&hoc.
Rex.
The?. Vol. 28. No. 2, pp. 153-158,
Printed in Great Britain.
All
1990
0005-7967
90 53.00 + 0.00
Copyright ,P 1990 Pergamon Press plc
rights reserved
THE NATURE OF WORRY IN GENERALIZED DISORDER: A PREDOMINANCE OF THOUGHT
ANXIETY ACTIVITY
T. D. BORKOVEC and JONATHANINZ The Pennsylvania State University, Department of Psychology, The Stress and Anxiety Disorders Institute, 415 Bruce V. Moore Building, University Park, PA 16802, U.S.A. (Received 2 August 1989)
Summary-Indirect evidence suggests that worry primarily involves thought, rather than imaginal, activity, a distinction within cognitive process that is potentially crucial to some theories of anxiety maintenance and modification. The present study constrasted the frequencies of reporting the presence of thoughts and images among generalized anxiety disorder clients and matched, nonanxious control subjects during a self-relaxation period and a worry period. Repetition of the assessment was conducted with clients after they completed 12 sessions of therapy. Sampling of mentation during these periods revealed that (a) during relaxation, nonanxious subjects reported a predominance of imagery whereas clients show equal amounts of thought and imagery, (b) nonanxious subjects shifted to a predominance of thought during worry, and (c) clients showed a normalization of thought and image frequencies after successful therapy. This combination of results suggests that worry is principally thought-like in content. The speculation is offered that worry may function as motivated avoidance of emotional imagery and its attendant somatic sensations.
is the central definitional feature of generalized anxiety disorder (DSM-III-R, American Psychiatric Association, 1987). Despite the construct’s importance, little is known about its nature and function. A common assumption, however, is that worry is a type of cognitive event. Factor analytical studies of anxiety questionnaires support this assumption, revealing two elements, one commonly labeled cognitive anxiety or worry and the other variously called somatic anxiety or emotionality, and interestingly it is the worry factor that correlates most strongly with other aspects of anxious behavior (Deffenbacher, 1980). There are, however, many possible processes subsumed under the label, ‘cognitive’. Two such processes may be particularly relevant to anxiety: abstract, conceptual, thinking activity and imaginal activity. Indirect evidence suggests that worry primarily involves thought: (a) when asked about their worry experience in interviews, anxious clients tend to report a predominance of ideation over imagery. (b) Induction of a worrisome state in normal Ss results in significantly greater increases in negative thought intrusion during a subsequent attention-focussing task than does neutral mood induction, whereas induction of a somatic anxiety state does not (York, Borkovec, Vasey & Stern, 1987). (c) No differences in peripheral physiological activation have been observed between worriers and nonworriers or between rest and instructed worry periods. Because thinking and verbal articulation of fear material have been found to produce significantly less cardiovascular response than images of the same content (Lang, 1988), this absence of physiological difference suggests a relative absence of imagery and likely predominance of ideation during worry. If worry is basically a conceptual activity, its process may have significant implication for the maintenance of anxiety states. Based on Lang’s (1985) theory and extant research on therapy with anxiety disorders, Foa and Kozak (1986) have argued that accessing the entire cognitive/affective fear structure stored in memory, via in citto exposure or imagery, is important for a complete processing of emotional material. Through such emotional processing, fear modification can occur. Degree of cardiovascular response to phobic presentations is seen as evidence for the degree to which such accessing has taken place. Thinking about feared material may thus be an inefficient method of fear structure elicitation and, moreover, may actually create a conceptual context that interferes with therapeutic emotional processing. Evidence for the latter effect can be seen in a recent study demonstrating that no cardiovascular reaction occurs in response to phobic imagery if the S engages in worrisome thinking just prior to image presentation (Borkovec & Hu, 1990). ‘Worry’
BRT X3,2--~
I53
154
T.D.
E~ORKOVEC
and
JOSATHANISZ
The present study was designed to obtain further information on the phenomenology of worry and whether it is primarily a conceptual, rather than imaginal, process. Significantly, this information was gathered from carefully diagnosed generalized anxiety disorder clients. During a more extensive physiological assessment session, clients and nonanxious controls were asked to relax themselves for a period and to worry during a subsequent period. During these phases. Ss were periodically contacted and asked (a) to report whether the mental content at the time of the contact was a thought, image, or both, or whether they were ‘unsure’, and (b) to rate the affective value of the content on a scale of pleasantness/unpleasantness and a scale of relaxation/ anxiousness. Because this assessment session was conducted again after the clients had received therapy, we were able also to assess whether changes occurred in their mentation once their anxiety problem had been reduced. METHOD Subjects
Clinical Ss were the first 13 consecutive admissions to an ongoing project on the treatment of generalized anxiety disorder (GAD). Two were referred from a clinical agency, and 11 had responded to a newspaper advertisement. Selection of GAD clients for the treatment project involved a 3-step screening process. A lengthy telephone contact established the possible presence of GAD and the absence of certain exclusion criteria. Second, the client underwent a structured diagnostic interview with a psychiatric assessor, following a revised version of the Anxiety Disorder Interview Schedule (ADIS, DiNardo, O’Brien, Barlow, Waddell & Blanchard, 1983) for the DSM-III-R differential diagnoses of anxiety disorders. Hamilton Anxiety and Depression Rating Scales (Hamilton, 1959, 1960) were also completed by the assessor as part of the Interview. A second ADIS was administered within 7_days by an independent psychiatric assessor. Only clients meeting criteria for GAD as the primary diagnosis in the judgment of both assessors and having an assessor severity rating of 4 (moderate anxiety) or greater on a 9-point scale were admitted. Further exclusion criteria included diagnosable panic disorder (even secondary), medication use, current involvement in therapy, major affective disorder, substance abuse or any other primary mental disorder. The nonanxious Control Ss were solicited through announcements, selected to match the age, education and gender characteristics of the GAD sample, and administered the ADIS by one of the psychiatric assessors. To be admitted to the experiment they had to have no diagnosable disorder. One S was excluded due to dysthymia and was replaced. The GAD group for the present experiment was composed of 7 women and 6 men, ranging in age from 19 to 48 yr (M = 32.1). The duration of their anxiety problem averaged 9.04 yr. Education levels included completion of high school only (I), completion of some college (7) and completion of some graduate study (5). One client was Black, one was from India, and the remainder ivere Caucasian. The Control group included 7 women and 6 men, age range = 19-48.5 yr (M = 32.6), with education levels matched to the GAD clients. Two Ss were Black, and the remainder were Caucasian. Procedure
After admission to the project through the above methods, the Ss completed a variety of self-report questionnaires in the office. These included the trait version of the State-Trait Anxiety Inventory (Spielberger, Gorsuch & Lushene, 1970) the Fear Questionnaire (Marks & Mathews, 1979), the Zung Self-Rating of Anxiety (Zung, 197.5), the Beck Depression Inventory (Beck, Ward, Mendelson, Mock & Erbaugh, 1961), the Reactions to Relaxation and Arousal Questionnaire (Heide & Borkovec, 1983) the Cognitive/Somatic Anxiety Inventory (Borkovec & Mathews, 1988), and the Worry Questionnaire (Meyer, 1988). They were then scheduled for an extensive psychophysiological assessment session. Since the assessor and assessment staff overlapped, they were not blind to S status. Lab procedures were highly standardized, however, and no staff were in the S room during data collection. After electrode attachment, Ss were seated alone in a recliner chair in a dimly lit private room adjacent to a central monitoring room.
Worry
in generalized anxiety disorder
155
The portions of the session relevant to the present report involved a lo-min period of self-relaxation and a lo-min period of worrying, both of which occurred just after a IO-min information processing task (Mathews & MacLeod, 1986). For the relaxation period, they were instructed via intercom to close their eyes and relax, focusing on their breathing and increasingly allowing their minds to wander as they relaxed. After a S-min rest period, they were then instructed to worry in their usual fashion about a topic of current concern to them; the topic had been previously selected at the ADIS interview. At 3, 7 and 10 min during both the relaxation period and worry period, the experimenter contacted them over the intercom and asked them to judge whether the mental content at the moment of the contact involved thought, image, both, or ‘unsure’. Finally, they rated the affective value of this content on 2, 9-point scales (pleasant/ unpleasant and relaxed/anxious), and were told to resume relaxing or worrying. The 13 GAD clients subsequently participated in 12 sessions of nondirective therapy (4) applied relaxation (4) or a combined package of applied relaxation, coping desensitization and cognitive therapy (5). During the following week, they completed the same questionnaire materials that were given before-therapy and a post-ADIS with their original assessor, yielding post-therapy Hamilton Anxiety and Depression scores and the Overall Assessor Severity Rating. Finally, they participated in a repetition of the physiological assessment session, including the self-relaxation and worry periods with the same mentation measures. RESULTS
Pre -session anxiety and depression scores
Table 1 presents means and standard deviations for the GAD and Control Ss on the various pre-session assessor ratings and self-report questionnaires. In each case, significant differences were found between the two groups [all ts_ (24) > 5.45, P < O.OOl]. Since norms on carefully screened, nonanxious Ss are not available for most of these measures, the Control data provided in Table 1 will allow in the future an estimation of local norms for the overall therapy outcome project. Also included in Table 1 are the posttherapy scores for GAD clients subsequent to their receiving 12 sessions of therapy [except for the Fear Questionnaire-Total, all pre-test vs post-test ts (24) > 2.66, P < 0.021. Thought/image
mentation reports
The frequencies of mental content samples that were reported by Ss to be ‘both thoughts and images’ or ‘unsure’ did not differ significantly between groups or between relaxation/worry periods (during relaxation, GAD vs Control frequencies for ‘both’ were 10 and 10, respectively, and for ‘unsure’ were 2 and 1; during worry, frequencies for ‘both’ were 14 and 11, and for ‘unsure’ were 2 and 1). Consequently, these data were disregarded in subsequent analysis. Figure 1 displays the percentage of the total sample reports that Ss labeled as ‘thought’ and as ‘image’, during relaxation and worry periods for GAD and Control Ss. Selected critical comparisons among groups and periods on differences between thought and image frequencies were Table I. Means and standard deviations on assessor ratings and questionnaires Ss at pre-session and for GAD Ss at post-therapy GADPre
Hamilton Anxiety Hamilton Depression Assessor Severity STAI RRAQ
FQ CSI Zung
WQ BDI
Control-Pre
for GAD and control GAD-Post
M
SD
M
SD
M
SD
20.16 13.79 4.7 59.46 27.92 42. I5 176.69 43.38 68.23 15.69
3.94 5.18 0.60 8.42 6.74 17.24 76.61 a.89 6.21 5.86
1.65 0.99 0.04 27.38 II.54 12.85 30.85 24.77 28.38 I.61
1.33 0.89 0.14 3.50 2.1 I 8.85 17.55 3.68 6.36 2.26
9.41 7.21 2.22 46.77 18.85 30.61 98.23 33.61 54.54 9.54
5.15 4.57 1.23 13.05 7.80 19.41 53.86 5.71 16.09 5.90
STAI, Trait version of the State-Trait Anxiety Inventory; RRAQ, Reactions to Relaxation and Arousal Questionnaire: FQ, Fear Questionnaire-Total Score; CSI, Cognitive/Somatic Anxiety Inventory-Total Score; Zung. Zung Self-Rating of Anxiety: WQ. Worry Questionnaire: BDI. Beck Depression Inventory.
T. D. BORKOVEC and JONATHANINZ
156
45%
30%
15%
0% Fig. I. Percentage
of reports
of ‘thought’
and ‘image’ during nonanxious control
relaxation Ss.
and worry
periods
for GAD
and
tested by x2, using Sime’s correction to Bonferroni adjustment for multiple tests (Simes, 1986). During relaxation, Controls showed a strong predominance of imagery with little thought, whereas GAD displayed nearly equal amounts of each mentation type, x’(l) = 4.34, P < 0.037. GAD and Control Ss were very similar when they worried, with thought predominating over image occurrence. For Controls, mentation type shifted from image predominance during relaxation to thought predominance during worry, x2( 1) = 9.75, P c 0.002; for GADS, percentages shifted from roughly equal thoughts and images during relaxation to increases in thought and decreases in imagery during worry, though not to a statistically significant degree. The greatest discrepancy between thought and image frequencies was found when contrasting Controls during relaxation and GADS during worry, x’(l) = 10.00, P < 0.002. Percentages of thought and image reports by GAD clients at the post-therapy assessment session indicated a normalization during the relaxation period: images (51.3%) and thoughts (12.8%) nearly matched the percentages found for the nonanxious Controls. These changes from pre- to post-therapy percentages approached significance after Bonferroni adjustment, x’(l) = 4.54, P < 0.033. Relative to the pre-therapy worry period, the GADS also showed a reversal away from predominance of thought during post-therapy worry (image = 30.8%; thought = 23.1%), but this change was not statistically significant. Mentation
afect
ratings
Table 2 presents means and standard deviations on the pleasant/unpleasant and relaxed/anxious rating scales for the two groups during each sample of the relaxation and worry periods at the pre-therapy assessment session and for the GAD group at the post-therapy session. A 3-way (Group by Relaxation/Worry Periods by Sample) repeated measures multivariate analysis of variance was applied to each of the rating scale scores from the pre-therapy session. Significant Sample, Period, and Sample by Period effects emerged from analysis of the pleasant/ unpleasant scale scores. Unpleasantness increased from relaxation to worry periods and increased during repeated sampling, especially during the relaxation period. However, no effect involving Group was found. The same significant within S effects were found on the relaxed/anxious scores, but the main effect of Group was also significant, F( 1,24) = 6.10, P < 0.03. Average anxiety rating of mental content was greater for the GADS than for the Controls collapsed across relaxation and worry periods, M = 5.08 vs 4.08, respectively. Three-way (Pre/Post-therapy by Relaxation/Worry Periods by Sample) repeated measures multivariate analyses of variance on scale scores for the GAD group assessed whether changes in
in generalized anxiety disorder
Worry
157
Table 2. Means and standard deviations on pleasant/unpleasant and relaxed,‘anxious rating scales for GAD and control Ss during each sample of the relaxation and worry periods and for the GAD Ss at oost-thcratw Assessment Relaxation Sample Pleasant/unpleasant GAD I44 (SD) Control M (SD) GAD-post (SD) Relaxed/anxious GAD &) Control :D) GA&post (SD)
I 2.68
wow
2
3
I
2
3
3.31
3.77 (2.09)
7.31 (1.11)
7.23 (1.74)
6.54 (1.39)
6.08
(1.81)
6.23 (2.01) 6.61 (2.22)
(1.49)
(1.80)
2.15 (1.28) 3.31 (1.70)
3.23 (2.49) 4.08
3.54 (1.71) 3.61
(2.46)
(1.98)
6.69 (1.32)
2.38
3.69 (2.02)
4.00 (1.68)
7.08
6.85
(1.45)
(0.95)
(1.14)
6.46 (1.27)
2.00 (1.22) 2.54 (1.71)
2.38 (1.56) 2.54 (I .90)
2.69 (1.89) 2.61 (1.94)
6.00 (1.63) 4.77 (2.35)
5.85 (2.15) 4.42 (2.27)
5.54 (2.07) 4.77 (2.3 I)
6.54
(2.66) 5.46 (2.07)
affective value of mental content samples occurred subsequent to effective therapy. No effects involving the Pre/Post-therapy factor were significant on the pleasant/unpleasant scores. On the anxiety scale, the main effect of Pre/Post-therapy was significant, F( 1, 12) = 16.8 1, P < 0.01, whereas the Pre/Post-therapy by Period interaction approached significance, F( 1, 12) = 4.20, P < 0.065. Recovered GAD clients reported less anxiety associated with their sampled mentation at post-therapy than at pre-therapy (M = 3.61-5.08), especially during the worry period.
DISCUSSION The results from the present study indicate that, in terms of phenomenological report, worry involves primarily thought as opposed to imaginal activity. This characterization of worry is supported in three different ways. First, the dispositional group comparison showed imagery to predominate the mental content of nonanxious Ss during a self-relaxation period, whereas equal amounts of each mentation type were reported by GAD clients who are chronic worriers by diagnostic definition. Second, manipulation of relaxation and worry periods revealed a shift among nonanxious Ss in the direction of thought predominance; for GADS, further increases in thought and decreases in imagery occurred, though not significantly so. Finally, effective therapy shifted GAD clients to a more normal predominance of imagery during relaxation with a similar, possibly compensatory (albeit nonsignificant) shift during worry. Not surprisingly, pre-therapy mentation samples were reported to have more anxious valence among the clients throughout the periods, and these also significantly shifted in the nonanxious direction subsequent to therapy. Whether S classification of mental content into ‘thoughts’ vs ‘images’ validly reflects actual differences in cognitive process remains to be determined. We do know, however, that distinctive electrocortical differences have been found between chronic worriers and nonworriers during worry periods. Specifically, worriers show increases in left hemisphere, frontal j? activation relative to nonworriers (Carter, Johnson & Borkovec, 1986). It remains to be seen whether these two sets of distinguishing cognitive activity (reported predominance of thought activity and electrocortical activity) are truly reflecting the same process. If worry is predominantly a conceptual, thought activity, what function might it serve? GAD clients are distinctive from other anxiety disorders in that fewer obvious, circumscribed environmental situations are overtly avoided. One could speculate that worrying is a method of attempting to avoid future catastrophe, a common attribution made by worriers to explain their worrying, and that such anticipation is frequently reinforced by the nonoccurrence of feared outcomes. The present data provide clues, however, that there may exist an even more basic function for worry in anxiety maintenance. The declines in imagery and increases in thought that seem to characterize worry in the present study may represent the motivated avoidance of imaginal process. If imagery
158
T.D.
B~RKOVEC
and JONATHAS
INZ
is a primary vehicle for the somatic activation of emotion (Lang, 1988) then one way to learn to avoid somatic anxiety is to avoid imagery by increasing the generation of, or attentional focus on, conceptual activity. Worry may thus represent an avoidance of affect in general or anxious emotional experience in particular. Under such circumstances, triggers for worry episodes are likely to become broadly generalized to any stimulus associated with any degree of emotional elicitation. At this point, such a hypothesized avoidance function for worry is very speculative. Further research will need to isolate the relationships between conceptual activity and worry, on the one hand, and imagery, imagery vividness, general affect, and anxious affect on the other. Given the importance of worry (or “apprehensive expectation”, Barlow, 1988) in each of the anxiety disorders, increased understanding of how ideational activity interacts with the imaginal and affective processing systems is likely to further our knowledge about the maintenance of these disorders. Acknoi&dgemenrs-This research was supported in part by NIMH Grant MH-39172 expressed to Teresa Elliott and Thomas Meyer who served as experimenters and management of the project.
to the first author. Appreciation is to Mildred Hopkins for her data
REFERENCES American Psychiatric Association (1987). Diugnosric and statisricol manual of mental disorders. Washington, D.C.: A.P.A. Barlow, D. H. (1988). Anxiefy and ifs disorders. New York: Guilford Press. Beck, A. T., Ward, C. H., Mendelson, M., Mock, J. & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-57 I. Borkovec, T. D. & Hu, S. (1990). The effect of worry on cardiovascular response to phobic imagery. Behatiour Research and Therapy, 28, 69-73. Borkovec, T. D. & Mathews, A. M. (1988). Treatment of nonphobic anxiety disorders: A comparison of nondirective, cognitive. and coping desensitization therapy. Journal of Consulting and Clinical Psychology, 56, 877-884. Carter, W. R., Johnson, M. C. & Borkovec, T. D. (1986). Worry: An electrocortical analysis. Advances in Behariour Research and Therapy, 8, 193-204. Deffenbacher. J. L. (1980). Worry and emotionalitv. In Sarason, I. G. (Ed.), Test anxiety: Theorv. research. anduDplicarions. . Hillsdale, N.J.: Erlbaum. _ DiNardo. P. A., O’Brien. G. T., Barlow, D. H., Waddell, M. T. & Blanchard, E. (1983). Reliability of DSM-III anxiety disorder categories using a new structured interview. Archices of General Psychiufry. 40, 1070-1078. Foa, E. B. & Kodak, M. J. (1986). Emotional processing of fear: Exposure to corrective information. Psychological BuNerin, 99, 20-35. Hamilton, M. (1959). The assessment of anxiety states by rating. British Journal of Medical Psychology, 32, 50-55. Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery and Psychiatry, 23, 56-62. Heide, F. J. & Borkovec, T. D. (1983). Relaxation-induced anxiety: Paradoxical anxiety enhancement due to relaxation training. Journal of Consulting and Clinical Psychology, 51, I7 l-182. Lang, P. J. (1985). The cognitive psychophysiology of emotion: Fear and anxiety. In Tuma, A. H. & Maser, J. D. (Eds), Anxiefy ond rhe onxiery disorders. Hillsdale, N.J.: Erlbaum. Lang, P. J. (1988). The image of fear: Emotion and memory. Invited address, Behavior Therapy World Congress, Edinburgh. Marks, I. M. & Mathews, A. M. (1979). Brief standard self-rating for phobic patients. Behaciour Research and Therapy, 17, 263-267. Mathews, A. M. & MacLeod, C. (1986). Discrimination of threat cues without awareness in anxiety states. Journal of Abnormal Psychology, 95, 131-138. Meyer. T. (1988). Development of a screening questionnaire to identify levels of chronic worry. Unpublished masters thesis, Pennsylvania State University, University Park, Pa. Simes, R. J. (1986). An improved Bonferroni procedure for multiple tests of significance. Biomefrika 73, 751-754. Speilberger, C. D., Gorsuch, R. L. & Lushene, R. F. (1970). Manualfor the Srare-Trait Anxierv Incenrory. Palo Alto, Calif.: Consulting Psychologists Press. York, D., Borkovec, T. D., Vasey, M. & Stern, R. (1987). Effects of worry and somatic anxiety induction on thought intrusions. subjective emotion, and physiological activity. Behaciour Research and Therapy, 25, 523-526. Zung, W. W. K. (1975). A rating instrument for anxiety disorders. Psychosomarics, 12, 371-379.
Rex.
The?. Vol. 28. No. 2, pp. 153-158,
Printed in Great Britain.
All
1990
0005-7967
90 53.00 + 0.00
Copyright ,P 1990 Pergamon Press plc
rights reserved
THE NATURE OF WORRY IN GENERALIZED DISORDER: A PREDOMINANCE OF THOUGHT
ANXIETY ACTIVITY
T. D. BORKOVEC and JONATHANINZ The Pennsylvania State University, Department of Psychology, The Stress and Anxiety Disorders Institute, 415 Bruce V. Moore Building, University Park, PA 16802, U.S.A. (Received 2 August 1989)
Summary-Indirect evidence suggests that worry primarily involves thought, rather than imaginal, activity, a distinction within cognitive process that is potentially crucial to some theories of anxiety maintenance and modification. The present study constrasted the frequencies of reporting the presence of thoughts and images among generalized anxiety disorder clients and matched, nonanxious control subjects during a self-relaxation period and a worry period. Repetition of the assessment was conducted with clients after they completed 12 sessions of therapy. Sampling of mentation during these periods revealed that (a) during relaxation, nonanxious subjects reported a predominance of imagery whereas clients show equal amounts of thought and imagery, (b) nonanxious subjects shifted to a predominance of thought during worry, and (c) clients showed a normalization of thought and image frequencies after successful therapy. This combination of results suggests that worry is principally thought-like in content. The speculation is offered that worry may function as motivated avoidance of emotional imagery and its attendant somatic sensations.
is the central definitional feature of generalized anxiety disorder (DSM-III-R, American Psychiatric Association, 1987). Despite the construct’s importance, little is known about its nature and function. A common assumption, however, is that worry is a type of cognitive event. Factor analytical studies of anxiety questionnaires support this assumption, revealing two elements, one commonly labeled cognitive anxiety or worry and the other variously called somatic anxiety or emotionality, and interestingly it is the worry factor that correlates most strongly with other aspects of anxious behavior (Deffenbacher, 1980). There are, however, many possible processes subsumed under the label, ‘cognitive’. Two such processes may be particularly relevant to anxiety: abstract, conceptual, thinking activity and imaginal activity. Indirect evidence suggests that worry primarily involves thought: (a) when asked about their worry experience in interviews, anxious clients tend to report a predominance of ideation over imagery. (b) Induction of a worrisome state in normal Ss results in significantly greater increases in negative thought intrusion during a subsequent attention-focussing task than does neutral mood induction, whereas induction of a somatic anxiety state does not (York, Borkovec, Vasey & Stern, 1987). (c) No differences in peripheral physiological activation have been observed between worriers and nonworriers or between rest and instructed worry periods. Because thinking and verbal articulation of fear material have been found to produce significantly less cardiovascular response than images of the same content (Lang, 1988), this absence of physiological difference suggests a relative absence of imagery and likely predominance of ideation during worry. If worry is basically a conceptual activity, its process may have significant implication for the maintenance of anxiety states. Based on Lang’s (1985) theory and extant research on therapy with anxiety disorders, Foa and Kozak (1986) have argued that accessing the entire cognitive/affective fear structure stored in memory, via in citto exposure or imagery, is important for a complete processing of emotional material. Through such emotional processing, fear modification can occur. Degree of cardiovascular response to phobic presentations is seen as evidence for the degree to which such accessing has taken place. Thinking about feared material may thus be an inefficient method of fear structure elicitation and, moreover, may actually create a conceptual context that interferes with therapeutic emotional processing. Evidence for the latter effect can be seen in a recent study demonstrating that no cardiovascular reaction occurs in response to phobic imagery if the S engages in worrisome thinking just prior to image presentation (Borkovec & Hu, 1990). ‘Worry’
BRT X3,2--~
I53
154
T.D.
E~ORKOVEC
and
JOSATHANISZ
The present study was designed to obtain further information on the phenomenology of worry and whether it is primarily a conceptual, rather than imaginal, process. Significantly, this information was gathered from carefully diagnosed generalized anxiety disorder clients. During a more extensive physiological assessment session, clients and nonanxious controls were asked to relax themselves for a period and to worry during a subsequent period. During these phases. Ss were periodically contacted and asked (a) to report whether the mental content at the time of the contact was a thought, image, or both, or whether they were ‘unsure’, and (b) to rate the affective value of the content on a scale of pleasantness/unpleasantness and a scale of relaxation/ anxiousness. Because this assessment session was conducted again after the clients had received therapy, we were able also to assess whether changes occurred in their mentation once their anxiety problem had been reduced. METHOD Subjects
Clinical Ss were the first 13 consecutive admissions to an ongoing project on the treatment of generalized anxiety disorder (GAD). Two were referred from a clinical agency, and 11 had responded to a newspaper advertisement. Selection of GAD clients for the treatment project involved a 3-step screening process. A lengthy telephone contact established the possible presence of GAD and the absence of certain exclusion criteria. Second, the client underwent a structured diagnostic interview with a psychiatric assessor, following a revised version of the Anxiety Disorder Interview Schedule (ADIS, DiNardo, O’Brien, Barlow, Waddell & Blanchard, 1983) for the DSM-III-R differential diagnoses of anxiety disorders. Hamilton Anxiety and Depression Rating Scales (Hamilton, 1959, 1960) were also completed by the assessor as part of the Interview. A second ADIS was administered within 7_days by an independent psychiatric assessor. Only clients meeting criteria for GAD as the primary diagnosis in the judgment of both assessors and having an assessor severity rating of 4 (moderate anxiety) or greater on a 9-point scale were admitted. Further exclusion criteria included diagnosable panic disorder (even secondary), medication use, current involvement in therapy, major affective disorder, substance abuse or any other primary mental disorder. The nonanxious Control Ss were solicited through announcements, selected to match the age, education and gender characteristics of the GAD sample, and administered the ADIS by one of the psychiatric assessors. To be admitted to the experiment they had to have no diagnosable disorder. One S was excluded due to dysthymia and was replaced. The GAD group for the present experiment was composed of 7 women and 6 men, ranging in age from 19 to 48 yr (M = 32.1). The duration of their anxiety problem averaged 9.04 yr. Education levels included completion of high school only (I), completion of some college (7) and completion of some graduate study (5). One client was Black, one was from India, and the remainder ivere Caucasian. The Control group included 7 women and 6 men, age range = 19-48.5 yr (M = 32.6), with education levels matched to the GAD clients. Two Ss were Black, and the remainder were Caucasian. Procedure
After admission to the project through the above methods, the Ss completed a variety of self-report questionnaires in the office. These included the trait version of the State-Trait Anxiety Inventory (Spielberger, Gorsuch & Lushene, 1970) the Fear Questionnaire (Marks & Mathews, 1979), the Zung Self-Rating of Anxiety (Zung, 197.5), the Beck Depression Inventory (Beck, Ward, Mendelson, Mock & Erbaugh, 1961), the Reactions to Relaxation and Arousal Questionnaire (Heide & Borkovec, 1983) the Cognitive/Somatic Anxiety Inventory (Borkovec & Mathews, 1988), and the Worry Questionnaire (Meyer, 1988). They were then scheduled for an extensive psychophysiological assessment session. Since the assessor and assessment staff overlapped, they were not blind to S status. Lab procedures were highly standardized, however, and no staff were in the S room during data collection. After electrode attachment, Ss were seated alone in a recliner chair in a dimly lit private room adjacent to a central monitoring room.
Worry
in generalized anxiety disorder
155
The portions of the session relevant to the present report involved a lo-min period of self-relaxation and a lo-min period of worrying, both of which occurred just after a IO-min information processing task (Mathews & MacLeod, 1986). For the relaxation period, they were instructed via intercom to close their eyes and relax, focusing on their breathing and increasingly allowing their minds to wander as they relaxed. After a S-min rest period, they were then instructed to worry in their usual fashion about a topic of current concern to them; the topic had been previously selected at the ADIS interview. At 3, 7 and 10 min during both the relaxation period and worry period, the experimenter contacted them over the intercom and asked them to judge whether the mental content at the moment of the contact involved thought, image, both, or ‘unsure’. Finally, they rated the affective value of this content on 2, 9-point scales (pleasant/ unpleasant and relaxed/anxious), and were told to resume relaxing or worrying. The 13 GAD clients subsequently participated in 12 sessions of nondirective therapy (4) applied relaxation (4) or a combined package of applied relaxation, coping desensitization and cognitive therapy (5). During the following week, they completed the same questionnaire materials that were given before-therapy and a post-ADIS with their original assessor, yielding post-therapy Hamilton Anxiety and Depression scores and the Overall Assessor Severity Rating. Finally, they participated in a repetition of the physiological assessment session, including the self-relaxation and worry periods with the same mentation measures. RESULTS
Pre -session anxiety and depression scores
Table 1 presents means and standard deviations for the GAD and Control Ss on the various pre-session assessor ratings and self-report questionnaires. In each case, significant differences were found between the two groups [all ts_ (24) > 5.45, P < O.OOl]. Since norms on carefully screened, nonanxious Ss are not available for most of these measures, the Control data provided in Table 1 will allow in the future an estimation of local norms for the overall therapy outcome project. Also included in Table 1 are the posttherapy scores for GAD clients subsequent to their receiving 12 sessions of therapy [except for the Fear Questionnaire-Total, all pre-test vs post-test ts (24) > 2.66, P < 0.021. Thought/image
mentation reports
The frequencies of mental content samples that were reported by Ss to be ‘both thoughts and images’ or ‘unsure’ did not differ significantly between groups or between relaxation/worry periods (during relaxation, GAD vs Control frequencies for ‘both’ were 10 and 10, respectively, and for ‘unsure’ were 2 and 1; during worry, frequencies for ‘both’ were 14 and 11, and for ‘unsure’ were 2 and 1). Consequently, these data were disregarded in subsequent analysis. Figure 1 displays the percentage of the total sample reports that Ss labeled as ‘thought’ and as ‘image’, during relaxation and worry periods for GAD and Control Ss. Selected critical comparisons among groups and periods on differences between thought and image frequencies were Table I. Means and standard deviations on assessor ratings and questionnaires Ss at pre-session and for GAD Ss at post-therapy GADPre
Hamilton Anxiety Hamilton Depression Assessor Severity STAI RRAQ
FQ CSI Zung
WQ BDI
Control-Pre
for GAD and control GAD-Post
M
SD
M
SD
M
SD
20.16 13.79 4.7 59.46 27.92 42. I5 176.69 43.38 68.23 15.69
3.94 5.18 0.60 8.42 6.74 17.24 76.61 a.89 6.21 5.86
1.65 0.99 0.04 27.38 II.54 12.85 30.85 24.77 28.38 I.61
1.33 0.89 0.14 3.50 2.1 I 8.85 17.55 3.68 6.36 2.26
9.41 7.21 2.22 46.77 18.85 30.61 98.23 33.61 54.54 9.54
5.15 4.57 1.23 13.05 7.80 19.41 53.86 5.71 16.09 5.90
STAI, Trait version of the State-Trait Anxiety Inventory; RRAQ, Reactions to Relaxation and Arousal Questionnaire: FQ, Fear Questionnaire-Total Score; CSI, Cognitive/Somatic Anxiety Inventory-Total Score; Zung. Zung Self-Rating of Anxiety: WQ. Worry Questionnaire: BDI. Beck Depression Inventory.
T. D. BORKOVEC and JONATHANINZ
156
45%
30%
15%
0% Fig. I. Percentage
of reports
of ‘thought’
and ‘image’ during nonanxious control
relaxation Ss.
and worry
periods
for GAD
and
tested by x2, using Sime’s correction to Bonferroni adjustment for multiple tests (Simes, 1986). During relaxation, Controls showed a strong predominance of imagery with little thought, whereas GAD displayed nearly equal amounts of each mentation type, x’(l) = 4.34, P < 0.037. GAD and Control Ss were very similar when they worried, with thought predominating over image occurrence. For Controls, mentation type shifted from image predominance during relaxation to thought predominance during worry, x2( 1) = 9.75, P c 0.002; for GADS, percentages shifted from roughly equal thoughts and images during relaxation to increases in thought and decreases in imagery during worry, though not to a statistically significant degree. The greatest discrepancy between thought and image frequencies was found when contrasting Controls during relaxation and GADS during worry, x’(l) = 10.00, P < 0.002. Percentages of thought and image reports by GAD clients at the post-therapy assessment session indicated a normalization during the relaxation period: images (51.3%) and thoughts (12.8%) nearly matched the percentages found for the nonanxious Controls. These changes from pre- to post-therapy percentages approached significance after Bonferroni adjustment, x’(l) = 4.54, P < 0.033. Relative to the pre-therapy worry period, the GADS also showed a reversal away from predominance of thought during post-therapy worry (image = 30.8%; thought = 23.1%), but this change was not statistically significant. Mentation
afect
ratings
Table 2 presents means and standard deviations on the pleasant/unpleasant and relaxed/anxious rating scales for the two groups during each sample of the relaxation and worry periods at the pre-therapy assessment session and for the GAD group at the post-therapy session. A 3-way (Group by Relaxation/Worry Periods by Sample) repeated measures multivariate analysis of variance was applied to each of the rating scale scores from the pre-therapy session. Significant Sample, Period, and Sample by Period effects emerged from analysis of the pleasant/ unpleasant scale scores. Unpleasantness increased from relaxation to worry periods and increased during repeated sampling, especially during the relaxation period. However, no effect involving Group was found. The same significant within S effects were found on the relaxed/anxious scores, but the main effect of Group was also significant, F( 1,24) = 6.10, P < 0.03. Average anxiety rating of mental content was greater for the GADS than for the Controls collapsed across relaxation and worry periods, M = 5.08 vs 4.08, respectively. Three-way (Pre/Post-therapy by Relaxation/Worry Periods by Sample) repeated measures multivariate analyses of variance on scale scores for the GAD group assessed whether changes in
in generalized anxiety disorder
Worry
157
Table 2. Means and standard deviations on pleasant/unpleasant and relaxed,‘anxious rating scales for GAD and control Ss during each sample of the relaxation and worry periods and for the GAD Ss at oost-thcratw Assessment Relaxation Sample Pleasant/unpleasant GAD I44 (SD) Control M (SD) GAD-post (SD) Relaxed/anxious GAD &) Control :D) GA&post (SD)
I 2.68
wow
2
3
I
2
3
3.31
3.77 (2.09)
7.31 (1.11)
7.23 (1.74)
6.54 (1.39)
6.08
(1.81)
6.23 (2.01) 6.61 (2.22)
(1.49)
(1.80)
2.15 (1.28) 3.31 (1.70)
3.23 (2.49) 4.08
3.54 (1.71) 3.61
(2.46)
(1.98)
6.69 (1.32)
2.38
3.69 (2.02)
4.00 (1.68)
7.08
6.85
(1.45)
(0.95)
(1.14)
6.46 (1.27)
2.00 (1.22) 2.54 (1.71)
2.38 (1.56) 2.54 (I .90)
2.69 (1.89) 2.61 (1.94)
6.00 (1.63) 4.77 (2.35)
5.85 (2.15) 4.42 (2.27)
5.54 (2.07) 4.77 (2.3 I)
6.54
(2.66) 5.46 (2.07)
affective value of mental content samples occurred subsequent to effective therapy. No effects involving the Pre/Post-therapy factor were significant on the pleasant/unpleasant scores. On the anxiety scale, the main effect of Pre/Post-therapy was significant, F( 1, 12) = 16.8 1, P < 0.01, whereas the Pre/Post-therapy by Period interaction approached significance, F( 1, 12) = 4.20, P < 0.065. Recovered GAD clients reported less anxiety associated with their sampled mentation at post-therapy than at pre-therapy (M = 3.61-5.08), especially during the worry period.
DISCUSSION The results from the present study indicate that, in terms of phenomenological report, worry involves primarily thought as opposed to imaginal activity. This characterization of worry is supported in three different ways. First, the dispositional group comparison showed imagery to predominate the mental content of nonanxious Ss during a self-relaxation period, whereas equal amounts of each mentation type were reported by GAD clients who are chronic worriers by diagnostic definition. Second, manipulation of relaxation and worry periods revealed a shift among nonanxious Ss in the direction of thought predominance; for GADS, further increases in thought and decreases in imagery occurred, though not significantly so. Finally, effective therapy shifted GAD clients to a more normal predominance of imagery during relaxation with a similar, possibly compensatory (albeit nonsignificant) shift during worry. Not surprisingly, pre-therapy mentation samples were reported to have more anxious valence among the clients throughout the periods, and these also significantly shifted in the nonanxious direction subsequent to therapy. Whether S classification of mental content into ‘thoughts’ vs ‘images’ validly reflects actual differences in cognitive process remains to be determined. We do know, however, that distinctive electrocortical differences have been found between chronic worriers and nonworriers during worry periods. Specifically, worriers show increases in left hemisphere, frontal j? activation relative to nonworriers (Carter, Johnson & Borkovec, 1986). It remains to be seen whether these two sets of distinguishing cognitive activity (reported predominance of thought activity and electrocortical activity) are truly reflecting the same process. If worry is predominantly a conceptual, thought activity, what function might it serve? GAD clients are distinctive from other anxiety disorders in that fewer obvious, circumscribed environmental situations are overtly avoided. One could speculate that worrying is a method of attempting to avoid future catastrophe, a common attribution made by worriers to explain their worrying, and that such anticipation is frequently reinforced by the nonoccurrence of feared outcomes. The present data provide clues, however, that there may exist an even more basic function for worry in anxiety maintenance. The declines in imagery and increases in thought that seem to characterize worry in the present study may represent the motivated avoidance of imaginal process. If imagery
158
T.D.
B~RKOVEC
and JONATHAS
INZ
is a primary vehicle for the somatic activation of emotion (Lang, 1988) then one way to learn to avoid somatic anxiety is to avoid imagery by increasing the generation of, or attentional focus on, conceptual activity. Worry may thus represent an avoidance of affect in general or anxious emotional experience in particular. Under such circumstances, triggers for worry episodes are likely to become broadly generalized to any stimulus associated with any degree of emotional elicitation. At this point, such a hypothesized avoidance function for worry is very speculative. Further research will need to isolate the relationships between conceptual activity and worry, on the one hand, and imagery, imagery vividness, general affect, and anxious affect on the other. Given the importance of worry (or “apprehensive expectation”, Barlow, 1988) in each of the anxiety disorders, increased understanding of how ideational activity interacts with the imaginal and affective processing systems is likely to further our knowledge about the maintenance of these disorders. Acknoi&dgemenrs-This research was supported in part by NIMH Grant MH-39172 expressed to Teresa Elliott and Thomas Meyer who served as experimenters and management of the project.
to the first author. Appreciation is to Mildred Hopkins for her data
REFERENCES American Psychiatric Association (1987). Diugnosric and statisricol manual of mental disorders. Washington, D.C.: A.P.A. Barlow, D. H. (1988). Anxiefy and ifs disorders. New York: Guilford Press. Beck, A. T., Ward, C. H., Mendelson, M., Mock, J. & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-57 I. Borkovec, T. D. & Hu, S. (1990). The effect of worry on cardiovascular response to phobic imagery. Behatiour Research and Therapy, 28, 69-73. Borkovec, T. D. & Mathews, A. M. (1988). Treatment of nonphobic anxiety disorders: A comparison of nondirective, cognitive. and coping desensitization therapy. Journal of Consulting and Clinical Psychology, 56, 877-884. Carter, W. R., Johnson, M. C. & Borkovec, T. D. (1986). Worry: An electrocortical analysis. Advances in Behariour Research and Therapy, 8, 193-204. Deffenbacher. J. L. (1980). Worry and emotionalitv. In Sarason, I. G. (Ed.), Test anxiety: Theorv. research. anduDplicarions. . Hillsdale, N.J.: Erlbaum. _ DiNardo. P. A., O’Brien. G. T., Barlow, D. H., Waddell, M. T. & Blanchard, E. (1983). Reliability of DSM-III anxiety disorder categories using a new structured interview. Archices of General Psychiufry. 40, 1070-1078. Foa, E. B. & Kodak, M. J. (1986). Emotional processing of fear: Exposure to corrective information. Psychological BuNerin, 99, 20-35. Hamilton, M. (1959). The assessment of anxiety states by rating. British Journal of Medical Psychology, 32, 50-55. Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery and Psychiatry, 23, 56-62. Heide, F. J. & Borkovec, T. D. (1983). Relaxation-induced anxiety: Paradoxical anxiety enhancement due to relaxation training. Journal of Consulting and Clinical Psychology, 51, I7 l-182. Lang, P. J. (1985). The cognitive psychophysiology of emotion: Fear and anxiety. In Tuma, A. H. & Maser, J. D. (Eds), Anxiefy ond rhe onxiery disorders. Hillsdale, N.J.: Erlbaum. Lang, P. J. (1988). The image of fear: Emotion and memory. Invited address, Behavior Therapy World Congress, Edinburgh. Marks, I. M. & Mathews, A. M. (1979). Brief standard self-rating for phobic patients. Behaciour Research and Therapy, 17, 263-267. Mathews, A. M. & MacLeod, C. (1986). Discrimination of threat cues without awareness in anxiety states. Journal of Abnormal Psychology, 95, 131-138. Meyer. T. (1988). Development of a screening questionnaire to identify levels of chronic worry. Unpublished masters thesis, Pennsylvania State University, University Park, Pa. Simes, R. J. (1986). An improved Bonferroni procedure for multiple tests of significance. Biomefrika 73, 751-754. Speilberger, C. D., Gorsuch, R. L. & Lushene, R. F. (1970). Manualfor the Srare-Trait Anxierv Incenrory. Palo Alto, Calif.: Consulting Psychologists Press. York, D., Borkovec, T. D., Vasey, M. & Stern, R. (1987). Effects of worry and somatic anxiety induction on thought intrusions. subjective emotion, and physiological activity. Behaciour Research and Therapy, 25, 523-526. Zung, W. W. K. (1975). A rating instrument for anxiety disorders. Psychosomarics, 12, 371-379.
