An expanded childhood anxiety sensitivity index: its factor structure, reliability, and validity in a non-clinical adolescent sample

Behaviour Research and Therapy 40 (2002) 299–311 www.elsevier.com/locate/brat

An expanded childhood anxiety sensitivity index: its factor structure, reliability, and validity in a non-clinical adolescent sample Peter Muris

*

Department of Medical, Clinical, and Experimental Psychology, Maastricht University, P.O. Box 616, 6200 MD Maastricht, The Netherlands Received 1 May 2000; received in revised form 30 October 2000

Abstract Anxiety sensitivity refers to the fear of anxiety-related bodily sensations that are interpreted as having potentially harmful somatic, psychological, or social consequences. The current study examined the factor analytic structure of anxiety sensitivity in a large sample of normal adolescents (N=518) using the revised childhood anxiety sensitivity index (CASI-R). Confirmatory factor analysis indicated that anxiety sensitivity as measured by the CASI-R can best be conceptualised as a hierarchical construct with four lower-order factors loading on a single higher-order factor. The lower-order factors were ‘fear of cardiovascular symptoms’, ‘fear of publicly observable anxiety reactions’, ‘fear of cognitive dyscontrol’, and ‘fear of respiratory symptoms’. An additional aim of the present study was to investigate the psychometric properties of the CASI-R. Results showed the CASI-R to be a reliable scale in terms of internal consistency. Furthermore, CASI-R scores were substantially related to levels of anxiety sensitivity as measured by the original index, trait anxiety, symptoms of anxiety disorders, in particular ‘panic disorder and agoraphobia’, and depression. Finally, some evidence was found for the validity of the CASI-R factor scores. That is, all factors convincingly loaded on symptoms of ‘panic disorder and agoraphobia’, whereas the factor ‘fear of publicly observable anxiety reactions’ was also strongly associated with symptoms of ‘social phobia’.  2002 Elsevier Science Ltd. All rights reserved. Keywords: Anxiety sensitivity; Factor structure; Childhood Anxiety Sensitivity Index — Revised; Non-clinical adolescents

* Tel.: +31-43-388-1264; fax: +31-43-367-0968. E-mail address: [email protected] (P. Muris).

0005-7967/02/$ - see front matter  2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 5 - 7 9 6 7 ( 0 0 ) 0 0 1 1 2 - 1

300

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

1. Introduction Anxiety sensitivity refers to the fear of anxiety-related bodily sensations that are interpreted as having potentially harmful somatic, psychological, or social consequences (e.g., Taylor, 1995). Research in adult populations has indicated that anxiety sensitivity plays a role in the aetiology and maintenance of anxiety disorders, in particular panic disorder (e.g., Rachman, 1998). There is evidence to show that anxiety sensitivity is also involved in fear and anxiety of children and adolescents. A number of studies have found that anxiety sensitivity in children and adolescents correlates in a theoretically meaningful way with other anxiety measures (Chorpita, Albano, & Barlow, 1996; Muris, Schmidt, Merckelbach, & Schouten, 2001; Silverman, Fleisig, Rabian, & Peterson, 1991; Weems, Hammond-Laurence, Silverman, & Ginsburg, 1998). Furthermore, Rabian, Peterson, Richters, and Jensen (1993) compared levels of anxiety sensitivity in children with anxiety disorders, children with disruptive disorders, and children with no diagnosis. Results showed that children with anxiety disorders displayed significantly higher anxiety sensitivity scores than children with no diagnosis, whereas children with disruptive disorders scored in between. Finally, Lau, Calamari, and Waraczynski (1996) examined the relationship between anxiety sensitivity and panic disorder symptoms in normal adolescents. These authors found significant associations between anxiety sensitivity and the number of experienced panic attacks, the level of distress caused by the panic attacks, and the judged seriousness of the attacks (for similar findings, see Kearney, Albano, Eisen, Allan, & Barlow, 1997; Mattis & Ollendick, 1997). In children and adolescents, anxiety sensitivity is measured by means of the Childhood Anxiety Sensitivity Index (CASI; Silverman et al., 1991), which is an age-downward modification of the Anxiety Sensitivity Index (ASI; Reiss, Peterson, Gursky, & McNally, 1986), the most widely used instrument for assessing anxiety sensitivity in adults. The CASI consists of 18 items such as “It scares me when I feel shaky”, “It scares me when my heart beats fast”, and “It scares me when I feel nervous”; children and adolescents are asked to rate the extent to which each item applies to them (none, some, or a lot). Previous studies have consistently shown that the CASI is a reliable and valid questionnaire for measuring anxiety sensitivity in both clinical and nonclinical samples of children and adolescents (e.g., Rabian, Embry, & MacIntyre, 1999; Silverman et al., 1991). There has been considerable debate on the factor analytic structure of anxiety sensitivity. Although there is consensus that anxiety sensitivity should be regarded as a hierarchically organised construct consisting of several lower-order factors which load on a single higher-order factor (see, e.g., Cox, Parker, & Swinson, 1996; Taylor & Cox, 1998a; Zinbarg, Barlow, & Brown, 1997), the exact number and nature of the lower-order factors remain to be disclosed. Identification of distinct anxiety sensitivity factors seems important because these factors may reflect specific mechanisms that make individuals prone to develop specific types of fear and anxiety (see Cox, 1996). For example, the factor ‘fear of cardiovascular symptoms’ may lead to a panic attack with catastrophic thoughts about dying, whereas the factor ‘fear of publicly observable symptoms’ may give rise to social anxiety with thoughts about personal weakness. Previous factor analytic studies seem to indicate that anxiety sensitivity in children and adolescents can be best conceptualised as a hierarchical model with either three or four lower-order factors loading on one higher-order factor (i.e., ‘anxiety sensitivity’). For example, on the basis of their study in normal and clinically children aged 7 to 16 years, Silverman, Ginsburg, and

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

301

Goedhart (1999) concluded that a hierarchical structure with three lower-order factors, i.e., ‘fear of physical symptoms’, ‘fear of mental incapacitation’, and ‘fear of social evaluation’, probably is the most plausible representation of the anxiety sensitivity construct. Highly similar factors were obtained by Muris et al. (2001) in their sample of normal adolescents aged 13 to 16 years. Meanwhile, it is important to note that in both studies some indications were found for a fourfactor solution providing an even better fit for the data. The problem, however, is that the CASI, just like its adult version (the ASI), simply contains too few items (i.e., 18 and 16, respectively) to examine this issue and to reliably detect the major factors of anxiety sensitivity. In an attempt to deal with this problem, Taylor and Cox (1998b) recently developed an expanded scale of anxiety sensitivity for adults, the Anxiety Sensitivity Index — Revised (ASIR), which consists of 36 items each referring to one of the major domains of anxiety sensitivity suggested by earlier studies. The factor structure of the ASI-R was examined in a sample of adult psychiatric outpatients (N=155). Results indicated a hierarchical structure with four lower-order factors loading on one higher-order factor. The lower-order factors were (1) ‘fear of cardiovascular symptoms’, (2) ‘fear of publicly observable anxiety reactions’, (3) ‘fear of cognitive dyscontrol’, and (4) ‘fear of respiratory symptoms’. Silverman et al. (1999) have noted that it would be of interest to examine whether this hierarchical four-factor structure of anxiety sensitivity can also be obtained in children and adolescents when using an expanded childhood measure of anxiety sensitivity. The main purpose of the present study was to investigate just this issue. Following Taylor and Cox (1998b), a revised version of the CASI, i.e., the CASI-R, was constructed. The factor structure of the CASI-R was examined by carrying out confirmatory factor analysis. This statistical technique makes it possible to examine to what extent the data are in line with a hypothesised factor structure. Two models had our special attention. First of all, the hierarchical model with four lower-order factors (see supra) loading on one higher-order factor as reported by Taylor and Cox (1998b) for the (adult) ASI-R was tested. Second, a more parsimonious three-factor model (in which ‘fear of cardiovascular symptoms’ and ‘fear of respiratory symptoms’ were combined to one factor labelled ‘fear of physical symptoms’), as suggested in previous research on childhood measures of anxiety sensitivity (Silverman et al., 1999; Muris et al., 2001), was investigated. An additional aim of the current study was to investigate the psychometric properties of the CASI-R. First, the reliability (internal consistency) of the various CASI-R factors/scales was investigated. Second, the convergent and discriminant validity of the CASI-R was addressed. More specifically, relationships between the CASI-R and the original childhood index of anxiety sensitivity (i.e., the 18-item CASI), the trait anxiety version of the State–Trait Anxiety Inventory for Children (STAIC; Spielberger, 1973), and measures of anxiety disorders symptoms and depression were examined. It was hypothesised that CASI-R scores would correlate strongly with the original CASI (i.e., r⬎0.90), moderately with trait anxiety (r in the 0.60 or 0.70 range; see Taylor & Cox, 1998b; Muris et al., 2001), and with all of the anxiety disorders but most strongly with symptoms of panic disorder and agoraphobia. Furthermore, in keeping with Muris et al. (2001), it was expected that anxiety sensitivity and trait anxiety both have independent power in predicting symptoms of anxiety disorders and depression, suggesting that both constructs should be viewed as distinct vulnerability factors to psychopathology.

302

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

2. Method 2.1. Participants and procedure Five-hundred-and-eighteen adolescents (239 boys and 279 girls; mean age=14.9 years, SD=1.9, range 12–18 years; percentages of adolescents per age level were 12.7% 12-year-olds, 17.6% 13year-olds, 13.3% 14-year-olds, 17.2% 15-year-olds, 12.2% 16-year-olds, 15.4% 17-year-olds, and 11.6% 18-year-olds) were recruited from a regular secondary school. Participants completed a set of questionnaires (see below) in their classrooms. The teacher and a research assistant were always available to provide assistance if necessary and to ensure confidential and independent responding. 2.2. Questionnaires As mentioned earlier, the CASI and the CASI-R are both scales for measuring anxiety sensitivity in children and adolescents. In the present study, adolescents completed a 44-item questionnaire consisting of the 18 original CASI and 26 new items taken from Taylor and Cox’s (1998) ASI-R measure. Items had to be rated on three-point scales with 1=none, 2=some, and 3=a lot. CASI and CASI-R scores were computed by summing across relevant items. For the final data analysis of the CASI-R, only 31 items, eight from the CASI and 23 from the ASI-R (see Table 1), were used. Two clinical psychologists familiar with the anxiety sensitivity construct and one secondary school teacher were approached to make this selection of CASI-R items. Only those items that were attributed by the two psychologists to one and the same factor of anxiety sensitivity and that were judged by the schoolteacher as understandable to the 12-year-olds in our sample were retained in the final questionnaire. Thus, items were excluded because they (a) contained non-specific terms and hence did not pertain specifically to one of the four hypothesised anxiety sensitivity factors (e.g., “Unusual feelings in my body scare me”), (b) did not refer to fear or harmful consequences of anxiety (e.g., “It embarrasses me when my stomach growls”), or (c) were judged as too difficult to understand for the younger children in our sample (e.g., “It frightens me when my surroundings seem strange or unreal”). The trait version of the STAIC contains 20 items that measure chronic symptoms of anxiety. The child/adolescent is asked to rate the frequency with which (s)he experiences anxiety symptoms such as “I am scared”, “I feel troubled”, and “I get a funny feeling in my stomach” using three-point scales: 1=almost never, 2=sometimes, and 3=often. A total trait anxiety score can be calculated by summing the ratings on all items. The Spence Children’s Anxiety Scale (SCAS; Spence, 1998) is a self-report questionnaire measuring anxiety disorders symptoms in children and adolescents. The scale contains 38 items that can be allocated to the following subscales: generalised anxiety disorder (six items — e.g., “I worry that something bad will happen”); separation anxiety disorder (six items — e.g., “I feel scared when I have to sleep on my own”); social phobia (six items — e.g., “I feel afraid that I will make a fool of myself in front of people”); panic disorder and agoraphobia (nine items — e.g., “All of a sudden I feel really scared for no reason at all”, “I am afraid of being in crowded places”); obsessive–compulsive disorder (six items — e.g., “I have to think of special thoughts to stop bad things from happening”); and physical-injury fears replacing specific phobias (e.g. — “I am scared of insects or spiders”). SCAS items are rated on four-point scales: 0=never, 1=some-

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

303

Table 1 Factor loadings of CASI-R items as obtained by confirmatory factor analysis for a hierarchical model with four lowerorder factors loading on a single higher-order factor 1 Factor/subscale 1: Fear of cardiovascular symptoms When heart skipping a beat worry something seriously 0.82 wrong a 0.80 Scares me when heart beats fast 0.75 When heart beats fast worry something wronga When dizzy worry something wrong with brain 0.69 0.65 When stomach upset worry might be seriously illa When face feels numb worry might be stroke 0.65 When pain in chest worry going to have heart attack 0.63 When head pounding worry could have a stroke 0.62 Scares me when feel tingling or prickling sensations in 0.60 hands When strong pain in stomach worry could be cancer 0.50 Factor/subscale 2: Fear of publicly observable anxiety reactions Worry other people notice my anxiety When tremble in the presence of others fear what people think of me When sweat in the presence of others people think negatively of me Scares me when feel like throw upa Believe would be awful to vomit in public Scares me when blush in front of people Important not to appear nervousa Think would be horrible to faint in public Factor/subscale 3: Fear of cognitive dyscontrol When feel strange worry might go crazy When thoughts speed up worry might go crazy When mind goes blank worry something terribly wrong with me When trouble thinking clearly worry something wrong with me When cannot keep mind on schoolwork might go crazya Scares me when cannot keep mind on task a Factor/subscale 4: Fear of respiratory symptoms When chest feels tight scared cannot breathe properly When throat feels tight scared could choke to death Scares me when have feeling of choking When trouble swallowing worry could choke When breathing irregular fear something bad will happen When feel like not getting enough air scared might suffocate Scares me when short of breatha Loading of factor on higher-order factor 0.86 Note. CASI-R=Childhood Anxiety Sensitivity Index — Revised. a Items from the original CASI, other items were taken from the ASI-R.

2

3

4

0.76 0.70 0.65 0.61 0.61 0.60 0.60 0.54 0.89 0.88 0.65 0.55 0.53 0.53 0.77 0.73 0.72 0.69 0.66 0.63 0.76

0.75

0.61 0.88

304

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

times, 3=often, or 4=always. SCAS total and subscale scores are computed by summing relevant items. The Children’s Depression Inventory (CDI; Kovacs, 1981) is a commonly used self-report measure of depression symptoms in children and adolescents 7 to 17 years of age. The scale has 27 items dealing with sadness, self-blame, loss of appetite, insomnia, interpersonal relationships, and school adjustment. CDI items have to be scored on three-point scales with 0=not true, 1=somewhat true, or 2=very true. A total CDI score can be calculated by summing all item scores. 2.3. Statistical analysis The Statistical Package for Social Sciences (SPSS) was used to compute descriptive statistics, t-tests, and correlations. For confirmatory factor analysis, LISREL 8 (Jo¨ reskog & So¨ rbom, 1996) was employed. LISREL produces a wide range of goodness-of-fit indices of which the following were used in the present study: (a) chi square divided by degrees of freedom (with large sample sizes as in the current study, this value should be smaller than 4.00; the lower this value, the better the fit), (b) the Root Mean Square Residual (RMR; this value should be 0.05 or lower; the lower this value, the better the fit), (c) the Root Mean Square Error of Approximation (RMSEA; this value should be around 0.05 or lower; the lower the value, the better the fit), (d) the Comparative Fit Index (CFI; this value should be 0.90 or higher for a good fit; the higher this value, the better the fit), (e) the Non-Normed Fit Index (NNFI; this value should be 0.90 or higher for a good fit; the higher this value, the better the fit), (f) the Goodness-of-Fit Index (GFI; this value should be 0.90 or higher for a good fit; the higher the value, the better the fit) (for a discussion of the cut-off values of the goodness-of-fit indices, see, e.g., Gerbing & Anderson, 1993). As mentioned in the Introduction, two models for the CASI-R were tested: (1) a model with four lower-order factors (i.e., ‘fear of cardiovascular symptoms’, ‘fear of respiratory symptoms’, ‘fear of publicly observable anxiety reactions’, and ‘fear of cognitive dyscontrol’) loading on one higher-order factor, and (2) a more parsimonious model with three lower-order factors (i.e., ‘fear of physical symptoms’ [combining ‘fear of cardiovascular symptoms’ and ‘fear of respiratory symptoms’], ‘fear of publicly observable anxiety reactions’, and ‘fear of cognitive dyscontrol’) loading on one higher-order factor. Factor analyses were carried out for the total sample, for boys and girls, and for 12- to 15year-olds and 16- to 20-year-olds, separately. As these analyses essentially revealed a similar pattern of findings, only the results for the total sample will be presented hereafter. 3. Results 3.1. Factor structure of the CASI-R Confirmatory factor analysis indicated that the hierarchical factor structure with four lowerorder factors (i.e., ‘fear of cardiovascular symptoms’, ‘fear of publicly observable anxiety reactions’, ‘fear of cognitive dyscontrol’, and ‘fear of respiratory symptoms’) loading on a single higher-order factor (i.e., ‘anxiety sensitivity’) provided a good fit for the CASI-R data. Goodnessof-fit indices for this model were: chi square/degrees of freedom=3.6, RMR=0.05, RMSEA=0.07,

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

305

CFI=0.95, NNFI=0.94, and GFI=0.92. The factor loadings of CASI-R items for the hierarchical four-factor model are shown in Table 1. Note that all items loaded convincingly (i.e., ⱖ0.50) on their respective factor and that the four factors loaded substantially (i.e., ⱖ0.75) on the higherorder factor. A further confirmatory factor analysis revealed that the hierarchical model with three lowerorder factors (i.e., ‘fear of physical symptoms’, ‘fear of publicly observable anxiety reactions’, and ‘fear of cognitive dyscontrol’) loading on one higher-order factor yielded a less satisfactory fit for the data. Goodness-of-fit indices for this model were: chi square/degrees of freedom=5.2, RMR=0.05, RMSEA=0.09, CFI=0.91, NNFI=0.89, and GFI=0.87. A chi square difference test comparing the fits of both models indeed showed that the four-factor model provided a significantly better fit to the data than the more parsimonious three-factor model [c2(1)=167.3, P⬍0.001]. 3.2. Reliability of the CASI-R Cronbach’s alphas for the various CASI-R scales are displayed in the left column of Table 2. As can be seen, the alpha for the total scale was 0.93, the internal consistency for the various factors ranged between 0.81 and 0.88. Girls displayed somewhat higher levels of anxiety sensitivity than boys, mean CASI-R total scores being 41.3 (SD=9.7) versus 37.9 (SD=8.1) [t(515.7, adjusted df)=4.3, P⬍0.001]. For the CASI-R factors, significant gender differences were found for ‘fear of publicly observable anxiety reactions’ [t(516.0, adjusted df)=6.4, P⬍0.001] and ‘fear of respiratory symptoms’ [t(513.6, adjusted df)=4.0, P⬍0.001] (see Table 2). In passing, it should be mentioned that the other questionnaires (i.e., the original CASI, STAIC, SCAS, and CDI) were also reliable. Only the SCAS physical-injury fears subscale had a Cronbach’s alpha that was insufficient: 0.55. Furthermore, in keeping with earlier studies, girls had higher levels of anxiety sensitivity (as indexed by the original CASI), trait anxiety, most anxiety disorders symptoms, and depression than boys. 3.3. Convergent and discriminant validity of the CASI-R 3.3.1. Relationship with original CASI and measure of trait anxiety Anxiety sensitivity as measured by the CASI-R total score was strongly associated with the total score of the original CASI: r=0.93, P⬍0.001. Furthermore, the CASI-R total score was also substantially connected to trait anxiety as indexed by the STAIC: r=0.73, P⬍0.001 (Table 3). Note, however, that this correlation was significantly smaller than that between the CASI-R and the CASI (Z=9.0, P⬍0.001), supporting the discriminant validity of the new scale. 3.3.2. Relationship with anxiety disorders symptoms and depression Table 3 also shows correlations (corrected for gender) between anxiety sensitivity as measured by the CASI-R, on the one hand, and anxiety disorders symptoms (SCAS) and depression (CDI), on the other hand. Three main conclusions can be drawn from these results. First, the CASI-R total score was significantly associated with symptoms of anxiety disorders (r values between 0.50 and 0.80, all P values⬍0.001] and depression (r=0.62, P⬍0.001). Second, the CASI-R total score was significantly more strongly related to the SCAS subscale ‘panic disorder and agora-

306

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

Table 2 Cronbach’s alphas, means (SDs between parentheses), and gender differences for the CASI-R and the other questionnaires a CASI-R Total score Fear of cardiovascular symptoms Fear of publicly observable anxiety reactions Fear of cognitive dyscontrol Fear of respiratory symptoms CASI (original scale) STAIC SCAS Total score Generalised anxiety disorder Separation anxiety disorder Social phobia Panic disorder and agoraphobia Obsessive–compulsive disorder Physical-injury fears CDI

Total group (N=518) Boys (n=239)

Girls (n=279)

0.93

39.7 (9.2)

37.9 (8.1)a

41.3 (9.7)b

0.88

11.6 (2.9)

11.5 (2.8)a

11.8 (3.2)a

0.85

11.6 (3.3)

10.6 (2.9)a

12.4 (3.4)b

0.81

7.2 (1.9)

7.0 (1.7)a

7.3 (2.1)a

0.88

9.4 (2.9)

8.8 (2.5)a

9.8 (3.1)b

0.87 0.91

20.8 (4.9) 30.4 (8.1)

19.7 (4.3)a 28.2 (7.4)a

21.8 (5.2)b 32.3 (8.2)b

0.93

17.1 (13.5)

13.1 (11.3)a

20.6 (14.1)b

0.81

3.9 (3.0)

3.1 (2.6)a

4.6 (3.2)b

0.65 0.75

2.0 (2.1) 3.7 (3.0)

1.4 (1.7)a 2.9 (2.7)a

2.5 (2.2)b 4.3 (3.0)b

0.84

2.3 (3.4)

1.5 (2.6)a

3.0 (3.8)b

0.77

3.0 (3.1)

2.8 (2.9)a

3.2 (3.2)a

0.55 0.88

2.3 (2.2) 8.1 (6.9)

1.5 (2.0)a 7.0 (6.0)a

3.0 (2.2)b 9.0 (7.4)b

Notes. CASI-R=Childhood Anxiety Sensitivity Index — Revised; CASI=Childhood Anxiety Sensitivity Index, STAIC=trait anxiety version of the State–Trait Anxiety Inventory for Children; SCAS=Spence Children’s Anxiety Scale; CDI=Children’s Depression Inventory. Means in the same row that do not share the same subscripts differ at P⬍0.05/16.

phobia’ than to any of the other SCAS subscales (all Z values⬎3.7, P values⬍0.001).1 Third, correlations between CASI-R factor scores and SCAS subscales revealed that all factors loaded most strongly on the SCAS subscale ‘panic disorder and agoraphobia’, except for the factor ‘fear of publicly observable anxiety reactions’ which was most convincingly associated with SCAS ‘social phobia’. Partial correlations (corrected for gender) calculated separately for anxiety sensitivity as indexed by the CASI-R total score and trait anxiety (STAIC) with measures of anxiety disorders symptoms and depression while holding the other constant are displayed in Table 4. Results showed that even when controlling for levels of trait anxiety, anxiety sensitivity was significantly associated with anxiety disorders symptoms (r values between 0.21 and 0.53, all P values⬍0.001) and depression (r=0.18, P⬍0.001). Furthermore, comparisons of the partial correlations revealed that 1

Correlations were compared using the method as described by Meng, Rosenthal, and Rubin (1992).

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

307

Table 3 Correlations (corrected for gender) among CASI-R variables and other measures CASI-R Total score CASI-R Total score Fear of cardiovascular symptoms (1) Fear of publicly observable anxiety reactions (2) Fear of cognitive dyscontrol (3) Fear of respiratory symptoms (4) CASI (original scale) STAIC SCAS Total score Generalised anxiety disorder Separation anxiety disorder Social phobia Panic disorder and agoraphobia Obsessive–compulsive disorder Physical-injury fears CDI

(1)

(2)

(3)

(4)

0.84 0.82

0.53

0.78 0.85 0.93 0.73

0.56 0.67 0.75 0.52

0.56 0.55 0.84 0.70

0.58 0.78 0.60

0.74 0.58

0.80 0.66 0.64 0.60 0.76 0.61 0.50 0.62

0.61 0.52 0.50 0.36 0.62 0.48 0.40 0.47

0.73 0.61 0.54 0.69 0.61 0.55 0.45 0.58

0.64 0.50 0.51 0.47 0.62 0.52 0.36 0.52

0.65 0.54 0.55 0.41 0.66 0.46 0.43 0.48

Notes. N=518. All correlations were significant at P⬍0.001. CASI-R=Childhood Anxiety Sensitivity Index — Revised; CASI=Childhood Anxiety Sensitivity Index; STAIC=trait anxiety version of the State–Trait Anxiety Inventory for Children; SCAS=Spence Children’s Anxiety Scale; CDI=Children’s Depression Inventory. Table 4 Partial correlations (corrected for gender) calculated separately for anxiety sensitivity (CASI-R total score) and trait anxiety (STAIC) with measures of anxiety disorders symptoms (SCAS) and depression (CDI) while holding the other constant

SCAS Total score Generalised anxiety disorder Separation anxiety disorder Social phobia Panic disorder and agoraphobia Obsessive–compulsive disorder Physical-injury fears CDI

CASI-R controlling for trait anxiety

STAIC controlling for anxiety sensitivity

0.53a 0.31a 0.36a 0.21a 0.53a 0.29a 0.25a 0.18a

0.50a 0.43a 0.34a 0.43b 0.28b 0.31a 0.19a 0.52b

Notes. N=518. All correlations were significant at P⬍0.001. CASI-R=Childhood Anxiety Sensitivity Index — Revised; STAIC=trait anxiety version of the State–Trait Anxiety Inventory for Children; SCAS=Spence Children’s Anxiety Scale; CDI=Children’s Depression Inventory. Correlations in the same row that do not share same subscripts differ at P⬍0.05/16.

308

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

anxiety sensitivity was more strongly connected to ‘panic disorder and agoraphobia’ (Z=9.8, P⬍0.001), whereas trait anxiety was more substantially associated with ‘social phobia’ (Z=7.3, P⬍0.001) and depression (Z=11.3, P⬍0.001). 4. Discussion The current data show that anxiety sensitivity in children and adolescents as measured by the CASI-R can best be conceptualised as a hierarchical construct with four lower-order factors loading on a single higher-order factor. The lower-order factors were ‘fear of cardiovascular symptoms’, ‘fear of publicly observable anxiety reactions’, ‘fear of cognitive dyscontrol’, and ‘fear of respiratory symptoms’. Note that these findings were highly similar to those reported by Taylor and Cox (1998b) for an extended version of the anxiety sensitivity measure for adults (i.e., the ASI-R). Note also that the factor structure of the CASI-R largely confirms the three main factors (‘fear of physical symptoms’, ‘fear of publicly observable anxiety reactions’, and ‘fear of cognitive dyscontrol’) that have emerged in previous studies on the anxiety sensitivity construct in children and adolescents (Muris et al., 2001; Silverman et al., 1999). An additional aim of the present study was to examine the psychometric properties of the CASI-R. Results indicated that the internal consistency of the CASI-R was good and that this was not only the case for the total scale but also for the separate dimensions of anxiety sensitivity. Furthermore, the CASI-R total score was substantially related to scores on the original anxiety sensitivity measure for children and adolescents (i.e., the CASI), symptoms of anxiety disorders, in particular ‘panic disorder and agoraphobia’, and depression. Finally, some evidence was found for the validity of the CASI-R dimensions. That is, all dimensions were strongly associated with symptoms of ‘panic disorder and agoraphobia’, whereas the dimension ‘fear of publicly observable anxiety reactions’ was also convincingly associated with symptoms of ‘social phobia’. While the 18-item CASI provides a brief and economical measure of anxiety sensitivity in children and adolescents, the expanded CASI-R may yield important additional information on the specific dimensions of this construct. By using the CASI-R, researchers could gain insight into the distinct mechanisms that play a role in the development and maintenance of specific types of anxiety and fear reactions. Putting it in Taylor and Cox’s (1998b, p. 481) words: “By identifying the factors of AS [anxiety sensitivity] (with the ASI-R or other measures) an important next step is to determine whether the factors differ in their psychopathological correlates. To illustrate, some AS factors may be better than others in predicting emotional responses to panicogenic agents. Fear of respiratory symptoms, for example, may be the best predictor of panic attacks induced by CO2 inhalation. Fear of publicly observable anxiety reactions may be the best predictor of panic attacks induced by agents that cause trembling or sweating (e.g., drinking strong, hot coffee). Thus, the importance of the ASI-R [and its child version the CASI-R] may lie in its potential to stimulate further research into the causes and consequences of AS”. In a similar vein, Silverman et al. (1999; p. 915) pointed at the treatment implications of a multidimensional scale for measuring anxiety sensitivity in youths: “it might be worthwhile to target certain facets of AS in some [psychopathologic] conditions and to target other facets in other conditions”. Thus, researchers and clinicians who want to have a global indication of children and adolescents’ level of anxiety sensitivity can certainly use the original CASI. However, those who are interested in a more fine-grained assessment of the construct should preferably use the CASI-R.

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

309

As an aside for those who intend to employ the CASI-R, it is important to note that children should not be too young. Although effort was made to simplify the scale as much as possible, it is important to control whether children properly understand each of the items. The adolescents in the present study appeared to understand the items of the CASI-R and we believe that this is probably also the case with children aged between 10 and 12 years. For younger children, however, some items of the CASI-R might be too difficult and for them the use of the original CASI is recommended. Both anxiety sensitivity and trait anxiety are regarded as vulnerability factors for the development of anxiety disorders (Rachman, 1998). While it is clear that anxiety sensitivity and trait anxiety are inter-correlated (in the present study: r=0.73, P⬍0.001) and that this is true for both adult and child populations (e.g., Taylor, Koch, & Crockett, 1991; Silverman et al., 1991), there is agreement among researchers that both constructs are conceptually different from each other. Trait anxiety has been defined as the tendency to react anxiously to potentially anxiety-provoking stimuli, whereas anxiety sensitivity is viewed as the more specific tendency to react anxiously to one’s own anxiety and anxiety-related sensations (for a brief discussion, see Taylor, 1995). In the current study, anxiety sensitivity and trait anxiety both accounted for unique proportions of the variance of anxiety disorders symptoms. Results also indicated that anxiety sensitivity was more convincingly associated with ‘panic disorder and agoraphobia’, whereas trait anxiety was more strongly connected to ‘social phobia’ (see also Muris et al., 2001). Although it should be acknowledged that the present study was cross-sectional in nature and hence does not allow for causal interpretation of the correlations, the data support the idea that anxiety sensitivity and trait anxiety should be viewed as distinct vulnerability factors that have independent predictive power in the aetiology of anxiety disorders. Several researchers have put forward that anxiety sensitivity may also be involved in depression (Otto, Pollack, Fava, Uccello, & Rosenbaum, 1995; Taylor, Koch, Woody, & McLean, 1996). The present study found a significant correlation between anxiety sensitivity and childhood depression (see also Muris et al., 2001; Weems, Hammond-Laurence, Silverman, & Ferguson, 1997). When controlling for levels of trait anxiety, this correlation clearly attenuated but still reached statistical significance. Taylor et al. (1996) have proposed that the relationship between anxiety sensitivity and depression is due to one specific component of anxiety sensitivity, i.e., ‘fear of loss of cognitive control’ (see also Taylor & Cox, 1998b). The present data do not support this idea: although depression was somewhat more strongly related to ‘fear of cognitive dyscontrol’ than to ‘fear of cardiovascular symptoms’ and ‘fear of respiratory symptoms’, the most substantial correlation was found with ‘fear of publicly observable anxiety reactions’. It is not clear what caused this aberrant finding and further research on the connections between anxiety sensitivity dimensions and depression in child and adolescent samples seems necessary to resolve this issue. In his review article, Taylor (1995, p. 255) called for new measures of anxiety sensitivity in order “to further our understanding of the fear of anxiety”. The CASI-R is such a new index that can be used to assess anxiety sensitivity in children and adolescents. The scale has a hierarchical factor structure with four lower-order factors loading on a single higher-order factor, which is in keeping with current conceptualisations of the anxiety sensitivity construct. Furthermore, the psychometric properties of the CASI-R are satisfactory, that is, its reliability is good and there is good evidence for its convergent and discriminant validity. Future employment of this scale

310

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

could provide a more detailed picture of the role of anxiety sensitivity in the development and maintenance of anxiety disorders in children and adolescents. References Chorpita, B. F., Albano, A. M., & Barlow, D. H. (1996). Child anxiety sensitivity index: considerations for children with anxiety disorders. Journal of Clinical Child Psychology, 25, 77–82. Cox, B. J. (1996). The nature and assessment of catastrophic thoughts in panic disorder. Behaviour Research and Therapy, 34, 363–374. Cox, B. J., Parker, J. D., & Swinson, R. P. (1996). Anxiety sensitivity: confirmatory evidence for a multidimensional construct. Behaviour Research and Therapy, 34, 591–598. Gerbing, D. W., & Anderson, J. C. (1993). Monte Carlo evaluations of goodness-of-fit indices for structural equation models. In K. A. Bollen, & J. S. Long (Eds.), Testing structural equation models (pp. 40–65). Newbury Park, CA: Sage. Jo¨ reskog, K., & So¨ rbom, D. (1996). LISREL 8: User’s reference guide. Chicago, IL: Scientific Software International. Kearney, C. A., Albano, A. M., Eisen, A. R., Allan, W. D., & Barlow, D. H. (1997). The phenomenology of panic disorder in youngsters: an empirical study of a clinical sample. Journal of Anxiety Disorders, 11, 49–62. Kovacs, M. (1981). Rating scales to assess depression in school-aged children. Acta Paedopsychiatrica, 46, 305–315. Lau, J. J., Calamari, J. E., & Waraczynski, M. (1996). Panic attack symptomatology and anxiety sensitivity in adolescents. Journal of Anxiety Disorders, 19, 355–364. Mattis, S. G., & Ollendick, T. H. (1997). Children’s cognitive responses to the somatic symptoms of panic. Journal of Abnormal Child Psychology, 25, 47–57. Meng, X. L., Rosenthal, R., & Rubin, D. B. (1992). Comparing correlated correlation coefficients. Psychological Bulletin, 111, 172–175. Muris, P., Schmidt, H., Merckelbach, H., & Schouten, E. (2001). Anxiety sensitivity in adolescents: factor structure and its relationship to trait anxiety and symptoms of anxiety disorders and depression. Behaviour Research and Therapy, 39, 89–100. 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, 117–123. Rabian, B., Embry, L., & MacIntyre, D. (1999). Behavioral validation of the Childhood Anxiety Sensitivity Index in children. Journal of Clinical Child Psychology, 28, 105–112. Rabian, B., Peterson, R. A., Richters, J., & Jensen, P. S. (1993). Anxiety sensitivity among anxious children. Journal of Clinical Child Psychology, 22, 441–446. Rachman, S. (1998). Anxiety. Hove, UK: Psychology Press. Reiss, S., Peterson, R. A., Gursky, D. M., & McNally, R. J. (1986). Anxiety sensitivity, anxiety frequency, and the prediction of fearfulness. Behaviour Research and Therapy, 24, 1–8. Silverman, W. K., Fleisig, W., Rabian, B., & Peterson, R. A. (1991). Childhood anxiety sensitivity index. Journal of Clinical Child Psychology, 20, 162–168. Silverman, W. K., Ginsburg, G. S., & Goedhart, A. W. (1999). Factor structure of the childhood anxiety sensitivity index. Behaviour Research and Therapy, 37, 903–917. Spence, S. H. (1998). A measure of anxiety symptoms among children. Behaviour Research and Therapy, 36, 545–566. Spielberger, C. D. (1973). Manual for the state–trait anxiety inventory for children. Palo Alto, CA: Consulting Psychologists Press. Taylor, S. (1995). Anxiety sensitivity: theoretical perspectives and recent findings. Behaviour Research and Therapy, 33, 243–258. Taylor, S., & Cox, B. J. (1998a). Anxiety sensitivity: multiple dimensions and hierarchic structure. Behaviour Research and Therapy, 36, 37–51. Taylor, S., & Cox, B. J. (1998b). An expanded anxiety sensitivity index: evidence for a hierarchic structure in a clinical sample. Journal of Anxiety Disorders, 12, 463–484.

P. Muris / Behaviour Research and Therapy 40 (2002) 299–311

311

Taylor, S., Koch, W. J., & Crockett, D. J. (1991). Anxiety sensitivity, trait anxiety, and the anxiety disorders. Journal of Anxiety Disorders, 6, 293–311. 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. Weems, C. F., Hammond-Laurence, K., Silverman, W. K., & Ferguson, C. (1997). The relation between anxiety sensitivity and depression in children and adolescents referred for anxiety. Behaviour Research and Therapy, 35, 961–966. Weems, C. F., Hammond-Laurence, K., Silverman, W. K., & Ginsburg, G. S. (1998). Testing the utility of the anxiety sensitivity construct in children and adolescents referred for anxiety disorders. Journal of Clinical Child Psychology, 27, 69–77. Zinbarg, R. E., Barlow, D. H., & Brown, T. A. (1997). Hierarchical structure and general factor saturation of the anxiety sensitivity index. Psychological Assessment, 9, 277–284.