رابطه بین نگرانی، علائم وسواس و باورهای فراشناختی

The present study had two principal aims. First, the overlap between the revised Padua Inventory (PI-WSUR) a measure of obsessive–compulsive symptoms (Burns et al., 1996) and worry was assessed. Second, the relationship between meta-cognitive beliefs, proneness to pathological worry and obsessive–compulsive symptoms was explored when controlling for the interdependency of worry and obsessive–compulsive measures. The results indicate that whilst the PI-WSUR shows reduced overlap with the Penn State Worry Questionnaire, there are still problematic levels of overlap with a more content-based measure of worry. Results of multiple regression analyses demonstrated that specific sets of meta-cognitive beliefs were associated with worry proneness and obsessive–compulsive symptoms. The present data are consistent with recent formulations of generalized anxiety disorder and obsessive–compulsive disorder.

Worry and obsessions are two types of intrusive cognitive phenomena associated with significant distress. They are characteristic features of particular emotional disorders such as generalized anxiety disorder (GAD) and obsessive–compulsive disorder (OCD). Worry and obsessions are also normally occurring phenomena (e.g. Rachman and de Silva, 1978; Salkovskis and Harrison, 1984; Wells and Morrison, 1994) and nonclinical samples can provide an adequate analogue for research on OCD (e.g. Burns et al., 1995). Several self-report instruments have been developed to assess worry. These include the Anxious Thoughts Inventory (AnTI; Wells, 1994b), the Penn State Worry Questionnaire (PSWQ; Meyer et al., 1990) and the Worry Domains Questionnaire (WDQ; Tallis et al., 1992). An instrument that is gaining popularity in the measurement of obsessive–compulsive (O–C) symptoms is the Padua Inventory (PI; Sanavio, 1988). However, it has been noted that measures of worry and obsessions present a number of limitations to researchers wishing to distinguish between worry and obsessions and wishing to explore the differences between these types of ideational events (e.g. Freeston et al., 1994; Burns et al., 1996). To facilitate research, self-report instruments that reliably distinguish between worry and obsessions are required. The nature of the common variance in existing measures should be determined for purposes of refinement of psychometric measurement and for the development of theories of worry and obsessions. Worry and obsessions share a number of features (Turner et al., 1992; Wells and Morrison, 1994; Clark and Claybourn, 1997). However, there are a number of differences. In a comparison of normal worries and normal obsessions, Wells and Morrison (1994)demonstrated that worries tended to be more verbal, realistic and voluntary and were associated with ratings of a greater compulsion to act than obsessions. Despite these differences, considerable similarity existed. For example, worry and obsessions did not differ in intrusiveness, controllability, dismissability, the extent to which they were resisted and how distracting or distressing the thoughts were. A possible difference between worry and obsessions which remains to be explored concerns the functional significance of these types of thoughts. Worry has been viewed as avoidance (Borkovec and Inz, 1991), as a form of problem-solving (Davey, 1994) and as a coping activity (Wells, 1994a and Wells, 1995), whilst no functional significance appears to have been assigned to obsessions. One of the differentiating characteristics between worry and obsessions that has been used to improve the distinguishability of these phenomena is thought content (APA, 1994; Burns et al., 1996). In particular, worries are considered to consist of apprehensive expectation about real-life events such as finances, relations and work (APA, 1994). According to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; APA, 1994, p. 435), ``Several features distinguish the excessive worry of Generalized Anxiety Disorder from the obsessional thoughts of Obsessive Compulsive Disorder. Obsessional thoughts are not simply excessive worries about everyday or real-life problems, but rather are ego-dystonic intrusions that often take the form of urges, impulses and images in addition to thoughts''. In summary, worry and obsessions are currently distinguished on the basis of content. The content of obsessions is considered to be ego-dystonic, that is to say, it is viewed as alien or inappropriate by the individual. Typical obsessions concern sexual, aggressive and contamination themes that are out of character. However, the content distinction is not free from difficulties. In particular, some obsessional symptoms such as contamination fears are not necessarily ego-dystonic and are likely to overlap with worry themes such as those concerning health. Moreover, a distinction of obsessional phenomena purely in terms of their ego-dystonicity could lead to the neglect of other idiosyncratic features that could reliably differentiate obsessions from worry. On the basis of possible content differences, Burns et al. (1996)set out to revise the Padua Inventory (PI) and reduce its overlap with worry. A problem with the initial PI was that items could be answered with reference to either worry or obsessions (e.g. ``when doubts and worries come to my mind, I cannot rest until I have talked them over with a reassuring person''). Most of these problematic items were on the `impaired control over mental activities' subscale of the PI. Burns et al. (1996)attempted to eliminate this problem by selecting 39 of the original 60 items of the PI and organising them into content categories relevant to obsessions and compulsions. However, the content of items retained in the `obsessional thoughts about harm to self/others' category may remain equally applicable to worry and obsessions (e.g. ``I think or worry at length about having hurt someone without knowing it''). Therefore, difficulties are likely in the independence of this category from measures of worry. For example, `worry at length about having hurt someone' could refer to social-evaluative worry or to obsessions. In order to test the discriminant validity of the revised PI (Padua Inventory-Washington State University Revision or PI-WSUR), Burns et al. (1996)computed correlations for the modified subscales with the Penn State Worry Questionnaire (PSWQ; Meyer et al., 1990). However, the PSWQ is not considered to be a content-based measure of worry but assesses general worry frequency and distress. Questions concerning the degree of overlap between the PI-WSUR and a content-based measure of worry remain to be addressed. The present study aimed to further explore the relationship between worry and O–C symptoms. We examined correlations between the PI-WSUR and the PSWQ and used the Anxious Thoughts Inventory (AnTI) to explore further the distinctiveness of the PI-WSUR. The AnTI is a multi-dimensional measure of worry assessing dimensions of health, social and meta-worry (worry about worry). For purposes of this study, only the social and health worry subscales were analyzed. In addition to exploring the overlap between the PI-WSUR and the AnTI, we tested relationships between worry, O–C symptoms and beliefs. Beliefs about one's own thoughts, or meta-cognitive beliefs, have been linked to vulnerability to emotional disorder (Wells and Matthews, 1994 and Wells, 1997Wells and Matthews, 1996a). In particular, Wells (1995), Wells (1997)has advanced a cognitive model of GAD in which positive and negative beliefs about worry are central to problem maintenance. Meta-cognitive beliefs have also been linked to OCD. For example, Clark and Purdon, 1993 and Clark and Purdon, 1995suggest that obsessional individuals hold dysfunctional beliefs concerning the need to control thoughts. In a related vein, it has been suggested that an inflated sense of responsibility plays a significant role in OCD (Rachman, 1976Rachman, 1993; Salkovskis, 1985). Wells (1997)suggests that responsibility appraisals are an emergent property of meta-cognitive beliefs about the power and consequence of obsessional thoughts. In an analysis of the structure of responsibility, Rachman et al. (1995)demonstrated that the Thought Action Fusion (TAF) subscale of the Responsibility Appraisal Questionnaire (RAQ) was particularly significant to O–C symptoms. TAF is the belief that thoughts can influence events or are almost equivalent to actions. Given the importance of meta-cognitive beliefs, we tested for relationships between a measure of pathological worry (PSWQ) and meta-cognitive beliefs. Relationships between meta-cognitive beliefs and O–C symptoms were also investigated. In doing so, we controlled for the overlap between worry and O–C symptoms. In this study, we tested the following specific hypotheses: (1) a content-based worry measure (AnTI) will show significant overlap with the PI-WSUR. (2) Meta-cognitive beliefs about worry and intrusions will be associated with a measure of pathological worry (PSWQ). More specifically, in accordance with Wells' (Wells, 1995 and Wells, 1997) cognitive model of GAD, positive and negative beliefs about worry will be positively associated with pathological worry. (3) Meta-cognitive beliefs about worry and intrusions will be associated with O–C symptoms even when the overlap with worry is controlled. (4) Since compulsive checking may be viewed as a way of preventing future danger (e.g. Tallis and de Silva, 1992), we predict that it has a similar function to worry and may be conceptualised as a behavioural variant of worry. Therefore, we hypothesised that compulsive checking would be positively associated with positive beliefs about worry. However, contamination and washing compulsions serve a somewhat different function, since in many instances the individual believes that they have already encountered the danger (contaminant). Therefore, we hypothesised that there would be a different pattern of meta-cognitive predictors of washing compulsions. We did not have more detailed a priori hypotheses concerning the precise nature of predictors in this case, or of the predictors of PI-WSUR impulses and dressing compulsions.

Apart from scores on the PSWQ and the AnTI social worry subscale, all other scores were skewed. A log transformation was performed on all of the skewed variables and this was successful in `normalising' their distributions. Since some subjects scored zero on a number of subscales of the PI-WSUR, a numerical constant of 1 was added to all scores on these particular subscales prior to performing log transformations. The anti-logged means (minus 1 where appropriate) and 95% confidence limits for respective variables are shown in Table 1. All Pearson correlations and regressions were computed using the appropriate logged or unlogged variables. Table 1. Means and 95% confidence limits (CL) for AnTI, PI-WSUR, MCQ and PSWQ Scale Subscales Mean 95% CL AnTI: Health worry 9.4 8.8, 9.9 Social worry 18.5 17.6, 19.4 PI-WSUR: ob. thoughts 1.5 1.1, 1.9 impulses 0.9 0.6, 1.1 washing 3.4 2.8, 4.1 checking 4.7 3.9, 5.6 dressing 0.8 0.5, 1.1 total 12.0 10.0, 14.4 MCQ: positive beliefs 32.0 30.2, 33.9 danger/Uncontr. 27.2 25.9, 28.7 cog. confidence 16.6 15.7, 17.5 super/responsibility 21.0 20.0, 22.1 consciousness 15.5 14.7, 16.3 PSWQ 45.5 43.2, 47.8 Note: AnTI=Anxious Thoughts Inventory; PI-WSUR=Padua Inventory, Washington University Revision; MCQ=Meta-Cognitions Questionnaire; PSWQ=Penn State Worry Questionnaire. Table options 3.1. Independence of worry from obsessions and compulsions The PI-WSUR subscales and total score were significantly correlated with the PSWQ and the AnTI subscales. These data are presented in Table 2. The magnitude of correlations between the PI-WSUR subscales and the PSWQ were similar to those reported by Burns et al. (1996), with the exception of washing compulsions and obsessional impulses. Burns et al. (1996)reported a correlation of 0.21 between PSWQ and washing compulsions but in the present study the correlation was 0.43. Similarly, Burns et al. (1996)reported an association between obsessional impulses and PSWQ of 0.08, but the correlation in the present study was 0.25. Table 2. Pearson correlations for measures of worry, obsessive–compulsive symptoms and meta-cognitive beliefs PI-WSUR AnTI Checking Washing Dressing Impulses Ob.thoughts Total PSWQ Health Social PSWQ: 0.32∗∗∗ 0.43∗∗∗ 0.23∗ 0.25∗ 0.40∗∗∗ 0.45∗∗∗ – – – AnTI: Health 0.38∗∗∗ 0.33∗∗∗ 0.24∗ 0.38∗∗∗ 0.50∗∗∗ 0.47∗∗∗ 0.40∗∗∗ – – Social 0.38∗∗∗ 0.27∗ 0.21∗ 0.22∗ 0.46∗∗∗ 0.38∗∗∗ 0.58∗∗∗ – – MCQ: Positive beliefs 0.47∗∗∗ 0.37∗∗∗ 0.37∗∗∗ 0.26∗ 0.53∗∗∗ 0.54∗∗∗ 0.45∗∗∗ 0.31∗∗∗ 0.40∗∗∗ Danger/uncontr. 0.46∗∗∗ 0.52∗∗∗ 0.37∗∗∗ 0.35∗∗∗ 0.58∗∗∗ 0.57∗∗∗ 0.57∗∗∗ 0.30∗∗ 0.48∗∗∗ Cog. confidence 0.36∗∗∗ 0.31∗∗∗ 0.22∗ 0.28∗ 0.40∗∗∗ 0.44∗∗∗ 0.22∗ 0.39∗∗∗ 0.38∗∗∗ Super/responsib 0.38∗∗∗ 0.37∗∗∗ 0.30∗∗ 0.23∗ 0.43∗∗∗ 0.46∗∗∗ 0.34∗∗∗ 0.09 0.26∗ Consciousness 0.33∗∗∗ 0.25∗ 0.37∗∗∗ 0.16 0.33∗∗∗ 0.38∗∗∗ 0.26∗ 0.18 0.25∗ Note: ∗∗∗p<0.001, ∗∗p<0.002, ∗p<0.03. After Bonferroni correction ∗∗∗(p<0.05). Table options In the present study, the shared variance between the PI-WSUR total score and the PSWQ was 20%. This degree of shared variance is less than that quoted for the original version of the Padua Inventory, in which the common variance was 34% (Freeston et al., 1994). However, the shared variance found in this study for the revised Padua Inventory is greater than the 12% reported by Burns et al. (1996). Although these results support the use of the revised inventory, problems remain because of the degree of overlap with worry measures. When the common variance with a content-based measure of worry is examined the shared variance between the revised Padua Inventory and health and social worry subscales is 22 and 14%, respectively. In summary, notwithstanding the content revisions of the Padua Inventory, overlap still remains between this measure and both content and non-content based measures of worry. 3.2. Relationships between worry, obsessive–compulsive symptoms and meta-cognitive beliefs To explore the meta-cognitive predictors of worry proneness and proneness to individual subtypes of O–C symptoms, we controlled for the statistical interdependence of measures of worry and O–C symptoms. Initially, bivariate correlations were computed to test the relationship between worry and meta-cognitive beliefs, and between O–C symptoms and meta-cognitive beliefs. These correlations are presented in Table 2. Consistent with our hypothesis derived from the cognitive model of GAD (Wells, 1995 and Wells, 1997), positive and negative beliefs about worry were significantly and positively correlated with pathological worry. In accord with our other theoretical predictions, MCQ subscales were positively correlated with a range of O–C symptoms. We tested further our principle hypotheses concerning associations between worry, O–C symptoms and meta-cognitive beliefs by computing multiple regression analyses. Before running these analyses we examined the intercorrelations between predictors. The magnitude of correlations suggested that multi-collinearity was not a problem and this was also supported by the tolerance levels which are reported for each variable in the regression tables. The intercorrelations of MCQ subscales ranged from 0.10 to 0.61, with all but two coefficients below 0.46. 3.3. Meta-cognitive predictors of pathological worry The PSWQ served as the dependent measure of pathological worry for the predictive analysis. For purposes of the regression the PI-WSUR total score was entered on the first step to control for the common variance between worry and O–C symptoms and the MCQ subscales were entered on step two. Our hypotheses then concern the independent MCQ predictors of PSWQ while controlling for the overlap between worry and O–C symptoms. Following the cognitive model of GAD (Wells, 1995 and Wells, 1997), it was predicted that positive beliefs about worry and negative beliefs concerning the uncontrollability and danger of worry would independently predict PSWQ scores. Statistics for each block of the regression equation and overall summaries for the final equation are presented in Table 3. The MCQ subscales accounted for a significant 19% of the variance in PSWQ in addition to the PI-WSUR total score. However, in the final equation PI-WSUR did not contribute significantly to PSWQ scores, when the MCQ subscales were entered. Consistent with predictions, only positive beliefs about worry and beliefs concerning uncontrollability and danger were uniquely associated with PSWQ. Table 3. Regression statistics for each block of variables, and summary statistics for the final equation with PSWQ as dependent variable Block of variables R adj. R2 change R2 F change df Sig of F change (1) PI-WSUR 0.45 0.19 0.20 25.89 1,103 0.0001 (2) MCQ scales 0.63 0.36 0.19 6.19 6,98 0.0001 Individual variables in final equation β Tolerance T Sig of T PI-WSUR 0.12 0.52 1.07 0.29 MCQ: Positive beliefs 0.26 0.56 2.45 0.02 Danger/uncontr. 0.51 0.49 4.55 0.0001 Cog. confidence −0.09 0.74 −1.03 0.31 Super/responsib −0.17 0.49 −1.52 0.13 Consciousness −0.05 0.77 0.53 0.59 Table options 3.4. Meta-cognitive predictors of obsessive–compulsive symptoms To test for the independent MCQ predictors of O–C symptoms we treated each PI-WSUR subscale as a different dependent variable. For purposes of these analyses, we controlled for the statistical overlap of worry with PI-WSUR subscales scores. In addition to PSWQ, the two AnTI subscales (social and health worry) were entered on the first step followed by the forced entry of MCQ subscales on step two. We entered the AnTI subscales in addition to the PSWQ in these analyses because we wanted to additionally control for the overlap with worry content. Statistics for each step of these regressions and for the overall equations are presented in Table 4, Table 5, Table 6, Table 7 and Table 8. With the exception of obsessional impulses, the MCQ subscales accounted for significant additional variance (13–20%) in all other PI-WSUR subscales, in addition to that explained by the worry measures. Thus, consistent with predictions, meta-cognitive factors were associated with a range of O–C symptoms and different patterns of predictors were obtained for O–C symptom subtypes. Consistent with our hypothesis that checking would be associated with positive beliefs about worry, only this MCQ subscale independently predicted checking. In contrast, positive beliefs about worry and negative beliefs about uncontrollability significantly predicted obsessional thoughts about harm to self/others. Health worry also made an independent and significant contribution to obsessional thoughts. Washing compulsions were only predicted by MCQ negative beliefs about danger and uncontrollability of thoughts and dressing and grooming compulsions were only significantly associated with cognitive self-consciousness. Table 4. Regression statistics for each block of variables and summary statistics for the final equation with PI-WSUR obsessional thoughts as dependent variable Block of variables R adj. R2 Change R2 F change df Sig of F change (1) PSWQ and AnTI 0.57 0.30 0.32 16.12 3,101 0.0001 (2) MCQ scales 0.72 0.48 0.20 7.96 8,98 0.0001 Individual variables in final equation β Tolerance T Sig of T PSWQ −0.10 0.50 −1.0 0.32 AnTI health 0.30 0.65 3.4 0.001 AnTI social 0.06 0.54 0.70 0.50 MCQ: Positive belief 0.24 0.54 2.50 0.01 Danger/uncontr. 0.34 0.42 3.10 0.003 Cog. confidence 0.05 0.68 0.58 0.57 Super/responsib 0.03 0.45 0.24 0.81 Consciousness 0.09 0.79 1.10 0.28 Table options Table 5. Regression statistics for each block, and summary statistics for the final equation with PI-WSUR washing as dependent variable Block of variables R adj. R2 Change R2 F change df Sig of F change (1) PSWQ and AnTI 0.46 0.19 0.21 9.04 3,101 0.0001 (2) MCQ scales 0.59 0.29 0.13 3.92 8,96 0.003 Individual variables in final equation β Tolerance T Sig of T PSWQ 0.19 0.50 1.63 0.11 AnTI health 0.16 0.65 1.55 0.12 AnTI social −0.16 0.54 −1.46 0.15 MCQ: Positive beliefs 0.08 0.54 0.69 0.50 Danger/uncontr. 0.33 0.42 2.50 0.01 Cog. confidence 0.09 0.68 0.86 0.39 Super/responsib 0.04 0.45 0.33 0.74 Consciousness 0.07 0.79 0.72 0.47 Table options Table 6. Regression statistics for each block, and summary statistics for the final equation with PI-WSUR checking as dependent variable Block of variable R adj. R2 Change R2 F change df Sig of F change (1) PSWQ and AnTI 0.45 0.18 0.20 8.38 3,101 0.0001 (2) MCQ scales 0.60 0.31 0.17 4.98 8,96 0.0004 Individual variables in final equation β Tolerance T Sig of T PSWQ −0.10 0.50 −0.78 0.44 AnTI health 0.18 0.65 1.70 0.09 AnTI social 0.07 0.54 0.67 0.50 MCQ: Positive belief 0.24 0.54 2.20 0.03 Danger/uncontr. 0.21 0.42 1.70 0.10 Cog. confidence 0.10 0.68 0.97 0.33 Super/responsib 0.03 0.45 0.28 0.78 Consciousness 0.13 0.79 1.41 0.16 Table options Table 7. Regression statistics for each block, and summary statistics for the final equation with PI-WSUR dressing/grooming as dependent variable Block of variables R adj. R2 Change R2 F change df Sig of F change (1) PSWQ and AnTI 0.28 0.05 0.08 2.93 3,101 0.04 (2) MCQ scales 0.50 0.19 0.17 4.31 8,96 0.001 Individual variables in final equation β Tolerance T Sig of T PSWQ −0.06 0.50 −0.47 0.64 AnTI health 0.09 0.65 0.85 0.40 AnTI social −0.06 0.54 −0.53 0.60 MCQ: Positive beliefs 0.21 0.54 1.73 0.09 Danger/uncontr. 0.25 0.42 1.80 0.07 Cog. confidence 0.04 0.68 0.39 0.70 Super/responsib −0.07 0.45 −0.50 0.62 Consciousness 0.26 0.79 2.60 0.01 Table options Table 8. Regression statistics for each block, and summary statistics for the final equation with PI-WSUR obsessional impulses as dependent variable Block of variables R adj. R2 Change R2 F change df Sig of F change (1) PSWQ and AnTI 0.40 0.13 0.16 6.25 3,101 0.0006 (2) MCQ scales 0.47 0.15 0.06 1.52 8,96 0.20 Individual variables in final equation β Tolerance T Sig of T PSWQ 0.13 0.64 1.15 0.26 AnTI health 0.34 0.74 3.20 0.002 AnTI social −0.02 0.59 −0.20 0.84 MCQ: Positive beliefs 0.06 0.54 0.49 0.62 Danger/uncontr. 0.22 0.42 1.56 0.12 Cog. confidence 0.07 0.68 0.60 0.55 Super/responsib 0.03 0.45 0.25 0.81 Consciousness 0.02 0.79 0.18 0.86