Anxiety Disorders 19 (2005) 877892

A Brief Measure of Worry Severity (BMWS): personality and clinical correlates of severe worriers
Gemma L. Gladstone*, Gordon B. Parker, Philip B. Mitchell, Gin S. Malhi, Kay A. Wilhelm, Marie-Paule Austin
School of Psychiatry, University of New South Wales and Mood Disorders Unit, Black Dog Institute, Prince of Wales Hospital, Randwick, NSW 2031, Australia Received 31 August 2004; received in revised form 26 October 2004; accepted 15 November 2004

Abstract This report describes the development of a brief and valid self-report measure to assess severe and dysfunctional worry (the Brief Measure of Worry Severity or BMWS). Using three independent subject groups (clinical and non-clinical), the measure was used to examine the differential severity of worry in depression and anxiety and to examine the clinical and personality correlates of severe worriers. Preliminary psychometric evaluation revealed that the BMWS possesses good construct and clinical discriminant validity. Subjects reporting greater worry severity tended to be more introverted and obsessional, but less agreeable and conscientious. Subjects with depression only, reported less problems with worrying compared to those with co-morbid anxiety disorders. However, among the anxiety disorders, severe and dysfunctional worry was not exclusively experienced by subjects with generalized anxiety disorder (GAD). This study suggests that pathological worry is not only relevant for patients with GAD, but may be an equally detrimental cognitive activity for patients with panic disorder and obsessivecompulsive disorder. # 2004 Elsevier Inc. All rights reserved.
Keywords: Worry; Anxiety disorders; Depression; Questionnaire

* Corresponding author. Fax: +61 2 9382 3712. E-mail address: g.gladstone@unsw.edu.au (G.L. Gladstone). 0887-6185/$ see front matter # 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.janxdis.2004.11.003

878

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

1. Introduction Worry is both a common cognitive activity as well as a symptom correlated with psychological disturbance capable of reaching marked intensity. Depending on severity, worry may either be a general marker for anxiety proneness or a signicant component of more clinically meaningful anxiety, such as generalized anxiety disorder (GAD) where it is recognized as the cardinal diagnostic feature (APA, 1994). Worry activity can also be conceptualized as a spectrum, from commonplace potentially useful worrying, which may possess a motivational quality, to problematic worry characterized by repetitive catastrophic speculation. which is detrimental to performance and debilitating. The perceived uncontrollability of worry, its pervasiveness and disruption to daily functioning, and the presence of meta-worry (or worrying about worry), are all features considered to characterize more severe or pathological worry (Gladstone & Parker, 2003). Interest in the phenomenology and function of worry has led the way to developments in its measurement. Thus, various self-report instruments have been designed to measure one or more components of worry or related cognitive phenomena (e.g., Cartwright-Hatton & Wells, 1997; Wells, 1994). Some instruments have focused on the assessment of what people actually worry about (content), like the Worry Domains Questionnaire (WDQ; Tallis, Eysenck, & Mathews, 1992), including those for specic populations, such as the Worry Scale (Wisocki, Haden, & Morse, 1986) for elderly respondents, and the Student Worry Scale (SWS; Davey, Hamptom, Farrell, & Davidson, 1992). Other instruments have focused on worry as a trait, such as the well-known Penn State Worry Questionnaire or PSWQ (Meyer, Miller, Metzger, & Borkovec, 1990), designed to assess the frequency and intensity of worrying in general. There is currently no single brief measure of worry, which includes items dening differential key markers for dysfunctional and severe worry (e.g., uncontrollability, thwarted problem solving, associated mood disturbance, meta-worry). The extent to which pathological worrying constitutes an important component of different psychological disorders has gained recent attention in the literature, with particular interest in the degree to which severe and uncontrollable worrying is experienced by patients with differing anxiety disorders. Although pathological worry is regarded as the hallmark feature of GAD (APA, 1994), anxious apprehension (Barlow, 1988), described as a negative, future-orientated emotional state, whereby feared outcomes are anticipated, is taken to be characteristic of all anxiety disorders (Brown, OLeary, & Barlow, 1993). Thus, worry has gained preferential status in GAD, with studies suggesting that the worry in GAD is both more severe (i.e., frequent and uncontrollable) (Brown, Antony, & Barlow, 1992) and more pervasive in focus (Gross & Eifert, 1990) than in other anxiety disorders. Chelkminski and Zimmerman (2003), for example, compared worrying (using PSWQ scores) in patients with different anxiety

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

879

disorders and found that patients with GAD scored signicantly higher on the PSWQ than those with social phobia, specic phobia, panic disorder and posttraumatic stress disorder, but not obsessivecompulsive disorder (OCD). Excessive and unprovoked worrying about minor matters, along with greater concern about lack of control over the worry, are usually taken to be factors which distinguish the worrying in GAD from similar cognitive processes in other anxiety disorders (Craske, Rapee, Jackel, & Barlow, 1989). Repetitive cognitive processes, like worry, are also commonly experienced by patients with major depression (Starcevic, 1994). Also described in terms of depressive rumination (Morrow & Nolen-Hoeksema, 1990), such repetitive thought processes usually involve preseverative and static cognitions (rather than elaborative worry) geared toward the depressive illness itself (e.g., Why did this happen to me?). One study (Diefenbach et al., 2001), however, identied that the worrisome thoughts of depressed patients were particularly related to aimless future themescognitions clearly linked to depressed mood. Studies have demonstrated that ruminations work to maintain and exacerbate depressed affect, both for those with experimentally induced low mood (Morrow & NolenHoeksema, 1990) and those with clinical depression (Nolen-Hoeksema & Morrow, 1993). However, the degree to which worrisome and other repetitive cognitions differ in severity for those with depressed and anxious mood states has received limited empirical attention, with the few existing studies showing contrasting results. One study by Starcevic (1994) using the PSWQ, found no differences in worry scores between patients with GAD compared to patients with major depression, concluding that not only was pathological worrying not restricted to GAD sufferers, but it was also a feature of depression, and was equally severe in both disorders. However, the study by Chelkminski and Zimmerman (2003), which also included a comparison between patients with GAD only and major depression only, found that worry posed a signicantly greater problem for those with GAD compared to those with depression. A small amount of research has examined the personality correlates of people who worry excessively. Severe worriers also tend to be more selfevaluative, more socially anxious and more perfectionistic. They also tend to feel more time-pressured, and report more obsessional symptoms (Meyer et al., 1990; Pruzinsky & Borkovec, 1990). One study (Pruzinsky & Borkovec, 1990) found a greater tendency for negative daydreaming, along with poorer attentional control among chronic worriers. This nding might be accounted for by the view that worrying requires considerable attentional resources and results in poorer problem solving efforts and interrupted task completion (Dugas, Letarte, Rheaume, Freeston, & Ladouceur, 1995). Further research is needed to better understand the relationship between worry and related constructs such as conscientiousness, perfectionism and obsessionality, particularly in relation to differentiating between so called functional and dysfunctional worrying.

880

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

The present study reports the development of a short (8-item) self-report questionnaire of worry severity and dysfunction, with items designed to measure the central components of pathological worry identied in the research literature as being particularly relevant in the demarcation between normal, volitional worrying and dysfunctional worry which is detrimental to performance and linked to emotional disturbance (see Gladstone & Parker, 2003). Using clinical and non-clinical samples, the aim of this study is to provide an initial assessment of the psychometric properties of the new measure (the Brief Measure of Worry Severity; BMWS); to further examine the differential severity of worry in major depression and differing anxiety disorders, and nally, to provide further insight into the clinical and personality correlates of severe worriers.

2. Method 2.1. Participants and procedures One-hundred and seventeen subjects including 28 medical students and 89 patients receiving psychiatric treatment, were used to develop a Brief Measure of Worry Severity. The mean age for the student group was 22.4 years (S.D. = 0.68) with 18 males and 10 females. The 89 patients included those attending our Mood Disorders Unit (MDU) for outpatient assessment (n = 35); MDU inpatients (n = 4) and those visiting MDU psychiatrists privately (n = 50), of whom most had anxious or depressive disorders, with a minority reporting adjustment disorders, stress, grief reactions, and disordered eating habits. Patients with a psychotic illness, dementia or language difculties were not included in the study. The mean age for the total patient group was 41.2 years (S.D. = 15.0) and 58 (65%) were female. In addition to that study group, two other samples (outlined below), were used to assess the validity of the Brief Measure of Worry Severity and examine its clinical utility. 2.1.1. Depression sample This was a separate clinical sample comprising 184 patients referred to the MDU for assessment of depression. All patients met DSM-IV criteria for major depression, 79 (43%) of whom met DSM-IV criteria for melancholia. Patients ranged in age from 17 to 68 with a mean age of 39.6 (S.D. = 12.7) years and 110 (60%) were female. 2.1.2. Antenatal sample This comprised a general community sample of 748 women attending a public antenatal clinic at the Royal Hospital for Women in Sydney, recruited from approximately 26 weeks gestation onwards, for a separate longitudinal study into

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

881

risk factors for post-natal depression. Women ranged in age from 14 to 47 with a mean age of 30.5 years (S.D. = 5.23). 2.2. Measures and assessments 2.2.1. Depression sample Patients completed the nal worry measure (BMWS; described shortly) as part of their routine assessment together with other self-reports including the 60-item NEO-PI (Costa & McCrae, 1985); the PSWQ (Meyer et al., 1990) and the sevenfactor Temperament and Character Inventory or TCI (Cloninger, Svrakic, & Przybeck, 1993), as well as a 142-item personality questionnaire which measures 15 personality styles underlying formalized personality disorders (see Parker, Hadzi-Pavlovic, & Wilhelm, 2000 for a full description of the measure). Questionnaires were completed by patients at their home prior to their clinic appointment. Patients were instructed to complete the measures based on how they felt or behaved usually and not simply for how they felt when depressed. Patients also underwent separate clinical research interviews with a psychiatrist and a research psychologist. The psychologist administered the Composite International Diagnostic Interview or CIDI (WHO, 1997) to assess lifetime prevalence of anxiety disorders. The psychiatrist assigned patients an MDU clinical diagnosis for their current depression (i.e., psychotic depression, melancholic depression, or non-melancholic depression), and completed the 17-item Hamilton Rating Scale for Depression or HRSD (Hamilton, 1967). The HRSD measures various depressive features and also includes (psychic and somatic) anxiety items, and is widely used as a valid clinician-rated assessment instrument for depression. The psychiatrist also rated the severity of patients current depression, using a global severity judgment (options being: 0: not depressed; 1: mild; 2: moderate; or 3: severe). Based on behavioral signs exhibited by the patient during interview, the psychiatrist was required to rate (using clinical judgment) the degree to which patients displayed observable anxiety and observable irritability and/or hostility. Both assessments were rated on a 4-point scale (options being: 0: nil; 1: slight; 2: moderate; and 3: marked). Finally, the psychiatrist also asked patients to provide a subjective judgment of how much of a worrier they were generally (that is, when you are not depressed) using a scale of: 0: not at all; 1: mild; 2: moderate; or 3: severe. 2.2.2. Antenatal sample As well as the BMWS, these women also completed the trait subscale of the State-trait Anxiety Inventory for Adults (Spielberger, 1983) as a measure of trait anxiety and a general questionnaire including questions relating to depression, anxiety, premenstrual stress and a tendency to worry question (i.e., would you consider yourself a worrier?) rated on a 5-point scale (15) from not at all to very much.

882

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

2.3. Construction of the Brief Measure of Worry Severity (Original sample n = 117) A set of 26 questions about worry activity and process was constructed. Questions were derived by review of the existing literature on worry (see Gladstone & Parker, 2003) and were designed to capture indicators of problematic worry, such as worry marked by intensity, high frequency and uncontrollability, as well as the perceived negative consequences of worry. Four of the questions required subjects to make broad judgments about their experience of worrying and were later compared to the remaining specic worry questions. The rst of these global questions required subjects to choose one of three options (scales 03) in response to the prompt: generally I am a person who, with options being: (i) does not really worry to any signicant degree, under any circumstances; (ii) worries only when there is something stressful going on in my life and only when life is particularly difcult; or (iii) worries on a fairly constant basis, and not just when faced with very stressful situations. The second global question required subjects to judge the truest statement from six options (scales 05): (i) I never worry; (ii) my worrying is only minor and doesnt concern me; (iii) my worrying is sometimes difcult to manage; (iv) my worrying is often difcult to stop once I start; (v) my worrying is mostly uncontrollable no matter what I try or (vi) my worrying is completely uncontrollable. The third global question required subjects to judge the personal acceptability of their general worrying with options (scales 03) being: (i) completely acceptable; (ii) somewhat unacceptable; (iii) moderately unacceptable; or (iv) denitely unacceptable. The nal global question required subjects to judge the intensity (or strength) of their worrying with option ranging from insignicant/or absent to severe (scales 03). The remaining 22 items were all rated on a 4-point scale and included questions about the rebounding, frequency and interference of worry throughout the day; the behavioral concomitants associated with worry (e.g., nail biting); the perceived effectiveness of worry (e.g., worrying as prevention of negative things happening); catastrophizing as part of worry; negative cognitive (e.g., problem solving) and affective (e.g., mood disturbance) consequences of worrying and meta-worry (i.e., worrying about worry).

3. Results Based on subjects responses to each of the four global worry questions, all subjects endorsed worrying in general at least to some degree. Interestingly, no subject reported they never worried. 3.1. Reduction of item set Different methods were used to reduce the 22-item development set. Firstly, all 26 questions (i.e., the 4 global items and 22 specic worry questions) were

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

883

correlated with each other and with the total score in order to assess the degree of commonality between items and total. Pearson correlation coefcients ranged from 0.39 to 0.83 for item-total correlations. Items that consistently correlated poorly with other items were considered for exclusion. Next, we examined for items which failed to discriminate between the two options in the rst global worry question [i.e., those who worried only when stressed (n = 61) compared to those who worried fairly constantly (n = 56)]. Of the 22 items, those that failed to discriminate these groups, after applying a Bonferroni correction for multiple comparisons (P < .002), were eliminated. The same procedure was conducted using worry acceptability as the dependent variable. Response options on this question were collapsed into two groups: mostly acceptable (i.e., completely acceptable and somewhat unacceptable) and mostly unacceptable (i.e., moderately unacceptable and denitely unacceptable). Again, those items with mean scores signicantly higher for the mostly unacceptable group were retained. These procedures yielded a residual set of 16 items, which were then entered into a principal components analysis (PCA) with eigenvalues over 1 extracted and with a varimax rotation. This analysis produced a single factor accounting for 61% of the variance. Eight items were selected from this single factor for the nal measure based on a number of decisions. Items with higher factor loadings and commonality coefcients, those with more face validity for severe worry and items that successfully predicted important variables were retained. Separate regression analyses were used to determine this nal decision. First, the 16 worry items were entered into a general linear model regression analysis as covariates (predictor variables) with subject status (patient vs. student) entered as the dependent variable. Items which signicantly predicted patient status were retained. Similar analyzes were conducted separately using uncontrollability of worry (scales 05), intensity of worry (scales 03) and perceived unacceptability of worry (scales 03) as dependent variables, with signicantly predictive items retained. The remaining 8 items were entered into a nal principal components analysis with items and component coefcients (cc) given in Table 1. The nal eight items were also factor analyzed in both the depression and antenatal samples in order to test the integrity of the single factor. In both samples a single worry factor was retained, constituting 67.6% of the variance for the depression sample, and 56.7% of the variance for the antenatal (non-clinical) sample. 3.2. Descriptives statistics and internal consistency The depression group (n = 184) was used to assess the measures internal consistency with Cronbachs (1951) coefcient a as the internal consistency estimate. The BMWS demonstrated strong internal consistency with a coefcient a of 0.92, and with itemtotal correlations ranging from 0.79 to 0.87. For the total group, scores ranged form 0 to 24 with a mean score of 15.5 (S.D. = 6.9) and

884

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

Table 1 Final 8-item Brief Measure of Worry Severity and component coefcients (cc) of single worry factor Worry item 1. When I worry, it interferes with my day-to-day functioning (e.g., stops me getting my work done, or organizing my day) 2. When I think I should be nished worrying about something, I nd myself worrying about the same thing over and over 3. My worrying leads me to feel down and depressed 4. When I worry, it interferes with my ability to make decisions or solve problems 5. I feel tense and anxious when I worry 6. I worry that bad things or events are certain to happen 7. I often worry about not be able to stop myself from worrying 8. As a consequence of my worrying, I tend to feel emotional unease or discomfort cc 0.85 Component of dysfunctional worry Impairment and interference

0.79

Uncontrollability

0.80 0.80 0.82 0.77 0.78 0.83

Associated mood disturbance Associated indecision; thwarted problem solving Associated anxiety Catastrophic cognitions Meta cognitions: concern about excessive worrying Negative emotional consequences

Instructions Below is a list of statements about worrying. Please read each statement and indicate how true each one is in describing your general or usual experience of worrying. Item rating scale: 0: not true at all; 1: somewhat true; 2: moderately true; and 3: denitely true.

median of 16.0 and with no differences in mean scores for males (M = 15.2, S.D. = 6.7) and females (M = 14.1, S.D. = 7.1). Total worry scores were also unrelated to patient age (r = .09). 3.3. Construct validity and personality correlates Construct validity and personality correlate data from both the depression and antenatal groups are presented in Table 2 in the form of Pearson correlations. In order to gain comparison data, correlations were also obtained for the PSWQ. As PSWQ data were available for only 113 subjects, correlations were carried out listwise in order to ensure that identical subjects were used in both sets of correlations. As anticipated, there was a strong correlation between the PSWQ and the BMWS. Both the BMWS and the PSWQ were only moderately correlated with the single worry question asked during interview, suggesting a reasonable degree of independence between worry severity assessed by a global question compared to a more elaborate measure. For the antenatal group, there was also a moderate correlation between the single worry item and the BMWS. For this sample, the BMWS was signicantly correlated with the trait subscale of the State-trait Anxiety Inventory. Personality correlates are also reported. As expected, both worry measures were signicantly correlated with the neuroticism subscale, however, the strength of the relationship was considerably weakened when the worry item was partialed out and then became negligible when all three of the anxiety items

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892 Table 2 Correlations between BMWS total scores and measures of worry and personality Worry or personality measure Depression sample (n = 184) Single worry question PSWQ NEO subscales Neuroticism Nworrya Nanxietyb Extroversion Openness to experience Agreeableness Conscientiousness TCI subscales Novelty seeking Harm avoidance Reward dependence Persistence Self-directedness Cooperative Self-transcendence Antenatal sample (n = 748) Single worry question Trait anxiety
a b

885

BMWS 0.39** 0.75**

PSWQ 0.39**

0.62** 0.54** 0.12 0.34** 0.04 0.26** 0.25**

0.57** 0.46** 0.01 0.33** 0.09 0.20* 0.13

0.04 0.42** 0.04 0.20** 0.48** 0.13 0.06

0.15 0.49** 0.02 0.17 0.43** 0.13 0.04

0.52** 0.68**

Partial correlation controlling for neuroticism worry item. Partial correlation controlling for neuroticism worry/anxiety items. * P < .05. ** P < .01 (two-tailed); listwise n = 113.

from the neuroticism subscale were partialed. Both worry measures were moderately inversely correlated with extroversion and unrelated to the construct of openness to experience. Both measures were also negatively associated with agreeableness (see Table 2). The BMWS was also negatively correlated with conscientiousness, while the PSWQ was not signicantly related to this trait. Two of the TCI subscales were most signicantly related to worry severity. Harm avoidance was moderately correlated with worry severity as measured by both the BMWS and the PSWQ, while self-directedness was moderately inversely correlated with worrying. Also, a weak but signicant relationship between worry (as measured by the BMWS) and the persistence subscale of the TCI was also observed. Remaining TCI subscale scores were unrelated to worry scores. We also conducted correlations between the personality styles measured by the DSM-derived 142 item self-report questionnaire described earlier and both of the

886

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

worry measures. The pattern of correlations between worry and personality styles was mostly equivalent across both the BMWS and the PSWQ. All of the personality styles except for narcissistic, sadistic, and schizoid were signicantly positively correlated with worry severity as measured by the BMWS, with correlation coefcients ranging from 0.24 (P < .01) for histrionic to 0.59 (P < .001) for anxious. The obsessional subscale was moderately correlated with worry severity (r = .40, P < .001).
Table 3 Means and standard deviations (S.D.) of BMWS scores for different study groups, with and without specied clinical criteria. Sample/clinical variable Original sample (n = 117) BMWS total scores Student (n = 28) Mean 6.3 Clinical sub-sample (n = 89) S.D. 4.1 Clinical (n = 89) Mean 13.8 S.D. 5.5 6.59*** t(70) Statistic (df) t(115)

Depression only (n = 56) Mean 13.9 S.D. 5.2

Anxiety only (n = 16) Mean 17.4 Present S.D. 5.7

2.35* F(6,177)

Depression sample (n = 184) Panic disorder (n = 61)a GAD (n = 30)a Social phobia (n = 44)a OCD (n = 31)a Agoraphobia (n = 24)a Specic phobia (n = 36)a Any anxiety disorder (n = 110) DSM-IV melancholia (n = 79) Clinical diagnosis melancholia (n = 72)b Antenatal sample (n = 748)

Absent Mean 13.1 13.9 14.6 14.6 13.6 16.3 11.8 14.4 15.5 Absent Mean S.D. 4.1 4.4 4.3 S.D. 6.7 6.9 6.9 7.1 6.8 6.9 6.8 6.7 6.3

Mean 17.4 18.1 14.3 19.3 14.7 15.6 16.4 14.7 13.0 Present Mean 8.5 11.4 10.1

S.D. 6.1 4.8 6.7 3.9 7.6 6.7 6.3 7.1 7.6 16.4*** 10.7** 0.1 5.1* 0.5 0.1 t(182) 4.8*** 0.3 2.4 t(745) S.D. 5.4 5.6 6.1 9.4*** 10.0*** 8.7***

Lifetime depressive episode (n = 285) Signicant PMS (n = 72)c Signicant current anxiety or depression (n = 107)d
a b

5.3 5.7 5.9

Analysis of variance controlling for other anxiety disorders. MDU clinical diagnosis melancholia: melancholic and/or psychotic. c History of signicant self-reported premenstrual sadness or anxiety. d Signicant self-reported anxiety or depression during current pregnancy. * P < .05. ** P < .01. *** P < .001.

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

887

3.4. Discriminant clinical validity Clinical validity data for the three study groups are presented in Tables 3 and 4. Firstly, means and standard deviations are given for the original sample, showing signicantly higher BMWS scores for the clinical compared to the nonclinical (student) group. Group comparisons for a subset of clinical subjects (those with depression only compared to those with anxiety only) revealed that patients with anxiety disorders had signicantly greater worry scores than patients with depression. A large proportion of the depression sample (60%) also had co-morbid anxiety disorders (see Table 3), and these patients had signicantly higher BMWS scores than patients without co-morbid anxiety disorders. Differences in mean scores between patients with and without anxiety co-morbidity were examined for each separate anxiety disorder (controlling for the inuence of overlapping co-morbid disorders). Depressed subjects with co-morbid panic disorder had signicantly greater BMWS scores than depressed subjects without co-morbid panic disorder. Similarly, patients with co-morbid GAD had greater worry scores compared to those without co-morbid GAD. Also, patients with co-morbid OCD had higher worry scores compared to those without comorbid OCD. There were no differences in worry scores between depressed patients with and without co-morbid social phobia, agoraphobia or specic phobias. There were no signicant differences in worry scores between patients with and without melancholia, dened both by DSM-IV and MDU clinical diagnosis criteria. Key clinical variables were then examined for the antenatal sample. As part of a questionnaire assessing risk factors for post-natal depression, women were asked three separate questions related to depression and anxiety. Women who reported a past episode of major depression had signicantly higher BMWS scores than those who did not. Similarly, women who reported problems with previous pre-menstrual stress as well as those who reported present (during current pregnancy) difculties with anxiety or depression had signicantly higher worry scores than those women who did not report such disturbances (see Table 3).
Table 4 BMWS and PSWQ correlated with different clinician-rated anxiety and depression measures Clinical measure Observable anxiety Observable irritability Depression severity score Total HRSD HRSDdepression HRSDanxiety
**

BMWS 0.25** 0.08 0.02 0.16 0.10 0.24**

PSWQ 0.17 0.08 0.05 0.13 0.03 0.15

P < .01 (two-tailed); listwise n = 113.

888

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

In the depression sample, correlations were conducted between the BMWS and the PSWQ on various clinician-rated measures of depression and observable behavioral anxiety and irritability. The PSWQ was not signicantly correlated with any of the clinician-judged estimates of anxiety or depression. Self-reported worry severity as measured by the BMWS was signicantly positively correlated with observable (clinician-judged) anxiety and with the anxiety subscale of the HRSD (see Table 4), while measures of irritability and depression were unrelated to worry severity scores.

4. Discussion This study reports the development and initial validation of a new Brief Measure of Worry Severity. Application of factor analytic techniques to a set of worry items produced a single worry factor accounting for most of the variance, with this factor structure retained when examined in two independent samples. The measure includes 8 items each measuring a key characteristic of pathological or dysfunctional worry, and each predictive of three global subjective judgments about worrying (i.e., uncontrollability, intensity, and unacceptability of worry). The BMWS was found to possess good construct validity. The measure was moderately independent of worry assessed by a single question and was signicantly associated with high trait anxiety in a normal sample. As expected, BMWS scores were signicantly associated with neuroticism scores, with the relationship between excessive worry and neuroticism mediated entirely by the inuence of worry and anxiety items. Correlations with other personality measures revealed some interesting associations. We found that subjects who reported excessive worrying also tended to have an introverted personality style. This relationship is consistent with the act of worrying itself, and the considerable degree of inward energy expended by an individual when engaged in worrying (Dugas et al., 1995). An interesting inverse relationship between BMWS and both agreeableness and conscientiousness was observed, providing validation for the measures ability to assess worry which has become an impediment to both general functioning and self-efcacy. The negative relationship between conscientiousness and worry (as measured by the BMWS), suggests that excessive worrying interferes with an individuals sense of self-competence, and organization, and ones sense of effective planning or problem solving. Only the BMWS was sensitive to this aspect of dysfunctional worry. Conversely, a moderate (positive) relationship between worry and (DSM-derived) obsessionality was observed. Thus, it is worth noting that severe worry is positively related to obsessional cognitions and behaviorswhich are dysfunctional, yet negatively related to conscientiousnesswhich is a functional trait. While previous research has linked worry to perfectionism (e.g., Pruzinsky & Borkovec, 1990), the current study contributes further by differentiating the associations excessive worrying has

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

889

with related personality constructs, and underscores the distinction between conscientiousness and obsessionality. A modest but signicant relationship was also found between worrying and personality qualities such as cynicism and suspiciousnessas evidenced by its inverse relationship with NEO agreeableness. This nding ts well with the idea that people who worry excessively (including GAD sufferers) are particularly concerned about potential dangers and threats to personal safety (Mathews, 1990), have a general heightened awareness of potential threat cues and are often preoccupied with catastrophic thinking. Similarly, based on TCI scores, worriers were more likely to identify with harm avoidance behaviors. Worry was also signicantly inversely related to self-directedness, a nding which also ts with research suggesting worriers have poor attentional control (Pruzinsky & Borkovic, 1990). Although worry (as measured by the BMWS only) was negatively related to the tendency to be persistent, suggesting that the thwarted problem solving effects of excessive worrying may lead to a felt sense of failure, the relationship was only weak. In a subset of clinical subjects, worry scores were signicantly higher for those with anxiety only, compared with depression only. This result contrasts the study by Starcevic (1994), who reported equivalent levels of worrying in anxious and depressed patients, but is consistent with ndings by Chelkminski and Zimmerman (2003) who reported higher worry scores in anxious, compared with depressed patients. Depressed patients with co-morbid anxiety disorders reported more severe worry compared to those without anxious disorders, which is further evidence for the greater relevance of worry in anxiety, compared with depression. Analyzes of individual co-morbid disorders revealed that worry was particularly relevant for not only patients with GAD, but also for those with panic disorder and OCD. As cognitive appraisal in these disorders primarily focus upon threats to personal safety (e.g., panic, worry about personal harm, contamination fears), it is not surprising that worry is highly salient for these patients. Patients with co-morbid social phobia did not report worrying more than those without this co-morbidity, despite previous research suggesting that worriers tend to be more socially anxious (Pruzinsky & Borkovec, 1990). The present results suggest that while severe worry appears more relevant for anxiety than depression, among the anxiety disorders it is not exclusively relevant for GAD. In the depressed sample, worry was also unrelated to depression severity (based on clinician judgments), and also unrelated to depression subtype (i.e., melancholic vs. non-melancholic depression). Further validity of the BMWS was demonstrated in two signicant correlations with clinician-rated anxiety measures, including the HRSD anxiety sub-scale score and clinician judgment of observable anxiety (both rated by psychiatrists independently of patient reported BMWS scores). This relationship between worry and behavioral signs of anxiety (found only with the BMWS) provides further evidence for the instruments clinical utility and is consistent with research linking worry with somatic anxiety symptoms, particularly muscle tension (Joormann & Stober, 1999). The PSWQ, however, was not signicantly correlated

890

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

with these clinical assessments of anxiety. The most likely explanation for this may be found in key differences between the BMWS and PSWQ. The PSWQ is a valid unifactorial instrument, with items predominately measuring the frequency and uncontrollability of worry. Although the BMWS was also found to be unifactorial, its 8 items are distinctly more varied, perhaps providing a broader assessment of worry severity, including features of associated emotional disturbance (e.g., felt anxiety and depression) and functional impairment (e.g., interference with daily tasks and decision making). The BMWS may, therefore, have greater potential to detect pathological worry and clinical anxiety, but this issue requires further investigation. The present study also found evidence for the role of worry as a potentially key vulnerability marker for depressed and anxious symptoms within a non-clinical sample. In a large community group of antenatal clinic attendees, higher worry scores discriminated between women with and without a self-reported history of depression. Furthermore, women who reported having had previous difculties with premenstrual stress (PMS) reported signicantly greater worry than those who did not report PMS problems. Signicantly higher worry scores were also returned by women reporting difculties with contemporaneous depression or anxiety. Thus, the tendency to worry, for this sample may constitute a general vulnerability marker, or alternatively might reect the womens current emotional state. Issues of vulnerability for this sample are, however, difcult to decipher due to the cross-sectional nature of the data. The present study supports the validity of a newly devised Brief Measure of Worry Severity and provides evidence of its clinical utility for future worryrelated research. The BMWS requires further psychometric examination, but shows promise as a succinct, easy to complete and valid self-report assessment of pathological or dysfunctional worrying. Finally, our ndings broaden the current empirical knowledge base on the personality characteristics of worriers while raising questions about the exclusive relevance of worry for GAD among the anxiety disorders.

Acknowledgments We thank the National Health and Medical Research Council (Program grant 2223208), the NSW Centre for Mental Health for infrastructure funding, Chris Boyd for data management and Karen Saint for collection of data for the antenatal clinic sample.

References
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorder. Washington, DC: Author.

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

891

Barlow, D.H. (1988). Anxiety and its disorders. New York, London: Guilford Press. Brown, T. A., Antony, M. M., & Barlow, D. H. (1992). Psychometric properties of the Penn State Worry Questionnaire in a clinical anxiety disorder sample. Behaviour Research and Therapy, 30, 3337. Brown, T. A., OLeary, T. A., & Barlow, D. H. (1993). Generalized anxiety disorder. In: D. H. Barlow (Ed.), Clinical handbook of psychological disorders (2nd ed., pp. 137188). New York: Guilford Press. Cartwright-Hatton, S., & Wells, A. (1997). Beliefs about worry and intrusions: the meta-cognitions questionnaire and its correlates. Journal of Anxiety Disorders, 11, 279296. Chelkminski, I., & Zimmerman, M. (2003). Pathological worry in depressed and anxious patients. Journal of Anxiety Disorders, 17, 533546. Cloninger, C. R., Svrakic, D. M., & Przybeck, T. R. (1993). A psychobiological model of temperament and character. Archives of General Psychiatry, 50, 975989. Costa, P. T., & McCrae, R. R. (1985). The NEO personality manual. Odessa, FL: Psychological Assessment Resources. Craske, M. G., Rapee, R. M., Jackel, L., & Barlow, D. H. (1989). Qualitative dimensions of worry in DSM-III-R generalized anxiety disorder subjects and nonanxious controls. Behaviour Research and Therapy, 27, 397402. Davey, G. C. L., Hampton, J., Farrell, J., & Davidson, S. (1992). Some characteristics of worrying: evidence for worrying and anxiety as separate constructs. Personality and Individual Differences, 13, 133147. Diefenbach, G. J., McCarthy-Larzelere, M. E., Williamson, D. A., Mathews, A., Manguno-Mire, G. M., & Bentz, B. G. (2001). Anxiety, depression and the content of worries. Depression and Anxiety, 14, 247250. Dugas, M. J., Letarte, H., Rheaume, J., Freeston, M. H., & Ladouceur, R. (1995). Worry and problem solving: evidence of a specic relationship. Cognitive Therapy and Research, 19, 109120. Gladstone, G., & Parker, G. (2003). Whats the use of worrying? Its function and its dysfunction. Australian and New Zealand Journal of Psychiatry, 37, 347354. Gross, P. R., & Eifert, G. H. (1990). Components of generalized anxiety: the role of intrusive thoughts vs. worry. Behaviour Research and Therapy, 28, 421428. Hamilton, M. (1967). Development of a rating scale for primary depressive illness. British Journal of Social and Clinical Psychology, 6, 278296. Joormann, J., & Stober, J. (1999). Somatic symptoms of generalized anxiety disorder from the DSMIV: associations with pathological worry and depressive symptoms in a nonclinical sample. Journal of Anxiety Disorders, 13, 491503. Mathews, A. (1990). Why worry? The cognitive function of anxiety. Behaviour Research and Therapy, 28, 455468. Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and validation of the Penn State Worry Questionnaire. Behaviour Research and Therapy, 28, 487495. Morrow, J., & Nolen-Hoeksema, S. (1990). Effects of responses to depression on the remediation of depressive affect. Journal of Personality and Social Psychology, 58, 519527. Nolen-Hoeksema, S., & Morrow, J. (1993). Effects of rumination and distraction on naturally occurring depressed mood. Cognition and Emotion, 7, 561570. Parker, G., Hadzi-Pavlovic, D., & Wilhelm, K. (2000). Modeling and measuring the personality disorders. Journal of Personality Disorders, 14, 189198. Pruzinsky, T., & Borkovec, T. D. (1990). Cognitive and personality characteristics of worriers. Behaviour Research and Therapy, 28, 507512. Spielberger, C. D. (1983). Manual for the State-Trait Anxiety Inventory (STAI). PaloAlto, CA: Consulting Psychologists Press. Starcevic, V. (1994). Pathological worry in major depression: a preliminary report. Behaviour Research and Therapy, 33, 5556. Tallis, F., Eysenck, M. W., & Mathews, A. (1992). A questionnaire for the measurement of nonpathological worry. Personality and Individual Differences, 13, 161168.

892

G.L. Gladstone et al. / Anxiety Disorders 19 (2005) 877892

Wells, A. (1994). A multi-dimensional measure of worry: development and preliminary validation of the anxious thoughts inventory. Anxiety, Stress, and Coping, 6, 289299. Wisocki, P. A., Handen, B., & Morse, C. K. (1986). The worry scale as a measure of anxiety among homebound and community active elderly. The Behavior Therapist, 5, 9195. World Health Organization (WHO). (1997). Composite International Diagnostic Interview, version 2.1 (CIDI-A). Sydney, Australia: WHO Research and Training Centre.