- Email: [email protected]
Anxiety sensitivity in agoraphobia
fournalofAnJu'edyDisorders,Vol. 4, pp. 325-333, 1990
0887-6185/90 $3.00 + .00 Copyright © 1990 Pergamon Press pie
Printed in the USA. All rights reserved.
Anxiety Sensitivity in Agoraphobia JANE WARDLE,PH.D., TAYSEERAI-IMAD,M.A., PETER HAYWARD, PH.D.
Department of Psychology, Institute of Psychiatry, University of London,UK Abstract--The factor structure of the Anxiety Sensitivity Index was assessed in 166 agoraphobic clients who had applied to a behavioural treatment programme and 120 age and sex-matched normal controls. In both samples a four factor solution emerged as the most useful, and it explained more than 60% of the total variance. The item loadings in the agoraphobicsample revealed a coherent theme with factors reflecting fear of heart and breathing symptoms, fear of loss of mental control, fear of gastrointestinal difficulties, and concern about other people detecting anxiety. The item loadings in the normal sample made less psychological sense. The results are discussed both in the light of the trait model of anxiety sensitivity and with respect to the clinical implications of anxiety sensitivity.
ANXIETY SENSITIVITY IN AGORAPHOBIA Cognitive conceptualizations o f anxiety and panic have increasingly implicated beliefs about anxiety (Goldstein & Chambless, 1978; Reiss & McNally, 1985; Beck & Emery, 1985; Clark, 1988). The concept o f "fear of fear" has been particularly central in the explanation o f the etiology o f agoraphobia and panic disorder and is thought both to motivate avoidance o f situations associated with anxiety and to amplify anxiety once it occurs (Goldstein & Chambless, 1978). Chambless, Caputo, Bright, and Gallagher (1984) developed two psychometric instruments to assess the extent of fear of fear. The "Agoraphobic Cognitions Questionnaire" (ACQ) contains a range of items which tap the supposed cognitions of agoraphobic patients and it has been shown to differentiate agoraphobic and panic patients from other psychiatric groups and normal controis (Chambless & Gracely, 1989). Factor analysis of this scale has revealed two subscales, one relating to thoughts of the physical consequences o f anxiety (e.g., heart attack, stroke) and one to thoughts of the behavioral and social consequences (e.g., acting foolishly, g o i n g crazy). The B o d i l y Sensations Questionnaire (BSQ) addresses the fear of the bodily sensations (e.g. heart palpitations, tingling in the fingertips). This scale also discriminated agoraphobic and panic patients from other groups (Chambless & Gracely, 1989). More recently, Reiss and McNally (1985) have proposed a similar construct, which they call "anxiety sensitivity," described as "an individual differReprint requests to Jane Wardle, Ph.D. Department of Psychology, Institute of Psychiatry, De CrespignyPark. LondonSE5 8AF,UK. 325
326
J. WARDLE, T. AHMAD, AND P. HAYWARD
ence variable consisting of beliefs that the experience of anxiety/fear causes illness, embarrassment or additional anxiety" (1985, p. 2), They argue that higher levels of anxiety sensitivity will increase alertness about the possibility of becoming anxious, increase worry about becoming anxious and increase motivation to avoid anxiety-provoking stimuli. The instrument they have developed to assess it is the "Anxiety Sensitivity Index" (ASI) (Reiss, Peterson, Gursky, & McNally, 1986) a sixteen-item inventory of statements designed to assess fear of, or embarrassment about, anxiety ("It scares me when I feel shaky") and negative health cognitions experienced during the anxious state ("When I notice my heart is beating rapidly, I worry that I might have a heart attack"). Reiss and his colleagues have argued that the ASI is a unidimensional scale on the basis of the finding of a single factor structure in their two normal population samples (Reiss et al., 1986). Regrettably there are no details in the published paper about the method of factor analysis used, the criterion for determining the number of factors, or the amount of variance accounted for by the first factor. Consequently the assertion of a single factor structure is hard to evaluate. In a re-analysis of these data, they show that if the ASI items are entered into a principal components analysis together with items from the Fear Survey Schedule-H, eight of the ASI items cluster together on Factor 3 (Reiss, Peterson, & Gursky, 1988). However, great significance cannot be attached to the finding that the correlations within ASI items are higher than the correlations between ASI items and items from another scale. In an attempt to replicate the single factor structure, Peterson and Heilbronner (1987) administered the ASI to a sample of 122 mildly anxious college students ("anxious during at least 40% of the day"). Using the Kaiser criterion for factor extraction, they found four factors that together accounted for 61% of the variance. Two of the factors were said to be associated with items related to fear of the consequences of anxiety and two with fear of the physical sensations themselves. No further details about item loadings were presented, but a two factor solution was then imposed. This had 11 out of 16 items loading on Factor 2, with most of those also loading in Factor 1. On this basis Peterson and Heilbronner argued, without being very convincing, that their analysis confirmed a "strong single factor." A further factor analytic study, again with data from college students studying psychology, gave less support to the single factor model and tended towards a multifactorial model for anxiety sensitivity (Telch, Shermis, & Lucas, 1989). This analysis again yielded a four factor solution, explaining 54% of the total variance. The authors made a valiant attempt to identify a plausible construct for each factor, but Factors 1 and 2 both included a mixture of items, Factor 3 only two items (appearing nervous and staying in control), and Factor 4 contained the two items about heart symptoms and one about shormess of breath. Because Peterson and Heilbronner (1989) had not presented the item loadings from their factor analysis, it was not possible to compare the factor structures with those obtained from previous studies. However Telch et al.'s (1989) work gave little support for a single factor structure and more evidence for what they termed "domain-specific appraisals of anxiety." The factor analytic studies described above have all recruited their samples
ANXIETY SENSITIVITY IN AGORAPHOBIA
327
from a college student population, with the unavoidable restriction of range of age, social status, and intelligence in these groups. Anxiety levels in such populations would be expected to be low and anxiety sensitivity might be predicted to be similarly rare. In fact the scores on the ASI in the normal samples have averaged around 20, which represents a mean item rating of 1.3 (i.e., "just a little"). This is entirely to be expected, because normal population sampies should be normal, but it reduces the chances of studying the nature of anxiety sensitivity in such a population. It is in patients with certain anxiety disorders, such as GAD, panic disorder, or agoraphobia, that concern about anxiety per se is thought to play an important underlying role. Data have been reported from clinical populations (McNally & Lorenz, 1987), but these have not been analysed using factor analytic procedures. On that basis, the factor analytic structure of the ASI in a clinical sample should be of some importance, and is addressed in the present study.
METHOD Subjects and procedure The anxious subjects were 142 female and 18 male clients applying for treatment in the Agoraphobic Treatment Programme at the Institute o f Psychiatry in the University of London. Normal controls were 100 women and 20 men recruited from the community around the Institute of Psychiatry, who were matched approximately for age with the patient sample. Anxious clients completed a range of assessment instruments including the Fear Survey S c h e d u l e (Wolpe and Lang, 1964) and the A g o r a p h o b i c Cognitions Questionnaire (Chambless et al., 1984). They were diagnosed as showing agoraphobic avoidance, with or without panic, on the basis of these assessments and from an interview either in person or by telephone with a clinical psychologist. Agoraphobic subjects had a mean age of 42 years (+11) and 80% were married or cohabiting. Normal controls had a mean age of 41 years (+16) and 61% were married or cohabiting. Subjects completed the ASI, along with other scales, in their homes. The Fear Survey Schedule Score for the patient sample was based on a 63 item version with a 20 item agoraphobic subscale. Taking into account the different number of items, the mean scores for the agoraphobic patients were closely similar to those reported by Mathews et al. (1976) and Emmelkamp et al. (1978). The result on the ACQ (mean score 2.44) was very close to that reported by Chambless et al. (1986).
RESULTS ASI means The agoraphobic sample had a higher score on the ASI than the normal con-
328
J. WARDLE,T. AHMAD, AND P. HAYWARD
trol sample (41.00 __.10.60vs. 15.0 +10.2) and this difference was highly significant (t = 18.9, p <.001). The score for the agoraphobics is comparable with Reiss et al.'s (1986) and McNally and Lorenz' (1987) small samples of agoraphobics who had mean scores of 38.2 and 36.2. The score in the normal group is slightly lower than, but comparable to, other female control groups.
Factor analysis A principal components analysis followed by a varimax rotation was used to identify the factor structure and factors with an eigen value greater than one were extracted (Kaiser's criterion). This was done for the anxious and normal samples separately.
Agoraphobic sample The principal components analysis resulted in a four factor solution that explained 60% of the total variance. Loadings of the items on the four rotated factors are shown in Table 1. Items contributing to a factor were selected on the basis that they should have a high loading on that factor and low loading on others. For most of the items, their factor position was very clear. Factor 1 loaded on items concerning the heart (6, 9), fear of shortness of breath (10), of having unusual bodily feelings (14). The items all had low relationships with the other three factors. Feeling shaky or faint (3, 4) also contributed, but were less specific, as they had lower loadings on Factor 1 and moderate, rather than low, ratings on other factors. Item 16 also had a moderate loading, but was equally highly loaded on Factor 2. Essentially, Factor 1 appeared to be a fear of physical sensation factor probably particularly in connection with cardiovascular health. Factor 2 loaded on worries that concentration problems meant going crazy (item 2), fear of being mentally ill (15), and fear of being unable to keep one's mind on a task (12). This factor clearly represented a fear of loss of mental control with the implications for psychiatric health. Factor 3 was a 'gastrointestinal' factor, loading on feeling nauseous (8), having stomach sounds (7), and being worded about stomach illness (11). Item 13 (other people notice when I feel shaky) was associated, but not so strongly or specifically as the other items. Factor 4 loaded on being concerned not to appear nervous (1) and being keen to stay in control of emotions (5) (i.e., it seemed to be a self-presentation factor). Items 13 (other people noticing shakiness) and 16 (being scared when nervous) and to a lesser extent items 3 and 4 (concerning trembling and faintness) had no specific relationship with any one factor, being loaded on more than one, and with relatively lower loadings. In view of the possibility that the Kaiser criterion would not produce the most psychologically meaningful factor structure, two and three factor solutions were also examined. In both of these solutions Factor 1 was similar, with Items 6, 9, 10, and 14 contributing strongly, and Items 3 and 4 also contributing, but with less specificity. Item 16 loaded on the first two factors, with a
ANXIETY SENSITIVITYIN AGORAPHOBIA
329
TABLE I FACTORLOADINGS.AGORAPHOBICSAMPLE(N ----160) FACTOR NUMBER VARIANCE EIGENVALUE
Factor 1 (34%) 5.49
.07 1. It is important to me not to appear nervous 2. When I cannot keep that .19 my mind on a task, I worry I might be going crazy 3. It scares me when *.53 I feel 'shaky' (trembling) 4. It scares me *.58 when I feel faint 5. It is important to me to .05 stay in conlrol of my emotions 6. It scares me when my **.71 heart beats rapidly 7. It embarrasses me .02 when my stomach growls 8. It scares me .13 when I am nauseous 9. When I notice that my that **.74 heart is beating rapidly, I worry I might have a heart attack 10. It scares me when I **.79 become short of breath ll. When my stomach .27 is upset, I worry that I might be seriously ill 12. It scares me when I am .11 unable to keep my mind on a task 13. Other people notice .20 when I feel shaky 14. Unusual body **.60 sensations scare me 15. When I am nervous, .21 I worry that I might be mentally ill 16. It scares me when I am nervous .43
Factor 2 (9%) 1.52
Factor 3 (9%) 1.39
Factor 4 (8%) 1.26
.11
-.01
**.82
**.86
.16
.04
.09
.25
.35
-.02
.32
.38
.36
.14
*.57
.23
.03
.00
.14
**.79
.02
.04
**.78
.12
.33
.13
-.24
.I0
.06
.04
.30
**.62
-.14
**.76
.26
.17
.15
*A7
.26
.12
.24
.27
**.80
.07
.19
.48
.12
.33
** high and specific factor loadings * moderate and less specific factor loadings s l i g h t l y h i g h e r l o a d i n g o n f a c t o r 2. I n t h e t h r e e f a c t o r s o l u t i o n , t h e t w o s e l f - p r e s e n t a t i o n i t e m s (1 a n d 5) w e r e a d d e d to the m e n t a l c o n t r o l i t e m s o n F a c t o r 2, b u t w i t h l o w e r f a c t o r l o a d i n g s , turning this into a fear of mental illness/self-presentation factor. Factor 3 r e m a i n e d a ' g a s t r i c ' f a c t o r ( I t e m s 7, 8, a n d 11). In the two factor solution, the other items were similar but the gastric items
330
J. WARDLE, T. AHMAD, AND P. HAYWARD
loaded on both factors. The results of these re-analyses did not improve the meaningfulness of the internal structure of the ASI. Consequently, the very clear factor structure that emerged from the 4-factor solution was judged to be most useful. In order to further validate the results, the analysis was repeated on two half-samples separately. In this way the factor structure emerging from one group could be regarded as exploratory and the second as confirmatory. Both of these analyses revealed factor structures closely similar to one another and to the structure described in Table 1. Items 3, 4 and 13 and 16, which were less clearly located on one factor in the original analysis, showed different loadlngs in the two sub-samples as well as items 14, which in one sub-sample was no longer located on Factor 1. However, the overall impression was one of strong support for the reliability of the factor structure.
Normal sample The principal components analysis again resulted in a four factor solution, which also explained just over 60% of the variance (61.7%). The eigen values and factor loading are shown in Table 2. A much less coherent picture emerged from the factor loadings on items. The items loading on the first factor concerned stomach sounds (7), fear of heart attack (9), fear of being out of breath (10), and fear of not concentrating (12). Factor 2 loaded on feeling faint (4) and other people noticing shaking (13). Factor 3 loaded on stomach illness (11) and mental illness (15), while Factor 4 loaded on concentration worries (2) being shaky (3), and staying in control (5). DISCUSSION The results of this study support the multidimensional structure of anxiety sensitivity. In the agoraphobic sample, a strong first factor linked together concerns about bodily symptoms, and included all the six items on that topic. The strongest items in this factor were those concerned with statements of breath and palpitations which parallel the major fears of heart attacks and strokes at the root of many of the catastrophic cognitions in panic disorder (Clark, 1988). Factor 2 loaded on the three gastro-intestinal items and Factor 4 loaded on the two items concerning other people's perception of symptoms. Fourteen out of the sixteen items appeared to be specific, loading highly on one factor and with low loadings on the other three. The factor structure was coherent and made good sense of what might otherwise have seemed to be a mixed group of worries. The first two factors, fear o f physical s y m p t o m s and fear o f the cognitive/behavioural consequences, reflected the same kinds of distinctions identified in the two factors of the ACQ, namely fear related to physical symptoms and physical illnesses and fear relating cognitive symptoms and psychiatric illness (Chambless et al., 1984). The third (gastric) factor is less prominent in the specific psychopathology of agoraphobia, although some
331
ANXIETY SENSITIVITYIN AGORAPHOBIA TABLE 2 FACTORLOADnqC,S. NORMALSAMPt~ (N = 120) FACTOR NUMBER VARIANCE
EIGENVALUE
Factor 1
(34%) 5.49
1. It is important to .33 me not to appear nervous 2. When I cannot keep .23 my mind on a task, I worry that I might be going crazy 3. It scares me .25 when I feel "shaky' (trembling) 4. It scares me when I feel faint .25 5. It is important to me to stay in .22 control of my emotions 6. It scares me when my *.54 heart beats rapidly 7. It embarrasses me when **.63 my stomach growls 8. It scares me when .16 I am nauseous 9. When I notice that my heart is *.67 beating rapidly, I worry that I might have a heart attack 10. It scares me when I become **.70 short of breath 1I. When my stomach is upset, .03 I worry that I might be seriously ill 12. It scares me when I am **.77 unable to keep my mind on a task 13. Other people notice .12 when I feel shaky 14. Unusual body sensations .40
Factor 2 (9%) 1.50
Factor 3 (9%) 1.37
Factor 4
.31
.10
*.56
-.23
.32
**.71
.37
.22
*.60
**.74 .39
.00 -.19
.10 *.60
*.50
.I1
.12
.11
.01
.26
**.70
.34
-.01
.40
.09
-.12
.33
.15
.06
.25
**.79
.02
-.18
.22
.35
**.64
.28
.15
.34
.45
.11
.07
**.80
.09
.34
.34
.24
(8%) 1.08
scare me
15. When I am nervous, .25 I worry that I might be mentally ill 16. It scares me when *.56 I am nervous
** high fact~ loadings * moderate and less specific factor loadlngs patients do report strong gastrointestinal symptoms and clinical experience s u g g e s t s t h a t i f t h e y do, t h e y s e e m to b e e s p e c i a l l y a v e r s e t o t h e m . T h e f o u r t h factor, c o n c e r n i n g s e l f - p r e s e r v a t i o n , is a g a i n n o t c o m m e n t e d o n s p e c i f i c a l l y i n connection with agoraphobia, but clinically appears very important to some p a t i e n t s . T h e t w o i t e m s n o t c o n n e c t e d i n as s p e c i f i c a l l y w i t h a n y o n e f a c t o r w e r e " O t h e r p e o p l e n o t i c e w h e n I feel s h a k y " (13) a n d "It s c a r e s m e w h e n I
332
J. WARDLE,T. AHMAD,AND P. HAYWARD
am nervous" (14). With respect to the first, there seems no obvious element of anxiety sensitivity per se; the item could be taken to reflect the extent of physical expression of anxiety. Interestingly, the closest connection was with the gastric item and one could speculate that htis related to visits to the bathroom. Item 16 is a curious one since being "scared when nervous" is almost tautological, as the words "scared" and "nervous" are essentially synonyms. The state of not being scared when being nervous (the implication of a low score) may represent a distinction between stimulus and response that is clear to the students of the concept of fear of fear, but could be too subtle to be drawn meaningfully by naive subjects. In the present study almost all subjects in both samples endorsed it positively. Given its relative lower factor loadings in any of the analyses, and its lack of specificity, it may not be all that useful. The normal sample analysis produced a quite different result. A four factor solution emerged, but the item structure had little obvious psychologyical meaning. It also failed to replicate Telch et al.'s (1989) factor structure, which in turn seemed to be different from Peterson and Heilbronner's (1987) reported structure. Indeed the two analyses with single factor solutions reported by Reiss et al. (1986) revealed four items that changed their loading substantially from one sample to another, and neither had any correspondence with the first factor in Telch et al.'s (1989) work or the present study. Why should the factor structure of anxiety sensitivity be unreliable and disorganised in a normal sample and cohernt in an agoraphobic sample? The most obvious answer lies in the differential meaning of anxiety sensitivity in the agoraphobic and normal population. In the present normal sample, as in other samples, the individual item scores averaged around one (representing the rating "a little"). Such low levels of endorsement are not likely to produce a meaningful structure. Indeed there seems little reason to predict anxiety sensitivity in a normal sample, since anxiety symptoms are often sought as part of a "thrill." By contrast, in an agoraphobic sample, anxiety is often both feared and avoided, but the factor analytic evidence presented here suggests that different groups of agoraphobics may be especially fearful of different aspects of anxiety, a finding supported by other work in the field (Chambless & Gracely, 1989).
REFERENCES Beck, A.T., & Emery, G. (1985). Anxiety Disorders and Phobias. New York: Basic Books. Chambless, D.L., Caputo, G.C., Bright, P., & Gallagher, R. (1984). Assessment of fear of fear in agoraphobics: The Body Sensations Questionnaire and the Agoraphobic Cognitions Questionnaire. Journal of Consulting and Clinical Psychology, 52, 1090-1097. Chambless, D.L. & Gracely, E.J. (1989). Fear of Fear and the Anxiety Disorders. Cognitive Therapy and Research, 13, 9-20. Clark, D. (1988). A cognitive model of panic attacks. In Rachman, J. & Maser, S. Panic: Psychological Perspectives. HiUdale, NJ: Laurence Erlbaum. Emmelkamp, P.M.G., Kuipers, A.C.M., & Eggeraat, J.B. (1978) Cognitive modification versus prolonged exposure in vivo: A comparison with agoraphobics as subjects. Behaviour Research and Therapy, 16, 33-41. Goldstein, A.J. & Chambless, D.L. (1978). A re-analysis of agoraphobia. Behavior Therapy, 9, 47-59.
ANXIETY SENSITIVITY IN AGORAPHIA
333
Mathews, A.M., Johnston, D.W., Lancashire, M., Munby, M., Shaw, P.M., & Gelder, M. G. (1976). Imaginal flooding and exposure to real phobic situations: Treatment outcome with agoraphobic patients. British Journal of Psychiatry, 129, 362-371. McNaily, R.J. & Lorenz, M. (1987). Anxiety Sensitivity in Agoraphobics. Journal of Behaviour Therapy and ExperimentalPsychiatry, 18, 3-11. Peterson, R.A. & Heilbronnet, R.L. (1987). The Anxiety Sensitivity Index. Journal of Anxiety Disorders, 1, 117-121. Reiss, S. & McNally, R_I. (1985). The expectancy model of fear. In Reiss, S. & Bootzin, R.R. (Eds.), Theoreticalissues in Behaviour Therapy. New York: Academic Press. Reiss, S., Peterson, R.A., & Gursky, D.M. (1988). Anxiety sensitivity, injury sensitivity and individual differences in fearfulness. BehaviourResearchand Therapy, 26, 341-345. Reiss, S., Peterson, R.A., Gursky, D.M., & McNally, RJ. (1986). Anxiety sensitivity, anxiety frequency and the prediction of fearfulness. BehaviourResearch and Therapy, 24, 1-8. Telch, MJ., Shermis, M.D., & Lucas, J.A. (1989). Anxiety Sensitivity: Unitary personality trait or domain-specific appraisals? Journal of An.xietyDisorders, 3, 22-32. Wolpe, J. & Lang, P. (1964). A Fear Survey Schedule for use in behavior therapy. Behaviour Research and Therapy, 2, 27-30.
0887-6185/90 $3.00 + .00 Copyright © 1990 Pergamon Press pie
Printed in the USA. All rights reserved.
Anxiety Sensitivity in Agoraphobia JANE WARDLE,PH.D., TAYSEERAI-IMAD,M.A., PETER HAYWARD, PH.D.
Department of Psychology, Institute of Psychiatry, University of London,UK Abstract--The factor structure of the Anxiety Sensitivity Index was assessed in 166 agoraphobic clients who had applied to a behavioural treatment programme and 120 age and sex-matched normal controls. In both samples a four factor solution emerged as the most useful, and it explained more than 60% of the total variance. The item loadings in the agoraphobicsample revealed a coherent theme with factors reflecting fear of heart and breathing symptoms, fear of loss of mental control, fear of gastrointestinal difficulties, and concern about other people detecting anxiety. The item loadings in the normal sample made less psychological sense. The results are discussed both in the light of the trait model of anxiety sensitivity and with respect to the clinical implications of anxiety sensitivity.
ANXIETY SENSITIVITY IN AGORAPHOBIA Cognitive conceptualizations o f anxiety and panic have increasingly implicated beliefs about anxiety (Goldstein & Chambless, 1978; Reiss & McNally, 1985; Beck & Emery, 1985; Clark, 1988). The concept o f "fear of fear" has been particularly central in the explanation o f the etiology o f agoraphobia and panic disorder and is thought both to motivate avoidance o f situations associated with anxiety and to amplify anxiety once it occurs (Goldstein & Chambless, 1978). Chambless, Caputo, Bright, and Gallagher (1984) developed two psychometric instruments to assess the extent of fear of fear. The "Agoraphobic Cognitions Questionnaire" (ACQ) contains a range of items which tap the supposed cognitions of agoraphobic patients and it has been shown to differentiate agoraphobic and panic patients from other psychiatric groups and normal controis (Chambless & Gracely, 1989). Factor analysis of this scale has revealed two subscales, one relating to thoughts of the physical consequences o f anxiety (e.g., heart attack, stroke) and one to thoughts of the behavioral and social consequences (e.g., acting foolishly, g o i n g crazy). The B o d i l y Sensations Questionnaire (BSQ) addresses the fear of the bodily sensations (e.g. heart palpitations, tingling in the fingertips). This scale also discriminated agoraphobic and panic patients from other groups (Chambless & Gracely, 1989). More recently, Reiss and McNally (1985) have proposed a similar construct, which they call "anxiety sensitivity," described as "an individual differReprint requests to Jane Wardle, Ph.D. Department of Psychology, Institute of Psychiatry, De CrespignyPark. LondonSE5 8AF,UK. 325
326
J. WARDLE, T. AHMAD, AND P. HAYWARD
ence variable consisting of beliefs that the experience of anxiety/fear causes illness, embarrassment or additional anxiety" (1985, p. 2), They argue that higher levels of anxiety sensitivity will increase alertness about the possibility of becoming anxious, increase worry about becoming anxious and increase motivation to avoid anxiety-provoking stimuli. The instrument they have developed to assess it is the "Anxiety Sensitivity Index" (ASI) (Reiss, Peterson, Gursky, & McNally, 1986) a sixteen-item inventory of statements designed to assess fear of, or embarrassment about, anxiety ("It scares me when I feel shaky") and negative health cognitions experienced during the anxious state ("When I notice my heart is beating rapidly, I worry that I might have a heart attack"). Reiss and his colleagues have argued that the ASI is a unidimensional scale on the basis of the finding of a single factor structure in their two normal population samples (Reiss et al., 1986). Regrettably there are no details in the published paper about the method of factor analysis used, the criterion for determining the number of factors, or the amount of variance accounted for by the first factor. Consequently the assertion of a single factor structure is hard to evaluate. In a re-analysis of these data, they show that if the ASI items are entered into a principal components analysis together with items from the Fear Survey Schedule-H, eight of the ASI items cluster together on Factor 3 (Reiss, Peterson, & Gursky, 1988). However, great significance cannot be attached to the finding that the correlations within ASI items are higher than the correlations between ASI items and items from another scale. In an attempt to replicate the single factor structure, Peterson and Heilbronner (1987) administered the ASI to a sample of 122 mildly anxious college students ("anxious during at least 40% of the day"). Using the Kaiser criterion for factor extraction, they found four factors that together accounted for 61% of the variance. Two of the factors were said to be associated with items related to fear of the consequences of anxiety and two with fear of the physical sensations themselves. No further details about item loadings were presented, but a two factor solution was then imposed. This had 11 out of 16 items loading on Factor 2, with most of those also loading in Factor 1. On this basis Peterson and Heilbronner argued, without being very convincing, that their analysis confirmed a "strong single factor." A further factor analytic study, again with data from college students studying psychology, gave less support to the single factor model and tended towards a multifactorial model for anxiety sensitivity (Telch, Shermis, & Lucas, 1989). This analysis again yielded a four factor solution, explaining 54% of the total variance. The authors made a valiant attempt to identify a plausible construct for each factor, but Factors 1 and 2 both included a mixture of items, Factor 3 only two items (appearing nervous and staying in control), and Factor 4 contained the two items about heart symptoms and one about shormess of breath. Because Peterson and Heilbronner (1989) had not presented the item loadings from their factor analysis, it was not possible to compare the factor structures with those obtained from previous studies. However Telch et al.'s (1989) work gave little support for a single factor structure and more evidence for what they termed "domain-specific appraisals of anxiety." The factor analytic studies described above have all recruited their samples
ANXIETY SENSITIVITY IN AGORAPHOBIA
327
from a college student population, with the unavoidable restriction of range of age, social status, and intelligence in these groups. Anxiety levels in such populations would be expected to be low and anxiety sensitivity might be predicted to be similarly rare. In fact the scores on the ASI in the normal samples have averaged around 20, which represents a mean item rating of 1.3 (i.e., "just a little"). This is entirely to be expected, because normal population sampies should be normal, but it reduces the chances of studying the nature of anxiety sensitivity in such a population. It is in patients with certain anxiety disorders, such as GAD, panic disorder, or agoraphobia, that concern about anxiety per se is thought to play an important underlying role. Data have been reported from clinical populations (McNally & Lorenz, 1987), but these have not been analysed using factor analytic procedures. On that basis, the factor analytic structure of the ASI in a clinical sample should be of some importance, and is addressed in the present study.
METHOD Subjects and procedure The anxious subjects were 142 female and 18 male clients applying for treatment in the Agoraphobic Treatment Programme at the Institute o f Psychiatry in the University of London. Normal controls were 100 women and 20 men recruited from the community around the Institute of Psychiatry, who were matched approximately for age with the patient sample. Anxious clients completed a range of assessment instruments including the Fear Survey S c h e d u l e (Wolpe and Lang, 1964) and the A g o r a p h o b i c Cognitions Questionnaire (Chambless et al., 1984). They were diagnosed as showing agoraphobic avoidance, with or without panic, on the basis of these assessments and from an interview either in person or by telephone with a clinical psychologist. Agoraphobic subjects had a mean age of 42 years (+11) and 80% were married or cohabiting. Normal controls had a mean age of 41 years (+16) and 61% were married or cohabiting. Subjects completed the ASI, along with other scales, in their homes. The Fear Survey Schedule Score for the patient sample was based on a 63 item version with a 20 item agoraphobic subscale. Taking into account the different number of items, the mean scores for the agoraphobic patients were closely similar to those reported by Mathews et al. (1976) and Emmelkamp et al. (1978). The result on the ACQ (mean score 2.44) was very close to that reported by Chambless et al. (1986).
RESULTS ASI means The agoraphobic sample had a higher score on the ASI than the normal con-
328
J. WARDLE,T. AHMAD, AND P. HAYWARD
trol sample (41.00 __.10.60vs. 15.0 +10.2) and this difference was highly significant (t = 18.9, p <.001). The score for the agoraphobics is comparable with Reiss et al.'s (1986) and McNally and Lorenz' (1987) small samples of agoraphobics who had mean scores of 38.2 and 36.2. The score in the normal group is slightly lower than, but comparable to, other female control groups.
Factor analysis A principal components analysis followed by a varimax rotation was used to identify the factor structure and factors with an eigen value greater than one were extracted (Kaiser's criterion). This was done for the anxious and normal samples separately.
Agoraphobic sample The principal components analysis resulted in a four factor solution that explained 60% of the total variance. Loadings of the items on the four rotated factors are shown in Table 1. Items contributing to a factor were selected on the basis that they should have a high loading on that factor and low loading on others. For most of the items, their factor position was very clear. Factor 1 loaded on items concerning the heart (6, 9), fear of shortness of breath (10), of having unusual bodily feelings (14). The items all had low relationships with the other three factors. Feeling shaky or faint (3, 4) also contributed, but were less specific, as they had lower loadings on Factor 1 and moderate, rather than low, ratings on other factors. Item 16 also had a moderate loading, but was equally highly loaded on Factor 2. Essentially, Factor 1 appeared to be a fear of physical sensation factor probably particularly in connection with cardiovascular health. Factor 2 loaded on worries that concentration problems meant going crazy (item 2), fear of being mentally ill (15), and fear of being unable to keep one's mind on a task (12). This factor clearly represented a fear of loss of mental control with the implications for psychiatric health. Factor 3 was a 'gastrointestinal' factor, loading on feeling nauseous (8), having stomach sounds (7), and being worded about stomach illness (11). Item 13 (other people notice when I feel shaky) was associated, but not so strongly or specifically as the other items. Factor 4 loaded on being concerned not to appear nervous (1) and being keen to stay in control of emotions (5) (i.e., it seemed to be a self-presentation factor). Items 13 (other people noticing shakiness) and 16 (being scared when nervous) and to a lesser extent items 3 and 4 (concerning trembling and faintness) had no specific relationship with any one factor, being loaded on more than one, and with relatively lower loadings. In view of the possibility that the Kaiser criterion would not produce the most psychologically meaningful factor structure, two and three factor solutions were also examined. In both of these solutions Factor 1 was similar, with Items 6, 9, 10, and 14 contributing strongly, and Items 3 and 4 also contributing, but with less specificity. Item 16 loaded on the first two factors, with a
ANXIETY SENSITIVITYIN AGORAPHOBIA
329
TABLE I FACTORLOADINGS.AGORAPHOBICSAMPLE(N ----160) FACTOR NUMBER VARIANCE EIGENVALUE
Factor 1 (34%) 5.49
.07 1. It is important to me not to appear nervous 2. When I cannot keep that .19 my mind on a task, I worry I might be going crazy 3. It scares me when *.53 I feel 'shaky' (trembling) 4. It scares me *.58 when I feel faint 5. It is important to me to .05 stay in conlrol of my emotions 6. It scares me when my **.71 heart beats rapidly 7. It embarrasses me .02 when my stomach growls 8. It scares me .13 when I am nauseous 9. When I notice that my that **.74 heart is beating rapidly, I worry I might have a heart attack 10. It scares me when I **.79 become short of breath ll. When my stomach .27 is upset, I worry that I might be seriously ill 12. It scares me when I am .11 unable to keep my mind on a task 13. Other people notice .20 when I feel shaky 14. Unusual body **.60 sensations scare me 15. When I am nervous, .21 I worry that I might be mentally ill 16. It scares me when I am nervous .43
Factor 2 (9%) 1.52
Factor 3 (9%) 1.39
Factor 4 (8%) 1.26
.11
-.01
**.82
**.86
.16
.04
.09
.25
.35
-.02
.32
.38
.36
.14
*.57
.23
.03
.00
.14
**.79
.02
.04
**.78
.12
.33
.13
-.24
.I0
.06
.04
.30
**.62
-.14
**.76
.26
.17
.15
*A7
.26
.12
.24
.27
**.80
.07
.19
.48
.12
.33
** high and specific factor loadings * moderate and less specific factor loadings s l i g h t l y h i g h e r l o a d i n g o n f a c t o r 2. I n t h e t h r e e f a c t o r s o l u t i o n , t h e t w o s e l f - p r e s e n t a t i o n i t e m s (1 a n d 5) w e r e a d d e d to the m e n t a l c o n t r o l i t e m s o n F a c t o r 2, b u t w i t h l o w e r f a c t o r l o a d i n g s , turning this into a fear of mental illness/self-presentation factor. Factor 3 r e m a i n e d a ' g a s t r i c ' f a c t o r ( I t e m s 7, 8, a n d 11). In the two factor solution, the other items were similar but the gastric items
330
J. WARDLE, T. AHMAD, AND P. HAYWARD
loaded on both factors. The results of these re-analyses did not improve the meaningfulness of the internal structure of the ASI. Consequently, the very clear factor structure that emerged from the 4-factor solution was judged to be most useful. In order to further validate the results, the analysis was repeated on two half-samples separately. In this way the factor structure emerging from one group could be regarded as exploratory and the second as confirmatory. Both of these analyses revealed factor structures closely similar to one another and to the structure described in Table 1. Items 3, 4 and 13 and 16, which were less clearly located on one factor in the original analysis, showed different loadlngs in the two sub-samples as well as items 14, which in one sub-sample was no longer located on Factor 1. However, the overall impression was one of strong support for the reliability of the factor structure.
Normal sample The principal components analysis again resulted in a four factor solution, which also explained just over 60% of the variance (61.7%). The eigen values and factor loading are shown in Table 2. A much less coherent picture emerged from the factor loadings on items. The items loading on the first factor concerned stomach sounds (7), fear of heart attack (9), fear of being out of breath (10), and fear of not concentrating (12). Factor 2 loaded on feeling faint (4) and other people noticing shaking (13). Factor 3 loaded on stomach illness (11) and mental illness (15), while Factor 4 loaded on concentration worries (2) being shaky (3), and staying in control (5). DISCUSSION The results of this study support the multidimensional structure of anxiety sensitivity. In the agoraphobic sample, a strong first factor linked together concerns about bodily symptoms, and included all the six items on that topic. The strongest items in this factor were those concerned with statements of breath and palpitations which parallel the major fears of heart attacks and strokes at the root of many of the catastrophic cognitions in panic disorder (Clark, 1988). Factor 2 loaded on the three gastro-intestinal items and Factor 4 loaded on the two items concerning other people's perception of symptoms. Fourteen out of the sixteen items appeared to be specific, loading highly on one factor and with low loadings on the other three. The factor structure was coherent and made good sense of what might otherwise have seemed to be a mixed group of worries. The first two factors, fear o f physical s y m p t o m s and fear o f the cognitive/behavioural consequences, reflected the same kinds of distinctions identified in the two factors of the ACQ, namely fear related to physical symptoms and physical illnesses and fear relating cognitive symptoms and psychiatric illness (Chambless et al., 1984). The third (gastric) factor is less prominent in the specific psychopathology of agoraphobia, although some
331
ANXIETY SENSITIVITYIN AGORAPHOBIA TABLE 2 FACTORLOADnqC,S. NORMALSAMPt~ (N = 120) FACTOR NUMBER VARIANCE
EIGENVALUE
Factor 1
(34%) 5.49
1. It is important to .33 me not to appear nervous 2. When I cannot keep .23 my mind on a task, I worry that I might be going crazy 3. It scares me .25 when I feel "shaky' (trembling) 4. It scares me when I feel faint .25 5. It is important to me to stay in .22 control of my emotions 6. It scares me when my *.54 heart beats rapidly 7. It embarrasses me when **.63 my stomach growls 8. It scares me when .16 I am nauseous 9. When I notice that my heart is *.67 beating rapidly, I worry that I might have a heart attack 10. It scares me when I become **.70 short of breath 1I. When my stomach is upset, .03 I worry that I might be seriously ill 12. It scares me when I am **.77 unable to keep my mind on a task 13. Other people notice .12 when I feel shaky 14. Unusual body sensations .40
Factor 2 (9%) 1.50
Factor 3 (9%) 1.37
Factor 4
.31
.10
*.56
-.23
.32
**.71
.37
.22
*.60
**.74 .39
.00 -.19
.10 *.60
*.50
.I1
.12
.11
.01
.26
**.70
.34
-.01
.40
.09
-.12
.33
.15
.06
.25
**.79
.02
-.18
.22
.35
**.64
.28
.15
.34
.45
.11
.07
**.80
.09
.34
.34
.24
(8%) 1.08
scare me
15. When I am nervous, .25 I worry that I might be mentally ill 16. It scares me when *.56 I am nervous
** high fact~ loadings * moderate and less specific factor loadlngs patients do report strong gastrointestinal symptoms and clinical experience s u g g e s t s t h a t i f t h e y do, t h e y s e e m to b e e s p e c i a l l y a v e r s e t o t h e m . T h e f o u r t h factor, c o n c e r n i n g s e l f - p r e s e r v a t i o n , is a g a i n n o t c o m m e n t e d o n s p e c i f i c a l l y i n connection with agoraphobia, but clinically appears very important to some p a t i e n t s . T h e t w o i t e m s n o t c o n n e c t e d i n as s p e c i f i c a l l y w i t h a n y o n e f a c t o r w e r e " O t h e r p e o p l e n o t i c e w h e n I feel s h a k y " (13) a n d "It s c a r e s m e w h e n I
332
J. WARDLE,T. AHMAD,AND P. HAYWARD
am nervous" (14). With respect to the first, there seems no obvious element of anxiety sensitivity per se; the item could be taken to reflect the extent of physical expression of anxiety. Interestingly, the closest connection was with the gastric item and one could speculate that htis related to visits to the bathroom. Item 16 is a curious one since being "scared when nervous" is almost tautological, as the words "scared" and "nervous" are essentially synonyms. The state of not being scared when being nervous (the implication of a low score) may represent a distinction between stimulus and response that is clear to the students of the concept of fear of fear, but could be too subtle to be drawn meaningfully by naive subjects. In the present study almost all subjects in both samples endorsed it positively. Given its relative lower factor loadings in any of the analyses, and its lack of specificity, it may not be all that useful. The normal sample analysis produced a quite different result. A four factor solution emerged, but the item structure had little obvious psychologyical meaning. It also failed to replicate Telch et al.'s (1989) factor structure, which in turn seemed to be different from Peterson and Heilbronner's (1987) reported structure. Indeed the two analyses with single factor solutions reported by Reiss et al. (1986) revealed four items that changed their loading substantially from one sample to another, and neither had any correspondence with the first factor in Telch et al.'s (1989) work or the present study. Why should the factor structure of anxiety sensitivity be unreliable and disorganised in a normal sample and cohernt in an agoraphobic sample? The most obvious answer lies in the differential meaning of anxiety sensitivity in the agoraphobic and normal population. In the present normal sample, as in other samples, the individual item scores averaged around one (representing the rating "a little"). Such low levels of endorsement are not likely to produce a meaningful structure. Indeed there seems little reason to predict anxiety sensitivity in a normal sample, since anxiety symptoms are often sought as part of a "thrill." By contrast, in an agoraphobic sample, anxiety is often both feared and avoided, but the factor analytic evidence presented here suggests that different groups of agoraphobics may be especially fearful of different aspects of anxiety, a finding supported by other work in the field (Chambless & Gracely, 1989).
REFERENCES Beck, A.T., & Emery, G. (1985). Anxiety Disorders and Phobias. New York: Basic Books. Chambless, D.L., Caputo, G.C., Bright, P., & Gallagher, R. (1984). Assessment of fear of fear in agoraphobics: The Body Sensations Questionnaire and the Agoraphobic Cognitions Questionnaire. Journal of Consulting and Clinical Psychology, 52, 1090-1097. Chambless, D.L. & Gracely, E.J. (1989). Fear of Fear and the Anxiety Disorders. Cognitive Therapy and Research, 13, 9-20. Clark, D. (1988). A cognitive model of panic attacks. In Rachman, J. & Maser, S. Panic: Psychological Perspectives. HiUdale, NJ: Laurence Erlbaum. Emmelkamp, P.M.G., Kuipers, A.C.M., & Eggeraat, J.B. (1978) Cognitive modification versus prolonged exposure in vivo: A comparison with agoraphobics as subjects. Behaviour Research and Therapy, 16, 33-41. Goldstein, A.J. & Chambless, D.L. (1978). A re-analysis of agoraphobia. Behavior Therapy, 9, 47-59.
ANXIETY SENSITIVITY IN AGORAPHIA
333
Mathews, A.M., Johnston, D.W., Lancashire, M., Munby, M., Shaw, P.M., & Gelder, M. G. (1976). Imaginal flooding and exposure to real phobic situations: Treatment outcome with agoraphobic patients. British Journal of Psychiatry, 129, 362-371. McNaily, R.J. & Lorenz, M. (1987). Anxiety Sensitivity in Agoraphobics. Journal of Behaviour Therapy and ExperimentalPsychiatry, 18, 3-11. Peterson, R.A. & Heilbronnet, R.L. (1987). The Anxiety Sensitivity Index. Journal of Anxiety Disorders, 1, 117-121. Reiss, S. & McNally, R_I. (1985). The expectancy model of fear. In Reiss, S. & Bootzin, R.R. (Eds.), Theoreticalissues in Behaviour Therapy. New York: Academic Press. Reiss, S., Peterson, R.A., & Gursky, D.M. (1988). Anxiety sensitivity, injury sensitivity and individual differences in fearfulness. BehaviourResearchand Therapy, 26, 341-345. Reiss, S., Peterson, R.A., Gursky, D.M., & McNally, RJ. (1986). Anxiety sensitivity, anxiety frequency and the prediction of fearfulness. BehaviourResearch and Therapy, 24, 1-8. Telch, MJ., Shermis, M.D., & Lucas, J.A. (1989). Anxiety Sensitivity: Unitary personality trait or domain-specific appraisals? Journal of An.xietyDisorders, 3, 22-32. Wolpe, J. & Lang, P. (1964). A Fear Survey Schedule for use in behavior therapy. Behaviour Research and Therapy, 2, 27-30.