A PSYCHOMETRIC AND CLINCIAL INVESTIGATION OF ANXIETY …eprints.qut.edu.au/15931/1/Kerry_Armstrong_Thesis.pdf · Revised in Australian Clinical and Normative Populations. European

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


5.2.1 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


xviii

Chapter 6: Examination of differences within diagnostic categories after cognitive behavioural therapy using the 21-item Anxiety Sensitivity Index


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


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



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


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


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.


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


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


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


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 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


– 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.



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




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 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)


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


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


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).


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).


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


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


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.


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


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


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.


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


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


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


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


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


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.


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


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


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


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


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


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


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


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


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.


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


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


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


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


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


(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.


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


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


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


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


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


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


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


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


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


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


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


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.


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,


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


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


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


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


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


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


(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.


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).


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


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


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


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


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.


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.


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


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


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


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.


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


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


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


(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


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


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


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.



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



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


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


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


(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


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


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 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,


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


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


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


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,


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.


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


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.


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




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


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.


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


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


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).


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


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


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


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


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


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


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


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.


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


(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.


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


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


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


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


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


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


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


hypothesised model for both populations (CFI range .586 to .833; RMSEA range

.95 to .21).


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


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


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)


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.


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.


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


Dimension


----------- 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




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


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


Dimension


Discriminant Measure

I II III IV

Self-Efficacy

.25** .10** .28** .13* .24**

COPE

.03 .02 .00 .04 .03




Cognitive Dyscontrol dimension. ** p < .01, *p < .05.


Table 4.13

Pearson Correlations among the 21-item ASI, Self-Efficacy, and COPE

questionnaires for the nonclinical group


Dimension


Discriminant Measure

I II III IV

Self-Efficacy

.13 .09 .06 .07 .19*

COPE

.10 .12 .09 .05 .01




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).


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.


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.


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


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


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


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.


Chapter Five: Differences in anxiety sensitivity between diagnostic and

nonclinical groups using the 21-item Anxiety Sensitivity Index


5.2 Method 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




5.3.3.6 Analysis of differences for the depression




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



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


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


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


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 (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.


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.


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


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


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


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.


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.


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)






30.52

5.11

12.54

4.46

8.41

8.29

4.10

4.72

3.66

5.94

PTSD (n = 21)






54.52

10.19

17.33

11.76

15.24

22.02

5.10

6.37

7.21

7.35

Depression (n = 57)






18.95

3.06

9.00

1.93

4.98

11.96

3.43

5.39

2.85

4.71

Nonclinical Control (n = 423)






19.08

5.38

8.99

2.13

2.58

11.49

3.97

5.41

3.18

3.52


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


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


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


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


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.


0

2

4

6

8

10

12


Mea

n 21

-item

AS

I 'Fe

ar o

f Res

pira

tory

Sym

ptom

s'

Dim

ensi

on


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


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.


0

2

4

6

8

10

12

14

16

18

20


Mea

n 21

-item

AS

I 'Fe

ar o

f Pub

licly

Obs

erva

ble

Sym

ptom

s'D

imen

sion



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


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.


0

2

4

6

8

10

12

14


Mea

n 21

-item

AS

I 'Fe

ar o

f Car

diov

ascu

lar/S

troke

S

ympt

oms'

Dim

ensi

on



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,


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


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.


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.


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


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


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’




(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’


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


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




Symptoms



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-


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


Cardiovascular/Stroke Symptoms’ and ‘Fear of Cognitive Dyscontrol’





dimensions and the PTSD group, η² = .591. The observed power was strong,

.979.



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


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




Symptoms



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


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.




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’


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


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




Symptoms



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


Observable Anxiety Symptoms’, ‘Fear of Cardiovascular/Stroke Symptoms’ and

‘Fear of Cognitive Dyscontrol’ dimensions as the within-subject variable.



(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.



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


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




Symptoms



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


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


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


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


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)


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.


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


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


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


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


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


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


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


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.


Chapter Six: Examination of differences within diagnostic categories after

cognitive behavioural therapy using the 21-item Anxiety Sensitivity Index


6.2 Method 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




6.3.4.4 Analysis of differences for the PTSD group

before and after CBT for the first-order



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



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


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


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


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


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


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


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


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


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


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.


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)






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)






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)






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)






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


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.


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


Cardiovascular/Stroke Symptoms’ and ‘Fear of Cognitive Dyscontrol’ factors as

the within-subject variable.





dimensions and the Panic disorder group, η² = .553. The observed power was

strong, 1.000.


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.


0

2

4

6

8

10

12

14

16

18

20


Fear of PubliclyObservable Symptoms


Symptoms



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.





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.


0

2

4

6

8

10

12

14




Symptoms



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


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



variable.

Using Wilks’ criterion, a non-significant multivariate effect was observed


(4, 7) = 3.638, p = .066. A profile plot illustrating the means of the 21-item ASI


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




Symptoms



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


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



variable.





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.


0

2

4

6

8

10

12




Symptoms



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


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


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


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


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.


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


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


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.



Chapter Seven: General discussion


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



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


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


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,


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


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.


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


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.


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


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


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 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


dissertation enhance our understanding of anxiety sensitivity and how the 21-

item ASI can be used to improve therapeutic interventions in clinical practice.


References

American Psychiatric Association. (2000). Diagnostic and statistical manual of

mental disorders (4th ed, Text Revision.). Washington, DC: Author.


mental disorders (4th ed.). Washington, DC: Author.


mental disorders (3rd ed, Revised.). Washington, DC: Author.


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.


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.


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.


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.


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


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.


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.



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.


Hegel, M. T., Ravaris, C. L., & Ahles, T. A. (1994). Combined Cognitive-

Behavioral and Time-Limited Alprazolam Treatment of Panic Disorder.


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.


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


Lumley, M. A., & Melamed, B. G. (1992). Blood phobics and nonphobics:

Psychological differences and affect during exposure. Behavior



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.


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.


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.


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.


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.


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.


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.


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.


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.


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.



Appendix A

Anxiety Sensitivity Index



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



Appendix B

Anxiety Sensitivity Index – Revised



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.


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



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


Appendix C

Items that form the 21-item Anxiety Sensitivity Index



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.


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



Appendix D

Questionnaire Battery



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.


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


Questionnaire Package

Confidential

Today’s Date: _____________________________________________


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 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.


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



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


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


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


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


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


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


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


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



Appendix E

Outline of the manual based cognitive behavioural therapy

for anxiety employed in the current dissertation



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


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


• 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


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.


Appendix F

Outline of the manual based cognitive behavioural therapy for depression employed in the current dissertation



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


• 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


• 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


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


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.



Appendix G

Preliminary screening sheet



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


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


(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



Appendix H

Handbook on the effects and treatment

of anxiety and anxiety disorders

halla

This appendix is not available online. Please consult the hardcopy thesis available from the QUT Library


Appendix I

Instruction Sheet for Participants



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

A PSYCHOMETRIC AND CLINCIAL INVESTIGATION OF ANXIETY …eprints.qut.edu.au/15931/1/Kerry_Armstrong_Thesis.pdf · Revised in Australian Clinical and Normative Populations. European

Documents