensions is identical to those identified by Taylor and Cox (1998) the number
of items unique to first-order dimension has changed because redundant or
problematic items were removed from the scale. There was also a change of one
of the dimension names from ‘Fear of Cardiac Sensations’ to ‘Fear of
Cardiovascular/Stroke Sensations’ to be more representative of the items
underlying that dimension. As a result of these changes the decision was made
to call this modified scale the 21-item ASI.
4.3.5 Comparisons of alternative models
In order to determine whether any alternative models could provide a
better fit of the data, the obtained 21-item four-factor hierarchical model was
compared with additional models. Firstly, a unifactorial model that contained all
of the items from the 21-item ASI was selected with the variance of the single
factor set to 1 in order to allow the individual items to vary freely. The second
alternative model tested was one that constrained the four first-order dimensions
of the 21-item ASI to be orthogonal. In order to set the scale, the path from the
first item for each of the first-order dimensions was set to 1 and the remaining
items were allowed to vary freely. Both the unifactorial and orthogonal models
failed to provide an adequate fit of the data in either the clinical or nonclinical
Anxiety Sensitivity 122
samples. In addition, a three-factor hierarchical model was also tested whereby
the ‘Fear of Respiratory Symptoms’ and ‘Fear of Cardiovascular/Stroke
Symptoms’ dimensions were combined to make one ‘Fear of Physiological
Symptoms’ dimension containing nine items (items 1, 2, 3, 4, 21, 22, 23, 24, and
27). The goodness-of fit indices for all three models for the clinical group are
reported in Table 4.5 and Table 4.6 for the nonclinical group. It can be seen that
while the orthogonal model is an improvement upon the unifactorial model, it too
does not meet the stringent criteria for adequate model fit as neither model was
capable of accounting for the increments in variance that is explained by the
hierarchical model. Likewise, the three-factor hierarchical model was not an
adequate fit of either the clinical or nonclinical datasets.
Table 4.5
Clinical groups CFA of the unifactorial, orthogonal 4-factor model and 3-factor
hierarchical models of the 21-item ASI.
Note. *p < .001
χ² SBχ² df Model AIC
NNFI CFI RMSEA
Unifactorial
21-item ASI 2266.93* 1926.17* 189 1888.98 .470 .523 .211
Multifactorial
21-item ASI Orthogonal 4-factor model
719.78*
624.71*
189
341.77
.865
.878
.107
Anxiety Sensitivity 123
Table 4.6
Nonclinical groups CFA of the unifactorial, orthogonal 4-factor model and 3-
factor hierarchical models of the 21-item ASI.
Note. *p < .001
4.3.6 Comparisons of the 21-item ASI and competing ASI models
In order to compare the 21-item ASI to previously validated ASI models
in the available literature, CFA was conducted on the unifactorial 16-item (Reiss
et al., 1986), 14-item (Taylor et al., 1992), 11-item (Blais et al., 2001), and 10-
item (Schmidt & Joiner, 2002) ASI models. Similarly, previous multifactorial
models of the ASI consisting of four-factors (Peterson & Heilbronner, 1987;
Telch et al., 1989), three-factors (Zinbarg et al., 1997) and two-factors (Blais et
al., 2001; Schmidt & Joiner, 2002) were also evaluated as orthogonal, oblique
and hierarchical structures. The goodness-of-fit indices for the clinical group are
reported in Table 4.7 and the goodness-of-fit indices for the nonclinical group are
reported in Table 4.8. It can be seen that for all models, the goodness-of-fit
indices were well below the cut-off criteria recommended for retaining a
χ² SBχ² df Model AIC
NNFI CFI RMSEA
Unifactorial
21-item ASI 2689.97* 1855.95* 189 2311.87 .462 .515 .175
Multifactorial
21-item ASI Orthogonal 4-factor model
826.31*
616.78*
189
448.31
.863
.876
.088
Anxiety Sensitivity 124
hypothesised model for both populations (CFI range .586 to .833; RMSEA range
.95 to .21).
Anxiety Sensitivity 125
Table 4.7
Clinical groups CFA of alternative hypothesised ASI models
χ²* SBχ²* df Model AIC
NNFI CFI RMSEA
Unifactorial
ASI16 728.18 608.97 104 520.18 .688 .730 .156 ASI14 (Taylor, 1992)
654.95
550.31
77
500.95
.676
.726
.174
ASI11 (Blais et al., 2001)
517.53
407.85
54
409.52
.683
.741
.186
ASI10 (Schmidt & Joiner, 2002)
390.88
321.88
35
320.88
.682
.753
.203
Multifactorial
Peterson and Heilbronner (1987) 4-factor model
Orthogonal 992.58 930.15 104 784.58 .556 .615 .186 Oblique 536.21 445.94 98 340.21 .768 .810 .135 Hierarchical 548.10 428.31 100 348.10 .767 .806 .135 Telch, Shermis, and Lucas (1989) 4-factor model
Orthogonal 969.88 884.09 104 761.88 .567 .625 .184 Oblique 510.47 424.35 98 314.47 .781 .821 .131 Hierarchical 533.59 418.14 100 333.59 .775 .812 .133 Zinbarg, Barlow, and Brown (1997) 3-factor model
Orthogonal 816.58 725.60 104 608.58 .644 .692 .167 Oblique 524.61 436.08 101 322.61 .782 .817 .130 Hierarchical 524.61 436.08 101 322.61 .782 .817 .130 Blais et al. (2001) 2-factor model
Orthogonal 466.16 379.06 44 378.16 .696 .757 .197 Oblique 333.39 258.26 43 247.39 .786 .833 .165 Hierarchical 333.39 249.70 42 249.39 .780 .832 .168 Schmidt and Joiner (2002) 2-factor model
Orthogonal 398.91 328.91 35 328.71 .675 .747 .205 Oblique 294.73 238.24 34 226.73 .760 .819 .176 Hierarchical 294.73 225.38 33 228.73 .752 .818 .179 Note. *p < .001
Anxiety Sensitivity 126
Table 4.8
Nonclinical groups CFA of alternative hypothesised ASI models
χ²* SBχ²* df Model AIC
NNFI CFI RMSEA
Unifactorial
ASI16 566.04 442.09 104 358.04 .621 .671 .117 ASI14 (Taylor, 1992)
475.93
368.15
77
321.93
.621
.680
.126
ASI11 (Blais et al., 2001)
326.64
227.60
44
238.64
.629
.704
.140
ASI10 (Schmidt & Joiner, 2002)
290.71
198.93
35
220.71
.596
.686
.150
Multifactorial
Peterson and Heilbronner (1987) 4-factor model
Orthogonal 686.15 554.43 104 478.15 .522 .586 .131 Oblique 415.57 330.85 98 219.57 .723 .774 .100 Hierarchical 419.28 334.00 100 219.28 .727 .773 .099 Telch, Shermis, and Lucas (1989) 4-factor model
Orthogonal 653.62 501.02 104 445.62 .549 .609 .128 Oblique 389.16 306.00 98 193.16 .746 .793 .096 Hierarchical 393.02 309.17 100 193.02 .750 .791 .095 Zinbarg, Barlow, and Brown (1997) 3-factor model
Orthogonal 604.08 482.94 104 396.08 .589 .644 .122 Oblique 438.70 346.34 101 .236.70 .714 .760 .102 Hierarchical 438.70 346.29 101 236.70 .714 .760 .102 Blais et al. (2001) 2-factor model
Orthogonal 298.19 211.68 44 210.19 .667 .733 .133 Oblique 235.41 167.41 43 .149.41 .742 .798 .117 Hierarchical 235.41 120.17 42 151.41 .734 .797 .119 Schmidt and Joiner (2002) 2-factor model
Orthogonal 264.51 195.65 35 194.51 .637 .718 .142 Oblique 212.66 154.69 34 144.66 .709 .780 .127 Hierarchical 212.66 90.01 33 146.66 .699 .779 .129 Note. *p < .001
Anxiety Sensitivity 127
Finally, the investigation also focused on examining the psychometric
properties and construct validity of any hypothesised model that resulted in
acceptable goodness-of-fit indices. As the only optimum model capable of
providing an adequate fit of either the clinical and nonclinical group datasets was
the 21-item four-factor hierarchical model, the adequacy of the psychometric
properties were determined for this model only.
4.3.7 Internal consistency and test-retest reliability of the 21-item ASI
Cronbach alphas and two-week test-retest reliability were calculated for
the four first-order dimensions for both the clinical and nonclinical samples.
Again, this time interval was chosen because it followed Reiss et al’s. (1986)
test-retest interval in their development and examination of the original ASI. It
can be seen (Table 4.9) that for both groups, the 21-item ASI can be considered
internally consistent and stable over time.
Table 4.9
Internal Consistency and Test-Retest Reliability of the 21-item ASI for the
Clinical and Nonclinical Groups
21-item ASI dimension
Clinical Group
(N = 248)
Nonclinical Group
(N = 435)
Total score .94 .90
Respiratory .92 .85
Publicly Observable .87 .85
Cardiovascular/Stroke .94 .88
Cognitive Dyscontrol .95 .91
Test-Retest Reliability (r = .76, n = 31) (r = .81, n = 39)
Anxiety Sensitivity 128
4.3.8 Concurrent Validity of the 21-item ASI
Correlations were calculated between the 21-item ASI, the four first-order
dimensions, and the total scores of the ASI, BAI, CCQ-M, FQ, Zung-SDS and
DASS-Stress Scale for both clinical group (see Table 4.10) and nonclinical group
(see Table 4.11). The first-order dimensions of the 21-item ASI were strongly
correlated with total score and the first-order dimensions were moderately
correlated with one another. There was also evidence of a strong relationship
between the 21-item ASI total score and the original ASI for the clinical and
nonclinical groups demonstrating that they both represent the anxiety sensitivity
construct. However, it is important to note that there are six items common to
each scale which account for such a large relationship.
The 21-item ASI total scores yielded mild to moderate correlations with
measures of anxiety symptoms (BAI), catastrophic cognitions (CCQ-M), fear of
anxiety (FQ), depression (Zung – SDS), and stress (DASS-Stress Scale). The
first-order dimensions were also small to moderately correlated with the BAI,
CCQ-M, FQ, Zung – SDS and DASS-Stress Scale. In most cases, the 21-item
ASI ‘fear of cognitive dyscontrol’ dimension was more strongly associated with
all of the criterion measures than the other dimensions.
Anxiety Sensitivity 129
Table 4.10
Pearson Correlations among the 21-item ASI, 16-item ASI, BAI, CCQ-M, FQ,
Zung – SDS, and DASS – Stress Scale for the clinical group
Measure
21-item ASI Second-Order
Dimension
21-item ASI First-Order Dimensions
----------- I II III IV
I
.75 --------
II
.71 .33 --------
III
.78 .57 .38 --------
IV
.83 .50 .50 .47 ---------
ASI (original)
.93 .65 .67 .69 .84
BAI
.67 .47 .49 .45 .64
CCQ-M
.57 .42 .39 .38 .57
FQ
.57 .39 .46 .43 .47
Zung – SDS
.42 .23 .29 .22 .49
DASS-Stress Scale
.49 .32 .39 .30 .51
Note. I = 21-item ASI Fear of Respiratory Symptoms dimension, II = 21-item
ASI Fear of Publicly Observable Anxiety Symptoms dimension, III = 21-item
Fear of Cardiovascular/Stroke Symptoms dimension, IV = 21-item ASI Fear of
Cognitive Dyscontrol dimension. All correlations are significant at p < .01.
Anxiety Sensitivity 130
Table 4.11
Pearson Correlations among the 21-item ASI, 16-item ASI, BAI, CCQ-M, FQ,
Zung – SDS, and DASS – Stress Scale for the nonclinical group
Measure
21-item ASI Second-Order
Dimension
21-item ASI First-Order Dimensions
----------- I II III IV
I
.65 --------
II
.77 .31 --------
III
.71 .33 .35 --------
IV
.74 .29 .40 .49 ---------
ASI (original)
.87 .53 .69 .63 .62
BAI
.48 .32 .41 .24 .36
CCQ-M
.45 .32 .35 .27 .33
FQ
.31 .25 .25 .17 .22
Zung – SDS
.30 .16 .22 .15 .32
DASS-Stress Scale
.40 .25 .38 .22 .26
Note. I = 21-item ASI Fear of Respiratory Symptoms dimension, II = 21-item
ASI Fear of Publicly Observable Anxiety Symptoms dimension, III = 21-item
Fear of Cardiovascular/Stroke Symptoms dimension, IV = 21-item ASI Fear of
Cognitive Dyscontrol dimension. All correlations are significant at p < .01.
4.3.9 Discriminant Validity of the 21-item ASI
Finally, there was evidence of the discriminant validity of the 21-item
ASI. Initially, correlations were calculated between the 21-item ASI and
Anxiety Sensitivity 131
measures of self-efficacy (Self-Efficacy Scale) and coping ability (COPE
Questionnaire), which revealed that the first and second-order dimensions of the
21-item ASI were capable of discriminating between conceptually unrelated
measures for both the clinical (see Table 4.12) and nonclinical groups (see Table
4.13).
Table 4.12
Pearson Correlations among the 21-item ASI, Self-Efficacy, and COPE
questionnaires for the clinical group
21-item ASI Second-Order
Dimension
21-item ASI First-Order Dimensions
Discriminant Measure
I II III IV
Self-Efficacy
.25** .10** .28** .13* .24**
COPE
.03 .02 .00 .04 .03
Note. I = 21-item ASI Fear of Respiratory Symptoms dimension, II = 21-item
ASI Fear of Publicly Observable Anxiety Symptoms dimension, III = 21-item
Fear of Cardiovascular/Stroke Symptoms dimension, IV = 21-item ASI Fear of
Cognitive Dyscontrol dimension. ** p < .01, *p < .05.
Anxiety Sensitivity 132
Table 4.13
Pearson Correlations among the 21-item ASI, Self-Efficacy, and COPE
questionnaires for the nonclinical group
21-item ASI Second-Order
Dimension
21-item ASI First-Order Dimensions
Discriminant Measure
I II III IV
Self-Efficacy
.13 .09 .06 .07 .19*
COPE
.10 .12 .09 .05 .01
Note. I = 21-item ASI Fear of Respiratory Symptoms dimension, II = 21-item
ASI Fear of Publicly Observable Anxiety Symptoms dimension, III = 21-item
Fear of Cardiovascular/Stroke Symptoms dimension, IV = 21-item ASI Fear of
Cognitive Dyscontrol dimension. *p < .05.
Additional evidence of the 21-item ASI ability to discriminate between
the clinical and nonclinical groups, not only for the total score or general
dimension but also for the four first-order dimensions pertaining to a ‘fear of
cardiovascular/stroke symptoms’, ‘fear of publicly observable anxiety reactions’,
‘fear of respiratory symptoms’ and ‘cognitive dyscontrol’ was also observed (see
Table 4.14). These findings are consistent with studies of the original ASI
showing that clinical populations typically exhibit significantly higher mean
scores when compared to nonclinical populations (see Cox et al., 1999 for
review).
Anxiety Sensitivity 133
Table 4.14
Between Groups Differences of the 21-item ASI and first-order dimensions:
Means, Standard Deviations and Univariate Results for the Clinical and
Nonclinical groups.
21-item ASI
dimension
Clinical Group
(N = 248)
M SD
Nonclinical Group
(N = 435)
M SD
Univariate F
(df 1, 681)
Total score 38.90 19.47 20.17 13.24 222.28*
Respiratory 6.94 5.03 5.52 4.11 15.92*
Publicly Observable 14.06 6.10 9.32 5.71 103.69*
Cardiovascular/Stroke 6.91 6.38 2.41 3.82 132.87*
Cognitive Dyscontrol 10.99 7.65 2.92 4.26 313.42*
Note. *p <. 0001.
4.4 Discussion
The purpose of the present investigation was to examine and validate the
ASI-R in a clinical and nonclinical sample. The results from CFA modelling of
both groups indicate that, through the removal of a number of problematic items,
the hypothesised model developed by Taylor and Cox (1998) can be improved
substantially by reducing the 36-item ASI-R to a 21-item index which is capable
of providing a more parsimonious and valid account of the anxiety sensitivity
construct. These dimensions were confirmed with a nonclinical sample of adults,
revealing that the 21-item ASI was capable of demonstrating adequate configural
and metric invariance.
Anxiety Sensitivity 134
Despite the numerous hypothesised models of anxiety sensitivity
measured, using either the original ASI or revised ASI-R that exist within the
available literature, it is important to note that not one of these models could be
confirmed using CFA in either a clinical or nonclinical sample. Therefore,
research in this area will continue to be problematic unless researchers move
away from the traditional exploratory factor analytic approach and adopt the
more stringent method of CFA which can identify the most robust items of a
scale and therefore the optimum identifiers of the construct that is under
measurement. The optimum model identified in the current investigation was the
21-item hierarchical four-factor model which was the only model capable of
providing an adequate fit of either the clinical and nonclinical group datasets.
It is important to note that there are sampling differences between the
participants who comprised the clinical sample for the present investigation and
those included in the Taylor and Cox (1998) study. First, the present
investigation employed a larger clinical sample size in order to reduce the impact
of instability on the covariance matrix that occurs with small sample sizes
(Tabachnick & Fidell, 2001). Secondly, while 21% of Taylor and Cox (1998)
participants were assessed following treatment, participants who formed the
clinical sample in the present investigation only included those individuals who
had been assessed before any treatment commenced. As such, it is proposed that
the methodology of the current investigation is more rigorous than that of Taylor
and Cox (1998). Further, these factors could have contributed to the overall
differences in models by improving the fit of theoretically relevant items and
reducing the impact of poor items on the model.
Anxiety Sensitivity 135
As the Taylor and Cox (1998) 36-item, four-factor hierarchical model
revealed the most promising fit statistics when compared to the models identified
by Deacon et al. (2003) or Zvolensky et al. (2003); it is important to note that the
current investigation sought to find the best possible model using the clinical
samples data because it was determined that this sample would be the most
representative sample to perform model modifications. The Lagrange Multiplier
test indicated that a number of items were significantly cross-loaded. Due to the
practical and theoretical implications of cross-loading items, the decision was
made to delete these items as they were too general or irrelevant to one of the
hypothesised anxiety sensitivity dimensions, or were so similar that individuals
had difficulty in separating the meaning of one item from another. For example,
it is clear that the item ‘It scares me when my body feels strange or different in
some way’ is not relevant to the specific anxiety sensitivity dimension ‘Fear of
Cardiac Sensations’ and is too general to tap into a specific anxiety sensitivity
dimension because a person could score highly on this variable for any number
of reasons. Conversely, evidence of large unmodelled covariance’s for a number
of items indicated that individuals had difficulty in separating the meaning of one
item from another. For example, it is probable that individuals had difficulty in
separating the fearful outcome of a publicly observable anxiety reaction
associated with fainting in public (item 15) with that of vomiting in public (item
14) as well as the fearful outcome of feeling faint because of a fear of respiratory
sensations (item 5).
Thus, while the current investigation reported the same number of lower-
order anxiety sensitivity dimensions as those identified by Taylor and Cox
(1998), the decision to remove redundant items resulted in a factor analytic
Anxiety Sensitivity 136
structure that was capable of providing an acceptable and reasonable account of
the anxiety sensitivity construct across two separate groups. The 21-item ASI
was more capable of accounting for increments in variance than either the
unifactorial or orthogonal models and did not perform well when forced into a
three-factor hierarchical model that attempted to replicate the dimensions
commonly reported when using the original ASI. It is proposed that if similar
results to those obtained in the current investigation can be confirmed in future,
then the 21-item ASI could be used in preference to the ASI-R and 16-item ASI.
Another aim of the current investigation was to examine the psychometric
properties of the 21-item ASI in both clinical and nonclinical populations. For
both samples, high Cronbach alphas were reported not only for the total score,
but also for the four first-order dimensions indicating that the 21-item ASI and its
dimensions are internally consistent. Similarly, high test-retest results for a
sample of both populations revealed that scores on the 21-item scale can be
considered stable over time. These results are similar to those reported by Reiss
et al., (1986) for use with the original ASI, indicating that the 21-item ASI is also
stable over time.
Concurrent validity was supported by the relationship between the 21-
item ASI and other conceptually related measures. Specifically, there was a
considerable relationship between the total score and scores on the original ASI
for both populations suggesting that both scales are similar measures of the
general anxiety sensitivity dimension. For the clinical population, all of the
dimensions were moderately correlated with the BAI, CCQ-M, FQ, Zung – SDS
and DASS-Stress scale. The dimension labelled ‘Cognitive Dyscontrol’ revealed
the highest association with the other measures highlighting the overlapping
Anxiety Sensitivity 137
cognitive processes of the scales. There were mild to moderate associations
between the 21-item ASI dimensions and the BAI, CCQ-M, FQ, Zung – SDS and
DASS – Stress scale for the nonclinical population. The pattern of correlations
for the 21-item ASI is the same as those published by Taylor and Cox (1998)
indicating that the nature of the scale has not been altered by the reduction of
problematic items. Similarly, discriminant validity was also supported by the
weak and non-significant relationship between the 21-item ASI and other
conceptually un-related measures of the Self-Efficacy scale and COPE
questionnaire, as well as the between-groups analysis that revealed the clinical
population scored significantly higher on the general and first-order anxiety
sensitivity dimensions than the nonclinical population. Similarly, fear of
manifesting anxiety symptoms in public was the most prominent feature for both
the clinical and nonclinical populations, whereas the fear associated with loss of
cognitive abilities was a salient feature of the clinical population, but not the
nonclinical population. All of these findings, when taken together, are consistent
with Reiss’s (1991) Expectancy Theory, which argues that anxiety sensitivity
augments anxiety responses.
4.5 Chapter Summary
This investigation is important, as it is the first to support the invariance
of the factor structure using 21 of the items contained in the expanded ASI-R
across a clinical and nonclinical sample. It is also the first to report the
psychometric properties of both samples adequately. Thus, while the 36-item
ASI-R was an important first step in investigating the multidimensional structure
of anxiety sensitivity, the 21-item ASI proposed in the present investigation
Anxiety Sensitivity 138
provides a more parsimonious, reliable and valid measure of anxiety sensitivity
in adults from both clinical and nonclinical populations.
Following on from these results, the second investigation is concerned
with the examination of differences in responding for various diagnostic groups
and the nonclinical control group on the dimensions measured by the 21-item
ASI. It is proposed that it is important to examine whether the measure
identified in the current investigation is capable of discriminating between and
within different diagnostic groups, as well as a nonclinical control group, in a
theoretically expected manner. As such, it is proposed that the following
investigation will provide further evaluation of the validity of this modified scale.
Anxiety Sensitivity 139
Chapter Five: Differences in anxiety sensitivity between diagnostic and
nonclinical groups using the 21-item Anxiety Sensitivity Index
5.1 Introduction 141
5.2 Method 144
5.2.1 Participants 144
5.2.2 Design 147
5.2.3 Measures 148
5.2.4 Procedure 148
5.3 Results 148
5.3.1 Data Screening and Cleaning 148
5.3.2 Preliminary Analyses 151
5.3.3 Tests of Hypotheses 153
5.3.3.1 Analysis of differences between
individuals with a primary diagnosis
of panic disorder, GAD, PTSD, depression,
and nonclinical controls on the total score
of the 21-item ASI 153
5.3.3.2 Analysis of differences between
individuals with a primary diagnosis
of panic disorder, GAD, PTSD, depression,
and nonclinical controls on the
specific dimensions of the 21-item ASI 155
Anxiety Sensitivity 140
5.3.3.3 Analysis of differences for the panic
disorder diagnostic group on the first-
order dimensions of anxiety sensitivity,
as assessed by the 21-item ASI 164
5.3.3.4 Analysis of differences for the GAD
diagnostic group on the first-order
dimensions of anxiety sensitivity, as
assessed by the 21-item ASI 166
5.3.3.5 Analysis of differences for the PTSD
diagnostic group on the first-order
dimensions of anxiety sensitivity, as
assessed by the 21-item ASI 167
5.3.3.6 Analysis of differences for the depression
diagnostic group on the first-order
dimensions of anxiety sensitivity, as
assessed by the 21-item ASI 169
5.3.3.7 Analysis of differences for the nonclinical
control group on the first-order dimensions
of anxiety sensitivity, as assessed by the
21-item ASI 171
5.4 Discussion 173
5.5 Chapter Summary 183
Anxiety Sensitivity 141
CHAPTER FIVE
DIFFERENCES IN ANXIETY SENSITIVITY BETWEEN AND WITHIN
DIAGNOSTIC AND NONCLINICAL GROUPS USING THE 21-ITEM
ANXIETY SENSITIVITY INDEX
5.1 Introduction
The previous chapter examined and validated the ASI-R in a clinical and
nonclinical sample using the method of confirmatory factor analysis and found
that, the ASI-R could be improved substantially by reducing the 36-item ASI-R
to a 21-item index through the removal of a number of problematic items. Thus,
while the 36-item ASI-R was an important first step in investigating the multi-
dimensional structure of anxiety sensitivity, the 21-item ASI provides a more
parsimonious, reliable and valid measure of anxiety sensitivity in adults from
both the clinical and nonclinical populations.
As seen in Section 2.8 of the literature review, anxiety sensitivity has
long been associated with different forms of psychopathology, including panic
disorder, social phobia, GAD, PTSD, OCD, specific phobia, and mood disorders.
The anxiety sensitivity construct has also been examined extensively using
nonclinical populations. However, while the research conducted in this area has
revealed important differences in anxiety sensitivity between different forms of
anxiety and mood pathology, it is important to note that all the investigations
conducted to date have been conducted using the 16-item ASI. As mentioned
previously (see Section 2.4), the original ASI has been subject to numerous
psychometric investigations, many of which have revealed that the original scale
does not contain adequate reliability or validity by virtue that it was never
Anxiety Sensitivity 142
designed to investigate the multidimensional structure of anxiety sensitivity. As
the previous investigation revealed that the 21-item ASI was capable of
providing an adequate account of the anxiety sensitivity construct across two
separate groups, it is considered important to conduct a further investigation that
focuses on examining the differences between the various aspects of anxiety
sensitivity and different anxiety and mood disorder classifications.
As previous investigations of anxiety sensitivity have focused on
comparing individuals with panic disorder to a mixed group of individuals with
other anxiety disorders (see Section 2.8.1), potentially important information has
been lost. The present investigation sought to compare groups of individuals
from either an anxiety disorder or depression diagnostic category as well as a
nonclinical control group, with respect to overall level of anxiety sensitivity.
While there is limited research available regarding the differences between
different diagnostic groups and the first-order dimensions of anxiety sensitivity,
it is still important to examine whether differences do exist. The current
investigation also sought to compare group responses on the first-order
dimensions of the 21-item ASI.
The first hypothesis of the current investigation was that there would be
significant differences in overall level of anxiety sensitivity between the
diagnostic groups as well as the nonclinical control group using the total score of
the 21-item ASI, while the second hypothesis states that there would be
significant differences in specific dimensions of anxiety sensitivity between the
diagnostic groups as well as the nonclinical control group using the four first-
order dimensions of the 21-item ASI. It was proposed that further examination
of the 21-item ASI ability to discriminate between groups in a theoretically
Anxiety Sensitivity 143
expected manner would lead to enhanced discriminant validity. This in turn
increases confidence in the construct validity of the scale. Thus, the present
investigation was also concerned with the theoretical and psychometric
significance of the measurement of anxiety sensitivity using the 21-item ASI.
A further issue concerns whether individual diagnostic groups endorse
particular facets of anxiety sensitivity over others in a theoretically expected
manner. As such, it was determined that the first-order dimensions of the 21-
item ASI should also be examined within each diagnostic group in order to
determine whether significant differences between the specific facets emerged
and, if so, whether they were consist with the theoretical expectations of each
diagnostic category. For example, as panic disorder is characterised as involving
a fear of physical and cognitive symptoms of anxiety (see Section 2.8.1 of the
literature review), it was hypothesised that individuals who meet the diagnostic
criteria for panic disorder with or without agoraphobia would be significantly
more fearful of the respiratory, cardiovascular/stroke and cognitive dyscontrol
dimensions of anxiety sensitivity than the publicly observable dimension of
anxiety sensitivity. Conversely, GAD is characterised as involving more
cognitive symptoms of anxiety than physiological (see Section 2.8.4 of the
literature review). As such, it was hypothesised that individuals who meet the
diagnostic criteria for GAD would be significantly more fearful of the cognitive
dyscontrol dimension of anxiety sensitivity than the respiratory, publicly
observable or cardiovascular/stroke dimensions.
The DSM-IV-TR (2000) characterises PTSD as involving a mixture of
physiological and cognitive symptoms (see Section 2.8.2 of the literature
review). Therefore, it was hypothesised that individuals who meet the diagnostic
Anxiety Sensitivity 144
criteria for PTSD would be significantly more fearful of the respiratory,
cardiovascular/stroke and cognitive dyscontrol dimensions of anxiety sensitivity
than the publicly observable anxiety sensitivity dimension.
Conversely, major depressive disorder and dysthymic disorder are
characterised as involving similar cognitive symptoms to those experienced by
individuals who meet the criteria for an anxiety disorder (see Section 2.8.6 of the
literature review). Therefore, it was hypothesised that individuals who meet the
diagnosis for these disorders and who are in a group called ‘depression’, would
be significantly more fearful of the cognitive dyscontrol dimension of anxiety
sensitivity than the fear of the respiratory, publicly observable and
cardiovascular/stroke anxiety sensitivity dimensions.
Finally there is no published literature that details how individuals from
the general community respond with reference to the first-order dimensions of
either the original 16-item ASI or 36-item ASI-R. As there is no theoretical
foundation on which to base hypotheses, the first-order dimensions for this group
were simply analysed and reported.
5.2 Method
5.2.1 Participants
A total of 648 individuals participated in the current study and were
separated into two main categories; diagnostic groups or nonclinical control
group. The first group consisted of individuals who were considered as having
met the criteria for a principal diagnosis of one of the following DSM-IV
diagnostic categories; Panic Disorder (n = 76), Generalised Anxiety Disorder
(GAD; n = 46), Posttraumatic Stress Disorder (PTSD; n = 21); Major Depression
or Dysthymic Disorder (Depression; n = 58), Social Phobia (n = 7) and Other
Anxiety Sensitivity 145
Anxiety (consisting of individuals who met the diagnostic criteria for Obsessive
Compulsive Disorder, Specific Phobia or Hypochondrias, [n = 5]).
As mentioned earlier (see Section 2.8.6), dysthymic disorder and major
depressive disorder are differentiated based on severity, chronicity, and
persistence, the differences between the two are particularly complicated because
they share similar symptoms and differences in onset, duration, persistence, and
severity that are not easy to evaluate retrospectively (APA, 2000). As a result, it
was maintained that it was legitimate to combine individuals who met the
diagnostic criteria for either disorder into a group called ‘depression’ for the
purposes of this dissertation.
Individuals that comprised the diagnostic groups were recruited from the
Cognitive-Behavioural Therapy Unit at the Toowong Private Hospital and the
School of Psychology and Counselling Therapy Clinic at the Queensland
University of Technology (see Section 3.4.1 for details). Table 5.1 displays the
mean age and range for the diagnostic group while Table 5.2 displays the general
demographic data for each diagnostic group. The participants who comprised the
nonclinical control group are the same as has been described earlier in the
methodology chapter (see Section 3.1.3 for details) and therefore will not be
repeated again here.
Anxiety Sensitivity 146
Table 5.1
Means, Standard Deviations and Age Range for diagnostic groups
Diagnostic
Group
N Mean Age
(in Years)
Standard Deviation Age Range
(in Years)
Panic Disorder 76 44.93 11.46 21-72
GAD 46 41.72 12.15 20-71
PTSD 21 44.05 10.77 23-57
Depression 58 43.72 11.41 19-67
Other Anxiety 5 49.00 18.03 29-66
Social Phobia 7 31.71 11.86 18-53
Note. GAD = Generalised Anxiety Disorder, PTSD = Posttraumatic Stress
Disorder.
Anxiety Sensitivity 147
Table 5.2
Sample size of Demographic Variables for the Diagnostic Groups
Panic
Disorder
GAD PTSD Depression Other
Anxiety
Social
Phobia
Demographic Categories
N (%) N (%) N (%) N (%) N (%) N (%)
Gender
Male
Female
19 (25.0)
57 (75.0)
19 (41.3)
27 (58.7)
12 (57.1)
9 (42.9)
19 (32.8)
39 (67.2)
2 (40.0)
3 (60.0)
2 (28.6)
5 (71.4)
Origin of Birth
Australia
Other than Australia
No Response
57 (75.0)
15 (19.4)
4 (5.3)
38 (82.6)
6 (13.0)
2 (4.3)
15 (71.4)
4 (19)
2 (9.5)
52 (89.7)
6 (10.3)
0 (0.0)
4 (80.0)
1 (20.0)
0 (0.0)
6 (85.7)
1 (14.3)
0 (0.0)
Language Regularly
Spoken at Home
English
Other than English
No Response
71 (93.4)
1 (1.3)
4 (5.3)
43 (93.5)
1 (2.2)
2 (4.3)
18 (85.7)
1 (4.8)
2 (9.5)
58 (100)
0 (0.0)
0 (0.0)
5 (100)
0 (0.0)
0 (0.0)
7 (100)
0 (0.0)
0 (0.0)
Education
High School
Certificate
TAFE Certificate
TAFE Diploma/
Ass. Diploma
University U/G
University P/G
Other
No Response
28 (36.8)
14 (18.4)
6 (7.9)
15 (19.7)
3 (3.9)
6 (7.9)
4 (5.3)
18 (39.1)
2 (4.3)
4 (8.7)
8 (17.4)
8 (17.4)
4 (8.7)
2 (4.3)
12 (57.1)
0 (0.0)
2 (9.5)
1 (4.8)
1 (4.8)
3 (14.3)
2 (9.5)
17 (29.3)
3 (5.2)
6 (10.3)
7 (12.1)
14 (24.1)
10 (17.2)
1 (1.7)
0 (0.0)
2 (40.0)
1 (20.0)
1 (20.0)
0 (0.0)
1 (20.0)
0 (0.0)
1 (14.3)
1 (14.3)
1 (14.3)
2 (28.6)
2 (28.6)
0 (0.0)
0 (0.0)
Note. Percentages are in parenthesis
5.2.2 Design
As it was determined that two diagnostic categories, other anxiety and
social phobia, did not contain a sample size large enough to be included in the
Anxiety Sensitivity 148
analyses, the current study was conducted using a between-groups design as well
as a within-groups design, with participants being separated into five groups:
panic disorder, GAD, PTSD, depression, and nonclinical controls. The
participants were measured on the 21-item ASI in order to determine how the
general and lower-order dimensions of anxiety sensitivity varied across and
within the DSM-IV anxiety disorders as well as the general community.
5.2.3 Measures
21-item Anxiety Sensitivity Index (see Appendix C). The measure has
been described in greater detail elsewhere (for details see Chapter 5). Briefly, the
21-item ASI possesses good reliability and validity in both clinical and
nonclinical populations and is regarded as internally consistent.
5.2.4 Procedure
The procedure for the participants used in the current study is the same as
has been described in detail elsewhere (see Sections 3.4.1 and 3.4.2). The only
difference to the procedure was that participants from the Anxiety Disorders
Clinic at the Wesley Hospital were not included in the present study as the
percentage of inter-rater reliability of diagnoses could not be established.
It is important to note that the clinical group was comprised of individuals
who were considered as having met the criteria for a principal diagnosis of one of
the anxiety diagnostic categories, major depression or dysthymia. Therefore,
some comorbidity amongst diagnostic groups was evident.
5.3 Results
5.3.1 Data Screening and Cleaning
Prior to analysis, responses on the 21-item ASI were examined using
SPSS Version 10 for errors in data entry, missing values, fit between
Anxiety Sensitivity 149
distributions and the assumptions of ANOVA and MANOVA. All variables
were examined separately for the diagnostic groups. It was determined that two
diagnostic categories; other anxiety and social phobia, did not contain a sample
size large enough to be included in the analyses. Therefore these two groups
were excluded from any further consideration in the current chapter and only the
panic disorder, GAD, PTSD, depression and nonclinical control groups will be
described and analysed from herein.
Examination of the data for all items contained in the 21-item ASI
revealed that there was no missing data. For the total scale items, seven cases,
one from the depression group and six from the nonclinical control group were
identified as univariate outliers because of their very large (> 3.29) z scores.
Univariate outlier’s were also observed for the ‘Fear of Respiratory Symptoms’
scale, with one case from the depression group; ‘Fear of Cardiovascular/Stroke
Symptoms’ scale, with one case from the depression group and seven cases from
the nonclinical control group; and ‘Fear of Cognitive Dyscontrol’ scale, with
seven cases from the nonclinical control group. Univariate outliers were not
observed for the ‘Fear of Publicly Observable’ scale. Using Mahalanobis
distance with p = .000, 13 cases, one from the depression group and seven from
the nonclinical control group, most of which were observed earlier as univariate
outliers, were also identified as multivariate outliers. Examination of these cases
separately revealed extremely high overall scores on the 21-item ASI, when
compared to other cases within the representative group. In order to reduce the
influence of Type I and Type II errors, cases that were identified as either a
univariate or multivariate outlier were deleted from the data matrix because it
was determined that they were not a part of the population for which the present
Anxiety Sensitivity 150
study was intended (Tabachnick & Fidell, 2001). These case deletions resulted
in 76 cases in the panic disorder group, 46 cases in the GAD group, 21 cases in
the PTSD group, 57 cases in the depression group, and 423 cases in the
nonclinical control group.
With the use of SPSS EXPLORE, all distributions were assessed for
normality by group which revealed moderate skewness on three subscales for the
depression group (21-item ASI ‘Fear of Respiratory Symptoms’, ‘Fear of
Cardiovascular/Stroke Symptoms’ and ‘Fear of Cognitive Dyscontrol’).
However as multivariate procedures of MANOVA were to be conducted,
Tabachnick and Fidell (2001) argue that if the departure from normality is due to
skewness rather than outliers, these analyses are robust to violations of normality
provided that there is at least 20 degrees of freedom for error in the univariate
ANOVA case and two-tailed tests are used.
Tests of homogeneity of variance revealed univariate breaches on three
subscales (‘Fear of Respiratory Symptoms’, ‘Fear of Cardiovascular/Stroke
Symptoms, and ‘Cognitive Dyscontrol’) as well as the total 21-item ASI scale.
Additionally, Box’s M was also significant (p = .000), indicating that the
variance-covariance matrix was not homogeneous. In order to deal with Type I
error inflation that is associated with breaches of homogeneity of variance, a
rigorous alpha level of .025 was used for all analyses.
Examination of the correlation matrices revealed that all variables were
moderately correlated, but not to such an extent as to make any one variable
redundant (see Section 4.3.8 for correlation table). Therefore, the assumption of
multicollinearity and singularity was met. Finally, testing for linearity revealed
that this assumption was also met.
Anxiety Sensitivity 151
5.3.2 Preliminary Analyses
Preliminary analyses were conducted using the total score and first-order
dimensions of the 21-item ASI with gender, age, and level of education attained
in order to determine whether a statistical difference existed between the four
diagnostic groups as well as the nonclinical control group. No significant
differences for any of the groups using these three variables were observed. This
excluded the possibility of these factors acting as confounds when comparing the
five groups.
In addition, further analyses were conducted using the variables of
gender, age, and level of education attained in order to determine whether a
statistical difference existed between the diagnostic groups recruited from both
the Cognitive-Behavioural Therapy Unit at the Toowong Private Hospital and the
School of Psychology and Counselling Therapy Clinic at the Queensland
University of Technology. Again, no significant differences between the two
groups using these three variables were observed. As such, it was reasonable to
combine the two clinical groups.
The means and standard deviations of the 21-item ASI total scale and
first-order dimension scores for the four diagnostic groups used in the current
study as well as the nonclinical control group are reported in Table 5.3. It can be
seen that the panic disorder group have higher mean scores on almost all total
and subscale scores, followed by the PTSD group, GAD group, depression and
nonclinical control group respectively.
Anxiety Sensitivity 152
Table 5.3.
Means and Standard Deviations for the Diagnostic and Nonclinical Control
Groups on the 21-item ASI dimensions
Mean Standard
Deviation
Panic Disorder (n = 76)
Total 21-item ASI scale
Respiratory Symptoms Dimension
Publicly Observable Symptoms Dimension
Cardiovascular/Stroke Symptoms Dimension
Cognitive Dyscontrol Symptoms Dimension
55.21
9.67
17.79
11.11
16.64
11.23
4.40
4.00
5.75
5.92
GAD (n = 46)
Total 21-item ASI scale
Respiratory Symptoms Dimension
Publicly Observable Symptoms Dimension
Cardiovascular/Stroke Symptoms Dimension
Cognitive Dyscontrol Symptoms Dimension
30.52
5.11
12.54
4.46
8.41
8.29
4.10
4.72
3.66
5.94
PTSD (n = 21)
Total 21-item ASI scale
Respiratory Symptoms Dimension
Publicly Observable Symptoms Dimension
Cardiovascular/Stroke Symptoms Dimension
Cognitive Dyscontrol Symptoms Dimension
54.52
10.19
17.33
11.76
15.24
22.02
5.10
6.37
7.21
7.35
Depression (n = 57)
Total 21-item ASI scale
Respiratory Symptoms Dimension
Publicly Observable Symptoms Dimension
Cardiovascular/Stroke Symptoms Dimension
Cognitive Dyscontrol Symptoms Dimension
18.95
3.06
9.00
1.93
4.98
11.96
3.43
5.39
2.85
4.71
Nonclinical Control (n = 423)
Total 21-item ASI scale
Respiratory Symptoms Dimension
Publicly Observable Symptoms Dimension
Cardiovascular/Stroke Symptoms Dimension
Cognitive Dyscontrol Symptoms Dimension
19.08
5.38
8.99
2.13
2.58
11.49
3.97
5.41
3.18
3.52
Anxiety Sensitivity 153
Examination of Table 5.3 reveals that, on the surface, there appears to be
a small but significant difference between the Depression and the baseline
nonclinical control groups. This is in contrast to previous investigations that
have found elevated levels of anxiety sensitivity in individuals with major
depression when using the original ASI. Therefore, it was decided to perform
another preliminary analysis in order to empirically show that the depression
group was different from the baseline nonclinical control group. Using the total
score scale of the Zung - SDS, an independent groups t-test was performed that
showed a statistically significant difference, t (330) = 10.28, p = .000, existed
between the depression and baseline nonclinical control groups. This is
important as the Zung – SDS is both a reliable and valid measure, capable of
discriminating between individuals with or without major depression or
something more than major depression. As there was a significant difference
between the depression group (M = 49.72, SD = 8.26) and nonclinical control
group (M = 37.99, SD = 7.56) self-rated depression, this shows that the
depression group’s mean score is consistent with a diagnosis of major depression
while the baseline nonclinical control group’s mean score is consistent with a
rating of ‘no depression’.
5.3.3 Tests of Hypotheses
5.3.3.1 Analysis of differences between individuals with a primary
diagnosis of panic disorder, GAD, PTSD, depression, and nonclinical controls
on the total score of the 21-item ASI. In order to examine whether any
significant differences exist between the different diagnostic group and
nonclinical controls, a between-subjects analysis of variance (ANOVA) with
follow-up post-hoc tests conducted using the total score of the 21-item ASI as the
Anxiety Sensitivity 154
dependent variable and diagnostic category, with five levels, as the independent
variable. Keeping in mind that the alpha level has been set to .025 because the
homogeneity of variance assumption has been violated, a significant difference
was found between diagnostic and nonclinical control groups on the dependent
measures, F (4, 618) = 190.201, p = .000. The results reflected a moderate
association between the diagnostic groups (panic disorder, GAD, PTSD,
depression, and nonclinical control) and the dependent variable (21-item ASI
total score), η² = .552. The observed power was strong, 1.000.
Examination of diagnostic differences by follow-up post-hoc tests using
Bonferroni adjustment were also conducted to determine where differences
existed. To control for Type I error, an alpha correction of p = .005 was set.
Significant differences were observed between the panic disorder and GAD
diagnostic groups, p = .000; panic disorder and depression diagnostic groups, p =
.000; panic disorder and nonclinical control groups, p = .000; GAD and PTSD
diagnostic groups, p = .000; GAD and depression diagnostic groups, p = .000;
GAD and nonclinical control groups, p = .000; PTSD and depression diagnostic
groups, p = .000; as well as the PTSD and nonclinical control groups, p = .000.
No significant differences were observed, however, between the panic disorder
and PTSD diagnostic groups, p = 1.000; and depression and nonclinical control
groups, p = 1.000. A profile plot illustrating the means of the 21-item ASI total
score as a function of the four diagnostic groups as well as the nonclinical control
group is provided in Figure 5.1. These results reveal that individuals with a
diagnosis of panic disorder or PTSD score significantly higher on the 21-item
ASI compared to individuals with a diagnosis of GAD or depression and
individuals from the nonclinical control group. Conversely, individuals with a
Anxiety Sensitivity 155
diagnosis of depression respond in a similar manner to those individuals from the
general community.
0
10
20
30
40
50
60
Panic GAD PTSD Depression NonclinicalDiagnostic Category
Mea
n 21
-item
AS
I tot
al it
em s
cale
Figure 5.1. Mean differences and standard error between panic disorder, GAD,
PTSD, depression, and nonclinical control groups and the 21-item ASI total
score.
5.3.3.2 Analysis of differences between individuals with a primary
diagnosis of panic disorder, GAD, PTSD, depression, and nonclinical controls
on the specific dimensions of the 21-item ASI. In order to examine whether there
were significant differences between the diagnostic groups as well as the
nonclinical control group on the four first-order 21-item ASI dimensions, a
between-subjects multivariate analysis of variance (MANOVA) with post-hoc
tests conducted using the ‘Fear of Respiratory Symptoms’, ‘Fear of Publicly
Observable Anxiety Symptoms’, ‘Fear of Cardiovascular/Stroke Symptoms’ and
‘Fear of Cognitive Dyscontrol’ dimensions as the dependent variables and
diagnostic group as the independent variable. Again, as the assumption of
Anxiety Sensitivity 156
homogeneity of variance-covariance was breached, a more stringent alpha level
of .025 was set for the between-groups multivariate effect. Using Wilks’
criterion, there was a significant multivariate effect between the diagnostic group
and the combined dependent measures, F (4, 16) = 48.531, p = .000. The results
reflected a small association between the diagnostic groups (panic disorder,
GAD, PTSD, depression, and nonclinical controls) and the combined dependent
variables, η² = .232. The observed power was strong, 1.000.
Examination of the between-subjects effect for each of the dependent
variables revealed significant differences between the diagnostic groups on the
21-item ASI ‘Fear of Respiratory Symptoms’ dimension, F (4, 618) = 31.993, p
= .000, η² = .172, power = 1.000; ‘Fear of Publicly Observable Symptoms’
dimension, F (4, 618) = 57.309, p = .000, η² = .271, power, = 1.000; ‘Fear of
Cardiovascular/Stroke Symptoms’ dimension, F (4, 618) = 118.404, p = .000, η²
= .434, power = 1.000; and ‘Fear of Cognitive Dyscontrol’ dimension, F (4, 618)
= 202.669, p = .000, η² = .567, power = 1.000.
To control for Type I error, an alpha correction of p = .006 was set for
follow-up post-hoc Bonferroni comparisons. Significant differences were
observed on the ‘Fear of Respiratory Symptoms’ dimension between the panic
disorder and GAD diagnostic groups, p = .000; panic disorder and depression
diagnostic groups, p = .000; panic disorder and nonclinical control groups, p =
.000; GAD and PTSD diagnostic groups, p = .000; PTSD and depression
diagnostic groups, p = .000; PTSD and nonclinical control groups, p = .000, and
depression and nonclinical control groups, p = .000. Significant differences were
not observed, however, between the panic disorder and PTSD diagnostic groups,
p = 1.000, GAD and depression diagnostic groups, p = .096, and GAD and
Anxiety Sensitivity 157
nonclinical control groups, p = 1.000. Accordingly, the results on this factor
revealed that individuals with a diagnosis of panic disorder appear to be
significantly more fearful of the respiratory symptoms of anxiety when compared
to individuals with a diagnosis of GAD, depression, or individuals from the
nonclinical control group. Not surprisingly, individuals with a diagnosis of
PTSD responded in a manner similar to that observed in individuals with a
diagnosis of panic disorder. Conversely, individuals with a diagnosis of GAD do
not appear to be significantly more fearful of the respiratory symptoms of anxiety
when compared to individuals with a diagnosis of depression or individuals from
the nonclinical control group. However, it is interesting to note that individuals
from the general community report significantly greater concern regarding these
symptoms when compared to individuals from the depression group. A profile
plot illustrating the means of the 21-item ASI ‘Fear of Respiratory Symptoms’
dimension as a function of the four diagnostic groups as well as nonclinical
control group is provided in Figure 5.2.
Anxiety Sensitivity 158
0
2
4
6
8
10
12
Panic GAD PTSD Depression NonclinicalDiagnostic Category
Mea
n 21
-item
AS
I 'Fe
ar o
f Res
pira
tory
Sym
ptom
s'
Dim
ensi
on
Figure 5.2. Mean differences and standard error between panic disorder, GAD,
PTSD, depression, and nonclinical control groups and the 21-item ASI ‘Fear of
Respiratory Symptoms’ dimension.
For the dimension, ‘Fear of Publicly Observable Symptoms’, significant
differences were observed between the panic disorder and GAD diagnostic
groups, p = .000; panic disorder and depression diagnostic groups, p = .000;
panic disorder and nonclinical control groups, p = .000; GAD and nonclinical
control groups, p = .000, PTSD and depression diagnostic groups, p = .000; and
PTSD and nonclinical control groups, p = .000. Conversely, significant
differences were not observed between the panic disorder and PTSD diagnostic
groups, p = 1.000; or the depression and nonclinical control groups, p = 1.000. It
is interesting to note that the significance level was approached between the
GAD and PTSD diagnostic groups, p = .006 as well as the GAD and depression
diagnostic groups, p = .007 on this dimension. However, if a less stringent alpha
level had been adopted, the differences between these groups would be
considered significant. As such, the results on this factor reveal that individuals
Anxiety Sensitivity 159
with a diagnosis of panic disorder appear to be significantly more fearful of
anxiety symptoms that are observable by others when compared to individuals
with a diagnosis of GAD, depression or individuals from the nonclinical control
group, while individuals with a diagnosis of GAD appear to be significantly
more concerned with the social consequences of anxiety symptoms when they
are compared to individuals from the nonclinical control group. Significant
differences were also observed between the PTSD and depression groups, with
individuals with a diagnosis of PTSD indicating greater concern on this
dimension than individuals with a diagnosis of depression. A similar result was
also observed when comparing the PTSD and nonclinical control groups.
Conversely, a significant difference was not observed between individuals with a
diagnosis of panic disorder and PTSD or individuals with a diagnosis of
depression when compared to individuals from the nonclinical control group. It
was interesting to note that individuals with a diagnosis of GAD did not differ
significantly from individuals with a diagnosis of PTSD or depression when a
conservative alpha level was adopted. A profile plot illustrating the means of the
21-item ASI ‘Fear of Publicly Observable Symptoms’ dimension as a function of
the four diagnostic groups and the nonclinical controls is provided in Figure 5.3.
Anxiety Sensitivity 160
0
2
4
6
8
10
12
14
16
18
20
Panic GAD PTSD Depression NonclinicalDiagnostic Category
Mea
n 21
-item
AS
I 'Fe
ar o
f Pub
licly
Obs
erva
ble
Sym
ptom
s'D
imen
sion
Figure 5.3. Mean differences and standard error between panic disorder, GAD,
PTSD, depression, and nonclinical control groups and the 21-item ASI ‘Fear of
Publicly Observable Symptoms’ dimension.
For the dimension ‘Fear of Cardiovascular/Stroke Symptoms’, significant
differences were observed between the panic disorder and GAD diagnostic
groups, p = .000; panic disorder and depression diagnostic groups, p = .000;
panic disorder and nonclinical control groups, p = .000; GAD and PTSD
diagnostic groups, p = .000; GAD and nonclinical control groups, p = .001;
PTSD and depression diagnostic groups, p = .000; and PTSD and nonclinical
control, p = .000. Again, significant differences were not observed between the
panic disorder and PTSD diagnostic groups, p = 1.000; depression and
nonclinical control groups, p = 1.000 or GAD and depression diagnostic groups,
p = .008. A profile plot illustrating the means of the 21-item ASI ‘Fear of
Cardiovascular/Stroke Symptoms’ dimension as a function of the four diagnostic
groups and nonclinical control group is provided in Figure 5.4. The results on
this factor reveal that individuals with a diagnosis of panic disorder and PTSD
Anxiety Sensitivity 161
appear to be significantly more fearful of anxiety symptoms that resemble
cardiovascular or stroke symptoms when compared to individuals with a
diagnosis of GAD, depression or individuals from the nonclinical control group.
Likewise, individuals with a diagnosis of GAD also appear to be significantly
more fearful of anxiety symptoms that resemble cardiovascular or stroke
symptoms when they are compared to individuals from the nonclinical control
group, whereas individuals from the depression group do not appear to be any
more concerned with cardiovascular/stroke symptoms when compared to
individuals from the nonclinical control group. Conversely, significant
differences were not observed between individuals with a diagnosis of panic
disorder and PTSD or individuals with a diagnosis of depression compared to
individuals from the nonclinical control group. It was interesting to note that
individuals with a diagnosis of GAD did not differ significantly from individuals
with a diagnosis of depression when a conservative alpha level was adopted.
Anxiety Sensitivity 162
0
2
4
6
8
10
12
14
Panic GAD PTSD Depression NonclinicalDiagnostic Category
Mea
n 21
-item
AS
I 'Fe
ar o
f Car
diov
ascu
lar/S
troke
S
ympt
oms'
Dim
ensi
on
Figure 5.4. Mean differences and standard error between panic disorder, GAD,
PTSD, depression, and nonclinical control groups and the 21-item ASI ‘Fear of
Cardiovascular/Stroke Symptoms’ dimension.
For the fourth first-order dimension, ‘ Fear of Cognitive Dyscontrol’
significant differences between the panic disorder and GAD diagnostic groups, p
= .000; panic disorder and depression diagnostic groups, p = .000; panic disorder
and nonclinical control groups, p = .000; GAD and PTSD diagnostic groups, p =
.000, GAD and depression diagnostic groups, p = .001; GAD and nonclinical
control groups, p = .000; PTSD and depression diagnostic groups, p = .000;
PTSD and nonclinical control groups, p = .000; and depression and nonclinical
control groups, p = .001 were observed. A significant difference was not
observed between the panic disorder and PTSD diagnostic groups, p = 1.000.
The results on this factor reveal that individuals with a diagnosis of panic
disorder appear to be significantly more fearful of phrenophobia or fear of
cognitive dyscontrol when compared to individuals with a diagnosis of GAD,
depression, or individuals from the nonclinical control group. Not surprisingly,
Anxiety Sensitivity 163
individuals with a diagnosis of PTSD also responded in a manner similar to that
observed in individuals with a diagnosis of panic disorder, whilst individuals
with a diagnosis of GAD appeared to be significantly more fearful of cognitive
dyscontrol when compared to individuals with a diagnosis of depression or
individuals from the nonclinical control group. Interestingly, individuals with a
diagnosis of depression also appeared to be significantly more fearful of loss of
cognitive control than individuals from the nonclinical control group. Significant
differences were not observed between individuals with a diagnosis of panic
disorder and PTSD. A profile plot illustrating the means of the 21-item ASI
‘Fear of Cognitive Dyscontrol’ dimension as a function of the four diagnostic
groups and the nonclinical controls’ is provided in Figure 5.5.
0
2
4
6
8
10
12
14
16
18
Panic GAD PTSD Depression NonclinicalDiagnostic Category
Mea
n 21
-item
AS
I 'Fe
ar o
f Cog
nitiv
e D
ysco
ntro
l' D
imen
sion
Figure 5.5. Mean differences and standard error between Panic Disorder, GAD,
PTSD, Depression, and nonclinical control groups and the 21-item ‘Fear of
Cognitive Dyscontrol’ dimension.
Anxiety Sensitivity 164
5.3.3.3. Analysis of differences for the panic disorder diagnostic group
on the first-order dimensions of anxiety sensitivity, as assessed by the 21-item
ASI. In order to determine whether there were significant differences between
the four first-order dimensions of the 21-item ASI for individuals with a
diagnosis of panic disorder, a within-groups multivariate analysis of variance
(MANOVA) with post-hoc tests was conducted using the ‘Fear of Respiratory
Symptoms’, ‘Fear of Publicly Observable Anxiety Symptoms’, ‘Fear of
Cardiovascular/Stroke Symptoms’ and ‘Fear of Cognitive Dyscontrol’
dimensions as the within-subject variable.
Using Wilks’ criterion, a significant multivariate effect was observed
within the combination of the four first-order dimensions of the 21-item ASI, F
(3, 73) = 59.285, p = .000. The result reflected a moderate to large association
within the dimensions and the panic disorder group, η² = .709. The observed
power was strong, 1.000.
To control for Type I error, an alpha correction of p = .0125 was set for
follow-up univariate tests. Significant differences were observed between the
first-order ‘Fear of Respiratory Symptoms’ and the ‘Fear of Publicly Observable
Symptoms’ dimensions, p = .000 and ‘Fear of Cognitive Dyscontrol’
dimensions, p = .000; the first-order ‘Fear of Publicly Observable Symptoms’
and the ‘Fear of Cardiovascular/Stroke Symptoms’ dimensions, p = .000; and the
first-order ‘Fear of Cardiovascular/Stroke Symptoms’ and the ‘Fear of Cognitive
Dyscontrol’ dimensions, p = .000. Significant differences were not observed,
however, between the first-order ‘Fear of Respiratory Symptoms’ and the ‘Fear
of Cardiovascular/Stroke Symptoms’ dimensions, p = .316 as well as the first-
order ‘Fear of Publicly Observable Symptoms’ and the ‘Fear of Cognitive
Anxiety Sensitivity 165
Dyscontrol’ dimensions, p = .826. A profile plot illustrating the means of the
21-item ASI first-order dimensions for the panic disorder group is provided in
Figure 5.6. It can be seen that individuals with a diagnosis of panic disorder
appear to be significantly more fearful of anxiety symptoms that are observable
by others as well as symptoms associated with cognitive dyscontrol when
compared to a fear of symptoms relating to either respiratory or
cardiovascular/stroke. Conversely, there is no difference between fear of
respiratory and cardiovascular/stroke symptoms for this group.
0
2
4
6
8
10
12
14
16
18
20
Fear of RespiratorySymptoms
Fear of Publicly ObservableSymptoms
Fear ofCardiovascular/Stroke
Symptoms
Fear of CognitiveDyscontrol
21-item ASI first-order Dimensions
Mea
n sc
ores
for t
he P
anic
Gro
up o
n th
e 4
first
-ord
er
dim
ensi
ons
of th
e 21
-item
AS
I
Figure 5.6. Mean differences and standard error for the panic disorder group and
the four first-order dimensions of the 21-item ASI.
5.3.3.4 Analysis of differences for the GAD diagnostic group on the first-
order dimensions of anxiety sensitivity, as assessed by the 21-item ASI. In order
to determine whether there were significant differences between the four first-
order dimensions of the 21-item ASI for individuals with a diagnosis of GAD, a
Anxiety Sensitivity 166
within-groups MANOVA with post-hoc tests was conducted using the ‘Fear of
Respiratory Symptoms’, ‘Fear of Publicly Observable Anxiety Symptoms’, ‘Fear
of Cardiovascular/Stroke Symptoms’ and ‘Fear of Cognitive Dyscontrol’
dimensions as the within-subject variable.
Using Wilks’ criterion, a significant multivariate effect was observed
within the combination of the four first-order dimensions of the 21-item ASI, F
(3, 43) = 27.812, p = .000. The result reflected a moderate association within the
dimensions and the GAD group, η² = .660. The observed power was strong,
1.000.
An alpha correction of p = .0125 was set for follow-up univariate tests.
Significant differences were observed between the first-order ‘Fear of
Respiratory Symptoms’ and the ‘Fear of Publicly Observable Symptoms’
dimensions, p = .000; the first-order ‘Fear of Publicly Observable Symptoms’
and the ‘Fear of Cardiovascular/Stroke Symptoms’ dimensions, p = .000 and
‘Fear of Cognitive Dyscontrol’ dimensions, p = .001; and the first-order ‘Fear of
Cardiovascular/Stroke Symptoms’ and the ‘Fear of Cognitive Dyscontrol’
dimensions, p = .000. Significant differences were not observed between the
first-order ‘Fear of Respiratory Symptoms’ and the ‘Fear of Cardiovascular/
Stroke Symptoms’ dimensions, p = 1.000. It is interesting to note that the
significance level was approached between the first-order ‘Fear of Respiratory
Symptoms’ and the ‘Fear of Cognitive Dyscontrol’ dimensions, p = .043.
However, if a less stringent alpha level had been adopted, the differences
between these dimensions would be considered significant.
Thus, the results for this diagnostic group reveal that individuals with a
diagnosis of GAD appear to be significantly more fearful of anxiety symptoms
Anxiety Sensitivity 167
that are observable by others compared to a fear of either the respiratory,
cardiovascular/stroke or cognitive dyscontrol symptom dimensions. However it
was also important to note that individuals with a diagnosis of GAD also report
being more fearful of the symptoms associated with cognitive dyscontrol when
compared to their fear of cardiovascular/stroke symptoms, and if a conservative
alpha level had not been adopted, a fear of respiratory symptoms as well.
Conversely, there is no difference between fear of respiratory and
cardiovascular/stroke symptoms for this group. A profile plot illustrating the
means of the 21-item ASI first-order dimensions for the GAD group is provided
in Figure 5.7.
0
2
4
6
8
10
12
14
Fear of RespiratorySymptoms
Fear of Publicly ObservableSymptoms
Fear ofCardiovascular/Stroke
Symptoms
Fear of CognitiveDyscontrol
21-item ASI first-order Dimensions
Mea
n sc
ores
for t
he G
AD
gro
up o
n th
e 4
first
-ord
er
dim
ensi
ons
of th
e 21
-item
AS
I
Figure 5.7. Mean differences and standard error for the GAD group and the four
first-order dimensions of the 21-item ASI.
5.3.3.5 Analysis of differences for the PTSD diagnostic group on the first-
order dimensions of anxiety sensitivity, as assessed by the 21-item ASI. In order
to determine whether there were significant differences between the four first-
Anxiety Sensitivity 168
order dimensions of the 21-item ASI for individuals with a diagnosis of PTSD, a
MANOVA with post-hoc tests was conducted using the ‘Fear of Respiratory
Symptoms’, ‘Fear of Publicly Observable Anxiety Symptoms’, ‘Fear of
Cardiovascular/Stroke Symptoms’ and ‘Fear of Cognitive Dyscontrol’
dimensions as the within-subject variable.
Using Wilks’ criterion, a significant multivariate effect was observed
within the combination of the four first-order dimensions of the 21-item ASI, F
(3, 18) = 8.654, p = .001. The result reflected a moderate association within the
dimensions and the PTSD group, η² = .591. The observed power was strong,
.979.
To control for Type I error, an alpha correction of p = .0125 was set for
follow-up univariate tests. Significant differences were observed between the
first-order ‘Fear of Respiratory Symptoms’ and the ‘Fear of Publicly Observable
Symptoms’ dimensions, p = .000 and ‘Fear of Cognitive Dyscontrol’
dimensions, p = .006; and the first-order ‘Fear of Publicly Observable
Symptoms’ and the ‘Fear of Cardiovascular/Stroke Symptoms’ dimensions, p =
.005. Significant differences were not observed, however, between the first-
order ‘Fear of Respiratory Symptoms’ and the ‘Fear of Cardiovascular/Stroke
Symptoms’ dimensions, p = .845; the first-order ‘Fear of Publicly Observable
Symptoms’ and the ‘Fear of Cognitive Dyscontrol’ dimensions, p = .718; or
between the first-order ‘Fear of Cardiovascular/Stroke Symptoms’ and ‘Fear of
Cognitive Dyscontrol’ Dimensions, p = .057. A profile plot illustrating the
means of the 21-item ASI first-order dimensions for the PTSD group is provided
in Figure 5.8. It can be seen that a significant difference does not exist for
individuals with a diagnosis of PTSD between fear of anxiety symptoms that are
Anxiety Sensitivity 169
observable by others and symptoms associated with cognitive dyscontrol, or
symptoms associated with a fear of respiratory and cardiovascular/stroke.
However, it is important to note that while there was no difference between
fearful symptoms associated with cardiovascular/stroke and cognitive dyscontrol,
this group did report to be significantly more fearful of anxiety symptoms that
are observable by others when compared to symptoms associated with a fear of
cardiovascular/stroke.
0
2
4
6
8
10
12
14
16
18
20
Fear of RespiratorySymptoms
Fear of Publicly ObservableSymptoms
Fear ofCardiovascular/Stroke
Symptoms
Fear of CognitiveDyscontrol
21-item ASI first-order Dimensions
Mea
n sc
ores
for t
he P
TSD
gro
up o
n th
e 4
first
-ord
er
dim
ensi
ons
of th
e 21
-item
AS
I
Figure 5.8. Mean differences and standard error for the PTSD group and the
four first-order dimensions of the 21-item ASI.
5.3.3.6 Analysis of differences for the depression diagnostic group on the
first-order dimensions of anxiety sensitivity, as assessed by the 21-item ASI. In
order to determine whether there were significant differences between the four
first-order dimensions of the 21-item ASI for individuals with a diagnosis of
depression, a within-groups MANOVA with post-hoc tests was conducted using
Anxiety Sensitivity 170
the ‘Fear of Respiratory Symptoms’, ‘Fear of Publicly Observable Anxiety
Symptoms’, ‘Fear of Cardiovascular/Stroke Symptoms’ and ‘Fear of Cognitive
Dyscontrol’ dimensions as the within-subject variable.
Using Wilks’ criterion, a significant multivariate effect was observed
within the combination of the four first-order dimensions of the 21-item ASI, F
(3, 54) = 30.468, p = .000. The result reflected a moderate association within the
dimensions and the depression group, η² = .629. The observed power was
strong, 1.000.
An alpha correction of p = .0125 was set for follow-up univariate tests.
Significant differences were observed between the first-order ‘Fear of
Respiratory Symptoms’ and the ‘Fear of Publicly Observable Symptoms’
dimensions, p = .000; the first-order ‘Fear of Publicly Observable Symptoms’
and the ‘Fear of Cardiovascular/Stroke Symptoms’ dimensions, p = .000 and
‘Fear of Cognitive Dyscontrol’ dimensions, p = .000; and the first-order ‘Fear of
Cardiovascular/Stroke Symptoms’ and the ‘Fear of Cognitive Dyscontrol’
dimensions, p = .000. Again, it is interesting to note that the significance level
was approached between the first-order ‘Fear of Respiratory Symptoms’ and the
‘Fear of Cardiovascular/Stroke Symptoms’ dimension, p = .013 and the ‘Fear of
Cognitive Dyscontrol’ dimension, p = .014. However, if a less stringent alpha
level had been adopted, the differences between these dimensions would be
considered significant. Thus, it can be seen that individuals from the depression
group are significantly more fearful of anxiety symptoms that are observable by
others when compared to a fear of either the respiratory, cardiovascular/stroke or
cognitive dyscontrol dimensions. It is also of interest to note that while this
groups fear of cognitive dyscontrol was greater than their fear of
Anxiety Sensitivity 171
cardiovascular/stroke symptoms, differences between their fear of respiratory and
cardiovascular/stroke symptoms as well as respiratory and cognitive dyscontrol
symptoms would have also been significant if a stringent alpha level has not been
implemented. A profile plot illustrating the means of the 21-item ASI first-order
dimensions for the depression group is provided in Figure 5.9.
0
2
4
6
8
10
12
Fear of RespiratorySymptoms
Fear of Publicly ObservableSymptoms
Fear ofCardiovascular/Stroke
Symptoms
Fear of CognitiveDyscontrol
21-item ASI first-order Dimensions
Mea
n sc
ores
for t
he D
epre
ssio
n gr
oup
on th
e 4
first
-ord
er
dim
ensi
ons
of th
e 21
-item
AS
I
Figure 5.9. Mean differences and standard error for the depression group and the
four first-order dimensions of the 21-item ASI.
5.3.3.7 Analysis of differences for the nonclinical control group on the
first-order dimensions of anxiety sensitivity, as assessed by the 21-item ASI. In
order to determine whether there were significant differences between the four
first-order dimensions of the 21-item ASI for individuals from the baseline,
nonclinical control group, a within-groups MANOVA with post-hoc tests was
conducted using the ‘Fear of Respiratory Symptoms’, ‘Fear of Publicly
Anxiety Sensitivity 172
Observable Anxiety Symptoms’, ‘Fear of Cardiovascular/Stroke Symptoms’ and
‘Fear of Cognitive Dyscontrol’ dimensions as the within-subject variable.
Using Wilks’ criterion, a significant multivariate effect was observed
within the combination of the four first-order dimensions of the 21-item ASI, F
(3, 420) = 290.118, p = .000. The result reflected a moderate association within
the dimensions and the nonclinical control group, η² = .675. The observed power
was strong, 1.000.
To control for Type I error, an alpha correction of p = .0125 was set for
follow-up univariate tests. Significant differences were observed between the
first-order ‘Fear of Respiratory Symptoms’ and ‘Fear of Publicly Observable
Symptoms’ dimensions, p = .000, ‘Fear of Cardiovascular/Stroke Symptoms’ p
= .000 and the ‘Fear of Cognitive Dyscontrol’ dimensions, p = .000; the first-
order ‘Fear of Publicly Observable Symptoms’ and the ‘Fear of
Cardiovascular/Stroke Symptoms’ dimensions, p = .000, and ‘Fear of Cognitive
Dyscontrol’ dimensions, p = .000. The significance level was approached
between the first-order ‘Fear of Cardiovascular/Stroke Symptoms’ and the ‘Fear
of Cognitive Dyscontrol’ dimensions, p = .036. However, if a less stringent
alpha level had been adopted, the differences between these dimensions would be
considered significant. Thus, it can be seen that, similar to the GAD and
depression groups, individuals from the nonclinical control group are
significantly more fearful of anxiety symptoms that are observable by others
when compared to a fear of either the respiratory, cardiovascular/stroke or
cognitive dyscontrol. It is also of interest to note that while this groups fear of
respiratory symptoms was greater than their fear of cardiovascular/stroke or
cognitive dyscontrol symptoms, the difference between their fear of cognitive
Anxiety Sensitivity 173
dyscontrol and cardiovascular/stroke symptoms would have also been significant
if a stringent alpha level had not been adopted. A profile plot illustrating the
means of the 21-item ASI first-order dimensions for the nonclinical group is
provided in Figure 5.10.
0
1
2
3
4
5
6
7
8
9
10
Fear of RespiratorySymptoms
Fear of Publicly ObservableSymptoms
Fear ofCardiovascular/Stroke
Symptoms
Fear of CognitiveDyscontrol
21-item ASI first-order Dimensions
Mea
n sc
ores
for t
he N
oncl
inic
al g
roup
on
the
4 fir
st-o
rder
dim
ensi
ons
of th
e 21
-item
AS
I
Figure 5.10. Mean differences and standard error for the nonclinical
control group and the four first-order dimensions of the 21-item ASI.
5.4 Discussion
The purpose of the current investigation was to examine whether
differences in the overall and specific dimensions of anxiety sensitivity existed
between as well as within the four diagnostic categories and a baseline
nonclinical control group, using the 21-item ASI, in order to determine the
construct, criterion and discriminant validity of the scale. Examination of
responses from diagnostic groups comprised of individuals who met the DSM-IV
Anxiety Sensitivity 174
criteria for panic disorder, GAD, PTSD or depression as well as a single,
baseline, nonclinical control group found not only significant differences but
similarities as well between as well as within all the groups at different levels.
The findings relating to the differences between the groups on the 21-item ASI
will be discussed first, followed by the findings relating to the differences within
each group.
Specifically, significant differences were not found between the panic
disorder and PTSD groups on any of the anxiety sensitivity dimensions. In most
cases, both the panic disorder and PTSD groups were associated with the highest
level of anxiety sensitivity on all dimensions when compared to the other
diagnostic and nonclinical control groups. The exception to this was on the ‘Fear
of Cardiovascular/Stroke Symptoms’ dimension, where the conservative alpha
level resulted in a non-significant difference between the panic disorder and
PTSD groups.
These results are consistent with Taylor et al’s. (1992) investigation
which found that, with the exception of individuals with PTSD, individuals with
panic disorder score significantly higher overall ASI when compared to
individuals with another anxiety disorder. As elevated levels of anxiety
sensitivity are more likely to lead to a vicious cycle of anxiety amplification that
has the potential to culminate in a panic attack, it is possible that individuals with
panic disorder or PTSD experience more intense psychological distress and
physiological reactivity than individuals with a diagnosis of GAD or depression
and individuals from the general population. This process has the potential to
assist in creating not only a fear of anxiety related symptoms but also elevated
anxiety sensitivity when compared to other diagnostic groups and is consistent
Anxiety Sensitivity 175
with Expectancy Theory. It provides evidence of criterion validity for the 21-
item ASI. Additionally, it is maintained that significant differences were not
observed between the two diagnostic groups because both panic disorder and
PTSD are sensitive to anxiety symptoms given the phenomenological similarity
between panic attacks and flashbacks. Davidson and Foa (1991) noted that both
panic attacks and flashbacks are transitory, acute events activated by internal or
external fear stimuli and share similar symptomatology, while Mellman and
Davis (1985) have argued that the symptoms associated with flashbacks meet the
DSM-III criteria for panic attacks. Thus flashbacks have an elemental course
similar to that observed in panic attacks, with both panic disorder and PTSD
possessing high arousal as an underlying common denominator.
Elevated levels of anxiety sensitivity were also observed in the GAD
diagnostic group, with some similarities and differences being observed between
all the groups at different levels. Specifically, individuals with a diagnosis of
GAD reported significantly greater anxiety sensitivity at the overall level as well
as on the dimension relating to a fear of cognitive dyscontrol, than individuals
who met the diagnostic criteria for a depressive disorder or individuals from the
nonclinical control group. This result is similar to the results of Taylor et al.’s
(1992) and Sandin et al.’s (1992) findings who also reported that individuals who
met the diagnostic criteria for GAD displayed elevated levels of anxiety
sensitivity generally when compared to a baseline, nonclinical control group.
However, it is a new finding that this diagnostic group also report greater
concern regarding a ‘Fear of Cognitive Dyscontrol’ when compared to either a
baseline nonclinical control group or individuals who met the diagnostic criteria
for depression. It is proposed that as GAD is distinguished by excessive worry
Anxiety Sensitivity 176
accompanied by symptoms relating to concentration difficulties or mind going
blank (APA, 2000); this investigation revealed that individuals tend to be fearful
of anxiety symptoms that are characterised as resulting in cognitive
incapacitation. This new finding provides further evidence of construct validity
of the 21-item ASI.
Individuals who met the diagnostic criteria for GAD also reported
significantly greater concern relating to a fear of both the publicly observable and
cardiovascular/stroke symptoms of an anxious episode than individuals from the
general community. There was no significant difference between the GAD and
depression groups on these two anxiety sensitivity dimensions, particularly when
a conservative alpha level was employed. This is a new finding when examining
differences in specific dimensions of anxiety sensitivity between individuals with
GAD and a baseline nonclinical control group and future investigations are
clearly warranted. Further, individuals with a diagnosis of GAD failed to report
greater concern regarding a fear of respiratory symptoms of anxiety when
compared to individuals with a diagnosis of depression as well as individuals
from the general community. Therefore, GAD was comparable to the baseline
nonclinical group for this anxiety sensitivity dimension. It is possible that as
GAD is characterised as involving increased cognitive symptoms of anxiety and
while it is associated with some somatic sensations such as sweating, nausea,
diarrhoea, trembling, twitching, feelings of shakiness, as well as muscular aches,
soreness and exaggerated startle response (APA, 2000), it is important to note
that respiratory symptoms, especially a fear of such symptoms, are not
characteristic of the disorder. As such, it is proposed that individuals with GAD
Anxiety Sensitivity 177
do not experience these sensations in the first place and therefore it follows that
they would be comparable to the nonclinical control group on this dimension.
Examination of individuals who met the diagnostic criteria for a
depressive disorder failed to demonstrate elevated levels of anxiety sensitivity
both generally as well as on the specific dimensions relating to a fear of publicly
observable and cardiovascular/stroke symptoms when compared to the baseline
nonclinical control group. It was also of interest to note that individuals from the
general community reported greater concern regarding a fear of respiratory
symptoms than did individuals from the depression group. It is evident why
individuals who met the diagnostic criteria for depression failed to report
elevated levels of anxiety sensitivity on these dimensions. Examination of the
DSM-IV-TR (APA, 2000) diagnostic criteria for major depressive disorder and
dysthymic disorder reveals that both are characterised by symptoms not
associated with fears of anxiety related sensations generally or fears associated
with anxiety related respiratory, cardiovascular/stroke symptoms and social
embarrassment. Thus, individuals in the depression diagnostic group did not
report elevated fear on these 21-item ASI dimensions because they are not
characteristic of the underlying disorder. It is contended that this is further
evidence of the construct validity on the 21-item ASI.
Conversely, individuals who met the diagnostic criteria for a depressive
disorder reported greater concern regarding a fear of cognitive dyscontrol or
phrenophobia than individuals who comprised the baseline nonclinical control
group. Previous investigations (see Cox et al., 2001; Taylor et al., 1996; Schmidt
et al., 1998) have demonstrated a link between depression and ‘Fear of Cognitive
Dyscontrol’; however it is not clear why this would occur. Taylor et al. (1996)
Anxiety Sensitivity 178
have argued that phrenophobia is a depression-specific form of anxiety
sensitivity, unrelated to anxiety and panic, while Schmidt et al. (1998) argue that
phrenophobia is related to both anxiety and depression. They argue that the
cognitive symptoms of phrenophobia such as impairment in concentration,
depersonalisation and derealisation are conceptually consistent with anxiety
sensitivity and have the potential to lead to specific fears (Schmidt et al., 1998).
Thus, individuals who meet the diagnostic criteria for depression report elevated
fear regarding symptoms that are characteristic of cognitive dyscontrol because
research (see Taylor et al., 1996) has shown that individuals who are clinically
depressed commonly report slowed thinking, concentration difficulties and
distractibility as part of the symptoms of the disorder.
In the examination of differences on the 21-item ASI dimensions within
each diagnostic classification as well as the nonclinical control group, one of the
most important findings concerned the 21-item ASI dimension ‘Fear of Publicly
Observable Symptoms’. Regardless of the category under examination, it was
observed that all four diagnostic categories as well as the nonclinical control
group indicated that they were significantly more fearful of anxiety symptoms
that are observable by others when compared to the other anxiety sensitivity
dimensions. As the theoretical expectations for each diagnostic group included
in the present investigation did not indicate that this dimension would be
significantly elevated compared to the other dimensions of the 21-item ASI, it is
important to critically examine why this finding might have occurred. One
explanation concerns the notion that this dimension of the 21-item ASI may be
measuring a related latent construct, namely fear of negative evaluation.
Anxiety Sensitivity 179
This argument is not unique to the expanded 36-item ASI-R or the
current, revised 21-item ASI. Taylor (1995) conjectured that the ‘Social
Concerns’ dimension of the original 16-item ASI might be more correctly
conceptualised as being a part of the domain of negative evaluation sensitivity
(NES) rather than the domain of anxiety sensitivity. In contrast, Zinbarg et al.
(1999) have argued that the ‘Social Concerns’ dimension was theoretically
distinct from NES. They contended that NES is differentiated as a fear of
negative evaluation that occurs from a range of behaviours that are not limited to
displays of anxiety symptoms, whereas the ‘Social Concerns’ domain of the 16-
item ASI is characterised as concerns regarding negative evaluation resulting
from displays of publicly observable anxiety symptoms. As a result of these
claims, Taylor, Rabain, and Fedoroff (1999) argued that the ‘Social Concerns’
dimension of the original ASI could in fact be a blend of two fundamental
sensitivities; anxiety sensitivity and NES. They have posited that a publicly
observable symptom of anxiety, such as trembling in the presence of others,
could be feared either because of the negative evaluation that could result from
such a symptom (NES) and/or because of the symptom is a sensation related to
anxiety (anxiety sensitivity) (Taylor et al., 1999).
In an investigation that was concerned with whether the social concerns
component of the ASI belonged to the domain of anxiety sensitivity or the
domain of NES, McWilliams, Stewart, and MacPherson (2000) found that while
measures of anxiety sensitivity and NES possess a common element at the
general or global level, their observations suggested that the ‘Social Concerns’
dimension of the 16-item ASI was distinct from NES as well as the other first-
order dimensions of the original ASI. Thus, while it is not intrinsically clear as
Anxiety Sensitivity 180
to why this finding occurred in the current investigation, it is speculated that the
‘Fear of Publicly Observable Symptoms’ dimension of the 21-item ASI is a
blend of both anxiety sensitivity and NES. However, this possible explanation
requires further validation and empirical support before it can be asserted.
Once the ‘Fear of Publicly Observable Symptoms’ dimension was taken
into consideration, elevated scores on the dimension ‘Fear of Cognitive
Dyscontrol’ were observed in the panic disorder group, relative to the scores
observed on dimensions relating to ‘Fear of Respiratory Symptoms’ and ‘Fear of
Cardiovascular/Stroke Symptoms’. This finding is in direct contrast to the
hypothesis for the panic disorder group. It was hypothesised that there would not
be a significant difference between the dimensions relating to the respiratory,
cardiovascular/stroke and cognitive dyscontrol dimensions of anxiety sensitivity
when compared to the fear of publicly observable dimension. As such it is not
clear as to why this finding occurred. One possible explanation concerns the
hierarchy of anxiety sensitivity for this diagnostic group. As panic disorder is
characterised by the DSM-IV-TR as the experience of repeated panic attacks as
well as a number of somatic and cognitive symptoms (APA, 2000), it is possible
that individuals in this diagnostic category regard symptoms associated with
mental incapacitation or phrenophobia as significantly more fearful than
symptoms associated with respiratory or cardiovascular/stroke events.
An alternative hypothesis is that the significant difference observed
between the ‘Fear of Cognitive Dyscontrol’ dimension and the ‘Fear of
Respiratory’ and ‘Fear of Cardiovascular/Stroke Symptoms’ dimensions may be
the direct result of the effects of the medication that individuals from this
diagnostic category were prescribed. As all participants from the diagnostic
Anxiety Sensitivity 181
categories were in receipt of medication for either an anxiety or depressive
disorder or both, it is possible that the medications had the effect of dulling or
reducing the experience of physiological arousal including symptoms associated
with respiratory or cardiovascular events when compared to incapacitating
cognitive symptoms. However, it is important to note that both possible
explanations require further validation and empirical support.
A similar finding to the panic disorder group was also observed in the
GAD and depression groups, with scores on the ‘Fear of Cognitive Dyscontrol’
dimension being significantly elevated relative to scores on the ‘Fear of
Cardiovascular/Stroke Dimension’. However, the pattern was different from
those in the panic disorder category in that responses on the dimension ‘Fear of
Respiratory Symptoms’ approached significance, particularly when a
conservative alpha level was employed, compared to scores on the ‘Fear of
Cognitive Dyscontrol’ dimension. This finding partially supports the hypothesis
for both the GAD and depression groups which stipulated that scores on the
cognitive dyscontrol dimension of anxiety sensitivity would be significantly
greater than scores observed on the respiratory, publicly observable or
cardiovascular/stroke dimensions. For instance, according to the DSM-IV-TR
(APA, 2000), GAD is distinguished by excessive worry and accompanied by
symptoms relating to concentration difficulties or mind going blank and less so
by symptoms of autonomic arousal. Similarly, as many individuals experiencing
a major depressive episode report an impaired capability to think, concentrate or
make decisions and have a tendency to report memory problems that are similar
to those observed in individuals with an anxiety disorder (APA, 2000), it is
proposed that this is why both the GAD and depression groups are more fearful
Anxiety Sensitivity 182
of anxiety symptoms that are characterised as resulting in cognitive
incapacitation than a respiratory or cardiovascular/stroke event, once the fear of
publicly observable symptoms has been taken into account. This is a new
finding in the literature and provides greater evidence of construct and
discriminant validity of the 21-item ASI.
Conversely to the GAD and depression groups, the reverse was observed
in the PTSD group. It was found that scores for this diagnostic category were
significantly elevated on the ‘Fear of Cognitive Dyscontrol’ dimension when
compared to scores on the ‘Fear of Respiratory Symptoms’ dimension and not on
the ‘Fear of Cardiovascular/Stroke Symptoms’ dimension. This finding has only
provided partial support for the PTSD groups’ hypothesis which predicted there
would not be a significant difference between the dimensions relating to the
respiratory, cardiovascular/stroke and cognitive dyscontrol dimensions of anxiety
sensitivity when compared to the fear of publicly observable dimension. Similar
to the panic disorder group, it is not inherently clear as to why this finding
occurred. Like panic disorder, the DSM-IV-TR (APA, 2000) characterises
PTSD as involving a combination of both physical and cognitive symptoms.
Therefore, while it is possible that individuals in this diagnostic category regard
symptoms associated with mental incapacitation significantly more fearful than
symptoms associated with respiratory events, it is not known why there was no
difference in the reporting of fear of cardiovascular/stroke symptoms. Clearly,
further investigations into this finding are warranted.
While there was no difference in scores between the ‘Fear of Cognitive
Dyscontrol’ and ‘Fear of Cardiovascular/Stroke Symptoms’ dimensions for the
nonclinical control group, a significant difference was observed between scores
Anxiety Sensitivity 183
on the ‘Fear of Respiratory Symptoms’ and ‘Fear of Cognitive Dyscontrol’
dimensions, with individuals from the nonclinical control group reporting greater
concern regarding fear of respiratory symptoms than cognitive dyscontrol
symptoms. This finding is surprising given that for all the diagnostic group, a
fear of mental incapacitation or cognitive dyscontrol was greater than or at least
equal to a fear of respiratory symptoms.
It is constructive to note that with the exception of the nonclinical control
group, a significant difference was not observed between the ‘Fear of Respiratory
Symptoms’ and ‘Fear of Cardiovascular/Stroke Symptoms’ for any of the
diagnostic groups. While it would appear on the surface that the diagnostic
groups do not differentiate between these two separate physiological dimensions,
it is important note that the previous CFA and the discriminate validity
investigations contained in the current dissertation found support for the claim
that they are independent dimensions. It is necessary to keep in mind that the
sample size in the present investigation has been significantly reduced compared
to the sample size utilised in the CFA investigation because the participants from
the Anxiety Disorders Clinic at the Wesley Hospital could not be included as
inter-rater reliability of diagnoses could not be established. Thus, while these
dimensions are similar in that they both measure a fear of physical concerns, it is
possible that the differences in the degree of specificity between these two
dimensions would increase as a result of an increase in sample size.
5.5 Chapter Summary
The results of the current investigation have revealed that the differences
between the diagnostic groups were consistent with theoretical expectations.
Specifically, significant differences were not found between the panic disorder
Anxiety Sensitivity 184
and PTSD groups on any of the anxiety sensitivity dimensions. Elevated levels
of anxiety sensitivity were observed in the GAD diagnostic group, with some
similarities and differences being observed between all the groups at different
levels. Individuals who met the diagnostic criteria for a depressive disorder
reported greater concern regarding a ‘Fear of Cognitive Dyscontrol’ or
phrenophobia than individuals who comprised the baseline nonclinical control
group. As such, there is sufficient evidence to determine that 21-item ASI
possesses not only construct validity but convergent and discriminant validity as
well.
However, the findings obtained regarding differences within each groups
clearly warrant further investigation. While it is acknowledged that the results
obtained are preliminary, it is proposed that there is sufficient evidence to reveal
that the cognitive pattern of anxiety sensitivity is different within the various
psychopathologies investigated. As such, the results obtained provide additional
support for multidimensional view of anxiety sensitivity and further evidence of
the construct validity of the 21-item ASI. Further, the current investigation has
not only increased our understanding of the aetiology of different forms of
anxiety and depressive psychopathology, but it is proposed that a valid,
multidimensional measure could be helpful in ascertaining whether there a
particular anxiety sensitivity dimension is more germane to a specific
psychopathology. Thus, a multidimensional measure such as the 21-item ASI
may stimulate further research into these areas.
It is important to note that one limitation of the current investigation
concerns the fact that there was some comorbidity among the diagnostic groups.
It is contended that this overlap existed because the participants who took part in
Anxiety Sensitivity 185
the current investigation were attending the clinics for the purposes of therapeutic
intervention rather than for the purposes of research. As such, the participants
who were selected to take part in the current investigation were from a
population that one would expect to see in an effectiveness setting, rather than an
efficacy setting and selected because they meet a ‘pure’ or singular anxiety or
depression diagnosis. Whilst comorbidity amongst anxiety and depressive
disorders has been well documented (see Australian Bureau of Statistics, 1998;
and Brown & Barlow, 1992 for review) it is important to note that these
disorders are also comprised of unique features. Therefore, it is proposed that it
is still applicable to view the differences among these disorders using the
construct of anxiety sensitivity. As the results of the current investigation are
valid and show important differences between the diagnostic groups when using
the 21-item ASI, the addition of comorbidity in the present investigation does not
alter the main findings. Clearly, one way to refine the current investigation
would be to include only those participants who meet with a ‘pure’ or singular
anxiety or depression diagnosis in order to determine if the 21-item ASI
discriminated between and within the diagnostic groups in a theoretically
expected manner.
A further limitation was that there were not enough participants who met
the diagnostic criteria for social phobia, OCD or specific phobia to include in the
present investigation. The inclusion of these three diagnostic groups could have
led to a richer understanding of how anxiety sensitivity varies across and within
all of the anxiety disorders and this warrants further investigation.
Previous investigations of the original ASI have demonstrated that
university students and individuals from the general community usually score
Anxiety Sensitivity 186
one to two standard deviations lower when compared to clinical populations (see
Cox et al., 1991; Lilienfeld, 1997; Peterson & Heilbronner, 1987; Reiss et al.,
1986; Stewart et al., 1997; Telch et al., 1989). While this previous finding was
upheld for the panic disorder, GAD and PTSD groups, the present investigation
observed that the mean scores between the depression and nonclinical control
groups, at the general or overall level, were comparable. As previous
investigations (Otto et al., 1995; Taylor et al., 1996) have found that individuals
with a diagnosis of major depression report elevated levels of anxiety sensitivity
when compared to a control group, it is not determined why the individuals who
comprised the depression group in the present investigation were comparable to
the baseline nonclinical control group. Clearly, further investigations are
warranted.
As a result of the current investigation, it is posited that there is further
evidence of discriminant validity for the total scale as well as the first-order
dimensions of the 21-item ASI between the different diagnostic groups examined
and the nonclinical control group. As the findings obtained the current
investigation reveal the significance of anxiety sensitivity in the different types
of anxiety and mood pathology, the construct is reflected in a more appropriate
manner through the use of a more refined scale than the original 16-item ASI or
expanded 36-item ASI-R. The final investigation is concerned with examining
the change within each diagnostic category after CBT in order to provide a
further test of validity for the revised 21-item ASI. It is proposed that it is
important to examine changes in 21-item ASI scores following CBT as anxiety
sensitivity is commonly employed as an outcome measure for the evaluation of
treatment programs.
Anxiety Sensitivity 187
Chapter Six: Examination of differences within diagnostic categories after
cognitive behavioural therapy using the 21-item Anxiety Sensitivity Index
6.1 Introduction 189
6.2 Method 191
6.2.1 Participants 191
6.2.2 Measures 194
6.2.3 Design 194
6.2.4 Treatment 194
6.3 Results 195
6.3.1 Data Screening and Cleaning 195
6.3.2 Reliability Analyses of Scale 197
6.3.3 Preliminary Analyses 197
6.3.4 Tests of Hypotheses 200
6.3.4.1 Analysis of differences for each diagnostic
group before and after CBT for overall
level of anxiety sensitivity 200
6.3.4.2 Analysis of differences for the panic
disorder group before and after CBT for
the first-order dimensions of anxiety
sensitivity, as assessed by the 21-item
ASI 201
6.3.4.3 Analysis of differences for the GAD group
before and after CBT for the first-order
dimensions of anxiety sensitivity, as
assessed by the 21-item ASI 203
Anxiety Sensitivity 188
6.3.4.4 Analysis of differences for the PTSD group
before and after CBT for the first-order
dimensions of anxiety sensitivity, as
assessed by the 21-item ASI 205
6.3.4.5 Analysis of differences for the depression
group before and after CBT for the first-
order dimensions of anxiety sensitivity,
as assessed by the 21-item ASI 206
6.3.4.6 Analysis of differences on the BAI,
CCQ-M, FQ and Zung – SDS after CBT 208
6.4 Discussion 209
6.5 Chapter Summary 213
Anxiety Sensitivity 189
CHAPTER SIX
EXAMINATION OF DIFFERENCES WITHIN DIAGNOSTIC CATEGORIES
AFTER COGNITIVE-BEHAVIOURAL THERAPY USING THE 21-ITEM
ANXIETY SENSITIVITY INDEX
6.1 Introduction
As a psychological construct, anxiety sensitivity has become associated
with understanding panic pathology in particular and is generally associated with
understanding negative emotional functioning found in other anxiety and mood
psychopathology. As seen in Section 5.4, the previous investigation found that
the differences between the diagnostic groups, using the 21-item ASI, were
consistent with theoretical expectations. While significant differences were not
found between the panic disorder and PTSD groups on any of the anxiety
sensitivity dimensions, elevated levels of anxiety sensitivity were observed in the
GAD diagnostic group, with some similarities and differences being observed
between all the groups at different levels. Conversely, individuals who met the
diagnostic criteria for a depressive disorder reported greater concern regarding a
fear of cognitive dyscontrol or phrenophobia than individuals who comprised the
baseline nonclinical control group. These results were similar to those published
previously (see Section 2.8.6 of the literature review). As such, it was argued
that the 21-item ASI possessed evidence of not only construct validity but also of
convergent and discriminant validity.
Similarly, the results from the previous investigation also found that the
differences within the diagnostic groups were, to some extent, consistent with
theoretical expectations. Regardless of the category under examination, all four
Anxiety Sensitivity 190
diagnostic categories as well as the nonclinical control group responded that they
were significantly more fearful of anxiety symptoms that are observable by
others when compared to the other anxiety sensitivity dimensions. Further, the
panic disorder group reported greater concern regarding cognitive dyscontrol
than respiratory and cardiovascular/ stroke symptoms. A similar pattern was
observed in both the PTSD, GAD and depression groups, while the nonclinical
group reported greater concern regarding fear of respiratory symptoms than
cognitive dyscontrol. It was argued that the results provided further evidence
that the 21-item ASI contains acceptable construct and convergent validity. The
final study is concerned with examining differences within each diagnostic
category before and after CBT in order to provide a further test of validity for the
revised 21-item ASI.
The goal of CBT for anxiety and anxiety related disorders is to uncover
the irrational and problematic thinking styles of the client whilst the behavioural
techniques follow the premise that maladaptive behaviours are learned and
therefore can be unlearned (Bourne, 1995). Taken together, the cognitive and
behavioural strategies create a balanced approach towards the understanding and
subsequent treatment of anxiety related disorders. As a result, CBT allows
clients to revisit situations and environments that were once avoided, armed with
positive cognitive processing and behavioural techniques.
The present investigation is concerned with the differences between the
pre and post-treatment scores on the 21-item ASI following CBT. As previous
studies (see Section 2.9 of the literature review) have revealed that CBT
interventions are capable of reducing individuals’ total ASI scores, it is proposed
that significant findings from this study will again demonstrate the important role
Anxiety Sensitivity 191
that anxiety sensitivity plays in various diagnostic classifications as well as
provide further evidence of validity for the measure. As such, it is hypothesised
that CBT will result in a significant decrease in total anxiety sensitivity scores
using the 21-item ASI, for each diagnostic group. However, what is not yet
known is whether CBT interventions are successful in reducing the first-order or
specific anxiety sensitivity dimensions. Examining individuals' scores both
before and after CBT provides potentially important information relating to the
construct validity of the 21-item ASI as well as anxiety sensitivity theory. Thus,
it is also hypothesised that CBT will result in a significant decrease in scores on
the four first-order dimensions of the 21-item ASI for individuals with a
diagnosis of panic disorder, GAD, PTSD, and depression. Finally, as it is
arguably important to examine whether CBT is capable of facilitating a reduction
in the cognitions, symptoms and fear of anxiety and depression globally, it was
hypothesised that CBT will result in a significant decrease in post-treatment
scores using the Beck Anxiety Inventory (BAI), Catastrophic Cognitions
Questionnaire - Modified (CCQ-M), Fear Questionnaire (FQ), and Zung – SDS.
6.2 Method
6.2.1 Participants
A total of 131 individuals participated in the current investigation, and
were classified according to the DSM-IV criteria for either Panic Disorder (n =
47), Generalised Anxiety Disorder (GAD; n = 30), Posttraumatic Stress Disorder
(n = 11); Major Depression or Dysthymic Disorder (Depression; n = 40), Social
Phobia (n = 2), or Other Anxiety Disorder (consisting of individuals who met the
diagnostic criteria for Obsessive Compulsive Disorder, Specific Phobia or
Anxiety Sensitivity 192
Hypochondrias, [n = 1]). Again, it is contended that it is acceptable to combine
individuals who meet the criteria for major depressive disorder and dysthymic
disorder into a group called ‘depression’ for the purposes of this study because
the two disorders share comparable symptoms that are not easy to differentiate.
Finally, all participants were recruited from the Cognitive-Behavioural Therapy
Unit at the Toowong Private Hospital. The mean age and range of each
diagnostic group is displayed in Table 6.1, while Table 6.2 presents the general
demographic data for each group.
Table 6.1.
Means, Standard Deviations and Age Range for Diagnostic Groups
Diagnostic
Group
N Mean Age
(in Years)
Standard Deviation Age Range
(in Years)
Panic Disorder 47 44.79 11.78 21-72
GAD 30 42.20 11.93 20-69
PTSD 11 47.36 10.35 25-57
Depression 40 45.00 11.28 19-67
Social Phobia 2 36.00 4.24 33-39
Other Anxiety 1 55.00 0.00 55-55
Anxiety Sensitivity 193
Table 6.2.
Sample size of Demographic Variables for the Diagnostic Groups
Panic
Disorder
GAD PTSD Depression Other
Anxiety
Social
Phobia
Demographic Categories
N (%) N (%) N (%) N (%) N (%) N (%)
Gender
Male
Female
8 (17.0)
39 (83.0)
10 (33.3)
20 (66.7)
8 (72.7)
3 (27.3)
16 (40.0)
24 (60.0)
1 (100)
0 (0.0)
1 (50.0)
1 (50.0)
Origin of Birth
Australia
Other than Australia
No Response
37 (78.7)
7 (14.9)
3 (6.4)
25 (83.3)
3 (10.0)
2 (6.7)
8 (72.2)
2 (18.2)
1 (9.1)
35 (87.5)
5 (12.5)
0 (0.0)
0 (0.0)
0 (0.0)
1 (100)
2 (100)
0 (0.0)
0 (0.0)
Language Regularly
Spoken at Home
English
Other than English
No Response
43 (91.5)
1 (2.1)
3 (6.4)
28 (93.3)
2 (6.7)
0 (0.0)
9 (81.8)
1 (9.1)
1 (9.1)
40 (100)
0 (0.0)
0 (0.0)
1 (100)
0 (0.0)
0 (0.0)
2 (100)
0 (0.0)
0 (0.0)
Education
High School
Certificate
TAFE Certificate
TAFE Diploma/
Ass. Diploma
University U/G
University P/G
Other
No Response
20 (42.6)
9 (19.1)
1 (2.1)
10 (21.3)
1 (2.1)
3 (6.4)
3 (6.4)
11 (36.7)
1 (3.3)
4 (13.3)
4 (13.3)
4 (13.3)
4 (13.3)
2 (6.7)
6 (54.5)
0 (0.0)
0 (0.0)
0 (0.0)
1 (9.1)
3 (27.3)
1 (9.1)
10 (25.0)
3 (7.5)
6 (15.0)
5 (12.5)
9 (22.5)
7 (17.5)
0 (0.0)
0 (0.0)
0 (0.0)
0 (0.0)
1 (100)
0 (0.0)
0 (0.0)
0 (0.0)
0 (0.0)
0 (0.0)
0 (0.0)
1 (50.0)
1 (50.0)
0 (0.0)
0 (0.0)
Note. Percentages are in parenthesis
Anxiety Sensitivity 194
6.2.2 Measures
The measures used in the present investigation have been described in
detail elsewhere. In brief they include the 21-item ASI, BAI, CCQ-M, FQ and
Zung – SDS (see Chapter 4 for details of the 21-item ASI and Section 3.3 of the
Methodology for other measures).
6.2.3 Design
The present investigation was conducted using a within-groups design at
two points of time. The participants were measured on the 21-item ASI both
before and after CBT in order to determine whether the revised scale was capable
of capturing any reduction for each diagnostic group’s fear of anxiety sensations
and belief that such sensations result in harmful consequences.
6.2.4 Treatment
The therapy (Oei, 1999) used in the present investigation for both the
anxiety and depression treatment groups have been described in detail elsewhere
(see Appendices E and F). In brief, the CBT Anxiety, Fear and Phobias and the
CBT Depression day treatment programs, are both manual driven and are
comprised of eight sessions each, held over two days per week. Each session
was approximately 5.5 hours, with follow-up sessions provided on a four-weekly
basis, as clinically required. The program components included mini-lectures,
group discussion, exercises, role plays, modelling, handouts and homework. For
the Anxiety program, the CBT emphasised informational interventions, cognitive
restructuring, breathing retraining, relaxation training, and in vivo situational and
interoceptive exposure. For the Depression program, the CBT emphasised
education, behavioural techniques, automatic thoughts identification, cognitive
restructuring, and problem solving and relapse prevention. Therapists for both
Anxiety Sensitivity 195
treatment programs were either a Professor of Clinical Psychology or a registered
psychological intern from the Clinical Psychology Masters or PhD programs at
the University of Queensland.
6.3 Results
6.3.1 Data Screening and Cleaning
Prior to analysis, responses on the 21-item ASI were examined using
SPSS Version 10 for errors in data entry, missing values, fit between
distributions and the assumptions of repeated measures ANOVA and MANOVA.
All variables were examined separately for the diagnostic groups. Again, it was
determined that the two diagnostic categories, social phobia and other anxiety,
did not contain a sample size large enough to be included in the analyses.
Therefore these two groups were excluded from any further consideration in the
current chapter and only the pre and post-treatment responses for the panic
disorder, GAD, PTSD and depression groups are described and analysed from
here on.
Assessment of pre and post-treatment items on all variables revealed a
small portion of the data was missing on the post data set only. As the pattern of
missing data was random, means were calculated by diagnostic category for each
item containing missing data and substituted for missing values (Tabachnick &
Fidell, 2001).
There were no cases identified as univariate outliers for the pre and post-
treatment 21-item ASI total scale items. However, one case from the depression
group was identified as a univariate outlier on the pre-treatment ‘Fear of
Respiratory Symptoms’ and ‘Fear of Publicly Observable Symptoms’ scales;
while one case from the GAD group was identified as a univariate outlier on the
Anxiety Sensitivity 196
post-treatment ‘Fear of Cardiovascular/Stroke Symptoms’ scale. These cases
were also identified as multivariate outliers by way of Mahalanobis distance with
p = .000. Examination of these cases separately revealed extremely high overall
scores on the 21-item ASI, when compared to other cases within the
representative group. In order to reduce the influence of Type I and Type II
errors, these cases were deleted from the data matrix because it was determined
that they were not a part of the population for which the present study was
intended (Tabachnick & Fidell, 2001). These case deletions resulted in a final
sample with 47 cases in the panic disorder group, 29 cases in the GAD group, 11
cases in the PTSD group and 39 cases in the depression group.
With the use of SPSS EXPLORE, all distributions were assessed for
normality by group which revealed moderate skewness on five subscales for the
Depression group (21-item ASI pre-treatment ‘Fear of Cardiovascular/Stroke
Symptoms’ and ‘Fear of Cognitive Dyscontrol’ scales, as well as the 21-item
ASI post-treatment ‘Fear of Respiratory Symptoms’, ‘Fear of
Cardiovascular/Stroke Symptoms’ and ‘Fear of Cognitive Dyscontrol’ scales).
Moderate skewness was also observed on the pre-treatment 21-item ASI total
scale for the PTSD group and on the post-treatment ‘Fear of
Cardiovascular/Stroke Symptoms’ scale for the GAD group. However,
Tabachnick and Fidell (2001) argue that multivariate repeated measures analysis
is robust to violations of normality in the same manner as MANOVA.
As repeated measures analyses are only to be conducted within groups
and not between groups, testing for the assumption underlying homogeneity of
variance-covariance was not warranted (Tabachnick & Fidell, 2001).
Examination of the correlation matrices revealed that all variables were
Anxiety Sensitivity 197
moderately correlated, but not to such an extent to make any one variable
redundant. Finally, tests for multicollinearity, singularity and linearity revealed
that these assumptions were met.
6.3.2 Reliability Analyses of Scale
Cronbach alphas for the 21-item ASI total and subscale scores have been
reported again as the clinical sample is smaller than has been reported elsewhere
(see Section 4.3.7). In addition, the Cronbach alphas for the 21-item ASI total
and subscale scores for the post-treatment condition are also reported. It can be
seen in both the pre and post-treatment conditions that the 21-item ASI remains
internally consistent, even with a reduced sample size (see Table 6.3).
Table 6.3
Internal Consistency of the 21-item ASI for the Clinical Group Pre and Post CBT
21-item ASI dimension
Pre-Treatment Clinical Group
(N = 126)
Post-Treatment Clinical Group
(N = 126)
Total ASI score
.94 .96
Fear of Respiratory Symptoms
.93 .92
Fear of Publicly Observable Symptoms
.90 .89
Fear of Cardiovascular/Stroke Symptoms
.92 .94
Fear of Cognitive Dyscontrol Symptoms
.95 .95
6.3.3 Preliminary Analyses In order to examine whether there were any significant differences
between those participants who completed the post-treatment questionnaire
Anxiety Sensitivity 198
battery and those who did not, responses were analyses using independent
samples t-test on the pre -treatment total scores for the completers (M = 37.37,
SD = 19.59) and non-completers (M = 42.60, SD = 21.18) on the 21-item ASI;
pre -treatment total scores for the completers (M = 22.55, SD = 12.96) and non-
completers (M = 23.87, SD = 13.51) on the BAI; pre -treatment total scores for
the completers (M = 57.52, SD = 17.21) and non-completers (M = 62.67, SD =
18.90) on the CCQ-M; pre -treatment total scores for the completers (M = 32.98,
SD = 22.82) and non-completers (M = 40.41, SD = 28.74) on the FQ; and pre -
treatment total scores for the completers (M = 52.32, SD = 9.58) and non-
completers (M = 53.03, SD = 8.64) on the Zung – SDS. Using a probability level
of .05 for all t-test analyses, significant differences were not observed between
the completers and non-completers on the 21-itm ASI, t (187) = -1.684, p = .094;
BAI, t (187) = -.653, p = .120; CCQ-M, t (185) = -1.871, p = .063; Fear
Questionnaire, t (99.471) = -1.756, p = .082; and Zung – SDS, t (178) = -.483, p
= .630. Thus, a possible confounding effect was not present between those who
completed the post-treatment questionnaire battery and those who did not.
Means and standard deviations of the pre and post-treatment 21-item ASI
total scale and first-order dimension scores for the four diagnostic groups are
reported in Table 6.4. It can be seen that the panic disorder group have higher
mean scores on most of 21-item ASI scales in both the pre and post-treatment
conditions, followed by the PTSD, GAD, and Depression groups respectively.
Anxiety Sensitivity 199
Table 6.4.
Means and Standard Deviations for the Diagnostic Groups before and after CBT Pre-
Treatment
Mean SD
Post-
Treatment
Mean SD
Panic Disorder (n = 47)
Total 21-item ASI scale
Respiratory Symptoms Dimension
Publicly Observable Symptoms Dimension
Cardiovascular/Stroke Symptoms Dimension
Cognitive Dyscontrol Symptoms Dimension
55.04 7.87
9.72 4.06
17.87 3.93
10.87 5.81
16.57 5.61
37.98 18.37
6.60 4.32
13.04 6.01
7.53 5.48
10.81 7.33
GAD (n = 29)
Total 21-item ASI scale
Respiratory Symptoms Dimension
Publicly Observable Symptoms Dimension
Cardiovascular/Stroke Symptoms Dimension
Cognitive Dyscontrol Symptoms Dimension
31.24 7.61
5.21 4.35
12.34 4.86
4.21 3.31
9.48 6.36
23.48 13.39
4.97 3.20
8.41 5.69
3.93 4.64
6.17 5.46
PTSD (n = 11)
Total 21-item ASI scale
Respiratory Symptoms Dimension
Publicly Observable Symptoms Dimension
Cardiovascular/Stroke Symptoms Dimension
Cognitive Dyscontrol Symptoms Dimension
51.91 23.81
10.36 5.59
15.64 6.99
11.64 6.45
14.27 7.10
43.27 24.56
9.45 5.87
12.27 7.68
10.36 6.42
11.18 7.17
Depression (n = 39)
Total 21-item ASI scale
Respiratory Symptoms Dimension
Publicly Observable Symptoms Dimension
Cardiovascular/Stroke Symptoms Dimension
Cognitive Dyscontrol Symptoms Dimension
16.53 8.62
2.28 2.72
8.69 5.46
1.74 2.34
3.82 3.56
11.56 7.47
1.79 2.09
6.28 4.83
1.03 1.72
2.46 2.67
Anxiety Sensitivity 200
6.3.4 Tests of Hypotheses
6.3.4.1 Analysis of differences for each diagnostic group before and after
CBT for overall level of anxiety sensitivity. To examine whether CBT reduced
overall anxiety sensitivity scores for each of the diagnostic groups, a repeated
measures analysis of variance was conducted using the pre and post-treatment
21-item ASI total scale scores as the within-subject variable. Significant
differences between pre and post-treatment 21-item ASI total scale scores were
observed for the panic disorder group, F (1, 46) = 49.371, p = .000, η² = .518,
power = 1.000; GAD group, F (1, 28) = 10.849, p = .003, η² = .279, power =
.889; and depression group, F (1, 38) = 23.254, p = .000, η² = .380, power =
.997. These results reveal that CBT is capable of providing a significant
reduction in the general or overall level anxiety sensitivity for individuals with a
diagnosis of panic, GAD, and depression.
Conversely, a significant difference was not observed between the pre
and post-treatment 21-item ASI total scale scores for the PTSD group, F (1, 10)
= 6.733, p = .027, η² = .402, power = .648. However it is important to note that
if a stringent alpha level had not been adopted, the difference between the total
scale scores would be significant, p = .027. Thus, while on the surface it would
appear that CBT was not useful in reducing the overall or general fear of anxiety
symptoms for individuals with a diagnosis of PTSD, it is proposed that this result
is a reflection of the very small sample size of this diagnostic group. A profile
plot illustrating the means of the 21-item ASI total scale scores as a function of
the four diagnostic groups before and after CBT is provided in Figure 6.1.
Anxiety Sensitivity 201
0
10
20
30
40
50
60
70
Panic GAD PTSD DepressionDiagnostic Category
Mea
n 21
-item
AS
I tot
al it
em s
cale
Pre CBTPost CBT
Figure 6.1. Mean differences and standard error on the 21-item ASI total
score for the Panic Disorder, GAD, PTSD and Depression diagnostic groups
before and after CBT.
6.3.4.2 Analysis of differences for the panic disorder group before and
after CBT for the first-order dimensions of anxiety sensitivity, as assessed by the
21-item ASI. In order to examine whether CBT resulted in a significant
reduction of scores on the four first-order dimensions of the 21-item ASI for
individuals with a diagnosis of panic disorder, a repeated measures analysis was
conducted using the pre and post-treatment scores of the ‘Fear of Respiratory
Symptoms’, ‘Fear of Publicly Observable Anxiety Symptoms’, ‘Fear of
Cardiovascular/Stroke Symptoms’ and ‘Fear of Cognitive Dyscontrol’ factors as
the within-subject variable.
Using Wilks’ criterion, a significant multivariate effect was observed
within the combination of the four first-order dimensions of the 21-item ASI, F
Anxiety Sensitivity 202
(4, 43) = 13.289, p = .000. The result reflected a moderate association within the
dimensions and the Panic disorder group, η² = .553. The observed power was
strong, 1.000.
To control for Type I error, an alpha correction of p = .0125 was set for
follow-up univariate tests. Significant differences between pre and post-
treatment scores were observed on the 21-item ASI ‘Fear of Respiratory
Symptoms’ dimension, F (1, 46) = 29.904, p = .000, η² = .394, power = 1.000;
‘Fear of Publicly Observable Symptoms’ dimension, F (1, 46) = 26.903, p =
.000, η² = .369, power = .999; ‘Fear of Cardiovascular/Stroke Symptoms’
dimension, F (1, 46) = 13.641, p = .001, η² = .229, power = .951; and ‘Fear of
Cognitive Dyscontrol’ dimension, F (1, 46) = 35.682, p = .000, η² = .437, power
= 1.000, for the panic disorder group. A profile plot illustrating the means of the
21-item ASI first-order dimensions before and after CBT for the panic disorder
group is provided in Figure 6.2. Thus, it can be seen that CBT is successful in
reducing scores on all the first-order dimensions of the 21-item ASI for
individuals with a diagnosis of panic disorder.
Anxiety Sensitivity 203
0
2
4
6
8
10
12
14
16
18
20
Fear of RespiratorySymptoms
Fear of PubliclyObservable Symptoms
Fear ofCardiovascular/Stroke
Symptoms
Fear of CognitiveDyscontrol
21-item ASI first-order Dimensions
Mea
n sc
ores
for t
he P
anic
Dis
orde
r gro
up o
n th
e 4
first
-ord
er
Dim
ensi
ons
of th
e 21
-item
AS
I bef
ore
and
afte
r CBT
Pre CBTPost CBT
Figure 6.2. Mean differences and standard error for the panic disorder group on
the four first order dimensions of the 21-item ASI before and after CBT.
6.3.4.3 Analysis of differences for the GAD group before and after CBT
for the first-order dimensions of anxiety sensitivity, as assessed by the 21-item
ASI. In order to examine whether CBT resulted in a significant reduction of
scores on the four first-order dimensions of the 21-item ASI for individuals with
a diagnosis of GAD, a repeated measures analysis was conducted using the pre
and post-treatment scores of the ‘Fear of Respiratory Symptoms’, ‘Fear of
Publicly Observable Anxiety Symptoms’, ‘Fear of Cardiovascular/Stroke
Symptoms’ and ‘Fear of Cognitive Dyscontrol’ factors as the within-subject
variable.
Using Wilks’ criterion, a significant multivariate effect was observed
within the combination of the four first-order dimensions of the 21-item ASI, F
(4, 25) = 7.794, p = .000. The result reflected a moderate association within the
Anxiety Sensitivity 204
dimensions and the GAD group, η² = .555. The observed power was strong,
.991.
Again, the alpha level was restricted to .0125 for follow-up univariate
tests. Significant differences were observed on the ‘Fear of Publicly Observable
Anxiety Symptoms’, F (1, 28) = 18.676, p = .000, η² = .400, power = .986; and
‘Fear of Cognitive Dyscontrol’ dimensions, F (1, 28) = 13.986, p = .001, η² =
.333, power = .950. However, significant differences were not observed on the
‘Fear of Respiratory Symptoms’, F (1, 28) = .058, p = .811, η² = .002, power =
.056; and ‘Fear of Cardiovascular/Stroke Symptoms’ dimensions, F (1, 28) =
.136, p = .715, η² = .005, power = .065. A profile plot illustrating the means of
the 21-item ASI first-order dimensions before and after CBT for the GAD group
is provided in Figure 6.3. It can be seen that, for individuals with a diagnosis of
GAD, CBT is successful in reducing scores relating to fearfulness of publicly
observable and cognitive dyscontrol anxiety symptoms.
Anxiety Sensitivity 205
0
2
4
6
8
10
12
14
Fear of RespiratorySymptoms
Fear of PubliclyObservable Symptoms
Fear ofCardiovascular/Stroke
Symptoms
Fear of CognitiveDyscontrol
21-item ASI first-order Dimensions
Mea
n sc
ores
for t
he G
AD G
roup
on
the
4 fir
st-o
rder
dim
ensi
ons
of th
e 21
-item
AS
I bef
ore
and
afte
r CBT
Pre CBTPost CBT
Figure 6.3. Mean differences and standard error for the GAD group on the four
first order dimensions of the 21-item ASI before and after CBT.
6.3.4.4 Analysis of differences for the PTSD group before and after CBT
for the first-order dimensions of anxiety sensitivity, as assessed by the 21-item
ASI. In order to examine whether CBT resulted in a significant reduction of
scores on the four first-order dimensions of the 21-item ASI for individuals with
a diagnosis of PTSD, a repeated measures analysis was conducted using the pre
and post-treatment scores of the ‘Fear of Respiratory Symptoms’, ‘Fear of
Publicly Observable Anxiety Symptoms’, ‘Fear of Cardiovascular/Stroke
Symptoms’ and ‘Fear of Cognitive Dyscontrol’ factors as the within-subject
variable.
Using Wilks’ criterion, a non-significant multivariate effect was observed
within the combination of the four first-order dimensions of the 21-item ASI, F
(4, 7) = 3.638, p = .066. A profile plot illustrating the means of the 21-item ASI
Anxiety Sensitivity 206
first-order dimensions before and after CBT for the PTSD group is provided in
Figure 6.4. One explanation for this null finding is that the sample size is too
small to detect a difference between the pre and post conditions.
0
2
4
6
8
10
12
14
16
18
20
Fear of RespiratorySymptoms
Fear of PubliclyObservable Symptoms
Fear ofCardiovascular/Stroke
Symptoms
Fear of CognitiveDyscontrol
21-item ASI first-order Dimensions
Mea
n sc
ores
for t
he P
TSD
Gro
up o
n th
e 4
first
-ord
er
Dim
ensi
ons
of th
e 21
-item
AS
I bef
ore
and
afte
r CBT
Pre CBTPost CBT
Figure 6.4. Mean differences and standard error for the PTSD group on the four
first order dimensions of the 21-item ASI before and after CBT.
6.3.4.5 Analysis of differences for the depression group before and after
CBT for the first-order dimensions of anxiety sensitivity, as assessed by the 21-
item ASI. In order to examine whether CBT resulted in a significant reduction of
scores on the four first-order dimensions of the 21-item ASI for individuals with
a diagnosis of depression, a repeated measures analysis was conducted using the
pre and post-treatment scores of the ‘Fear of Respiratory Symptoms’, ‘Fear of
Publicly Observable Anxiety Symptoms’, ‘Fear of Cardiovascular/Stroke
Symptoms’ and ‘Fear of Cognitive Dyscontrol’ factors as the within-subject
variable.
Anxiety Sensitivity 207
Using Wilks’ criterion, a significant multivariate effect was observed
within the combination of the four first-order dimensions of the 21-item ASI, F
(4, 35) = 5.792, p = .001. The result reflected a moderate association within the
dimensions and the Depression group, η² = .398. The observed power was
strong, .966.
For follow-up univariate tests, an alpha level restriction of .0125 was set.
Significant differences were observed on the ‘Fear of Publicly Observable
Symptoms’, F (1, 38) = 18.987, p = .000, η² = .333, power = .989; ‘Fear of
Cardiovascular/Stroke Symptoms’, F (1, 38) = 7.497, p = .009, η² = .165, power
= .761; and ‘Fear of Cognitive Dyscontrol’ dimensions, F (1, 38) = 8.858, p =
.005, η² = .189, power = .826. However, a significant difference was not
observed on the ‘Fear of Respiratory Symptoms’ dimension, F (1, 38) = 2.072, p
= .158, η² = .052, power = .289. A profile plot illustrating the means of the 21-
item ASI first-order dimensions before and after CBT for the depression group is
provided in Figure 6.5. It can be seen that, for individuals with a diagnosis of
depression, CBT is successful in reducing already low scores even further for
those dimensions measuring fearfulness of publicly observable, cardiovascular/
stroke and cognitive dyscontrol anxiety symptoms.
Anxiety Sensitivity 208
0
2
4
6
8
10
12
Fear of RespiratorySymptoms
Fear of PubliclyObservable Symptoms
Fear ofCardiovascular/Stroke
Symptoms
Fear of CognitiveDyscontrol
21-item ASI first-order Dimensions
Mea
n sc
ores
for t
he D
epre
ssio
n G
roup
on
the
4 fir
st-o
rder
D
imen
sion
s of
the
21-it
em A
SI b
efor
e an
d af
ter C
BT
Pre CBTPost CBT
Figure 6.5. Mean differences and standard error for the depression group
on the four first order dimensions of the 21-item ASI before and after CBT.
6.3.4.6 Analysis of differences on the BAI, CCQ-M, FQ and Zung – SDS
after CBT. Finally, in order to examine whether CBT was capable of facilitating
a reduction in participant’s scores on well-established measures, responses were
analysed using a paired samples t-test on the pre (M = 22.41, SD = 12.90) and
post (M = 16.01, SD = 12.54) treatment scores of the BAI; the pre (M = 57.54,
SD = 17.17) and post (M = 50.08, SD = 18.17) treatment scores of the CCQ-M;
the pre (M = 32.86, SD = 22.83) and post (M = 26.87, SD = 21.28) treatment
scores of the Fear Questionnaire; and pre (M = 52.24, SD = 9.56) and post (M =
45.98, SD = 10.92) treatment scores of the Zung – SDS. Using a probability
level of .05 for all t-test analyses, significant differences were observed between
the participants pre and post-treatment scores on the BAI, t (127) = 6.572, p =
.000; CCQ-M, t (124) = 5.994, p = .000; Fear Questionnaire, t (121) = 5.359, p =
.000; and Zung – SDS, t (127) = 7.817, p = .000. Thus, it can be seen that the
Anxiety Sensitivity 209
manual driven CBT utilised in the current investigation was also capable of
providing significant reductions in the symptoms, cognitions and fear of anxiety
and depression globally.
6.4 Discussion
The purpose of the current investigation was to examine whether
differences in the overall and specific dimensions of anxiety sensitivity existed
within four diagnostic categories both before and after CBT in order to provide a
further test of the validity of the revised 21-item ASI. Examination of responses
from diagnostic groups comprised of individuals who met the DSM-IV criteria
for panic disorder, GAD, PTSD or depression found significant differences
within each group with the exception of the PTSD group. In addition, significant
differences between pre and post CBT scores were observed using the BAI,
CCQ-M, FQ and Zung-SDS. As such, independent of the changes in scores on
the 21-item ASI, the manual driven CBT utilised in the current investigation was
effective as it was capable of facilitating significant reductions in the symptoms,
cognitions and fear of anxiety and depression globally. It is proposed that
changes in the anxiety sensitivity scores among diagnostic groups are not an
artifact of the therapy utilised in the current investigation. Therefore it is posited
that the results of the current investigation provide support for the construct
validity of the 21-item ASI. Each will be discussed in turn.
Significant differences were found between pre and post-treatment scores
for the panic disorder, GAD and depression groups using the total score of the
21-item ASI. The exception to this was for the PTSD group who failed to show
a significant difference between the pre and post-treatment scores. These
findings are consistent with results from Hazen et al. (1996), McNally and
Anxiety Sensitivity 210
Lorenz (1987), Penava et al. (1997), Shear et al. (1994), and Telch et al. (1993),
who also reported significant reduction in scores using the original ASI
following manual driven CBT for individuals with an anxiety disorder. Further,
while on the surface it would appear that CBT was not useful in reducing the
overall or general fear of anxiety symptoms for individuals with a diagnosis of
PTSD, it is possible that this result is a reflection of the very small sample size of
this diagnostic group.
Once the pre and post-treatment differences of the second-order
dimension or total scale score of the 21-item ASI were taken into consideration,
differences within the first-order or specific anxiety sensitivity dimensions were
examined. As it was, to the author’s knowledge, unknown whether CBT
interventions would be successful in facilitating a reduction in the first-order or
specific dimensions of anxiety sensitivity, it was hypothesised that examination
of this for each diagnostic group would provide potentially important information
relating to the construct validity of the 21-item ASI as well as anxiety sensitivity
theory.
For the panic disorder group, significant differences were observed
between the pre and post CBT scores for all four of the 21-item ASI first-order
dimensions. The largest therapeutic gain following CBT was observed on the
‘Fear of Cognitive Dyscontrol’ dimension followed by the ‘Fear of Publicly
Observable Symptoms’ dimension. Approximately equal therapeutic gain was
observed post CBT for the ‘Fear of Respiratory Symptoms’ and ‘Fear of
Cardiovascular/Stroke Symptoms’ dimensions. As panic disorder is
characterised by the DSM-IV-TR as the experience of repeated panic attacks as
well as a number of somatic and cognitive symptoms (APA, 2000), it is
Anxiety Sensitivity 211
theoretically consistent that a treatment that includes components designed to
address fears related to both cognitive and physiological arousal would be
capable of facilitating a change in anxiety sensitivity, not only globally but also
on specific dimensions.
A similar finding to the panic disorder group was also observed for the
GAD diagnostic group, with significant differences between the pre and post
CBT scores on the ‘Fear of Cognitive Dyscontrol’ and ‘Fear of Publicly
Observable Symptoms’ dimensions. As such, it can be seen that for individuals
with a diagnosis of GAD, CBT is successful in reducing scores relating to
fearfulness of publicly observable and cognitive dyscontrol anxiety symptoms.
However, it is interesting to note that significant differences post-treatment were
not observed on the ‘Fear of Respiratory Symptoms’ and ‘Fear of
Cardiovascular/Stroke Symptoms’ dimensions. One possible explanation for this
null finding is the argument that the cognitive features of GAD are more
prominent and even though sensitivity to arousal did not reduce, sensitivity to
cognitive dyscontrol decreased. Secondly, the pre-treatment scores for the
respiratory and cardiovascular/stroke dimensions were not elevated to begin
with. As such, it is possible that individuals who meet the diagnostic criteria for
GAD are not sensitive or fearful of these symptoms and as such therapy aimed at
reducing fear of symptoms associated with these dimensions is not warranted in
this diagnostic group nor was it specifically provided.
Conversely, significant differences between the pre and post CBT scores
on the ‘Fear of Publicly Observable Symptoms’ ‘Fear of Cardiovascular/Stroke
Symptoms’ and ‘Fear of Cognitive Dyscontrol’ dimensions were observed for
individuals from the depression diagnostic group. As such, it is evident that for
Anxiety Sensitivity 212
individuals with a diagnosis of depression, CBT is successful in reducing already
low scores even further for these anxiety sensitivity dimensions. However, it is
interesting to note that a significant post-treatment difference was not observed
on the ‘Fear of Respiratory Symptoms’ dimension. The failure to find a
significant difference on the ‘Fear of Respiratory Symptoms’ dimension may be
the result of this diagnostic group learning about these symptoms and knowing
not to fear them rather than ever being fearful of such symptoms in the first
instance.
Finally, it is interesting to note that for the PTSD group, significant
differences were not found between the pre and post CBT scores for any of the
four 21-item ASI first-order dimensions. As the theoretical expectations for this
diagnostic group did not indicate that CBT would be ineffective, it is important
to critically examine why this finding might have occurred. As mentioned in the
previous chapter, the DSM-IV-TR (APA, 2000) characterises PTSD as involving
a combination of both physical and cognitive symptoms. Therefore, it would be
logical to hypothesise that a treatment that includes components designed to
address fears related to both cognitive and physiological arousal would be
capable of facilitating a change in anxiety sensitivity, not only globally but also
on specific dimensions. As such, it is not entirely clear why there was a failure
to observe any changes post therapy. One explanation for this null finding is that
the sample size was too small to detect a difference between the pre and post-
treatment conditions. Therefore, further studies using a larger sample size are
warranted.
Anxiety Sensitivity 213
6.5 Chapter Summary
The results of the current investigation have revealed that generalised,
manual driven CBT is capable of facilitating a reduction in both global and
specific anxiety sensitivity scores measured using the 21-item ASI for
individuals with a diagnosis of panic disorder, GAD and depression, but not for
individuals with a diagnosis of PTSD. As anxiety sensitivity is defined as a
cognitive, individual difference variable that is delineated by the individual’s fear
of anxiety related sensations and based on the belief that such sensations result in
harmful consequences (Reiss, 1991; Reiss & Havercamp, 1996; Reiss &
McNally, 1985), the findings of the current investigation provide further
important evidence for the construct validity of the 21-item ASI. Further
important information regarding anxiety sensitivity theory was also obtained in
that manual-driven, generalised CBT was capable of facilitating a reduction in
some or all of the specific, first-order dimensions of the 21-item ASI for the
panic disorder, GAD, and depression diagnostic groups. These findings are
further support of the multidimensional nature and thus construct validity of
anxiety sensitivity. Further, the results obtained in the present investigation are
also unique in that previous investigations have, for the most part, focused on
comparing treatment effects between individuals with panic disorder and a mixed
group of individuals with other anxiety disorders. This practice has led to
potentially important information being lost because the uniqueness of each
anxiety disorder has not been examined previously.
One limitation of the current investigation was that there were not enough
participants who met the diagnostic criteria for social phobia, OCD or specific
phobia to include in the present investigation. The inclusion of these three
Anxiety Sensitivity 214
diagnostic groups could have led to a richer understanding of how anxiety
sensitivity varies across all of the anxiety disorders and this warrants further
investigation. Additionally, the sample size of the PTSD diagnostic group was
very small and is also regarded as a limitation in that it was not large enough to
detect a difference following CBT. As such, the effectiveness of CBT for this
diagnostic category is not clear and further studies, using a much larger sample
size are warranted.
A further limitation of the current investigation involves the therapy
itself. It is important to note that the manual driven CBT utilised in the current
investigation was generalisable to all anxiety or depressive disorders rather than
specifically targeted to a particular diagnostic category. Thus, further
investigations that utilise manual driven CBT for PTSD, GAD and major
depression and dysthymia specifically should be used to examine both general
and specific changes in anxiety sensitivity.
By employing a waitlist control group, the question of whether anxiety
sensitivity is a stable, personality variable could also have been tested. Findings
from Hazen et al. (1996) revealed that while their data indicated that anxiety
sensitivity remained relatively stable in the absence of treatment, they found that
the construct was responsive to CBT. As a result, Hazen et al. (1996) argued that
anxiety sensitivity should be regarded as a ‘cross-situational, dispositional’
variable rather than a stable personality trait. Thus, the inclusion of a waitlist
control group in the current investigation would have been able to re-test this
hypothesis in all four diagnostic groups. Further, it would have been beneficial
to include a waitlist control group in order to determine the magnitude of
Anxiety Sensitivity 215
improvement in anxiety sensitivity that may occur over time independent of CBT
intervention.
Thus, the results from the current investigation suggest that, at an overall
level, the 21-item ASI is sensitive enough to capture clinical improvements.
Similarly, while the findings suggest that anxiety sensitivity is responsive to
CBT, the specific components of treatment which reduce anxiety sensitivity
remain to be determined. As such, it is concluded that while the current
investigation provides further evidence of construct validity of not only the total
scale but also of the first order dimensions of anxiety sensitivity among various
diagnostic groups, it also provides important information pertaining to anxiety
sensitivity theory as well.
Anxiety Sensitivity 216
Anxiety Sensitivity 217
Chapter Seven: General discussion
7.1 Introduction 219
7.2 Measurement of Anxiety Sensitivity 219
7.3 Clinical Utility 221
7.4 Therapy 222
7.5 Theoretical Contribution 222
7.6 Limitations of the Research 224
7.7 Future Directions 226
7.8 Conclusion 229
Anxiety Sensitivity 218
Anxiety Sensitivity 219
CHAPTER SEVEN
GENERAL DISCUSSION
7.1 Introduction
The goal of the current dissertation was to conduct a series of empirical
investigations that examined the multidimensional nature of anxiety sensitivity as
well as the psychometric properties of an expanded measure in order to provide a
substantial theoretical and empirical contribution to the field of research. The
main findings and strengths of each empirical investigation, the theoretical
contribution, and limitations of the research, as well as future directions will be
examined and discussed in the following sections.
7.2 Measurement of Anxiety Sensitivity
The purpose of the first empirical investigation was to evaluate the factor
structure of the 36-item ASI-R for both a clinical and nonclinical sample by
using confirmatory factor analysis on the hypothesised models identified by
Deacon et al. (2003); Taylor and Cox (1998); and Zvolensky et al. (2003). The
findings from this investigation demonstrated that, through the removal of a
number of problematic items, the hypothesised model developed by Taylor and
Cox (1998) could be improved substantially by reducing the 36-item ASI-R to a
21-item index which was capable of providing a more parsimonious and valid
account of the anxiety sensitivity construct, that also showed evidence of both
configural and metric invariance using both a clinical and nonclinical sample.
The decision to remove redundant items resulted in a factor analytic structure
that was capable of providing an acceptable and reasonable account of the
anxiety sensitivity construct across two separate groups. As long as there are
Anxiety Sensitivity 220
enough items present to tap the full range of anxiety sensitivity dimension under
investigation, the presence of additional items does little to improve the construct
validity of the scale. Further, evidence of acceptable internal consistency, test-
retest reliability, and concurrent and discriminant validity was observed for both
the clinical and nonclinical samples. It is proposed that, by using the more
stringent method of CFA, the first investigation revealed the 21-item ASI
provides a more parsimonious, reliable and valid measure of anxiety sensitivity
in adults from both clinical and nonclinical populations that has the ability to
encourage future research in the area of anxiety sensitivity and contribute to
clinical practice. Thus, the main strength of the current dissertation is that it
represents, to the author’s knowledge, the only investigation to have provided a
substantial investigation of both clinical and nonclinical groups scores using an
expanded measure of anxiety sensitivity that is capable of adequately describing
the major anxiety sensitivity dimensions. As discussed throughout this
dissertation, the 21-item ASI has repeatedly revealed evidence of construct
validity of not only the total scale but also of the first-order dimensions among
both groups, which also provides important information pertaining to the anxiety
sensitivity construct and Expectancy Theory.
It is also noted that a further strength of the current dissertation can be
found in the analytic choice to employ CFA over EFA. As the goal of CFA is to
confirm a theory regarding the internal structure of a specified measurement
model, it is a more robust evaluation to employ once initial exploratory work has
been done. Therefore, as there were several measurement models available in
the literature, it was decided to employ the use of CFA in order to test the
multidimensional theory of anxiety sensitivity rather than add to the confusion
Anxiety Sensitivity 221
and contradictions that have been created through the use of EFA studies of the
original 16-item ASI. Thus, the current dissertation is the first to provide an in-
depth account of the psychometric properties of an expanded measure of anxiety
sensitivity in both a clinical and nonclinical sample.
7.3 Clinical Utility
The purpose of the second investigation was to compare the differences
both between and within different diagnostic group’s responses on the general
and first-order dimensions of the 21-item ASI. It was proposed that further
examination of the 21-item ASI’s ability to discriminate between groups in a
theoretically expected manner as well as to investigate whether individual
diagnostic groups would endorse particular facets of anxiety sensitivity in a
theoretically expected manner, would provide further evidence of the construct
validity of the scale. Thus, the second investigation found that differences
between the diagnostic groups were consistent with theoretical expectations and
were also consistent to previously published results. Further, a combination of
significant differences and similarities within each group was also reported. It is
important to note that main strength of the second investigation was that has
increased our understanding of the aetiology of different forms of anxiety and
depressive psychopathology as well as provided further evidence of construct
validity of the 21-item ASI. As such, the current dissertation’s examination the
contribution of each dimension to a particular psychopathology has assisted with
the theoretical clarification of the anxiety sensitivity construct by revealing the
important role each dimension plays in the different psychopathologies
examined. Further, it is proposed that the examination of differences between
and within the various psychopathologies, as well as nonclinical control group,
Anxiety Sensitivity 222
has provided useful answers for researchers and clinicians alike by revealing the
unique fear-based cognitive content of each anxiety and depression
psychopathology investigated.
7.4 Therapy
The purpose of the third and final investigation was to examine
differences within each diagnostic category both before and after CBT in order to
provide a further test of validity for the revised 21-item ASI. It was found that
generalised, manual driven CBT was capable of facilitating significant reductions
in anxiety sensitivity for all groups that contained adequate sample sizes, which
in turn, provided further support for the multidimensional nature and construct
validity of the 21-item ASI and its ability to be a useful tool for measuring
cognitive change. It is further proposed that the main strength of the third
investigation is that it has the potential to assist in the design of enhanced
treatment by developing a CBT program that is tailored towards both anxiety
sensitivity and individual diagnostic classifications. Such a program may be
beneficial in assisting in greater cognitive and behavioural change in individuals
who present for therapy for a specific anxiety or depressive disorder. Finally, the
examination of both general and specific anxiety sensitivity dimensions before
and after CBT has arguably provided important information regarding the role
anxiety sensitivity plays in the aetiology of the various psychopathologies that
were investigated.
7.5 Theoretical Contribution
The current dissertation has made a significant and valid theoretical
contribution to the field of anxiety sensitivity in particular and anxiety research
in general. For instance, as discussed in Section 1.3, the central tenet of Reiss
Anxiety Sensitivity 223
and McNally’s Expectancy Theory (1985) is the notion of fundamental fears, or
sensitivities. Reiss and McNally (1985) proposed that there are three
fundamental fears; anxiety sensitivity; fear of illness, injury and death; and fear
of negative evaluation. Further, they defined anxiety sensitivity as a cognitive,
individual difference variable that is delineated by an individual’s fear of anxiety
related sensations and based on the belief that such sensations result in harmful
consequences (Reiss & McNally, 1985). Thus, when they first proposed the
concept of anxiety sensitivity, they believed it to be a generalised fear-based on
belief that non-specific anxiety symptoms have harmful outcomes. In other
words, anxiety sensitivity was regarded as unidimensional and non-specific. It is
postulated that the results from the numerous investigations contained in the
current dissertation provide enough evidence to support the view that anxiety
sensitivity is hierarchically arranged and as such, current theoretical
conceptualisation of anxiety sensitivity, as defined by Reiss and McNally (1985)
and could benefit by incorporating such a perspective.
By incorporating the notion that anxiety sensitivity is multidimensional,
or comprised of both general and specific components, the current dissertation
has made an important theoretical contribution by suggesting that the original
definition of anxiety sensitivity is too broad and should be revised in order to
incorporate the notion of a hierarchical structure. For example, it is proposed
that the definition of anxiety sensitivity could be made specific by referring to it
as a cognitive, individual difference variable that is delineated by an individual’s
fear of anxiety sensations related to respiratory, publicly observable,
cardiovascular/stroke or cognitive concerns and based on the belief that such
sensations result in harmful consequences.
Anxiety Sensitivity 224
Further, it would appear logical that anxiety sensitivity would be
multidimensional rather than unidimensional when we consider that anxiety and
anxiety disorders are multidimensional. For instance, it is well documented that
anxiety is comprised of both cognitive and somatic symptoms that manifest in
either involuntary physiological arousal (Barlow, 1991; Bourne, 1995; Stein &
Bouwer, 1997) or catastrophic misinterpretation of the physical or psychosocial
experience of anxiety as dangerous (Clark, 1986; Khawaja & Oei, 1992; 1998).
However, examination of specific diagnostic categories reveals that all of the
anxiety and even some mood disorders are manifested by specific physiological
and cognitive symptoms that are unique to each pathology. Therefore, it makes
intuitive sense that anxiety sensitivity, which is a fear of anxiety symptoms,
would also be multidimensional in order to take into account the different
physiological symptoms and thus, cognitive interpretations, that are associated
with the large array of anxiety and mood psychopathologies.
7.6 Limitations of the Research
Although there are a number of strengths to be found in the current
dissertation, it is worth noting that there are also weaknesses present. Firstly,
whilst the exploratory nature of the dissertation is a key strength, it also limits the
conclusions and generalisations that can be made about the area. Other
weaknesses arise from specific sampling issues that were beyond the control of
the author. In the first instance, there was evidence of comorbidity among the
diagnostic groups. As it has been argued that comorbidity results in a substantial
impact on the treatment and prevalence of the principal disorder (see Brown &
Barlow, 1992 for review), the inclusion of pure diagnostic groups in the current
dissertation would have lead to greater confidence when describing the
Anxiety Sensitivity 225
differences between and within the various diagnostic groups as well as after
CBT. However, it has been noted that in effectiveness settings rather than in
efficacy settings, it is very difficult to obtain pure diagnostic groups because
individuals tend to seek therapy when they feel they are no longer able to
continue functioning. As such, the individuals who take part in generalised
psychological therapy may have been living with a combination of anxiety, mood
and substance use disorders, to name a few, for many years. Conversely,
participants included in efficacy settings tend to be a part of a well controlled,
clinical research trial and have been included as part of the research methodology
because they conform to a strict selection criteria. Regardless of whether they
have been living with the condition for a number of years, they are selected
because they fit the selection criteria. As such, comorbidity is a common
phenomenon for many individuals who are part of research from effectiveness
settings and the results obtained in the current dissertation generalise to real
world settings.
A further limitation is the lack of participants to include diagnostic
groups comprising of social phobia, OCD or specific phobia. The inclusion of
such groups is clearly warranted as the addition of these diagnostic groups would
have led to a richer understanding of how anxiety sensitivity varies across all of
the anxiety disorders. Similarly, the small number of individuals who had a
principal diagnosis of PTSD and who also completed the therapeutic component
was also a limitation and a larger sample size may have identified whether CBT
was capable of facilitating a change in anxiety sensitivity scores following
treatment.
Anxiety Sensitivity 226
Finally, it is worth reiterating that all of the participants who comprised
the clinical sample were in receipt of prescription medication at the time of
participation. While Oei, Llamas, and Evans (1997) have reported that
individuals with panic disorder with or without agoraphobia who received
manual driven CBT simultaneously with a pre-existing treatment of anti-anxiety,
anti-depressant or combination of both medications do not differ significantly
from one another or from individuals who did not receive medication either at
pre-treatment, it is still not known whether scores obtained in the current
dissertation were lower than what would have been obtained if the participants
who formed the clinical group were un-medicated at the time of participation.
7.7 Future Directions
Future research should attempt to address the limitations posed by
replicating the investigations conducted in the current dissertation in order to
determine stability and generalisability. It is proposed that the main
contributions of the current dissertation is that it has found evidence that anxiety
sensitivity is arranged hierarchically, with four-first order dimensions that load
onto a general, second-order dimension and an improved measure of anxiety
sensitivity has been produced, which can more accurately measure these
dimensions. However, it should be noted that the empirical investigations
contained in the current dissertation are all derived from one large sample of
clinical and nonclinical adults. Retesting the 21-item ASI, using diverse
populations has the ability to contribute to the convergent validity of the 21-item
ASI as well as the growing research interest in anxiety sensitivity. The results
and conclusions derived in the current dissertation are an important first step
towards providing an adequate and psychometrically sound instrument that is
Anxiety Sensitivity 227
capable of capturing the multidimensional structure of anxiety sensitivity in both
clinical and nonclinical groups.
Like any other assessment tool, it is important that the 21-item ASI
undergo further revisions and refinement. This will ensure that the psychometric
properties of the measure are kept up to date as well as provide on-going validity
of the measure. In addition, as the results contained in the current dissertation
have revealed that the measure is an improvement upon the 16-item ASI and 36-
item ASI-R, it is important to develop norms, not only for clinical and normative
samples, but also for other valid samples such as other ethnic cultures and
groups. There has been recent interest in examining psychological constructs in
non-western and non-English speaking cultures in order to understand the
phenomena of mental health issues between and within diverse cultural groups
(see Arrindell et al., 2004; Cintrón, Carter, Suchday, Sbrocco, & Gray, 2004;
Norton, De Coteau, Hope, & Anderson, 2003; Zvolensky et al., 2003 for review).
Further, it is also of interest to examine whether there is configural and metric
invariance between diverse cultural groups using the improved 21-item ASI or
whether diverse samples respond entirely differently to Westernised and English
speaking samples. Further it allows for the explicit comparison of Australian
samples with other diverse cultural groups in order to determine whether similar
or diverse findings emerge.
Although the differences observed between the different diagnostic
groups as well as the nonclinical group were consistent with theoretical
expectations and findings from previous investigations of anxiety sensitivity, the
current dissertation has identified an empirically and theoretically sound
instrument that is capable of not only measuring specific domains of anxiety
Anxiety Sensitivity 228
sensitivity that is also sensitive enough to capture the differences between and
within several psychopathologies. Thus, future research conducted using the 21-
item ASI has the potential to lead to greater understanding of how anxiety
sensitivity can contribute towards the development of particular
psychopathologies as well as provide important information relating to the
specific fear-based cognitive content of the particular psychopathology under
investigation. As anxiety sensitivity has been examined widely, it is of interest
to examine how the construct, using the 21-item ASI, is linked to other clinical
problems such as asthma, hypochondriasis, chronic pain, and substance use to
name a few.
It is further proposed that the results obtained regarding differences in
anxiety sensitivity scores following CBT has important implications for future
research in terms of intervention. Future research into generalised CBT for
anxiety and mood disorders could benefit from the inclusion of a component that
deals with anxiety sensitivity in general and the specific fear-based domains in
particular. For example, Taylor (2003) has argued that anxiety sensitivity may
play a key role in understanding PTSD and that therapeutic interventions that
directly target anxiety sensitivity could improve treatment outcome, particularly
if the treatment is applied before the commencement of trauma-related exposure
therapy. Thus, it is proposed that interventions that include a specific anxiety
sensitivity component could be more successful than a generalised CBT
program.
While the findings from the third investigation reveal that anxiety
sensitivity is responsive to general CBT, it remains to be seen whether the
change is maintained following therapy. Could it be that elevated levels of
Anxiety Sensitivity 229
anxiety sensitivity at post-treatment indicate those who are at risk of relapse?
Future research is clearly warranted in order to determine if post-treatment scores
on the 21-item ASI are maintained over time, and also to examine whether 21-
item ASI scores are capable of predicting relapse on other clinical indicators.
Finally, as all of the participants who comprised the clinical sample were
in receipt of prescription medication at the time of participation, future research
is needed to determine whether changes in all the anxiety sensitivity dimensions
is seen when pure groups containing only non-medicated participants who
receive CBT and vice versa are responsive to other forms of treatment.
7.8 Conclusion
The current dissertation is, to the author’s knowledge, the first to
establish that the 16-item ASI and 36-item ASI-R could be improved upon by
using 21 items contained in the 36-item ASI-R. It is the first to provide an in-
depth account of the psychometric properties of an expanded measure of anxiety
sensitivity in both a clinical and nonclinical samples. It has been revealed that
the theoretical definition of anxiety sensitivity is perhaps too broad to capture the
specific facets of anxiety sensitivity and could be refined in order to incorporate
the notion of a hierarchical structure. Further, it has provided useful answers for
researchers and clinicians alike by revealing the unique fear-based cognitive
content of each psychopathology investigated. Thus, the current dissertation is
capable of assisting clinicians in the assessment process and it has the potential
to assist in the development of more comprehensive therapeutic interventions for
particular psychopathologies. In conclusion, it is proposed that the overall
findings relating to the empirical investigations discussed in the current
Anxiety Sensitivity 230
dissertation enhance our understanding of anxiety sensitivity and how the 21-
item ASI can be used to improve therapeutic interventions in clinical practice.
Anxiety Sensitivity 231
References
American Psychiatric Association. (2000). Diagnostic and statistical manual of
mental disorders (4th ed, Text Revision.). Washington, DC: Author.
American Psychiatric Association. (1994). Diagnostic and statistical manual of
mental disorders (4th ed.). Washington, DC: Author.
American Psychiatric Association. (1987). Diagnostic and statistical manual of
mental disorders (3rd ed, Revised.). Washington, DC: Author.
American Psychiatric Association. (1980). Diagnostic and statistical manual of
mental disorders (3rd ed.). Washington, DC: Author.
Arrindell, W. A., Eisemann, M., Oei, T. P. S., Caballo, V. E., Sanavio, E.,
Sica, C., Bages, N., Feldman, L., Torres, B., Iwawaki, S., Hatzichristou,
C., Castro, J., Canalda, G., Furnham, A., van der Ende, J. (in press).
Phobic anxiety in 11 nations: Part II. Hofstede’s dimensions of national
cultures predict national level variations. Personality and Individual
Differences.
Australian Bureau of Statistics. (1998). National survey of mental health and
wellbeing of adults: Profile of adults (No. 4326.0). Canberra, Australian
Capital Territory: Author.
Anderson, J. C., & Gerbing, D.W. (1988). Structural Equation Modeling in
Practice: A Review and Recommended Two-Step Approach.
Psychological Bulletin, 103 (3), 411-423.
Anxiety Sensitivity 232
Antony, M. M., Bieling, P. J., Cox, B. J., Enns, M. W., & Swinson, R. P.
(1998). Psychometric properties of the 42-item and 21-item versions of
the Depression Anxiety Stress Scales (DASS) in clinical groups and a
community sample. Psychological Assessment, 10, 176-181.
Apfledorf, W. J., Shear, M. K., Leon, A. C., & Portera, L. (1994). A brief screen for panic disorder. Journal of Anxiety Disorders, 8, 71–78.
Asmundson, G. J. G. (1999). Anxiety sensitivity and chronic pain: Empirical
findings, clinical implications, and future directions. In S. Taylor (Ed.),
Anxiety Sensitivity: Theory, Research, and Treatment of the Fear of
Anxiety (pp. 269-285). New Jersey: Lawrence Erlbaum Associates.
Asmundson, G. J. G., & Norton, G. R. (1993). Anxiety sensitivity and its
relationship to spontaneous and cued panic attacks in college students.
Behavior Research and Therapy, 31, 199-201.
Ball, S. G., Otto, M. W., Pollack, M H., Uccello, R. & Rosenbaum, J. F.
(1995). Differentiating social phobia and panic disorder: A test of core
beliefs. Cognitive Therapy and Research, 19 (4), 473-482.
Barlow, D. H. (1991). Disorders of emotion: Clarification, elaboration, and
future directions. Psychological Inquiry, 2 (1), 97-105.
Beck, A. T., Emery, G., & Greenberg, R. L. (1985). Anxiety Disorders and
Phobias. New York: Basic Books, Inc.
Beck, A. T., & Emery, G. (1979). Cognitive therapy of anxiety and phobic
disorders. Philadelphia: Centre for Cognitive Therapy.
Beck, A. T., Epstein, N., Brown, G., & Steer, R. A. (1988). An inventory for
measuring clinical anxiety: Psychometric properties. Journal of
Consulting and Clinical Psychology, 56 (6), 893-897.
Anxiety Sensitivity 233
Beck, A. T., & Steer, R. A. (1993). Beck Anxiety Inventory Manual. San
Antonio: Harcourt Brace and Company.
Bentler, P. M. (1990). Comparative fit indices in structural models.
Psychological Bulletin, 107, 238-246.
Bentler, P. M. (1995). EQS Structural equations program manual. Encino,
CA: Multivariate Software, Inc.
Biggs, J. T., Wylie, L. T. & Ziegler, V. E. (1978). Validity of the Zung Self-
Rating Depression Scale. British Journal of Psychiatry, 132, 381–385.
Blais, M. A., Otto, M. W., Zucker, B. G., McNally, R. J., Schmidt, N. B.,
Fava, M., & Pollack, M. H. (2001). The Anxiety Sensitivity Index:
Item analysis and suggestions for refinement. Journal of Personality
Assessment, 77(2), 272-294.
Bonin, M. F., Norton, G. R., Asmundson, G. J. G., Dicurzio, S., & Pidlubney,
S. (2000). Drinking away the hurt: The nature and prevalence of PTSD
in substance abuse patients attending a community-based treatment
program. Journal of Behavior Therapy and Experimental Psychiatry, 31
(1), 55-66.
Bourne, E. J. (1995). The anxiety and phobia workbook (2nd ed.). Oakland,
CA: New Harbinger Publications, Inc.
Brown, T. A., Antony, M. M., & Barlow, D. H. (1995). Diagnostic Comorbidity
in Panic Disorder: Effect on Treatment Outcome and Course of
Comorbid Diagnosis Following Treatment. Journal of Consulting and
Clinical Psychology, 63 (3), 408-418.
Anxiety Sensitivity 234
Brown, T. A., & Balow, D. H. (1992). Comorbidity among anxiety disorders:
Implications for treatment and DSM-IV. Journal of Consulting and
Clinical Psychology, 60 (6), 835-844.
Brown, T. A., & Deagle, E. A. (1992). Structured interview of assessment of
nonclinical panic. Behavior Therapy, 23 (1), 75-85.
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit.
In: K. A. & J. S. Long (Eds.), Testing Structural Equation Models (pp.
136-162). Newbury Park: Sage Publications.
Bruce, T. J., Spiegel, D. A., Gregg, S. F., & Nuzzarello, A. (1995). Predictors
of Alprazolam Discontinuation With and Without Cognitive Behavior
Therapy in Panic Disorder. American Journal of Psychiatry, 152, (8),
1156-1160.
Burnstein, A. (1985). Posttraumatic Flashbacks, Dream Disturbances, and
Mental Imagery. Journal of Clinical Psychiatry, 46, 374-378.
Carver, C. S., Scheier, M. F., & Weintraub, J. J. (1989). Assessing coping
strategies: A theoretically based approach. Journal of Personality and
Social Psychology, 65, 375-390.
Cintrón, J. A., Carter, M. M., Suchday, S., Sbrocco, T., & Gray, J. (in
press). Factor structure and construct validity of the Anxiety Sensitivity
Index among island Puerto Ricans. Journal of Anxiety Disorders.
Clark, D. M. (1986). A cognitive model of panic. Behavior Research and
Therapy, 24, 461-470.
Clum, G. A. (1990). Coping with Panic: A drug-free approach to dealing with
anxiety attacks. Pacific Grove, CA: Brooks/Cole.
Anxiety Sensitivity 235
Cox, B. J. (1996). The nature and assessment of catastrophic thoughts in panic
disorder. Behavior Research and Therapy, 34 (4), 363-374.
Cox, B. J., Borger, S. C., & Enns, M. W. (1999). Anxiety sensitivity and
emotional disorders: Psychometric studies and their theoretical
implications. In S. Taylor (Ed.), Anxiety Sensitivity: Theory, Research,
and Treatment of the Fear of Anxiety (pp. 115-148). New Jersey:
Lawrence Erlbaum Associates.
Cox, B. J., Endler, N. S., Norton, G. R., & Swinson, R. P. (1991). Anxiety
sensitivity and nonclinical panic attacks. Behavior Research and
Therapy, 29 (4), 367-369.
Cox, B. J., Endler, N. S., & Swinson, R. P. (1995). Anxiety sensitivity and
panic attack symptomatology. Behavior Research and Therapy, 33 (7),
833-836.
Cox, B. J., Enns, M. W., & Taylor, S. (2001). The effect of rumination as a
mediator of elevated anxiety sensitivity in major depression. Cognitive
Research and Therapy, 25 (5), 525-534.
Cox, B. J., Parker, J. D. A., & Swinson, R. P. (1996). Anxiety Sensitivity:
Confirmatory Evidence for a Multidimensional Construct. Behavior
Research and Therapy, 34 (7), 591-598.
Cox, B. J., Swinson, R. P., & Shaw, B. F. (1991). Value of the Fear
Questionnaire in differentiating agoraphobia and social phobia. Journal
of Psychiatry, 159, 842-845.
Creamer, M., Foran, J., & Bell (1995). The Beck Anxiety Inventory in a
nonclinical sample. Behavior Research and Therapy, 33 (4), 477-485.
Anxiety Sensitivity 236
Davidson, J. R. T., & Foa, E. B. (1991). Diagnostic issues in Posttraumatic
Stress Disorder: Considerations for the DSM-IV. Journal of Abnormal
Psychology, 100 (3), 346-355.
Deacon, B. J., Abramowitz, J. S., Woods, C. M., & Tolin, D. F. (2003). The
Anxiety Sensitivity Index – Revised: Psychometric properties and factor
structure in two nonclinical populations. Behavior Research and
Therapy.
Deacon, B. J., & Valentiner, D. P. (2001). Dimensions of anxiety sensitivity and
their relationship to nonclinical panic. Journal of Psychopathology and
Behavioral Assessment, 23, 25–33.
Donnell, C. D., & McNally, R. J. (1990). Anxiety sensitivity and history of
panic as predictors of response to hyperventilation. Behavior Research
and Therapy, 27 (4), 325-332.
Ehlers, A., (1995). A 1-Year Prospective Study of Panic Attacks: Clinical
Course and Factors Associated with Maintenance. Journal of Abnormal
Psychology, 104 (1), 164-172.
Eke, M., & McNally, R. J. (1996). Anxiety sensitivity, suffocation fear, trait
anxiety, and breath-holding duration as predictors of response to carbon
dioxide challenge. Behavior Research and Therapy, 34 (8), 603-607.
Eysenck, M. W. (1982) Attention and arousal: Cognition and performance. New
York: Springer-Verlag.
Eysenk, H. J. (1968). A theory of the incubation of fear/anxiety response.
Behavior Research and Therapy, 6, 319-321.
Anxiety Sensitivity 237
Fedoroff, I. C., Taylor, S., Asmundson, G. J. G., & Koch, W. J. (2000).
Cognitive factors in traumatic stress reactions: Predicting PTSD
symptoms from anxiety sensitivity and beliefs about harmful events.
Behavioural and Cognitive Psychotherapy, 28, 7-17.
Fenichel. (1945). The psychoanalytic theory of neurosis. New York:
W. W. Norton.
Frankl, V. E. (1975). Paradoxical intention and dereflection. Psychotherapy
Theory, Research and Practice, 12(3), 226-237.
Freud, S. (1995). Inhibitions, symptoms and anxiety. In J. Strachey (ed. and
Trans.), The standard edition of the complete psychological works of
Sigmund Freud (Vol. 20, pp. 87-172). London: Hogarth Press.
(Original work published 1926).
Gabrys, J. B., & Peters, K. (1985). Reliability, discriminant and predictive
validity of the Zung Self-rating Depression Scale. Psychological Reports,
57, 1091–1096.
Gerbing, D. W., & Hamilton, J. G. (1996). Viability of exploratory factor
analysis as a precursor to confirmatory factor analysis. Structural
Equation Modeling, 3, 62-72.
Goldstein, A. J., & Chambless, D. L. (1978). A reanalysis of agoraphobia.
Behavior Therapy, 9, 47-59.
Hazen, A. L., Walker, J. R., & Eldridge G. D. (1996). Anxiety Sensitivity and
Treatment Outcome in Panic Disorder. Anxiety, 2, 34-39.
Hazen, A. L., Walker, J. R., & Stein, M. B. (1995). Comparison of anxiety
Sensitivity in panic disorder and social phobia. Anxiety, 1, 298-301.
Anxiety Sensitivity 238
Hegel, M. T., Ravaris, C. L., & Ahles, T. A. (1994). Combined Cognitive-
Behavioral and Time-Limited Alprazolam Treatment of Panic Disorder.
Behavior Therapy, 25, 183-195.
Hewitt, P. L., & Norton, G. R. (1993). The Beck Anxiety Inventory: A
psychometric analysis. Psychological Assessment, 5 (4), 408-412.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indices in covariance
structure analysis: Conventional criteria versus new alternatives.
Structural Equation Modeling, 6(1), 1-55.
Jack, M. S., Heimberg, R. G., & Mennin, D. S. (1999). Situational panic
attacks: Impact on distress and impairment among patients with social
phobia. Depression and Anxiety, 10, 112-118.
Khawaja, N. G., & Oei, T. P. S. (1998). Catastrophic cognitions in panic
disorderwith the without agoraphobia. Clinical Psychology Review,
18(3), 341-365.
Khawaja, N. G., Oei, T. P., & Baglioni, A. J. (1994). Modification of the
Catastrophic Cognitions Questionnaire (CCQ-M) for normal and patients:
Exploratory and LISREL analyses. Journal of Psychopathology and
Behavioural Assessment, 16(4), 325-342.
Khawaja, N. G., & Oei, T. P. S. (1992). Development of a Catastrophic
Cognitions Questionnaire. Journal of Anxiety Disorders, 6, 305-318.
Kleinknecht, R. A. (1988). Specificity and psychosocial correlates of
blood/injuryfear and fainting. Behavior Research and Therapy, 26, 303-
309.
Anxiety Sensitivity 239
Lang, A. J., Kennedy, C. M., & Stein, M. B. (2002). Anxiety sensitivity and
PTSD Among Female Victims of Intimate Partner Violence. Depression
and Anxiety, 16, 77-83.
Lehman, C. L., & Cheung, R. C. (2002). Depression, anxiety, post-traumatic
stress, and alcohol-related problems among veterans with chronic
hepatitis C. The American Journal of Gastroenterology, 97 (10), 2640-
2646.
Lewinsohn, P. M., Gotlib, I. H., Lewinsohn, M., Seeley, J. R., & Allen, N. B.
(1998). Gender differences in anxiety disorders and anxiety symptoms
in adolescents. Journal of Abnormal Psychology, 107(1), 109-117.
Lilienfeld, S. O. (1997). The relation of anxiety sensitivity and fear of fear to
higher-and lower-order personality dimensions: Implications for the
etiology of panic attacks. Journal of Abnormal Psychology, 106, 539-
544.
Lilienfeld, S. O., Jacob, R. G., & Turner, S. M. (1989). Comment of
Holloway and McNally’s (1987) “Effects of anxiety sensitivity on the
response to hyperventilation”. Journal of Abnormal Psychology, 98, 100-
102.
Lilienfeld, S. O., Turner, S. M., & Jacob, R. G. (1993). Anxiety sensitivity:
An examination of theoretical and methodological issues. Advances in
Behavior Research and Therapy, 15, 147-183.
Lumley, M. A., & Melamed, B. G. (1992). Blood phobics and nonphobics:
Psychological differences and affect during exposure. Behavior
Research and Therapy, 30 (5), 425-434.
Anxiety Sensitivity 240
Maidenberg, E., Chen, E., Craske, M., Bohn, P., & Bystritsky, A. (1996).
Specificty of attentional bias in panic disorder and social phobia. Journal
of Anxiety Disorders, 10, 529-541.
Maller, R. G., & Reiss, S. (1992). A behavioral validation of the Anxiety
Sensitivity Index. Journal of Anxiety Disorders, 1, 265-272.
Marks, I. M., & Mathews, A. M. (1979). Case histories and shorter
communications. Behavior Research and Therapy, 17, 263-267.
Mathews, A. M., Gelder, M. G., & Johnston, D. W. (1981). Agoraphobia:
Nature and Treatment. New York : The Guilford Press.
Mavissakalian, M. R., Perel, J. M., Talbott-Green, M., & Sloan, C. (1998).
Gauging the Effectiveness of Extended Imipramine Treatment for Panic
Disorder with Agoraphobia. Biological Psychiatry, 43, 848-854.
McNally, R. J. (2002). Anxiety sensitivity and panic disorder. Society of
Biological Psychiatry, 52, 938-946.
McNally, R. J. (1999). Theoretical approaches to the fear of anxiety. In S.
Taylor (Ed.), Anxiety Sensitivity: Theory, Research, and Treatment of the
Fear of Anxiety (pp. 3-16). New Jersey: Lawrence Erlbaum Associates.
McNally, R. J. (1994). Panic disorder: A critical analysis. New York: The
Guilford Press.
McNally, R. J. (1989). Is anxiety sensitivity distinguishable from trait anxiety?
Reply to Lilienfeld, Jacob, and Turner (1989). Journal of Abnormal
Psychology, 98, 193-194.
McNally, R. J., & Lorenz, M. (1987). Anxiety sensitivity in agoraphobics.
Journal of Behavior Research and Therapy, 18 (1), 3-11.
Anxiety Sensitivity 241
McWilliams, L. A., & Cox, B. J. (2001). How distinct is anxiety sensitivity
from trait anxiety? A re-examination from a multidimensional
perspective. Personality and Individual Differences, 38, 813-818.
McWilliams, L. A., Stewart, S. H., & MacPherson, P. S. R. (2000). Does the
social concerns component of the Anxiety Sensitivity Index belong to the
domain of anxiety sensitivity or the domain of negative evaluation
sensitivity? Behavior Research and Therapy, 38, 985-992.
Mellman, T. A., & David, G. C. (1985). Combat-related flashbacks in
Posttraumatic stress disorder. Phenomenology and similarity to panic
attacks. Journal of Clinical Psychiatry, 46, 379-382.
Norton, P. J., De Coteau, T. J., Hope, D. A., & Anderson, J. (in press). The
factor structure of the Anxiety Sensitivity Index among Northern Plains
Native Americans. Behavior Research and Therapy.
Oei, T. P., Llamas,M., & Evans, L. (1997). Does concurrent drug intake affect
the long-term outcome of group cognitive behaviour therapy I panic
disorder with or without agoraphobia? Behavior Research and Therapy,
35, 851-857.
Oei, T. P., Moylan, A., & Evans, L. (1991). Validity and clinical utility of the
Fear Questionnaire for anxiety-disorder patients. Psychological
Assessment, 3, 391-397.
Otto, M. W., Pollack, M. H., Fava, M., Uccello, R., & Rosenbaum, J. F.
(1995). Elevated Anxiety Sensitivity Index scores in patients with major
depression: Correlates and changes with antidepressant treatment.
Journal of Anxiety Disorders, 9 (2), 117-123.
Anxiety Sensitivity 242
Otto, M. W., Pollack, M. H., Meltzer-Brody, S., & Rosenbaum, J. F. (1992).
Cognitive-Behavioral Therapy for Benzodiazepine Discontinuation in
Panic Disorder Patients. Psychopharmacology Bulletin, 28 (2), 123-130.
Penava, S. J., Otto, M. W., Maki, K. M., & Pollack, M. H. (1998). Rate of
improvement during cognitive-behavioral group treatment for panic
disorder. Behavior Research and Therapy, 36, 665-673.
Peterson, R. A., & Heilbronner, R. L. (1987). The Anxiety Sensitivity Index:
Construct validity and factor analytic structure. Journal of Anxiety
Disorders, 1(2), 117-121.
Peterson, R. A., & Plehn, K. (1999). Measuring Anxiety Sensitivity. In S.
Taylor (Ed.), Anxiety Sensitivity: Theory, Research and Treatment of the
Fear of Anxiety (pp. 61-82). New Jersey: Lawrence Erlbaum
Associates.
Plehn, K., & Peterson, R. A. (2002). Anxiety sensitivity as a predictor of the
development of panic symptoms, panic attacks, and panic disorder: A
prospective study. Journal of Anxiety Disorders, 16, 455-474.
Rapee, R. M., Ancis, J. R., & Barlow, D. H. (1988). Emotional reactions to
physiological sensations: panic disorder patients and non-clinical Ss.
Behavior Research and Therapy, 26 (3), 265-269.
Rapee, R. M., Brown, T. A., Antony, M. M., & Barlow, D. H. (1992). Response
to hyperventilation and inhalation of 5.5% carbon dioxide-enriched air across the DSM-III-R anxiety disorders. Journal of Abnormal Psychology, 101, 538–552.
Anxiety Sensitivity 243
Rapee, R. M., & Medoro, L. (1994). Fear of psychical sensations and trait
anxiety as mediators of the response to hyperventilation in nonclincial
subjects. Journal of Abnormal Psychology, 103 (4), 693-699.
Razan, G. (1961). The observable unconscious and the inferable conscious in
current Soviet psychophysiology: Interoceptive conditioning, semantic
conditioning, and the orienting reflex. Psychological Review, 68, 81-147.
Reiss, S. (1991). Expectancy theory of fear, anxiety, and panic. Clinical
Psychology Review, 11, 141-153.
Reiss, S., & Havercamp, S. (1996). The sensitivity theory of motivation:
Implications for psychopathology. Behavior Research and Therapy, 34,
621-632.
Reiss, S., & McNally, R. J. (1985). The expectancy model of fear. In S.
Reiss., & R. R. Bootzin (Eds.), Theoretical issues in Behaviour Therapy
(pp. 107-121). NY: Academic Press.
Reiss, S., Peterson, R. A., Gursky, D. M., & McNally, R. J. (1986). Anxiety
sensitivity, anxiety frequency and prediction of fearfulness. Behavior
Research and Therapy, 24, 1-8.
Sandin, B., Chorot, P., & McNally, R. J. (1996). Anxiety Sensitivity Index:
Normative data and its differentiation from trait anxiety. Behavior
Research and Therapy, 39, 213-219.
Schmidt, N. B., & Joiner, T. E. (2002). Structure of the Anxiety Sensitivity
Index psychometrics and factor structure in a community sample.
Journal of Anxiety Disorders, 16(1), 33-49.
Anxiety Sensitivity 244
Schmidt, N. B., Lerew, D. R., & Jackson, R. J. (1997). The role of anxiety
sensitivity in the pathogenesis of panic: Prospective evaluation of
spontaneous panic attacks during acute stress. Journal of Abnormal
Psychology, 106(3), 355-364.
Schmidt, N. B., Lerew, D. R., & Joiner, T. E. (1998). Anxiety sensitivity and
the pathogenesis of anxiety and depression: Evidence for symptom
specificity. Behavior Research and Therapy, 36, 165-177.
Schmidt, N. B., & Telch, M. J. (1994). Role of fear of fear and safety
information in moderating the effects of voluntary hyperventilation.
Behavior Therapy, 25, 197-208.
Scott, E. L., Heimberg, R. G., & Jack, M. S. (2000). Anxiety sensitivity in
social phobia: Comparison between social phobics with and without
panic attacks. Depression and Anxiety, 12, 189-192.
Sherer, M., Maddux, J. E., Mercadante, B., Prentice-Dunn, S., Jacobs, B., &
Rogers, R. W. (1982). The self-efficacy scale: Construction and
Validation. Psychological Reports, 51, 663-671.
Shear, M. K., Pilkonis, P. A., Cloitre, M., & Leon, A. (1994). Cognitive
Behavioral Treatment Compared With Nonprescriptive Treatment of
Panic Disorder. Archives of General Psychiatry, 51, 395-401.
Shostak, B. B., & Peterson, R. A. (1990). Effects of anxiety sensitivity on
emotional response to a stress task. Behavior Research and Therapy, 28,
513-521.
Anxiety Sensitivity 245
Spiegel, D. A., Bruce, T. J., Gregg, S. F., & Nuzzarello, A. (1994). Does
Cognitive Behavior Therapy Assist Slow-Taper Alprazolam
Discontinuation in Panic Disorder? American Journal of Psychiatry, 151,
876-881.
Stein, D. J., & Bower, C. (1997). A neuro-evolutionary approach to the anxiety
disorders. Journal of Anxiety Disorders, 11, 4, 409-429.
Stewart, S. H., Samoluk, S. B., & MacDonald, A. B. (1999). Anxiety
sensitivity and substance use and abuse. In S. Taylor (Ed.), Anxiety
Sensitivity: Theory, Research, and Treatment of the Fear of Anxiety (pp.
287-319). New Jersey: Lawrence Erlbaum Associates.
Stewart, S. H., Taylor, S., & Baker, J. M. (1997). Gender differences in
dimensions of anxiety sensitivity. Journal of Anxiety Disorders, 11(2),
179-200.
Tabachnick, B. G., & Fidell, L. S. (2001). Using Multivariate Statistics (4th
ed.). NY: Harper Collins Publishers, Inc.
Taylor, S. (2003). Anxiety Sensitivity and its implications for understanding
and treating PTSD. Journal of Cognitive Psychotherapy, 17 (2), 179-186.
Taylor, S. (1996). Nature and measurement of anxiety sensitivity: Reply to
Lilienfeld, Turner, and Jacob (1996). Journal of Anxiety Disorders, 5,
425-451.
Taylor, S. (1995). Anxiety sensitivity: Theoretical perspectives and recent
findings. Behavior Research and Therapy, 33(3), 243-258.
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(5), 463-483.
Anxiety Sensitivity 246
Taylor, S., Koch, W. J., & Crockett, D. J. (1991). Anxiety sensitivity, trait
anxiety, and the anxiety disorders. Journal of Anxiety Disorders, 5, 293-
311.
Taylor, S., Koch, W. J., & McNally, R. J. (1992). How does anxiety
sensitivity vary across the anxiety disorders? Journal of Anxiety
Disorders, 6, 249-259.
Taylor, S., Koch, W. J., McNally, R. J., & Crockett, D. J. (1992).
Conceptualizations of anxiety sensitivity. Psychological Assessment, 4,
245-250.
Taylor, S., Koch, W. J., Woody, S., & McLean, P. (1996). Anxiety sensitivity
and depression: How are they related? Journal of Abnormal Psychology,
105, 474-479.
Taylor, S., Rabian, B., & Fedoroff, I. C. (1999). Anxiety Sensitivity: Progress,
Prospects, and Challenges. In S. Taylor (Ed.), Anxiety Sensitivity:
Theory, Research and Treatment of the Fear of Anxiety (pp. 339-353).
New Jersey: Lawrence Erlbaum Associates.
Telch, M. J., Lucas, J. A., Schmidt, N. B., Hanna, H. H., Jaimez, T., & Lucas,
R. A. (1993). Group cognitive-behavioral treatment of panic disorder.
Behavior Research and Therapy, 31, 279-87.
Telch, M. J., Shermis, M. D., & Lucas, J. A. (1989). Anxiety Sensitivity:
Unitary personality trait or domain-specific appraisal? Journal of Anxiety
Disorders, 3, 25-32.
Wardle, J., Ahmad, T., & Hayward, P. (1990). Anxiety sensitivity in
agoraphobia. Journal of Anxiety Disorders, 4, 325-333.
Anxiety Sensitivity 247
Wells, A. (1997). Cognitive therapy of anxiety disorders: A practice manual
andconceptual guide. NY: John Wiley & Sons.
Westra, H. A., Stewart, S. H., & Conrad, B. E. (2002). Naturalistic manner of
benzodiazepine use and cognitive behavioral therapy outcome in panic
disorder with agoraphobia. Anxiety Disorders, 16, 233-246.
Wolman, B. B., & Stricker, G. (Eds.). (1994). Anxiety and related disorders: A
handbook. New York: John Wiley & Sons, Inc.
Zeitlin, S. B., & McNally, R. J. (1993). Alexithymia and anxiety sensitivity in
panic disorder and obsessive- compulsive disorder. American Journal of
Psychiatry, 150, 658-660.
Zinbarg, R. E., Barlow, D. H., & Brown, T. A. (1997). Hierarchical structure
and general factor saturation of the Anxiety Sensitivity Index: Evidence
and implications. Psychological Assessment, 9(3), 277-284.
Zinbarg, R. E., Mohlman, J., & Hong, N. N. (1999). Dimensions of anxiety
sensitivity. In S. Taylor (Ed.), Anxiety Sensitivity: Theory,
Research and Treatment of the Fear of Anxiety (pp. 83-114). New
Jersey: Lawrence Erlbaum Associates.
Zung, W. W. K. (1965). A self-rating depression scale. Archives of General
Psychiatry, 12, 63-70.
Zvolensky, M. J., Arrindell, W. A., Taylor, S., Bouvard, M., Cox, B. J.,
Stewart, S. H., Sandin, B., Cardenas, S. J., & Eifert, G. H. (2003).
Anxiety sensitivity in six countries. Behavior Research and Therapy,
41(7), 841-859.
Anxiety Sensitivity 248
Anxiety Sensitivity 249
Appendix A
Anxiety Sensitivity Index
Anxiety Sensitivity 250
Anxiety Sensitivity 251
Anxiety Sensitivity Index The items listed below reveal fears and worries, which we sometimes experience. Please read each item carefully and using the scale given below, rate it by circling the appropriate answer.
Very A little Some Much Very Little Much
1. It is important for me not to appear nervous 0 1 2 3 4 2. When I cannot keep my mind on a task, I worry that I might be going crazy 0 1 2 3 4 3. It scares me when I feel “shaky” (trembling) 0 1 2 3 4 4. It scares me when I feel faint 0 1 2 3 4 5. It is important to me to stay in control of my emotions 0 1 2 3 4 6. It scares me when my heart beats rapidly 0 1 2 3 4 7. It embarrasses me when my stomach growls 0 1 2 3 4 8. It scares me when I am nauseous 0 1 2 3 4 9. When my heart is beating rapidly, I worry that I might be having a heart attack 0 1 2 3 4 10. It scares me when I become short of breath 0 1 2 3 4 11. When my stomach is upset, I worry that I might be seriously ill 0 1 2 3 4 12. It scares me when I am unable to keep my mind on a task 0 1 2 3 4 13. Other people usually notice when I feel shaky 0 1 2 3 4 14. Unusual body sensations scare me 0 1 2 3 4 15. When I am nervous, I worry that I might be mentally ill 0 1 2 3 4 16. It scares me when I am nervous 0 1 2 3 4
Anxiety Sensitivity 252
Anxiety Sensitivity 253
Appendix B
Anxiety Sensitivity Index – Revised
Anxiety Sensitivity 254
Anxiety Sensitivity 255
Anxiety Sensitivity Index - Revised The items listed below reveal fears and worries, which we sometimes experience. Please read each item carefully and using the scale given below, rate it by circling the appropriate answer.
Very A little Some Much Very Little Much
1. When I feel like I’m not getting enough air, I get scared that I might suffocate 0 1 2 3 4 2. Smothering sensations scare me 0 1 2 3 4 3. It scares me when I become short of breath 0 1 2 3 4 4. When my chest feels tight, I get scared that I won’t be able to breath properly 0 1 2 3 4 5. It scares me when I feel faint 0 1 2 3 4 6. When my throat feels tight, I worry that I could choke to death 0 1 2 3 4 7. It scares me when my heart beats rapidly 0 1 2 3 4 8. When my breathing become irregular, I fear that something bad will happen 0 1 2 3 4 9. It scares me when I feel “shaky” (trembling) 0 1 2 3 4 10. When I have trouble swallowing, I worry that I could choke 0 1 2 3 4 11. It frightens me when my surroundings seem strange or unreal 0 1 2 3 4 12. It scares me when my body feels strange or different in some way 0 1 2 3 4 13. It is important for me not to appear nervous 0 1 2 3 4 14. I believe it would be awful to vomit in public 0 1 2 3 4 15. I think that it would be horrible for me to faint in public 0 1 2 3 4 16. I worry that other people will notice my anxiety 0 1 2 3 4 17. When I tremble in the presence of others I fear what people might think of me 0 1 2 3 4 18. When I begin to sweat in a social situation, I fear that people will think negatively of me 0 1 2 3 4 19. It scares me when I blush in front of people 0 1 2 3 4 20. When I feel a strong pain in my stomach, I worry it could be cancer 0 1 2 3 4 21. When my head is pounding, I worry I could have a stroke 0 1 2 3 4 22. When my heart is beating rapidly, I worry that I might be having a heart attack 0 1 2 3 4 23. When my face feels numb, I worry that I might be having a stroke 0 1 2 3 4 24. When I feel a pain in my chest, I worry that I’m going to have a heart attack 0 1 2 3 4 25. When I feel dizzy, I worry that there is something wrong with my brain 0 1 2 3 4 26. When my stomach is upset, I worry that I might be seriously ill 0 1 2 3 4
Anxiety Sensitivity 256
Very A little Some Much Very Little Much
27. When I notice my heart skipping a beat, I worry that something is seriously wrong with me 0 1 2 3 4 28. When I get diarrhoea, I worry that I might have something wrong with me 0 1 2 3 4 29. It scares me when I am nauseous 0 1 2 3 4 30. It scares me when I feel tingling or prickling sensation in my hands 0 1 2 3 4 31. When I feel “spacey” or spaced out I worry that I may be mentally ill 0 1 2 3 4 32. When my thoughts seem to speed up, I worry that I might be going crazy 0 1 2 3 4 33. When I have trouble thinking clearly, I worry that there is something wrong with me 0 1 2 3 4 34. When I cannot keep my mind on a task, I worry that I might be going crazy 0 1 2 3 4 35. It scares me when I am unable to keep my mind on a task 0 1 2 3 4 36. When my mind goes blank I worry that there is something terribly wrong with me 0 1 2 3 4
Anxiety Sensitivity 257
Appendix C
Items that form the 21-item Anxiety Sensitivity Index
Anxiety Sensitivity 258
Anxiety Sensitivity 259
Items that form the 21-item Anxiety Sensitivity Index The items listed below reveal fears and worries, which we sometimes experience. Please read each item carefully and using the scale given below, rate it by circling the appropriate answer.
Very A little Some Much Very Little Much
1. When I feel like I’m not getting enough air, I get scared that I might suffocate 0 1 2 3 4 2. Smothering sensations scare me 0 1 2 3 4 3. It scares me when I become short of breath 0 1 2 3 4 4. When my chest feels tight, I get scared that I won’t be able to breath properly 0 1 2 3 4 5. It is important for me not to appear nervous 0 1 2 3 4 6. I believe it would be awful to vomit in public 0 1 2 3 4 7. I worry that other people will notice my anxiety 0 1 2 3 4 8. When I tremble in the presence of others I fear what people might think of me 0 1 2 3 4 9. When I begin to sweat in a social situation, I fear that people will think negatively of me 0 1 2 3 4 10. It scares me when I blush in front of people 0 1 2 3 4 11. When my head is pounding, I worry I could have a stroke 0 1 2 3 4 12. When my heart is beating rapidly, I worry that I might be having a heart attack 0 1 2 3 4 13. When my face feels numb, I worry that I might be having a stroke 0 1 2 3 4 14. When I feel a pain in my chest, I worry that I’m going to have a heart attack 0 1 2 3 4 15. When I notice my heart skipping a beat, I worry that something is seriously wrong with me 0 1 2 3 4 16. When I feel “spacey” or spaced out I worry that I may be mentally ill 0 1 2 3 4 17. When my thoughts seem to speed up, I worry that I might be going crazy 0 1 2 3 4 18. When I have trouble thinking clearly, I worry that there is something wrong with me 0 1 2 3 4 19. When I cannot keep my mind on a task, I worry that I might be going crazy 0 1 2 3 4 20. It scares me when I am unable to keep my mind on a task 0 1 2 3 4 21. When my mind goes blank I worry that there is something terribly wrong with me 0 1 2 3 4
Anxiety Sensitivity 260
Anxiety Sensitivity 261
Appendix D
Questionnaire Battery
Anxiety Sensitivity 262
Anxiety Sensitivity 263
CONSENT FORM
Chief Investigator Kerry Ann Armstrong School of Psychology and Counselling (Ph. 3864-4686) Associate Investigator Dr Nigar Khawaja School of Psychology and Counselling (Ph. 3864-4757) An Investigation of Anxiety Sensitivity and the role it plays in the Development of Anxiety and Anxiety Disorders You are invited to participate in a research project which aims to advance our understanding Anxiety Sensitivity and the role it plays in the development and maintenance of anxiety and anxiety disorders. If you agree to participate you will be asked to complete a series of questionnaires. These questionnaires will consist of questions regarding how you felt and what you thought at a time when you possibly experienced an episode of anxiety. Although your participation may have no direct benefit to you, it is hoped that the findings will be useful in advancing our understanding Anxiety Sensitivity and the factors that contribute to the development and maintenance of this construct as well as add to the current research on this topic. We anticipate that the questionnaire package take approximately 30 minutes to complete. There are no right or wrong answers when completing the questionnaire material. Additionally there is no time limit for responses; therefore you can work at a pace that is comfortable to yourself. However, do not spend too much time on any one question and please ensure that all questions are answered before returning your responses. Your participation in this project is entirely voluntary and you are able to discontinue your involvement in this study at anytime without explanation or penalty. Your confidentiality will be preserved and no identifying information will be made public, as only your student number will be recorded on the questionnaire material. Copies of the research report will be made available for interested persons upon request. This study is being conducted as part of a PhD project by Kerry Ann Armstrong, and is supervised by Dr Nigar Khawaja. You may contact Kerry Armstrong or Dr Khawaja during the study if any matter of concern arises on the phone number given above. You may also contact the secretary of the QUT Research Ethics Committee on 3864-2902 if you have any further concerns about the ethical conduct of this research.
Anxiety Sensitivity 264
I, _________________________________________________ (name) consent to participate in the research described above. I have read the information provided above and I have had the opportunity to ask questions. I also understand that I am able to withdraw from this study at any time without explanation, and any information I provide is treated as confidential. I am aged 18 years or over at the time of this study. _____________________________________________________ Signature of participant Date Kerry Ann Armstrong Researcher’s Name Signature of Researcher Date
Anxiety Sensitivity 265
Questionnaire Package
Confidential
Today’s Date: _____________________________________________
Anxiety Sensitivity 266
Background - Personal
1. Your gender? (Circle one number only)
Female ................................................................................ 1 Male .................................................................................... 2
2. Your age in years? ..............................................................
3. Your country of birth? (Circle one number only)
Australia .............................................................................. 1
Other (please specify) ......................................................... 2
4. Are you Aboriginal or Torres Strait Islander? (Circle one number only)
No ....................................................................................... 0 Yes, Aboriginal ................................................................... 1 Yes, Torres Straight Islander .............................................. 2
5. In the home do you regularly speak a language other than English? (Circle one number only) No ....................................................................................... 0
Yes (please specify which language) .................................
............................................................................................ 1
6. What is Your Highest Level of Education Attained? (Circle one number only)
High School.......................................................................... 1 Tafe Certificate..................................................................... 2 Tafe Diploma / Associate Diploma....................................... 3 University Undergraduate.................................................... 4 University Postgraduate....................................................... 5 Other (please specify).......................................................... 6
Anxiety Sensitivity 267
Anxiety Sensitivity Index and Anxiety Sensitivity Index - Revised The items listed below reveal fears and worries, which we sometimes experience. Please read each item carefully and using the scale given below, rate it by circling the appropriate answer.
Very A little Some Much Very Little Much
1. When I feel like I’m not getting enough air, I get scared that I might suffocate 0 1 2 3 4 2. Smothering sensations scare me 0 1 2 3 4 3. It scares me when I become short of breath 0 1 2 3 4 4. When my chest feels tight, I get scared that I won’t be able to breath properly 0 1 2 3 4 5. It scares me when I feel faint 0 1 2 3 4 6. When my throat feels tight, I worry that I could choke to death 0 1 2 3 4 7. It scares me when my heart beats rapidly 0 1 2 3 4 8. When my breathing become irregular, I fear that something bad will happen 0 1 2 3 4 9. It scares me when I feel “shaky” (trembling) 0 1 2 3 4 10. When I have trouble swallowing, I worry that I could choke 0 1 2 3 4 11. It frightens me when my surroundings seem strange or unreal 0 1 2 3 4 12. It scares me when my body feels strange or different in some way 0 1 2 3 4 13. It is important for me not to appear nervous 0 1 2 3 4 14. I believe it would be awful to vomit in public 0 1 2 3 4 15. I think that it would be horrible for me to faint in public 0 1 2 3 4 16. I worry that other people will notice my anxiety 0 1 2 3 4 17. When I tremble in the presence of others I fear what people might think of me 0 1 2 3 4 18. When I begin to sweat in a social situation, I fear that people will think negatively of me 0 1 2 3 4 19. It scares me when I blush in front of people 0 1 2 3 4 20. When I feel a strong pain in my stomach, I worry it could be cancer 0 1 2 3 4 21. When my head is pounding, I worry I could have a stroke 0 1 2 3 4 22. When my heart is beating rapidly, I worry that I might be having a heart attack 0 1 2 3 4 23. When my face feels numb, I worry that I might be having a stroke 0 1 2 3 4 24. When I feel a pain in my chest, I worry that I’m going to have a heart attack 0 1 2 3 4 25. When I feel dizzy, I worry that there is something wrong with my brain 0 1 2 3 4 26. When my stomach is upset, I worry that I might be seriously ill 0 1 2 3 4
Anxiety Sensitivity 268
Very A little Some Much Very Little Much
27. When I notice my heart skipping a beat, I worry that something is seriously wrong with me 0 1 2 3 4 28. When I get diarrhea, I worry that I might have something wrong with me 0 1 2 3 4 29. It scares me when I am nauseous 0 1 2 3 4 30. It scares me when I feel tingling or prickling sensation in my hands 0 1 2 3 4 31. When I feel “spacey” or spaced out I worry that I may be mentally ill 0 1 2 3 4 32. When my thoughts seem to speed up, I worry that I might be going crazy 0 1 2 3 4 33. When I have trouble thinking clearly, I worry that there is something wrong with me 0 1 2 3 4 34. When I cannot keep my mind on a task, I worry that I might be going crazy 0 1 2 3 4 35. It scares me when I am unable to keep my mind on a task 0 1 2 3 4 36. When my mind goes blank I worry that there is something terribly wrong with me 0 1 2 3 4 37. It is important to me to stay in control of my emotions 0 1 2 3 4 38. It embarrasses me when my stomach growls 0 1 2 3 4 39. Other people usually notice when I feel shaky 0 1 2 3 4 40. Unusual body sensations scare me 0 1 2 3 4 41. When I am nervous, I worry that I might be mentally ill 0 1 2 3 4 42. It scares me when I am nervous 0 1 2 3 4
Anxiety Sensitivity 269
Beck Anxiety Inventory
Listed below are a variety of symptoms. Please read each symptom and by using the given scale rate how much you have been bothered by it over the past week. Read each item carefully and encircle the appropriate answer.
Not at all A Little Quite Severely, I Could Barely Stand It
1. Numbness or tingling 0 1 2 3
2. Feeling hot 0 1 2 3
3. Wobbliness in legs 0 1 2 3
4. Unable to relax 0 1 2 3
5. Fear of worst happening 0 1 2 3
6. Dizzy or lightheaded 0 1 2 3
7. Heart pounding or racing 0 1 2 3
8. Unsteady 0 1 2 3
9. Terrified 0 1 2 3
10. Nervous 0 1 2 3
11. Feeling of choking 0 1 2 3
12. Hands trembling 0 1 2 3
13. Shaky 0 1 2 3
14. Fear of losing control 0 1 2 3
15. Difficulty in breathing 0 1 2 3
16. Fear if dying 0 1 2 3
17. Scared 0 1 2 3
18. Indigestion or discomfort 0 1 2 3
19. Faint 0 1 2 3
20. Face flushed 0 1 2 3
21. Sweating not due to heat 0 1 2 3
Anxiety Sensitivity 270
COPE Questionnaire
We are interested in how people respond when they confront difficult or stressful events in their lives. There are a lot of ways to try and deal with stress. This questionnaire asks you to indicate whet you generally do and feel, when you experience stressful events. Obviously, different events bring out somewhat different responses, but think about what you usually do when you are under a lot of stress. Please respond to each of the following items by using the given scale. Choose your answers thoughtfully, and make your answers as true FOR YOU as you can. Please answer every item. There are no “right” or “wrong” answers, so choose the most accurate answer for YOU – not what you think “most people” would say or do. Indicate what YOU usually do when YOU experience a stressful event.
I usually don’t I usually do I usually do I usually don’t do this do this a this a medium do this a at all little bit amount lot
1. I try to grow as a person as a result of the experience 1 2 3 4 2. I turn to work or other substitute activities to take my mind off things 1 2 3 4 3. I get upset and let my emotions out 1 2 3 4 4. I try to get advice from someone about what to do 1 2 3 4 5. I concentrate my efforts on doing something about it 1 2 3 4 6. I say to myself “this isn’t real” 1 2 3 4 7. I admit to myself that I can’t deal with it, and quit trying 1 2 3 4 8. I restrain myself from doing anything too quickly 1 2 3 4 9. I discuss my feelings with someone 1 2 3 4 10. I get used to the idea that it happened 1 2 3 4 11. I talk to someone to find out more about the situation 1 2 3 4 12. I keep myself from getting distracted by other thoughts or activities 1 2 3 4 13. I daydream about things other than this 1 2 3 4 14. I get upset, and I am really aware of it 1 2 3 4 15. I seek God’s help 1 2 3 4 16. I make a plan of action 1 2 3 4 17. I accept that this has happened and that it can’t be changed 1 2 3 4 18. I put off doing anything until the situation permits 1 2 3 4 19. I try to get emotional support from friends and relatives 1 2 3 4 20. I just give up trying to reach my goal 1 2 3 4 21. I take additional action to try and get rid of the problem 1 2 3 4 22. I refuse to believe that it has happened 1 2 3 4
Anxiety Sensitivity 271
I usually don’t I usually do I usually do I usually don’t do this do this a this a medium do this a at all little bit amount lot
23. I let my feelings out 1 2 3 4 24. I try to see it in a different light, to make it seem more positive 1 2 3 4 25. I talk to someone who could do something concrete about the problem 1 2 3 4 26. I sleep more than usual 1 2 3 4 27. I try to come up with a strategy about what to do 1 2 3 4 28. I focus on dealing with this problem, and if necessary let other things slide a little 1 2 3 4 29. I get sympathy and understanding from someone 1 2 3 4 30. I give up the attempt to get what I want 1 2 3 4 31. I look for something good in what is happening 1 2 3 4 32. I think about how I might best handle the problem 1 2 3 4 33. I make sure not to make matters worse by acting too soon 1 2 3 4 34. I try hard to prevent other things from interfering 1 2 3 4 35. I go to movies or watch TV, to think about it less 1 2 3 4 36. I accept the reality of the fact that it happened 1 2 3 4 37. I ask people who have had similar experiences 1 2 3 4 38. I feel emotional distress and express those feelings 1 2 3 4 39. I force myself to wait for the right time to do something 1 2 3 4 40. I reduce the amount of effort I’m putting into solving the problem 1 2 3 4 41. I learn to live with it 1 2 3 4 42. I put aside other activities in order to concentrate on this 1 2 3 4 43. I think hard about what steps to take 1 2 3 4 44. I act as though it hasn’t ever happened 1 2 3 4 45. I do what has to be done, one step at a time 1 2 3 4
46. I learn something from the experience 1 2 3 4 47. I take direct action to get around the problem 1 2 3 4
Anxiety Sensitivity 272
DAS S – Stress Sub Scale
Please read each statement and circle a number 0, 1, 2 or 3 which indicates how much the statement applied to you over theThere are no right or wrong answers. Do not spend too much time on any statement.
The rating scale is as follows:
0 Did not apply to me at all; 1 Applied to me to some degree, or some of the time; 2 Applied to me to a considerable degree, or aof time; 3 Applied to me very much, or most of the time
I found it hard to wind down 0 1 2 3 I tended to over-react to situations 0 1 2 3 I felt that I was using a lot of nervous energy 0 1 2 3
I found myself getting agitated 0 1 2 3
I found it difficult to relax 0 1 2 3
I was intolerant of anything that kept me from getting on with what I was doing 0 1 2 3
I felt that I was rather touchy 0 1 2 3
Anxiety Sensitivity 273
Zung - SDS
Please indicate for each of these questions which answer best describes how you have been feeling during the past 4 weeks. Circle the most appropriate response. Circle only one response in each line. Please answer all the questions.
Rarely Some of A good part All or most the time of the time of the time
1. I feel downhearted, blue and sad 1 2 3 4
2. Morning is when I feel best 1 2 3 4
3. I have crying spells or feel like it 1 2 3 4
4. I have trouble sleeping through the night 1 2 3 4
5. I eat as much as I used to 1 2 3 4
6. I enjoy looking at, talking to and being with attractive men/women 1 2 3 4
7. I notice that I am losing weight 1 2 3 4
8. I have trouble with constipation 1 2 3 4
9. My heart beats faster than usual 1 2 3 4
10. I get tired for no reason 1 2 3 4
11. My mind is as clear as it used to be 1 2 3 4
12. I find it easy to do the things I used to 1 2 3 4
13. I am restless and can’t keep still 1 2 3 4
14. I feel hopeful about the future 1 2 3 4
15. I am more irritable than usual 1 2 3 4
16. I find it easy to make a decision 1 2 3 4
17. I feel that I am useful and needed 1 2 3 4
18. My life is pretty full 1 2 3 4
19. I feel that others would be better off if I were dead 1 2 3 4
20. I still enjoy the things I used to 1 2 3 4
Anxiety Sensitivity 274
The Self-Efficacy Scale To what extent do each of the following statements apply to you. Please circle the number that indicates your level of agreement.
Strongly Somewhat Neither Somewhat Strongly Agree Agree Disagree Disagree
1. When I make plans I am certain I can make them work 1 2 3 4 5 2. I feel insecure about my ability to do things 1 2 3 4 5 3. When I set important goals for myself, I rarely achieve them 1 2 3 4 5 4. I give up on things before completing them 1 2 3 4 5 5. If something looks too complicated, I will not even bother to try it 1 2 3 4 5 6. One of my problems is that I cannot get down to work when I should 1 2 3 4 5 7. I avoid facing difficulties 1 2 3 4 5 8. When I have something unpleasant to do, I stick to it until I finish it 1 2 3 4 5 9. I give up easily 1 2 3 4 5 10. When I decide to do something, I go right to work on it 1 2 3 4 5 11. When trying to learn something new, I soon give up if I am not initially successful 1 2 3 4 5 12. When unexpected problems occur, I do not handle them very well 1 2 3 4 5 13. I avoid trying to learn new things when they look too difficult for me 1 2 3 4 5 14. Failure just makes me try harder 1 2 3 4 5 15. If I can’t do a job the first time, I keep trying until I can 1 2 3 4 5 16. I am a self-reliant person 1 2 3 4 5 17. I do not seem capable of dealing with most problems in life 1 2 3 4 5
Anxiety Sensitivity 275
Fear Questionnaire (How You Feel in Certain Situations)
Choose a number from the scale below to show how much you would avoid each of the situations listed because of fear or other unpleasant feelings. Please circle the number opposite each situation.
0 1 2 3 4 5 6 7 8 Would not Slightly Definitely Markedly Always avoid it avoid it avoid it avoid it avoid it
1. Main phobia (fear) you want treated (describe in your own words and rate) 0 1 2 3 4 5 6 7 8 2. Injections or minor surgery 0 1 2 3 4 5 6 7 8 3. Eating or drinking with other people 0 1 2 3 4 5 6 7 8 4. Hospitals 0 1 2 3 4 5 6 7 8 5. Traveling alone by bus or coach 0 1 2 3 4 5 6 7 8 6. Walking alone in busy streets 0 1 2 3 4 5 6 7 8 7. Being watched or stared at 0 1 2 3 4 5 6 7 8 8. Going into crowded shops 0 1 2 3 4 5 6 7 8 9. Talking to people in authority 0 1 2 3 4 5 6 7 8 10. Sight of blood 0 1 2 3 4 5 6 7 8 11. Being criticised 0 1 2 3 4 5 6 7 8 12. Going alone far from home 0 1 2 3 4 5 6 7 8 13. Thought of injury or illness 0 1 2 3 4 5 6 7 8 14. Speaking or acting to an audience 0 1 2 3 4 5 6 7 8 15. Large open spaces 0 1 2 3 4 5 6 7 8 16. Going to the dentist 0 1 2 3 4 5 6 7 8 17. Other situations (describe and rate) 0 1 2 3 4 5 6 7 8
Anxiety Sensitivity 276
Anxiety Sensitivity 277
Appendix E
Outline of the manual based cognitive behavioural therapy
for anxiety employed in the current dissertation
Anxiety Sensitivity 278
Anxiety Sensitivity 279
OUTLINE OF THE MANUAL BASED COGNITIVE BEHAVIOURAL THERAPY FOR ANXIETY EMPLOYED IN THE CURRENT
DISSERTATION Day 1 Schedule
• Philosophy of the Program
• What is Anxiety?
• Main Types of Anxiety Disorders Panic Disorder (with or without Agoraphobia)
• Other Types of Anxiety Disorders Generalised Anxiety Disorder Simple Phobia Social Phobia Obsessive Compulsive Disorder (OCD) Post Traumatic Stress Disorder (PTSD) Differential Diagnosis
• Organic Disorders and Conditions Associated with Panic Diagnostic Criteria for Social Phobia Diagnostic Criteria for Generalised Anxiety Disorder Comorbidity Who Suffers from Panic Disorder? How can Panic Disorder be treated? Well-Established Treatments for Anxiety Disorders Symptom Control
• Understanding Anxiety I: Cognitions
• Understanding Anxiety II: Symptoms
• Understanding Panic Attacks
• Whys and the Why
• Understanding Fear: The Fire Analogy
• The Fire Analogy: Explanation
• Day 1 Homework
Anxiety Sensitivity 280
Day Two Schedule
• The ABC model of emotion • Symptoms Vs Fear Vs Cognition
• Three Phases of a Panic Attack
Phase One Phase Two Phase Three
• Different Techniques for each phase
Techniques for Phase One Techniques for Phase Two Techniques for Phase Three
• Body, Behaviour, Cognition
• Learning to Control Physical Sensations Slow Breathing exercise Relaxation Techniques – SECTION 1
Relaxation Techniques 1 – Progressive Muscular Relaxation (PMR)
Relaxation Techniques 2 – Rag Doll Technique Relaxation Techniques 3 – Relaxation by using Imagery Relaxation Techniques 4 – Autogenic Training When and Where Should Relaxation Strategies be Used?
• Day Two Homework Day Three Schedule
• The Flip-Flop Model
• Dangerous Mind Games that We Play: A Lose-Lose game
• A New Mind Game that I Need to Play: A Win-Win Game
• Day Three Homework Day Four Schedule
• Hyperventilation and Breathing Retraining • Hyperventilation
• Breathing Retraining
• Cognitive Distortions
Anxiety Sensitivity 281
• Questions to Help You Challenge Negative Thinking
• Positive Self-Statements
1. Preparation for the Event or Situation 2. Confronting the Event or Situation 3. Coping with Fear 4. Evaluation of Coping and Rewards for Successful Coping
• Cognitive Rehearsal • Problem Solving
• Day Four Homework
Day Five Schedule
• In Vivo Situational Exposure
• Interoceptive Exposure
• More Advanced Relaxation Techniques
• Day Five Homework
Day Six Schedule
• Changing the Way You Think
• Five Falsity Tests for Beliefs
• The Vertical Arrow Technique
• Distraction Techniques
• Thought Stopping
• Past, Present and Future
• Balance Sheet
• More Advanced Relaxation Techniques Instructions for Quick Isometric Relaxation
• Relaxation by Recall
• Day Six Homework
Anxiety Sensitivity 282
Day Seven Schedule
• Building a Support System
• Sources of Support
• Day Seven Homework
Day Eight Schedule
• Maintenance
• Relapse Prevention
• Day 8 Homework
All Information Contained in the Full Manual (including this Appendix) is
Protected by Copywrite to the Cognitive Behavioural Therapy (CBT) Day
Treatment Program at the Toowong Private Hospital 2000.
Anxiety Sensitivity 283
Appendix F
Outline of the manual based cognitive behavioural therapy for depression employed in the current dissertation
Anxiety Sensitivity 284
Anxiety Sensitivity 285
OUTLINE OF THE MANUAL BASED COGNITIVE BEHAVIOURAL THERAPY FOR DEPRESSION EMPLOYED IN THE CURRENT
DISSERTATION Day 1 Schedule
• Introduction and Ground Rules • Philosophy of the Program
• Cognitive Behavioural Therapy (CBT)
• Spring Cleaning
• Four Levels of Change
• Depression: Types and variations
• Depression is treatable
• Evidence based treatment for depression
• Criteria for Major Depressive Episode (from DSM-IV)
• Diagnostic criteria for Dysthymic Disorder (from DSM-IV)
• The 3 Ingredients of Emotion
• Common Errors regarding Emotion
• “You” and “Events / Others”
• Past events cannot be changed, but our memories of events can be
changed
• Pharmacotherapy
• Medication: Regular Review
• Medication
• Information about Medication
• Some Adverse Effects of Pharmacological Agents
Anxiety Sensitivity 286
• Medication commonly used in the Treatment of Depression and their
side-effects
• Pharmacologic Properties of Antidepressants
• Preparations and Dosage of Antidepressants
• Benzodiazepines used in the Treatment of Depression and Anxiety
Disorders
• Imagery Exercise – Day One
• Homework: Do Not A Do, A Dodo
Day 2 Schedule
• The Behavioural Component: Behavioural Do
• Some Activities are Useful to Combat Depression
• Reasons for Planning an Doing Activities that Combat Depression
• “Cookie Jar” Activities
• Suggestions for Pleasant Events
• Daily Schedule of Activities
• Exercise: Do It
• Daily Exercise Record
• Decision Making about Problems
• Homework Activity
Day 3 Schedule
• Cognitions
• Problems of Social Comparisons
Anxiety Sensitivity 287
• Present Characteristics Vs Future Goals
• Overwhelmed by Responsibility
• Automatic Thoughts (AT)
• The Link between Schemas and Automatic Thoughts
• Cognitive Triad and Schema
• Information from your Environment
• Automatic Thoughts Link Together or Talk to Each Other
• Fighting Automatic Thoughts
• Cognitive Distortions
• Thought that Cause People Problems
• Dangerous Mind Games that We Play: A lose-lose game
• Whys and the Why
• New Mind Games That I Need to Play: The win-win Game
• Thought Processes Vs Thought Content
• Daily Beliefs Monitoring Schedule Daily Record of Dysfunctional Thoughts 1 Example: Daily Record of Dysfunctional Thoughts 1
• Day 3 Homework
Day 4 Schedule
• Cognitive Restructuring Questions
• Counters
• Types of Counters
• Practice Counters
• Thoughts that Cause People Problems
• Mantras
Anxiety Sensitivity 288
Day 5 Schedule
• Five Falsity Tests for Beliefs
• Balance Sheet
• Changing our Memory of Unpleasant Events
• A Process of Change: Yes I Can
Day 6 Schedule
• The Vertical Arrow Technique
• The Automatic Thoughts – Iceberg Analogy
• The Characteristics of the Vertical Arrow
• Counters to Combat Automatic Thoughts
• Process of Combating Automatic Thoughts: Planting a New Tree
Analogy
• Reasons for Turning Counters to Mantras
• Logical Analysis Logical Analysis Worksheet Logical Analysis Worksheet Example 1 Logical Analysis Worksheet Example 2 Day 7 Schedule
• Problem Solving
• Past, Present, and Future
• Balance Sheet
• Building a Support System
• Sources of Support
• The Importance of Understanding Cost
Anxiety Sensitivity 289
Day 8 Schedule
• Relapse Prevention
• A Smorgasbord of Strategies to Life Depression
• Reading List
• Self Help books about Depression and Bipolar Disorder
• Autobiographical Accounts of Depression or Bipolar Disorder
• For your Supporters
• Self help books about Stress and anxiety
• Author Publications
All Information Contained in the Full Manual (including this Appendix) is
Protected by Copywrite to the Cognitive Behavioural Therapy (CBT) Day
Treatment Program at the Toowong Private Hospital 2000.
Anxiety Sensitivity 290
Anxiety Sensitivity 291
Appendix G
Preliminary screening sheet
Anxiety Sensitivity 292
Anxiety Sensitivity 293
PRELIMINARY SCREENING SHEET Date: …………………………………………………………………. Name: …………………………………………………………………. Sex: M F Age: …………… Contact Phone No: (W)…………………………….. (H)…………………………. Address:…………………………………………………………………… …………………………………………………………………….. Q1: Do you have a major physical problem? Y N If yes, what is the problem……………………………………………….. Are you receiving any treatment for it …..……………………..………... Q2: Do you have any other problem (emotional, neurological)? Y N If yes, what is the problem……………………………………………….. Are you receiving any treatment for it .…………………………………... Q3: Have you been hospitalised recently? Y N If yes, what was the problem…………………………………………….. Are you receiving any treatment for it..…………………………………... Q4: Do you drink alcohol? Y N If yes, has alcohol affected your health? Y N your family? Y N work? Y N If yes, are you receiving treatment for it?………………………………..
CONTINUE OVER PAGE
Anxiety Sensitivity 294
Q5: Do you take any social or recreational drugs? Y N
If Yes, which one/s……………………………………………………….
Has taking this drug made you confused? Y N See things? Y N Hear Voices? Y N Have strange ideas? Y N
Any other?………………………………………………………………… Did you have to go to hospital because of taking this drug? Y N
Q6: Are you depressed? Y N If yes, how is your daily life affected? …………………………………… (a) Do you have a loss of appetite Y N (b) Do you have trouble sleeping Y N (c) Do you wake up early in the morning Y N (d) Do you feel worthless Y N (e) Do you feel like life is not worth living Y N
Q7: Do you feel elated (hyper)? Y N If yes, have you been… (a) Over talkative Y N (b) Shouting at people Y N (c) Arguing with people Y N (d) Experiencing thoughts racing through your mind Y N (e) Involved in reckless activity Y N Q8: Do you feel very high and very low at times? Y N Q9: Now I am going to ask you about some unusual experiences which people
sometimes have. (a) Do you hear voices/sounds that others can’t hear? Y N (b) Do you see things that other’s can’t see? Y N (c) Do you smell things that other’s can’t smell? Y N (d) Do you think that you are especially important
in some way and have powers to do things that others can’t do? Y N
Anxiety Sensitivity 295
(e) Do you think that people are taking special notice of you, or talking about you? Y N
(f) Is anybody going out of the way to give you a hard time, or to hurt you Y N
(g) Do you receive special messages from other people, newspapers, radio or T.V? Y N
Notes on Yes responses …………………………………………………………………………………… …………………………………………………………………………………… …………………………………………………………………………………… …………………………………………………………………………………… Q 10: Do senseless thoughts keep on recurring Y N Do you keep on repeating acts such as: (a) washing self/object (s) Y N (b) checking objects Y N (c) counting objects Y N (d) any other? ………………………………………… Y N Q11: Do you ever have a sudden rush of anxiety? Y N If yes, do you at that time feel (a) shaky / trembling Y N (b) sweating Y N (c) out of breath Y N (d) choking Y N (e) detached from surroundings Y N (f) fear of dying Y N (g) fear of going crazy Y N (h) fear of going out of control Y N Q12: Are you anxious when (a) being alone Y N (b) going to crowded shopping centres Y N (c) using public transport Y N (d) using lifts / elevators Y N (e) crossing a busy street Y N (f) any other?...................................................................Y N
Anxiety Sensitivity 296
Anxiety Sensitivity 297
Appendix H
Handbook on the effects and treatment
of anxiety and anxiety disorders
Anxiety Sensitivity 327
Appendix I
Instruction Sheet for Participants
Anxiety Sensitivity 328
Anxiety Sensitivity 329
Instructions for Participants
Thank you for appearing for this study. My name is Kerry Ann Armstrong
and the current project is being conducted for the purposes of my PhD
research, which is supervised by Dr Nigar Khawaja. The study is
concerned with the measurement of anxiety sensitivity and its relationship
to other measures.
As part of this study your task is to complete the questionnaires in the
order they appear in the booklet. As there is no set time limit, you may
work at a speed which you feel comfortable. There are no right or wrong
answers to items in the questionnaires. Your responses will be
completely confidential. When you have completed the questionnaires, I
will initial your ‘Applied Research Experience’ sticker and you are free to
leave. There will be a 20 minute debriefing session after all
questionnaires have been returned for those of you who are interested.
Does anyone have any questions before we start?
If you have any questions about the task or materials as you are
completing the questionnaire items, please raise your hand and I will
attempt to answer your question.
Again, thank you for participating in this study.
Queensland University of Technology School of Psychology and Counselling
Questionnaire Study