Behaviour Research and Therapy 67 (2015) 30e40

Contents lists available at ScienceDirect

Behaviour Research and Therapy journal homepage: www.elsevier.com/locate/brat

Anxiety sensitivity, catastrophic misinterpretations and panic selfefficacy in the prediction of panic disorder severity: Towards a tripartite cognitive model of panic disorder
nchez-Arribas b, Paloma Chorot a, Rosa M. Valiente a Bonifacio Sandin a, *, Carmen Sa a b


n a Distancia, Juan del Rosal 10, 28040 Madrid, Spain Facultad de Psicología, Universidad Nacional de Educacio Hospital General de Úbeda, Úbeda, Ja
en, Spain

a r t i c l e i n f o

a b s t r a c t

Article history: Received 12 July 2014 Received in revised form 18 January 2015 Accepted 22 January 2015 Available online 3 February 2015

The present study examined the contribution of three main cognitive factors (i.e., anxiety sensitivity, catastrophic misinterpretations of bodily symptoms, and panic self-efficacy) in predicting panic disorder (PD) severity in a sample of patients with a principal diagnosis of panic disorder. It was hypothesized that anxiety sensitivity (AS), catastrophic misinterpretation of bodily sensations, and panic self-efficacy are uniquely related to panic disorder severity. One hundred and sixty-eight participants completed measures of AS, catastrophic misinterpretations of panic-like sensations, and panic self-efficacy prior to receiving treatment. Results of multiple linear regression analyses indicated that AS, catastrophic misinterpretations and panic self-efficacy independently predicted panic disorder severity. Results of path analyses indicated that AS was direct and indirectly (mediated by catastrophic misinterpretations) related with panic severity. Results provide evidence for a tripartite cognitive account of panic disorder. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Panic disorder Anxiety sensitivity Catastrophic misinterpretations Self-efficacy Cognitive models

Introduction Cognitive accounts of panic disorder (PD) have emphasized the role of different cognitive factors, including catastrophic misinterpretation of bodily sensations (Clark, 1986), anxiety sensitivity (McNally, 2002; Reiss & McNally, 1985), and panic self-efficacy (Casey, Oei, & Newcombe, 2004). Clark's model has generally been the dominant cognitive model of PD. The central characteristic of this model is the occurrence of catastrophic misinterpretation of benign bodily sensations (a catastrophe, in this context, was conceptualized by Clark as an immediate, impending disaster). This theory has been valuable in explaining the repetitive nature of panic attacks through a circular mechanism of cognition and perception, and it has demonstrated cognitive driven treatment applications (Craske & Barlow, 2008). Even though some studies have not provided a clear support for the relationship between catastrophic misinterpretation of bodily sensations and PD (Fentz et al., 2013; Richards, Richardson, & Pier, 2002), clinical evidence tends to support the role of catastrophic misinterpretation of bodily sensations in the prediction of PD severity and in the mechanisms

* Corresponding author. Tel.: þ34 91 3986254, þ34 639 123 638. E-mail address: [email protected] (B. Sandin). http://dx.doi.org/10.1016/j.brat.2015.01.005 0005-7967/© 2015 Elsevier Ltd. All rights reserved.

of change in cognitive-behavior therapy for PD (Casey, Newcombe, & Oei, 2005; Casey, Oei, Newcombe, & Kenardy, 2004; Clark, 1999; Hofmann et al., 2007; Khawaja & Oei, 1998). Likewise, prior research has suggested that panic disordered patients are more likely than people with other anxiety disorders to interpret bodily sensations as physical or psychological catastrophes (Austin & Richards, 2006; Clark et al., 1997; Khawaja & Oei, 1998). Anxiety sensitivity (AS) is a construct conceptually distinct from trait anxiety (McNally, 1994; Sandin, Chorot, & McNally, 2001; Taylor, 1999) and closely related to the concept of catastrophic misinterpretations of bodily symptoms. It was first described by Reiss and McNally (1985) as a dispositional variable defined by the fear of anxiety symptoms, arising from beliefs that the experience of fear/anxiety and related physical sensations has harmful somatic, psychological or social consequences. Even though AS has shown to be a relevant factor across all the anxiety disorders (NaragonGainey, 2010; Wheaton, Deacon, McGrath, Berman, & Abramowitz, 2012), convergent evidence from cross-sectional and longitudinal studies has revealed positive relationships between AS and: (a) a diagnosis of PD (Chorot, Sandin, Valiente, Santed, & Romero, 1997; Olatunji & Wolitzky-Taylor, 2009; Sandin, Chorot, & McNally, 1996; Taylor, 1999; Taylor, Koch, & McNally, 1992), (b) measures of PD symptoms (Deacon & Valentiner, 2001), (c) development of new panic attacks (Li & Zinbarg, 2007; Schmidt,

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40


rez-Benítez Lerew, & Jackson, 1999), (d) course of PD severity (Pe et al., 2009), and (e) changes in PD severity during cognitivebehavior therapy for PD (Gallagher et al., 2013) (see NaragonGainey, 2010, for a recent meta-analysis). Cross-cultural research has shown that the structure of AS is multifactorial, consisting of a superordinate factor (general AS) and three lower order correlated factors (physical concerns, cognitive concerns, and social concerns) (Sandin, Chorot, Valiente, Santed, & Lostado, 2004; Taylor, 1999; Taylor et al., 2007; Zinbarg, Barlow, & Brown, 1997). In addition, it has been suggested that the lowerorder dimensions of AS are specifically related to particular types of anxiety symptoms and disorders. Accordingly, physical concerns appears to be most closely related with PD and/or PD symptoms whereas social concerns were found to be strongly related to fear of negative evaluation and social phobia (Deacon & Abramowitz,
rez-Benítez et al., 2009; 2006; Olthuis, Watt, & Stewart, 2014; Pe Rodríguez, Bruce, Pagano, Spencer, & Keller, 2004; Taylor et al., 2007; Wheaton et al., 2012). Although results concerning the specificity of the cognitive dimension are mixed, recent works suggest a strong association between AS cognitive concerns with panic symptoms (Naragon-Gainey, 2010) and with a diagnosis of PD and/or generalized anxiety disorder (Taylor et al., 2007; Wheaton et al., 2012). However, no relevant associations have been found between AS social and PD (diagnosis or symptom severity). Both AS and catastrophic misinterpretation of bodily sensations are negative cognitions related to fear of interoceptive symptoms. Generally, much of the past research has equated these two constructs, being considered at least as overlapping constructs. According to some authors (Cox, 1996; Taylor, 1995, 2000), Clark's cognitive theory could be subsumed within the broader theory of fears, phobia and panic developed by Reiss and McNally (1985). However, such authors distinguished these two concepts in terms of state-trait, describing anxiety sensitivity as a trait (i.e., an individual difference variable) that predisposes to make catastrophic misinterpretation of bodily sensations (i.e., a state cognitive construct). That is to say, people with high AS have an “enduring tendency” to make catastrophic misinterpretations (catastrophic arousal beliefs ei.e., a state). In a similar vein, McNally (1994) stated that whereas AS is a dispositional construct, catastrophic misinterpretation of bodily sensations is an episodic concept. However, apart from this state-trait distinction, McNally (2002) also outlined more core conceptual differences between AS and catastrophic misinterpretation of bodily symptoms. Perhaps the most relevant one was that, in contrast to Clark's model, the AS approach does not imply the existence of catastrophic attributions. As McNally stated, the anxiety-sensitivity hypothesis assumes that people with high AS may dread bodily sensations as merely signaling another panic or fear of panic, not necessarily in the catastrophic way described by Clark (1986). Therefore, according to this author a main difference between these two constructs is that AS may operate on panic process and severity without the intervention of catastrophic misinterpretations. In line with this theoretical differentiation between these two cognitive factors involved in PD, one could argue that AS is a dispositional variable different from catastrophic misinterpretation of bodily symptoms. Given the lack of previous works concerning this question, a first purpose of the present study was to examine the unique association between AS and catastrophic misinterpretation in predicting panic severity. Based on prior research, we assumed that both AS and catastrophic misinterpretation of bodily symptoms are critical in determining panic severity. Also, based on the conceptual separation between AS and catastrophic misinterpretation of bodily sensations, we expected that AS should predict panic severity after controlling for the effects of catastrophic misinterpretation of bodily sensations. Thus, high levels of AS and

31

high levels of catastrophic misinterpretation of bodily sensations should independently predict high panic severity. Assuming the three components of AS and catastrophic misinterpretation (i.e., physical, cognitive/mental, and social), we hypothesized that AS physical concerns should predict PD severity partialling out the effects of the physical dimension of catastrophic misinterpretation. Although the association of panic with AS cognitive concerns is less consistent than with AS physical concerns, based on recent research (Naragon-Gainey, 2010; Taylor et al., 2007; Wheaton et al., 2012) we expected a similar pattern; in contrast, no significant effects were expected for AS social. Also, based on past theories concerning relationships between AS and catastrophic misinterpretations (Cox, 1996; McNally, 1990, 1994; Taylor, 1995, 2000), a second goal was to investigate whether catastrophic misinterpretation meditates the association between AS and PD severity. It was expected that AS should be positively associated with panic severity, and that this relationship might be mediated by catastrophic misinterpretations (indirect effect of AS). According to theory (Fava & Morton, 2009; McNally, 2002; Pilecki, Arentoft, & McKay, 2011), we also hypothesized a direct relationship between AS and panic severity, independent of the influence of catastrophic misinterpretations. Recently, Casey, Oei, and Newcombe (2004) emphasized the implication of panic self-efficacy (a positive cognitive factor) in the psychopathology of PD. Panic self-efficacy refers to the individual's perceived ability to cope with or control perceived danger in relation to panic attacks (Casey, Oei, Newcombe, et al., 2004). These authors proposed a new cognitive model of panic (named “integrated cognitive model of panic disorder”) in which both catastrophic misinterpretation of bodily sensations and panic self-efficacy independently contribute to maintenance of panic severity. Catastrophic misinterpretation of bodily sensations and a lack of panic self-efficacy are assumed to play central roles in the psychopathology of PD (Casey, Oei, & Newcombe, 2004). A few recent works have examined the role of panic self-efficacy and catastrophic misinterpretations as predictors of PD severity in patients with a diagnosis of PD. Some recent studies have provided preliminary empirical support for a relationship between these constructs and PD severity (Casey, Oei, Newcombe, et al., 2004; Fentz et al., 2013; Gallagher et al., 2013; Richards et al., 2002). Casey, Oei, Newcombe, et al. (2004) investigated the role of self-efficacy and catastrophic misinterpretation of bodily sensations, concluding that these two constructs independently predicted panic severity. Gallagher et al. (2013) provided evidence that changes in self-efficacy and AS during cognitive-behavior therapy influence subsequent changes in panic symptoms. Fentz et al. (2013) found that panic self-efficacy was significantly related to anxiety symptoms during the course of cognitive-behavior therapy for panic disordered patients, and Richards et al. (2002) found slight support for a predictive power of panic self-efficacy on panic severity in a sample of people with PD. Thus a third aim of the present study was to examine simultaneously the role of these three core cognitive factors (AS, catastrophic misinterpretation of bodily sensations, and panic self-efficacy) in predicting panic severity. We hypothesize that AS should predict panic severity after controlling for the effects of catastrophic misinterpretation of bodily sensations and panic self-efficacy. It was hypothesized a similar pattern to panic selfefficacy and catastrophic misinterpretation of bodily sensations; each of these variables should predict panic severity after controlling for the remaining two cognitive factors. Likewise, we expected a similar pattern including congruent domains of AS and catastrophic misinterpretation, except for the social domain.

32

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40

Method Participants The sample consisted of 168 adult individuals seeking treatment at a Spanish health center (Hospital San Juan de la Cruz de Úbeda,
n) with a diagnosis of PD with/without agoraphobia, according Jae to DSM-IV criteria (American Psychiatric Association, 1994). We used the following inclusion criteria: age 18 years, a principal diagnosis of PD with/without agoraphobia according to DSM-IV criteria, duration of the current episode 3 months, and, if medicated, to have had a stable dose for 3 months prior to the entry. The exclusion criteria were as following: an organic mental disorder, current psychotic symptoms, current and/or past schizophrenia or bipolar affective disorder, substance abuse, and severe depressive disorder that require immediate intervention. The mean age of the sample was 40.1 years (range 18e66 years; SD ¼ 10.6). The characteristics of the sample are shown in Table 1.

Measures
stico de los Trastornos de Ansiedad Entrevista para el Diagno [Anxiety Disorders Diagnostic Interview] ADIS-M (Valiente, Sandin, & Chorot, 2003). The ADIS-M is a structured clinical interview designed for the assessment of anxiety disorders and other comorbid emotional disorders. It is a Spanish modification of the Anxiety Disorders Interview ScheduleeRevised (ADIS-R; Di Nardo & Barlow, 1988) adapted to DSM-IV criteria. The ADIS-M is a reliable tool for diagnosing PD and agoraphobia. In addition to the diagnostic information, it was used to assess the following five indicators of PD severity by means of 5 items of the ADIS-M: frequency of panic attacks (last month), intensity of panic attacks (the mean for the 13 items of the panic attack symptoms scale), concerns about having a new panic attack, changes in behaviors related to the panic attacks, and impairment in family, occupational and social functioning due to the panic. Each item was rated on 5-point rating scale (ranged from 0 to 4), with higher score representing greater severity of each item. A composite score, representing overall panic disorder severity, was the sum of the five indicators, ranging from 0 to 20. In this study, Cronbach's a and test-retest reliability (one month) for the 5-item clinical rating and the 13item panic attack symptoms scale were .73/.71 and .85/.70, respectively.
nico y Agoraphobia [Panic Disorder and Cuestionario de Pa Agoraphobia Questionnaire] (CPA; Sandin, Chorot, Valiente,
nchez-Arribas, & Santed, 2004). The CPA is a self-report instruSa ment aimed at providing a comprehensive measure for the assessment and diagnosis of PD and agoraphobia according to the DSM-IV criteria. The structure of the CPA is based on the Panic Attack Questionnaire (Cox, Norton, & Swinson, 1992). By means of Table 1 Demographics and clinical characteristics of the sample (N ¼ 168). Age (years, mean/SD) Gender (women, n/%) Marital status (married, n/%) Educational level (high school or higher, n/%) Working status (full-time employed, %) Primary diagnosis (PD, n/%) Primary diagnosis (PDA, n/%) Comorbidity (present/not present, n/%) Comorbid anxiety disorder (current, n/%) Comorbid mood disorder (current, n/%) Psychotropic medication (current, n/%) PD history (years, mean/SD)

40.1 126 114 103 143 54 114 73/95 62 11 94 6.0

(10.6) (75.6) (67.9) (61.3) (85.2) (32.1) (67.9) (43.5/56.5) (36.9) (6.5) (56) (5.6)

40 items, the CPA provides detailed information on the clinical characteristics of PD and agoraphobia, including history, situational triggers, frequency, and intensity of panic attack symptoms, strategies for coping with panic attacks, anticipatory anxiety, functional impairment, catastrophic cognitions, interoceptive and agoraphobic avoidance, personal and family history of stress and psychopathology, and demographic data. For purposes of the present study, the CPA provided information concerning PD severity, catastrophic misinterpretation of bodily sensations, and panic selfefficacy. An overall composed score of panic disorder severity (ranging from 0 to 24) was calculated by summing the ratings of the following 6 items: frequency of panic attacks (last month), intensity of panic attacks (mean for a subscale scoring of 13 items representing the panic attack symptoms; Cronbach's a in the present study ¼ .88, and test-retest ¼ .68), concerns about having a new panic attack, worry about possible effects of panic attacks, changes in behaviors related to the attacks, and impairment in family, occupational and social functioning due to the panic (mean for items 25e27 of impairment due to the panic); each item was rated on 5-point rating scale (ranging from 0 to 4), with higher score representing greater severity of each item. Reliability of the 6-item scale in the present study was .76 (Cronbach's alpha) and .72 (one month test-retest). The scale is highly correlated with Houck, Spiegel, Shear, and Rucci (2002) Panic Disorder Severity Scale
nchez-Arribas, Valiente, & Sandin, 2014). (PDSS) (r ¼ .82) (Chorot, Sa The PDSS is a widely used and validated international scale of PD severity. Catastrophic misinterpretation of bodily sensations was assessed by means of the Panic Catastrophic Misinterpretations Scale (PCMS), a CPA 14-item frequency scale, ranging from 0 (never) to 4 (always); it was based on the Catastrophic Cognitions Questionnaire € (1993) con(Khawaja & Oei, 1992), the work of Westling and Ost cerning relationship between panic attacks and catastrophic cognitions in patients with PD, and our clinical experience. The items describe common catastrophic beliefs about what will happen during the attack, including physical (e.g., to have a heart attack), mental (e.g., to lose the control), and social (e.g., that people will look at you). Higher scores denote greater endorsement of catastrophic pathological cognitions. Data concerning reliability and
nchez-Arribas, validity of the PCMS have been reported (Valiente, Sa
nchez-Arribas, 2006). Chorot, & Sandin, 2014; Valiente, Sandin, & Sa For example, based on confirmatory factor analysis we found that the PCMS consists of three separate correlated factors labeled physical, mental, and social catastrophes. Also, we found a high correlation between the PCMS and the Panic Appraisal InventorydPanic Consequences (PAI-PC; Telch, Brouillard, Telch, Agras, & Taylor, 1989) total scores (r ¼ .82) and empirical support for convergent and discriminant validity based on intercorrelations between the subscales of both questionnaires (e.g., correlations between similar subscales ranged from .75 to .66) (Valiente et al., 2014). The alpha coefficient of the PCMS in the present study was .89, and one month test-retest reliability was .84. Subscales alpha and test-retest coefficients were, respectively: .84/.70 (physical), .85/.84 (mental), and .75/.74 (social). Panic self-efficacy was measured through the Panic Self-efficacy Scale (PSES), a CPA 4-item scale related to the level of confidence in the ability to cope effectively with panic attacks, based on current self-report measures of self-efficacy for PD (Taylor & Arnow, 1988; Telch et al., 1989). The PSES lists four situations related to self-confidence in effective coping with panic attacks and panic symptoms: (a) at panic attack onset, (b) during the full-blown attack, (c) to prevent escape or agoraphobic avoidance, and (d) to handle catastrophic misinterpretations. Each item is scored from 0 (not at all confident) to 4 (absolutely confident). Higher ratings reflect greater degree of confidence in coping with panic symptoms

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40

and panic attacks. The measure has demonstrated sound psychometric properties (Chorot et al., 2014; Olmedo, Sandin, Santed, & García-Campayo, 2006). For example, PSES strongly correlated with Telch et al.'s (1989) PAI-Panic Coping (r ¼ .66), a scale of panic self-efficacy (Chorot et al., 2014). The PSES had an alpha coefficient of .87 in the present study; test-retest (one month) reliability ¼ .80. Anxiety Sensitivity Indexe3 (ASI-3; Taylor et al., 2007). The ASI-3 is an 18-item self-report instrument that assesses the fear of anxiety-related sensations. Participants rate their concerns of anxiety-related sensations on a 5-point Likert scale, ranging from 0 (“very little”) to 4 (“very much”). The ASI-3 is composed of three 6-item subscales: Physical Concerns (fears of physical symptoms), Cognitive Concerns (fears of cognitive dyscontrol), and Social Concerns (fears of negative evaluation). Both a total score (range ¼ 0e72) and three separate subscale scores can be obtained (range ¼ 0e24). The Spanish adaptation of the ASI-3 appears to be a valid instrument, internally consistent, and test-retest reliable (Sandin, Valiente, Chorot, & Santed, 2007; Taylor et al., 2007). In the present study, Cronbach's alpha and test-retest (one month) reliability of the ASI-3 were .90 and .88, respectively. Subscales alpha and test-retest coefficients were, respectively: .87/.85 (physical), .91/.90 (mental), and .89/.88 (social). Procedure Participants were consecutive outpatients with a principal diagnosis of PD with or without agoraphobia who were seeking psychological treatment. Some of the selected patients were referred to mental health from other units of the hospital. The patients were initially screened by clinical psychologists from the health center staff. At this step, each participant underwent a psychiatric semistructured interview based on the DSM-IV criteria, and he/she was referred to us if found with a possible diagnosis of PD with/without agoraphobia. As a second step, after informed written consent was obtained, selected patients were interviewed again and diagnosed by the second author, who used the ADIS-M. The interviewer had experience with anxiety disorders and extensive training in ADIS-M administration and scoring. Primary diagnosis was established using a clinical severity rating scale that indicates the degree of distress and impairment related to the disorder, ranging from 0 (none) to 8 (very severely disturbing/ disabling). A sample of 10% of the interviews was audiotaped and was reassessed by the remaining authors who were blind to the original diagnoses. There was a 100% of agreement on primary diagnosis. When a diagnosis was unclear, consensus among the authors was taken as the criterion. After the ADIS-M interview, participants completed a battery of self-report measures, including the ones described in the present study. Data analysis overview Firstly, means, standard deviations, reliability (Cronbach's alpha and test-retest) and Person's correlations among all study variables were calculated. Secondly, a series of hierarchic linear regression analyses were carried out to examine the unique contribution of AS, catastrophic misinterpretations, and panic self-efficacy in the prediction of PD severity. These hierarchic regression analyses were run to examine the extent to which each of the three core cognitive factors (AS, catastrophic misinterpretations, and panic self-efficacy) is effective in the prediction of PD severity. Thirdly, a series of multiple linear regression analyses were conducted in which the independent variables (panic self-efficacy and selected subscales of AS and catastrophic misinterpretations) were entered simultaneously in each equation. These regression analyses were carried out to examine the specificity of panic self-efficacy and the

33

subscales of AS (physical, cognitive, and social) and catastrophic misinterpretation (physical, mental, and social) in the prediction of PD severity. Fourthly, in order to examine the mediation hypotheses (catastrophic misinterpretations as mediator of the effect of AS on panic severity), we conducted a series of path analyses based on the PROCESS Procedure for SPSS 2.11 (Hayes, 2013); in order to facilitate the interpretation of the results, prior to computations all variables were converted to standardized values (z-scores). Given that some independent variables were highly correlated, before calculating the regression analyses we checked for the issue of multicollinearity between the predictors. We examined the variance inflation factor (VIF) and the two criteria suggested by Tabachnick and Fidell (2007), i.e., condition index and variance proportions associated with each variable. In all of the regression analyses, values of VIF were less than 10 (VIFs ranged from 1.01 to 1.84), suggesting low levels of multicollinearity. Moreover, in all of the analyses the conditioning indexes for each dimension were less than 30 (condition indexes ranged from 11.41 to 11.78) and there were no variance proportions greater than .50 for at least two different variables. Thus, since none of our predictors violated the criteria, it appears that multicollinearity is not a threat to the validity of our analyses. Before conducting the multivariate regression analyses, a series of univariate regressions was conducted to check if clinical variables (comorbidity, psychotropic medication, and duration of PD) exerted a differential impact on PD severity, AS, catastrophic misinterpretations and panic self-efficacy. None of these variables significantly predicted any of the focal variables of the study (comorbidity: R2's  .008, F's  1.23, p's > .05; psychotropic medication: R2's  .038, p's > .05, F's  3.49, p's > .05; and duration of disorder: R2's  .009, F's  1.24, p's > .05). Thus, such variables were not statistically controlled in the current hierarchical regression analyses. In order to control for the influence of agoraphobia, a dichotomous variable (1 vs. 0) denoting presence or absence of agoraphobia was included in all the regression analyses. Agoraphobia was controlled for by entering it as predictor in all of the regression analyses (it was entered at step 1 in the hierarchic regression analyses). We examined the distribution of each of the predictor and criterion measures, finding a lack of skewness and kurtosis (values were less than 1 for all variables). Results Correlations and descriptive statistics The correlations, means, standard deviations, alpha coefficients, and test-retest (one month) for all study variables are reported in Table 2. As can be seen, all correlations between negative variables (AS and catastrophic misinterpretations) and PD severity were statistically significant, ranging from .24 to .61. As expected, correlations of panic self-efficacy with PD severity were negative and statistically significant. Self-reported and clinical-rated PD severity were strongly correlated (r ¼ .67, p < .001). There were no significant gender differences in PD severity (self-reported and clinicalrated) and in the remaining focal variables of the study (AS, catastrophic misinterpretation, and panic self-efficacy); F's ranged from 1.52 (AS) to .10 (self-reported PD severity). Thus, gender was not included in subsequent analyses. Anxiety sensitivity, catastrophic misinterpretations, and panic selfefficacy as predictors of panic disorder severity A series of multiple linear regression analyses was carried out to examine the relationship of AS, catastrophic misinterpretations and panic self-efficacy with PD severity. Separate analyses were run to

34

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40

Table 2 Means, standard deviations, reliability (Cronbach's alpha and test-retest) and zero-order correlations for panic disorder severity, panic self-efficacy, and subscales of anxiety sensitivity and catastrophic misinterpretations. Measure

1

1. Panic disorder severity (ADIS-M) 2. Panic disorder severity (CPA) 3. ASI-3 Total 4. ASI-3 Physical 5. ASI-3 Cognitive 6. ASI-3 Social 7. PCMS Total 8. PCMS Physical 9. PCMS cognitive 10. PCMS Social 11. Panic self-efficacy (PSES)

e

Mean SD Alpha Test-retest (one month)

12.23 3.40 .73 .71

2

.67 .47 .42 .48 .24 .48 .43 .44 .27 .27

3

4

5

6

7

8

9

10

11

e .60 .51 .57 .41 .61 .60 .53 .30 .36 15.26 5.98 .76 .72

e .76 .87 .77 .64 .52 .55 .41 .21 31.71 15.79 .90 .88

e .55 .32 .45 .58 .31 .14 .23 11.74 6.33 .87 .85

e .55 .61 .45 .63 .32 .27

e

9.85 6.68 .91 .90

10.68 6.64 .89 .88

.48 .27 .34 .57 .11

e .79 .84 .72 .23

e .49 .36 .25

e .43 .18

e .08

9.30 4.63 .84 .70

10.46 5.47 .85 .84

8.61 4.10 .75 .74

28.37 11.28 .89 .84

5.89 3.24 .87 .80


nico y Agorafobia [Panic Disorder and Note. ADIS-M ¼ Anxiety Disorders Interview ScheduledModified; ASI-3 ¼ Anxiety sensitivity Indexd3. CPA ¼ Cuestionario de Pa Agoraphobia Questionnaire]; PCMS ¼ Panic Catastrophic Misinterpretations Scale; PSES ¼ Panic Self-efficacy Scale. Correlations  .23, p < .01; correlations  .18, p < .05.

predict self-reported (CPA; see Table 3) and clinician-rated (ADISM; see Table 4) PD severity. The core predictor variables were AS, catastrophic misinterpretations, and panic self-efficacy. Each of the six regression analyses were run in four steps. Firstly, in order to control for the influence of the presence vs. absence of agoraphobia, for each of the equations the main effects of agoraphobia was entered in Step 1. Secondly, the main effects of panic self-efficacy (Equation (1)), AS (Equation (2)), and catastrophic misinterpretations (Equation (3)) were entered in Step 2. And thirdly, to examine the incremental prediction of each of the three main predictors, AS (Equation (1)), catastrophic misinterpretations (Equation (2)), and panic self-efficacy (Equation (3)) were entered in Step 4 (see Tables 3 and 4). Results of the hierarchical regressions predicting self-reported PD severity are reported in Table 3. As shown in this table, significant main effects were found for each predictor variable adding an important portion of explained variance. At Step 4, we examined

the possibility that each of the three focal predictors may provide an incrementally predictive variance above the effect of the remaining independent variables. AS, catastrophic misinterpretations, and panic self-efficacy were incrementally predictive of self-reported PD severity. Thus, it appears that the three variables significantly added to the prediction of self-reported panic severity, after controlling for agoraphobia (see Table 3). In the analyses of clinician-rated panic severity, in general the pattern of results was similar to the one found for self-reported PD severity. Each of the core independent variables was able to predict at Step 4 a statistically significant portion of the variance controlling for the effect of the remaining predictors (see Table 4). In order to examine whether the influence of the three core constructs (AS, catastrophic misinterpretations and panic selfefficacy) could be understood in terms of a single higher-order construct representing the moderating influence of these factors (Casey, Oei, Newcombe, et al., 2004), we tested three two-way

Table 3 Summary of hierarchical regression analyses examining anxiety sensitivity, catastrophic misinterpretations of bodily sensations, and panic self-efficacy in the prediction of self-reported panic severity (CPA).

Table 4 Summary of hierarchical regression analyses examining anxiety sensitivity, catastrophic misinterpretations of bodily sensations, and panic self-efficacy in the prediction of clinician-rated panic severity (ADIS-M).

Steps and predictor variables Equation 1 Step 1: Presence/absence of agoraphobia Step 2: Panic self-efficacy (PSES) Step 3: Catastrophic misinterpretations (PCMS) Step 4: Anxiety sensitivity (ASI-3) Equation 2 Step 2: Anxiety sensitivity (ASI-3) Step 3: Panic self-efficacy (PSES) Step 4: Catastrophic misinterpretations (PCMS) Equation 3 Step 2: Catastrophic misinterpretations (PCMS) Step 3: Anxiety sensitivity (ASI-3) Step 4: Panic self-efficacy (PSES)

R2

DR2

B

SE B b

.11*** .11*** 4.30 .99 . 22*** .11*** .59 .14 .42*** .20*** .27 .04

.34*** .32*** .50***

.48**

.06*** .12

.31***

.36*** .42*** .48**

.25*** .21 .02 .06*** .42 .12 .06*** .17 .04

.53*** .23*** .32***

.37***

.26*** .30

.55***

.44*** .48**

.07*** .13 .03 .04** .35 .11

.03

.03

.34*** .20**

Note. Step 1 for Equations 2 and 3 is omitted; it is the same as for Equation 1. ASI
nico y Agorafobia [Panic 3 ¼ Anxiety sensitivity Indexd3; CPA ¼ Cuestionario de Pa Disorder and Agoraphobia Questionnaire]; PCMS ¼ Panic Catastrophic Misinterpretations Scale; PSES ¼ Panic Self-efficacy Scale. **p < .01, ***p < .001.

Steps and predictor variables Equation 1 Step 1: Presence/absence of agoraphobia Step 2: Panic self-efficacy (PSES) Step 3: Catastrophic misinterpretations (PCMS) Step 4: Anxiety sensitivity (ASI-3) Equation 2 Step 2: Anxiety sensitivity (ASI-3) Step 3: Panic self-efficacy (PSES) Step 4: Catastrophic misinterpretations (PCMS) Equation 3 Step 2: Catastrophic misinterpretations (PCMS) Step 3: Anxiety sensitivity (ASI-3) Step 4: Panic self-efficacy (PSES)

R2

DR2

B

SE B

b

.11*** 17*** .27***

.11*** .06** .11***

2.41 .26 .11

.56 .07 .02

.33*** -.25*** .35***

.30***

.03*

.05

.02

.21*

.24*** .27*** .30***

.13*** .03* .03*

.08 .18 .07

.02 .07 .03

.39*** -.18* .23*

.25***

.14***

.12

.03

.40***

.28*** .30***

.03* .02*

.05 .16

.02 .08

.23* -.15*

Note. Step 1 for Equations 2 and 3 is omitted; it is the same as for Equation 1. ADISM ¼ Anxiety Disorders Interview ScheduledModified; ASI-3 ¼ Anxiety sensitivity Indexd3; PCMS ¼ Panic Catastrophic Misinterpretations Scale; PSES ¼ Panic Selfefficacy Scale. *p < .05, **p < .01, ***p < .001.

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40

interactions and a three-way interaction between these specific factors, after controlling for agoraphobia. We ran a hierarchical simultaneous regression analysis in which agoraphobia was controlled in Step 1, main effects of AS, catastrophic misinterpretations and panic self-efficacy were controlled in Step 2, and all two-way interactions and the three-way interaction were entered in Step 3. Prior to computation all independent variables were converted to standardized z-scores. Neither results for selfreported panic severity, nor results for clinician-rated panic severity revealed any significant two-way (b's  .03, t's  .33) or three-way (b's  .10, t's  1.04) interaction between AS, catastrophic misinterpretation and panic self-efficacy. Thus, each of the interactions failed to significantly increase the amount of variance in PD severity above the main effects of the specific factors. These null interactions could be due to insufficient power. Specificity of self-efficacy and dimensions of anxiety sensitivity and catastrophic misinterpretations as predictors of panic disorder severity We conducted a series of multiple linear regression analyses to examine the extent to which comparable subscales of AS and catastrophic misinterpretations uniquely predicted PD severity. Six separate regression analyses were run to test the prediction of selfreported (first three runs; Table 5) and clinician-rated PD severity (last three runs; Table 6). The four predictor variables entered into each analysis were agoraphobia, panic self-efficacy, and subscales of AS and catastrophic misinterpretations (physical, cognitive/ mental, or social). The predictors were entered simultaneously in the equation. This approach allowed us to account for overlap between the predictors and provided a stringent test of incremental validity. In the first regression analysis (Equation (1)), the incremental validity of physical subscales of AS and catastrophic misinterpretations in the prediction of PD severity was examined. Together the four variables explained a significant percentage of the variance in self-reported panic severity (44%). After controlling for the other variables, each predictor accounted for unique variance in panic severity. In the second analysis (Equation (2)), we examined the unique relations between cognitive/mental subscales

Table 5 Specificity of panic self-efficacy and subscales of catastrophic misinterpretation and anxiety sensitivity in predicting self-reported panic severity (CPA) controlling for agoraphobia. Dependent and predictor variable

R2

Equation 1 Presence/absence of agoraphobia ASI-3 Physical PCMS Physical Panic self-efficacy (PSES)

.44***

Equation 2 Presence/absence of agoraphobia ASI-3 Cognitive PCMS Mental Panic self-efficacy (PSES)

.46***

Equation 3 Presence/absence of agoraphobia ASI-3 Social PCMS Social Panic self-efficacy (PSES)

.32***

B

SE B

b

Partial r

1.76 .19 .47 .39

.85 .07 .10 .11

.14* .21** .36*** .21**

.17 .22 .35 .26

2.99 .25 .30 .39

.80 .07 .08 .12

.23*** .29** .27** .21**

.29 .28 .27 .26

.3.25 .26 .04 .56

.89 .07 .12 .06

.26*** .29** .03 .31***

.28 .27 .03 .35

Note. Independent variables were entered together in each equation. ASI
nico y Agorafobia [Panic 3 ¼ Anxiety sensitivity Indexd3; CPA ¼ Cuestionario de Pa Disorder and Agoraphobia Questionnaire]; PCMS ¼ Panic Catastrophic Misinterpretations Scale; PSES ¼ Panic Self-efficacy Scale. *p < .05, **p < .01, ***p < .001.

35

Table 6 Specificity of panic self-efficacy and subscales of catastrophic misinterpretation and anxiety sensitivity in predicting clinician-rated panic severity (ADIS-M) controlling for agoraphobia. Dependent and predictor variables

R2

Equation 1 Presence/absence of agoraphobia ASI-3 Physical PCMS Physical Panic self-efficacy (PSES)

.28***

Equation 2 Presence/absence of agoraphobia ASI-3 Cognitive PCMS Mental Panic self-efficacy (PSES)

.33***

Equation 3 Presence/absence of agoraphobia ASI-3 Social PCMS Social Panic self-efficacy (PSES)

.20***

B

SE B

b

Partial r

1.26 .11 .47 .17

.55 .05 .10 .07

.17* .20* .22** .17*

.18 .18 .19 .18

.1.73 .14 .11 .15

.49 .04 .05 .07

.24** .27** .18* .15*

.27 .24 .16 .17

1.95 .03 .09 .26

.56 .04 .07 .07

.26** .07 .12 .24**

.27 .06 .11 .26

Note. Independent variables were entered together in each equation. ADISM ¼ Anxiety Disorders Interview ScheduledModified; ASI-3 ¼ Anxiety sensitivity Indexd3; PCMS ¼ Panic Catastrophic Misinterpretations Scale; PSES ¼ Panic Selfefficacy Scale. *p < .05, **p < .01, ***p < .001.

of AS and catastrophic misinterpretations and self-reported PD severity. As in Equation (1), together the four variables explained a significant amount of the variance (46%). Also, after controlling for the other variables, each predictor accounted for unique variance in panic severity. Finally, the incremental validity of the social dimension of AS and catastrophic misinterpretations was examined in the third regression analysis (Equation (3)). In this analysis, the social dimension of catastrophic misinterpretations did not account for unique variance in self-reported panic severity. In the three regression analyses relating to clinician-rated PD severity we obtained similar results than the ones found on self-reported scores (see Table 6). Catastrophic misinterpretations as a mediator of the effect of anxiety sensitivity on panic severity To examine the hypothesis that catastrophic misinterpretations could mediate the effects of AS on PD severity, we tested the direct and indirect effects of AS on panic severity by means of the simple mediation model depicted in Fig. 1. As diagramed in the figure, we proposed that AS (X) leads to increased panic severity (Y) through a direct effect on Y and as a result of activation of catastrophic misinterpretations (M), which in turns yields to an increase of panic severity. That is to say, it was predicted that the experience of AS may influence panic severity by two ways, i.e., a direct effect and an indirect effect mediated by catastrophic misinterpretations (the more such interpretations result, the greater the increase in panic severity). The main results of the mediation analyses are summarized in Fig. 1. The coefficients (a, b and c0 ) represent the fully standardized regression coefficients for each antecedent variable in the statistical model of the consequent. As can be seen, all regression coefficients were statistically significant, both for self-reported and clinicianrated PD severity. Likewise, the indirect effect of X on Y was statistically different from zero, as revealed by a 95% bias-corrected bootstrap confidence interval and normal theory-based Sobel test, both for self-reported (bootstrap IC ¼ .13e.38; Sobel test: Z ¼ 4.32, p < .001) and clinician-rated panic severity (bootstrap IC ¼ .06e.28; Sobel test: Z ¼ 2.81, p < .01). As shown in Fig. 1, results are very similar for self-reported and clinician-rated PD severity. The total

36

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40

Fig. 1. Conceptual and statistical diagram of the mediation model for the direct and indirect effects of anxiety sensitivity on self-reported panic severity and clinician-rated panic severity (data concerning clinician-rated are shown in parenthesis). X ¼ independent, Y ¼ dependent, M ¼ mediator. Regression coefficients (fully standardized): a ¼ effect of X on M, b ¼ effect of M on Y, c0 ¼ effect of X on Y. ASI-3 ¼ Anxiety sensitivity Indexd3; PCMS ¼ Panic Catastrophic Misinterpretation Scale. ***p < .001; **p < .01.

effect of X on Y was statistically significant for self-reported panic severity (R2 ¼ .35, F(1, 143) ¼ 76, 94, p < .001; coeff. ¼ .60, p < .001), as well as for clinician rated panic severity (R2 ¼ .21, F(1, 140) ¼ 36,21, p < .001; coeff. ¼ .45, p < .001), indicating that two people who differ by one unit in AS are estimated to differ by .60 and .45 units in PD severity (self-reported and clinician-rated, respectively). Analogous mediational models were tested to examine the hypothesis that dimensions of catastrophic misinterpretations could mediate the effects of congruent dimensions of SA on PD severity. In a first model (Fig. 2), we proposed that AS physical (X) leads to increased panic severity (Y) through a direct effect on Y and as a result of activation of physical catastrophic misinterpretations (M), which in turns yields to an increase of panic severity (the model

and the main results are shown in Fig. 2). All effects were statistically significant. The indirect effect was statistically significant on both measures of panic severity, i.e., self-reported severity (effect size ¼ . 27; bootstrap CI ¼ .18e.37; Sobel test: Z ¼ 4.95, p < .001) and clinician-rated severity (effect size ¼ . 17; bootstrap CI ¼ .07e.27; Sobel test: Z ¼ 3.11, p < .01). The total effect of X on Y was statistically significant for self-reported panic severity (R2 ¼ .26, F(1, 154) ¼ 52.12, p < .001; coeff. ¼ .51, p < .001), as well as for clinician rated panic severity (R2 ¼ .17, F(1, 151) ¼ 26,21, p < .001; coeff. ¼ .42, p < .001). In a second model, the physical dimension of SA and catastrophic misinterpretations were replaced by the cognitive/mental dimension. As can be seen in Fig. 3, all regression parameters were statistically significant on both measures of PD severity. The

Fig. 2. Conceptual and statistical diagram of the mediation model for the direct and indirect effects of anxiety sensitivity (physical concerns subscale) on self-reported panic severity and clinician-rated panic severity (data concerning clinician-rated are shown in parenthesis). X ¼ independent, Y ¼ dependent, M ¼ mediator. Regression coefficients (fully standardized): a ¼ effect of X on M, b ¼ effect of M on Y, c0 ¼ effect of X on Y. ASI-3 ¼ Anxiety sensitivity Indexd3; PCMS ¼ Panic Catastrophic Misinterpretation Scale. ***p < .001; **p < .01.

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40

indirect effect size of X on Y for self-reported panic severity ¼ .20 (bootstrap CI ¼ .08e.30; Sobel test: Z ¼ 3.44, p < .001) and for clinician rated ¼ .12 (bootstrap CI ¼ .02e.23; Sobel test: Z ¼ 2.05, p < .05). Total effects of X on Y: R2 ¼ .32, F(1, 151) ¼ 71.09, p < .001; coeff. ¼ .57, p < .001(self-reported panic severity), and R2 ¼ .24, F(1, 148) ¼ 71.09, p < .001; coeff. ¼ .47, p < .001 (clinician-rated panic severity). Lastly, a third model was defined by the social facet of AS (independent) and catastrophic misinterpretation (mediator). There was no evidence of a significant indirect effect of the social dimension of AS on PD severity, either of the self-reported (effect size ¼ .05, n.s.) or clinician-rated (effect size ¼ .09, n.s.) PD severity. Discussion The present study investigated the contribution of AS, catastrophic misinterpretation of panic symptoms, and coping selfefficacy as predictors of PD severity in a clinical sample of patients with a diagnosis of PD. A firs goal of the study was to examine the relative influence of AS and catastrophic misinterpretation in the prediction of PD. Results revealed that AS significantly predicted PD controlling for the influence of catastrophic misinterpretation. Likewise, catastrophic misinterpretation was found to predict PD after controlling for AS. Thus, it appears that AS and catastrophic misinterpretation are uniquely related to PD. These results provide preliminary empirical support for the assumption that AS and catastrophic misinterpretation of panic-like symptoms are differentiable constructs which may have separate effects on PD (McNally, 2002). This author stated that the anxiety sensitivity perspective of PD does not require that an individual misconstrues sensations as catastrophic for panic to be highly aversive (McNally, 2002). Results are also consistent with empirical evidence that catastrophic misinterpretations are not necessary for the experience of panic (McNally, 1994) and that not all individuals with PD € experience catastrophic cognitions (Westling & Ost, 1993). As suggested by Fava and Morton (2009), AS and catastrophic misinterpretation may act on PD at different levels of the system underling the vicious circle of the panic attack. In addition, based on the fact that anxiety sensitivity and catastrophic misinterpretation may be understood as multidimensional constructs, we investigated the unique association of the

37

dimensions of AS and catastrophic misinterpretations with PD severity, i.e., we examined whether equivalent components of these factors (physical, cognitive/mental, and social) are independently associated with PD severity. In this respect, we found specific relationships of panic severity with the physical and cognitive domains, but not with the social domain. That is to say, AS physical concerns were effective to predict panic severity beyond the effect of the physical domain of catastrophic misinterpretation and, conversely, the physical dimension of catastrophic misinterpretation predicted panic severity after controlling for the influence of AS physical concerns. Similar results were found concerning the cognitive dimension. However, the social domain failed to provide unique contribution to the prediction of panic severity. From these findings two important conclusions can be derived. Firstly, besides that AS and catastrophic misinterpretation, as global constructs, appear to act as different cognitive factors, the physical and cognitive components of these two constructs also appear to work as distinct cognitive phenomena in the prediction of panic severity. Namely, the physical component of AS seems to be different from the physical domain of catastrophic misinterpretation, and the cognitive dimension of AS is distinct from the mental dimension of catastrophic misinterpretation. Secondly, the present findings suggest that the cognitive/mental domain is also important in the prediction of panic severity. Considering that past research suggested that AS physical concerns, as compared to cognitive or social ones, are more closely linked to PD (e.g., Deacon & Abramowitz, 2006; Grant, Beck, & Davila, 2007; Olthuis et al.,
rez-Benítez et al., 2009; Rodriguez et al., 2004) this 2014; Pe conclusion is somewhat surprising. However, recent evidence also found that panic is most closely related to both the physical and the cognitive dimensions of AS (Naragon-Gainey, 2010; Rector, Szacun-Shimizu, & Leybman, 2007; Taylor et al., 2007; Wheaton et al., 2012). Additional support comes from several works which found AS cognitive concerns to be the only significant prospective predictor of panic, after shared variance among the AS domains was controlled (Li & Zinbarg, 2007; Schmidt et al., 1999; see Naragon-Gainey, 2010, for a meta-analysis). Our results also are in line with the suggestion that some patients with a diagnostic of PD primarily fear the potential cognitive consequences of panic attacks (Cox, Swinson, Endler, & Norton, 1994; Wheaton et al., 2012).

Fig. 3. Conceptual and statistical diagram of the mediation model for the direct and indirect effects of anxiety sensitivity (cognitive concerns subscale) on self-reported panic severity and clinician-rated panic severity (data concerning clinician-rated are shown in parenthesis). X ¼ independent, Y ¼ dependent, M ¼ mediator. Regression coefficients (fully standardized): a ¼ effect of X on M, b ¼ effect of M on Y, c0 ¼ effect of X on Y. ASI-3 ¼ AS Indexd3; PCMS ¼ Panic Catastrophic Misinterpretation Scale. ***p < .001; **p < .01; * p < .05.

38

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40

A second goal of the present study was to examine the extent to which catastrophic misinterpretations mediate the relationship between AS and PD severity. We found that AS was able to predict panic severity in a dual way. Firstly, by means of an indirect effect AS was related to an increased level of panic severity as a result of the tendency for those patients under relative more AS to misinterpret panic-like symptoms catastrophically, which in turn translates into greater panic severity. Secondly, AS may also influence levels of panic severity in a direct way; i.e., patients who tend to experience more fear of symptoms of anxiety are more likely to exhibit high levels of panic severity, with independence of their possible catastrophic misinterpretations of the panic-like bodily sensations. The ratio of indirect-to-direct effect of AS on self-reported and clinician-rated panic severity was 42% and 37%, respectively, indicating that the direct effect appears to be slightly larger than the indirect effect. To our knowledge, this is the first empirical evidence concerning the mediational role of catastrophic misinterpretations in the relationship between SA and PD severity. This basic mediation model was also applied to examine to what extent each component of catastrophic misinterpretation (physical, cognitive/mental, and social) was able to mediate the relationship between each equivalent component of AS and panic severity. Also, it was expected that components of AS should predict panic severity independently of the mediation effect. Results concerning the physical and cognitive/mental dimensions were very similar to those found for AS and catastrophic misinterpretation total scores. The ratio of indirect-to-direct effect of physical AS on self-reported and clinician-rated panic severity was 53% and 40%, and the ratio of cognitive AS was 35% and 26%, respectively, suggesting a slightly preponderance of the direct effect for the cognitive component of AS. The lack of mediation effect found for the social component is consistent with its low and not significant association with panic severity as showed in our regression analyses.

Finally, a third main objective was to investigate the unique association between the three core cognitive factors (i.e., AS, catastrophic misinterpretation, and panic self-efficacy) and PD severity. Consistent with Casey, Oei, Newcombe, et al. (2004), catastrophic misinterpretations significantly predicted PD severity after controlling for the influence of panic self-efficacy and, conversely, panic self-efficacy predicted panic severity after controlling for catastrophic misinterpretation. These results are in accordance with findings reported by Gallagher et al. (2013) suggesting that both panic self-efficacy and AS uniquely predict PD severity throughout cognitive-behavior therapy treatment, as well as with those of Casey et al. (2005) who found that catastrophic misinterpretation and panic self-efficacy may be mechanisms of change of cognitive-behavior therapy for PD. However, by simultaneously examining the contribution of these three main cognitive factors, the present study advances upon this previous research in demonstrating that the three factors uniquely predict PD severity. Our findings showed that (a) AS was found to predict panic severity partialling out catastrophic misinterpretations and panic selfefficacy, (b) catastrophic misinterpretations predicted panic severity after controlling for AS and panic self-efficacy, and (c) panic self-efficacy predicted panic severity after controlling for AS and catastrophic misinterpretations. Thus, these results extend findings reported in previous research on relations between cognitive positive and negative factors of PD severity (Casey et al., 2005; Casey, Oei, Newcombe, et al., 2004; Fentz et al., 2013; Gallagher et al., 2013; Richards et al., 2002). The present findings support the cognitive model of panic described by Casey, Oey and Newcombe (2004), which integrates positive (panic self-efficacy) and negative (catastrophic misinterpretation) cognitive factors. However, our data suggest a more comprehensive tripartite cognitive model of panic that integrates the three main cognitive factors, i.e., AS, catastrophic misinterpretations, and panic self-efficacy (see Fig. 4). The model

Fig. 4. Tripartite cognitive model of panic disorder.

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40

depicted in Fig. 4 is a descriptive model based on our results and prior research on relationships between the three core cognitive factors and panic severity reviewed in the present study, as well as on past theory (Casey, Oei, & Newcombe, 2004; Fava & Morton, 2009; McNally, 1990, 1994, 2002; Pilecki et al., 2011; Taylor, 1995, 2000). As shown in the figure, it consists of an extension of Casey et al.'s model by taking into account that AS, in addition to catastrophic misinterpretations and panic self-efficacy, independently contributes to severity of panic as a mediating cognition in Clark's (1986) circular account of panic. Thus AS may influence the ongoing occurrence of panic by means of direct and/or indirect effects mediated by catastrophic misinterpretations. According to this proposal, when high AS is coupled with low panic self-efficacy the individual may overreact to internal threats (bodily sensations) and this may lead to an increased experience of fear and to a panic attack with or without the intervention of catastrophic misinterpretations (Fava & Morton, 2009; McNally, 1994). Therefore, in some instances Clark's vicious circle of panic would work without catastrophic misinterpretations. On the other hand, AS may also operate via catastrophic misinterpretations to facilitate the panic attack. In addition, we suggest that not only the physical component but also the cognitive component of both AS and catastrophic misinterpretations are relevant in the vicious circle of the panic attack. Taken together, the findings of the present study have several clinical implications. Given the core role of the three cognitive factors on PD severity, cognitive behavior therapy could be specifically targeted to reducing both the increased levels of AS and catastrophic misinterpretations, as well to increasing low levels of panic self-efficacy. Second, understanding the unique contributions of the nuclear cognitive factors described in the tripartite model could help to target main vulnerabilities in order to implement intervention programs. For example, they may promote positive cognitions related to the perception of bodily symptoms and/or regarding control or coping with panic-like symptoms. Third, the findings highlight the multidimensional nature of AS and catastrophic misinterpretation of bodily symptoms and suggest that the physical and cognitive/mental components of AS and/or catastrophic misinterpretation could represent unique treatment targets in cognitive behavior therapy. For example, when physical concerns are dominant, interoceptive exposure exercises, such as hyperventilation or physical exercise (Smits et al., 2008) may be useful. When cognitive concerns predominate, treatment might be maximized by focusing on promoting thinking skills (Craske & Barlow, 2007). Four, results allow for the prediction of possible specific subtypes of panic attacks. For example, apart from the physical subtypes, a cognitive subtype could be seen to develop characterized by the presence of the cognitive components of AS and catastrophic misinterpretations. This subtype may present common features with the cognitive subtype described by Kircanski, Craske, Epstein, and Wittchen (2009), which has been characterized as the presence of excessive fear in the absence of elevated somatic or physiological responding during a panic attack. Finally, it would be worthwhile to know the extent to which the three core cognitive factors investigated in the present study correspond to distinct phenomenological characteristics and/or clinical severity course of PD, which could help to identify specific vulnerabilities to PD. The present study has some limitations that should be noted. A first limitation was that the measure of PD severity is novel in representing of various scales, and the measures of catastrophic misinterpretations and panic self-efficacy are scales integrated in a more comprehensive questionnaire on PD and agoraphobia. However, the self-report PD severity measure was recently validated using correlations with the Panic Disorder Severity Scale (PDSS),

39

which is considered an evidence-based assessment instrument (Keller & Craske, 2008). Likewise, this potential weakness was not a threat because both scales have been tested for internal (factor structure) and external validity (correlations with the dimensions of the Panic Appraisal Inventory, a standard measure of catastrophic misinterpretations and panic self-efficacy). Second, method variance may have inflated the relationship between cognitive factors and self-reported PD severity. Third, given the substantial number of analyses executed, we are aware of the potential for Type I error. Finally, the tripartite model presented here predicts PD severity by means of an independent contribution of AS, catastrophic misinterpretations, and panic self-efficacy. However, the cross-sectional nature of the study limits the conclusions that could be drawn concerning the role of these cognitive factors in the etiology of PD. Future longitudinal research based on these cognitive constructs as causal or risk factors may permit possible etiological inferences to be made. For example, future longitudinal studies could assess or monitor PD characteristics over longer time intervals to allow for changes in panic symptoms associated to changes in the cognitive factors. Also, future prospective research could determine the temporal association between cognitive factors and the onset of panic attacks and PD. Based on the predictions of the model, it may be worthwhile to use longitudinal studies to determine possible specific vulnerabilities based on the three components of AS and catastrophic misinterpretation. A final recommendation is that the model suggests ways in which research may focus on understanding possible mechanisms of change for cognitive behavior therapy for PD. For example, through techniques such as virtual reality (Botella et al., 2014) and latent class growth analysis or latent growth mixture modeling (Gallagher et al., 2013), we may examine in future studies to what extent changes in SA, catastrophic misinterpretations, and panic self-efficacy predict distinct patterns of response during treatment.

Conflict of interest The authors declared that there is no conflict of interest.

References American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Austin, D. W., & Richards, J. C. (2006). A test of core assumptions of the catastrophic misinterpretation model of panic disorder. Cognitive Therapy and Research, 30, 53e68.
n-Lo
pez, J., Serrano, B., García-Palacios, A., Quero, S., & Ban ~ os, R. Botella, C., Breto (2014). Treatment of flying phobia using virtual reality exposure with or without cognitive restructuring: participants' preferences. Revista de Psicopatología y Psicología Clínica, 19, 157e169. Casey, L. M., Newcombe, P. A., & Oei, T. P. S. (2005). Cognitive mediation of panic severity: the role of castastrophic misinterpretation of bodily sensations and panic self-efficacy. Cognitive Therapy and Rerearch, 29, 187e200. Casey, L. M., Oei, T. P., & Newcombe, P. A. (2004). An integrated cognitive model of panic disorder: the role of positive and negative cognitions. Clinical Psychology Review, 24, 529e555. Casey, L. M., Oei, T. P. S., Newcombe, P. A., & Kenardy, J. (2004). The role of catastrophic misinterpretation of bodily sensations and panic self-efficacy in predicting panic severity. Journal of Anxiety Disorders, 18, 325e340.
nchez-Arribas, C., Valiente, R. M., & Sandin, B. (2014, AugustChorot, P., Sa September). Coping with panic: The Panic Self-efficacy Scale (PSES). In Poster session to be presented at the 45th Annual European Association for Behavioural and Cognitive Therapies, Jerusalem, Israel (Submitted). Chorot, P., Sandín, B., Valiente, R. M., Santed, M. A., & Romero, M. (1997). Actitud
tica en pacientes con hacia la enfermedad, ansiedad y sintomatología soma
nico e hipocondría [Illness attitude, anxiety, and somatic trastorno de pa symptoms in panic disorder and hypochondriasis]. Revista de Psicopatología y Psicología Clínica, 2, 123e136. Clark, D. M. (1986). A cognitive approach to panic. Behaviour Research and Therapy, 24, 461e470. Clark, D. M. (1999). Anxiety disorders: why they persist and how to treat them. Behaviour Research and Therapy, 37, S5eS27.

40

B. Sandin et al. / Behaviour Research and Therapy 67 (2015) 30e40

€ L. G., Breitholtz, E., Koehler, K. A., Westling, B. E., Clark, D. M., Salkovskis, P. M., Ost, et al. (1997). Misinterpretation of body sensations in panic disorder. Journal of Consulting and Clinical Psychology, 65, 203e213. Cox, B. J. (1996). The nature and assessment of catastrophic thoughts in panic disorder. Behaviour Research and Therapy, 34, 363e374. Cox, B. J., Norton, G. R., & Swinson, R. P. (1992). Panic Attack QuestionnairedRevised. Toronto, ON: Clarke Institute of Psychiatry. Cox, B. J., Swinson, R. P., Endler, N. S., & Norton, G. R. (1994). The symptom structure of panic attacks. Comprehensive Psychiatry, 35, 349e353. Craske, M. G., & Barlow, D. H. (2007). Mastery of your anxiety and panic: Therapist guide. New York, NY: Oxford University Press. Craske, M. G., & Barlow, D. H. (2008). Panic disorder and agoraphobia. In D. H. Barlow (Ed.), Clinical handbook of psychological disorders (4th ed.). (pp. 1e64). New York: The Guilford Press. Deacon, B., & Abramowitz, J. (2006). Anxiety sensitivity and its dimensions across the anxiety disorders. Journal of Anxiety Disorders, 20(7), 837e857. Deacon, B. J., & Valentiner, D. P. (2001). Dimensions of anxiety sensitivity and their relationship to nonclinical panic. Journal of Psychopathology and Behavioral Assessment, 23, 25e33. Di Nardo, P. A., & Barlow, D. H. (1988). Anxiety Disorders Interview ScheduleeRevised (ADIS-R). Albany, NY: Phobia and Anxiety Disorders Clinic, State University of New York. Fava, L., & Morton, J. (2009). Causal modeling of panic disorder theories. Clinical Psychology Review, 29, 623e637. Fentz, H. N., Hoffart, A., Jensen, M. B., Arendt, M., O'Toole, M. S., Rosenberg, N. K., et al. (2013). Mechanisms of change in cognitive behaviour therapy for panic disorder: the role of panic self-efficacy and catastrophic misinterpretations. Behaviour Research and Therapy, 51, 579e587. Gallagher, M. W., Payne, L. A., White, K. S., Shear, K. M., Woods, S. W., Gorman, J. M., et al. (2013). Mechanisms of change in cognitive behavioral therapy for panic disorder: the unique effects of self-efficacy and anxiety sensitivity. Behaviour Research and Therapy, 51, 767e777. Grant, D. M., Beck, J. G., & Davila, J. (2007). Does anxiety sensitivity predict symptoms of panic, depression, and social anxiety? Behaviour Research and Therapy, 45, 2247e2255. Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guilford Press. Hofmann, S. G., Meuret, A. E., Rosenfield, D., Suvak, M. K., Barlow, D. H., Gorman, J. M., et al. (2007). Preliminary evidence for cognitive mediation during cognitive-behavioral therapy of panic disorder. Journal of Consulting and Clinical Psychology, 75, 374. Houck, P. R., Spiegel, D. A., Shear, M. K., & Rucci, P. (2002). Reliability of the selfreport version of the Panic Disorder Severity Scale. Depression and Anxiety, 15, 183e185. Keller, M. L., & Craske, M. G. (2008). Panic disorder and agoraphobia. In J. Hunsley, & E. J. Mash (Eds.), A guide to assessments that work (pp. 229e253). New York, NY: Oxford University Press. Khawaja, N. G., & Oei, T. P. (1998). Catastrophic cognitions in panic disorder with and without agoraphobia. Clinical Psychology Review, 18, 341e365. Khawaja, N. G., & Oei, T. P. S. (1992). Development of a catastrophic cognitions questionnaire. Journal of Anxiety Disorders, 6, 305e318. Kircanski, K., Craske, M. G., Epstein, A. M., & Wittchen, H. U. (2009). Subtypes of panic attacks: a critical review of the empirical literature. Depression and Anxiety, 26, 878e887. Li, W., & Zinbarg, R. E. (2007). Anxiety sensitivity and panic attacks a 1-year longitudinal study. Behavior Modification, 31, 145e161. McNally, R. J. (1990). Psychological approaches to panic disorder: a review. Psychological Bulletin, 108, 403. McNally, R. J. (1994). Panic disorder: A critical analysis. New York: Guilford Press. McNally, R. J. (2002). Anxiety sensitivity and panic disorder. Biological Psychiatry, 52, 938e946. Naragon-Gainey, K. (2010). Meta-analysis of the relations of anxiety sensitivity to the depressive and anxiety disorders. Psychological Bulletin, 136, 128. Olatunji, B. O., & Wolitzky-Taylor, K. B. (2009). Anxiety sensitivity and the anxiety disorders: a meta-analytic review and synthesis. Psychological Bulletin, 135, 974e999. Olmedo, M., Sandin, B., Santed, M. A., & García-Campayo, J. (2006, July). Self-efficacy to control panic attacks: current questionnaires and a brief measure. In B. Sandin (Ed.), New self-report instruments for assessing panic-related constructs. 26th International Congress of Applied Psychology, Athens, Greece. Olthuis, J. V., Watt, M. C., & Stewart, S. H. (2014). Anxiety Sensitivity Index (ASI-3) subscales predict unique variance in anxiety and depressive symptoms. Journal of Anxiety Disorders, 28, 115e124.
rez-Benítez, C. I., Shea, M. T., Raffa, S., Rende, R., Dyck, I. R., Ramsawh, H. J., et al. Pe (2009). Anxiety sensitivity as a predictor of the clinical course of panic disorder: a 1-year follow-up study. Depression and Anxiety, 26, 335e342. Pilecki, B., Arentoft, A., & McKay, D. (2011). An evidence-based causal model of panic disorder. Journal of Anxiety Disorders, 25, 381e388.

Rector, N. A., Szacun-Shimizu, K., & Leybman, M. (2007). Anxiety sensitivity within the anxiety disorders: disorder-specific sensitivities and depression comorbidity. Behaviour Research and Therapy, 45, 1967e1975. Reiss, S., & McNally, R. J. (1985). Expectancy model of fear. In S. Reiss, & R. R. Bootzin (Eds.), Theoretical issues in behavior therapy (pp. 107e121). San Diego, CA: Academic Press. Richards, J. C., Richardson, V., & Pier, C. (2002). The relative contributions of negative cognitions and self-efficacy to severity of panic attacks in panic disorder. Behaviour Change, 19, 102e111. Rodriguez, B. F., Bruce, S. E., Pagano, M. E., Spencer, M. A., & Keller, M. B. (2004). Factor structure and stability of the Anxiety Sensitivity Index in a longitudinal study of anxiety disorder patients. Behaviour Research and Therapy, 42, 79e91. Sandin, B., Chorot, P., & McNally, R. J. (1996). Validation of the Spanish version of the Anxiety Sensitivity Index in a clinical sample. Behaviour Research and Therapy, 34, 283e290. Sandin, B., Chorot, P., & McNally, R. J. (2001). Anxiety Sensitivity Index: normative data and its differentiation from trait anxiety. Behaviour Research and Therapy, 39, 213e219.
nchez-Arribas, C., & Santed, M. A. (2004). Sandin, B., Chorot, P., Valiente, R. M., Sa
nico y Agorafobia (CPA): Características de los ataques de Cuestionario de Pa p
anico no clínicos [Panic and Agoraphobia Questionnaire: Characteristics of nonclinical panic attacks]. Revista de Psicopatología y Psicología Clínica, 9, 139e161. Sandin, B., Chorot, P., Valiente, R. M., Santed, M. A., & Lostao, L. (2004). Dimensiones de la sensibilidad a la ansiedad: Evidencia confirmatoria de la estructura
rquica [Dimensions of anxiety sensitivity: Confirmatory evidence of a hijera erarchic structure]. Revista de Psicopatología y Psicología Clínica, 9, 19e33. Sandin, B., Valiente, R. M., Chorot, P., & Santed, M. A. (2007). ASI-3: Nueva escala
n de la sensibilidad a la ansiedad [ASI-3: A new scale for the para la evaluacio assessment of anxiety sensitivity]. Revista de Psicopatología y Psicología Clínica, 12, 91e104. Schmidt, N. B., Lerew, D. R., & Jackson, R. J. (1999). Prospective evaluation of anxiety sensitivity in the pathogenesis of panic: replication and extension. Journal of Abnormal Psychology, 108, 532e537. Smits, J. A., Berry, A. C., Rosenfield, D., Powers, M. B., Behar, E., & Otto, M. W. (2008). Reducing anxiety sensitivity with exercise. Depression and Anxiety, 25, 689e699. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics. Boston, MA: Pearson. Taylor, S. (1995). Anxiety sensitivity: theoretical perspectives and recent findings. Behaviour Research and Therapy, 33, 243e258. Taylor, S. (1999). Anxiety sensitivity: Theory, research, and treatment of the fear of anxiety. Mahwah, NJ: Erlbaum. Taylor, S. (2000). Understanding and treating panic disorder: Cognitive-behavioural approaches. New York: John Wiley & Sons Ltd. Taylor, C. B., & Arnow, B. (1988). The nature and treatment of anxiety disorders. New York: Free Press. Taylor, S., Koch, W. J., & McNally, R. J. (1992). How does anxiety sensitivity vary across the anxiety disorders? Journal of anxiety disorders, 6, 249e259. Taylor, S., Zvolensky, M. J., Cox, B. J., Deacon, B., Heimberg, R. G., Ledley, D. R., et al. (2007). Robust dimensions of anxiety sensitivity: development and initial validation of the Anxiety Sensitivity Index-3 (ASI-3). Psychological Assessment, 19, 176e188. Telch, M. J., Brouillard, M., Telch, C. F., Agras, W. S., & Taylor, C. B. (1989). Role of cognitive appraisal in panic-related avoidance. Behaviour Research and Therapy, 27, 373e383. Valiente, R. M., S
anchez-Arribas, C., Chorot, P., & Sandin, B. (2014, AugustSeptember). Assessing catastrophic misinterpretation of panic symptoms: The Panic Catastrophic Misinterpretations Scale (PCMS). In Poster session to be presented at the 45th Annual European Association for Behavioural and Cognitive Therapies, Jerusalem, Israel (Submitted).
stico de las Valiente, R. M., Sandin, B., & Chorot, P. (2003). Entrevista para el Diagno Trastornos de Ansiedad [Anxiety Disorders Diagnostic Interview]. ADIS-M. Madrid: Klinik.
nchez-Arribas, C. (2006). Assessment of catastrophic Valiente, R. M., Sandin, B., & Sa misinterpretations of bodily sensations: a new brief self-report scale. In B. Sandin (Ed.), New self-report instruments for assessing panic-related constructs. 26th International Congress of Applied Psychology, Athens, Greece. € Westling, B. E., & Ost, L. G. (1993). Relationship between panic attack symptoms and cognitions in panic disorder patients. Journal of Anxiety Disorders, 7, 181e194. Wheaton, M. G., Deacon, B. J., McGrath, P. B., Berman, N. C., & Abramowitz, J. S. (2012). Dimensions of anxiety sensitivity in the anxiety disorders: evaluation of the ASI-3. Journal of Anxiety Disorders, 26, 401e408. 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, 277e284.