&urno/ ofAnr;en. Disorders, Vol. I, pp. IS&160. Prmted in the USA. Ail right%%reserved.
1987 CopyrIght
0887-61X5/87 $3.00 + .lW Q 1987 Pergamon Journals. Lid.
The Relationships Among Panic, Fear, and Avoidance MICHELLE. G. CRASKE,PH.D.*, WILLIAM. C. SANDERSON,AND DAVID. H. BARLOW, PH.D. State University of New York at Albany
Abstract-Interview data from 57 patients assigned a diagnosis of panic disorder with agoraphobia were analyzed in terms of panic, and fear and avoidance of specific situations. Patients were categorized into three avoidance groups (mild, moderate, and severe) in accordance with revisions proposed for DSM-III. Severe avoiders were reportedly experiencing panic as frequently as mild and moderate avoiders, suggesting that panic frequency is not the major determinant of avoidance behavior. Panic frequency was more closely related to fear ratings than to avoidance ratings. Fear of panic was characteristic of all avoidance groups. Previous history of panic was unrelated to current patterns of fear and avoidance. A long history of repeated panic was frequently unaccompanied by a pattern of extensive avoidance. The relevance of the data to an interoceptive conditioning model of panic in agoraphobia and to the role of unexpected panic attacks is discussed.
The revision of the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders (DSM-III: American Psychiatric Association. 1980) will eliminate the label of agoraphobia with panic attacks. The new category of panic disorder with agoraphobia (PDA) will consist of subcategories based upon the level of avoidance and the frequency of panic attacks (DSM-III Revised: APA, 1986). When diagnosing PDA, the assessor will designate the severity of avoidance (or endurance with dread) exhibited by the patient as mild, moderate, or severe. Thus, the DSM-III diagnosis of agoraphobia with panic attacks will become panic disorder with agoraphobia: severe agoraphobic avoidance. Hence, the central role of panic attacks in the agoraphobic disorder is recognized (Barlow, 1986: Spitzer & Williams, 1985) and more exact specification of the level of avoidance is allowed. This shift in classification appears to be consistent with the behavioral model set forth by Goldstein and Chambless (1978) who proposed that a “fear of fear” (i.e., fear of having a panic attack), rather than a fear of * Address correspondence and reprint requests to Dr. Michelle G. Craske, Center for Stress & Anxiety Disorders. State University of New York at Albany, Albany. NY 12203. 157
1.54
M. G. CRASKE
ET AL.
places avoided, is the central phobic component in agoraphobia. According to their model, situations which are avoided develop properties of conditioned stimuli and thus elicit anxiety in a secondary fashion. Also, recent research from our clinic and elsewhere, has shown that patients report the development of varying degrees of avoidance behavior after, rather than before, experiencing one or more panic attacks (Uhde. Roy-Byrne, Boulenger, Vittone, & Post, 1985). This is consistent with Klein’s (1981) original theoretical notion that agoraphobia develops secondarily to repeated panic attacks. Additional support for this notion is found in Swinson’s (1986) report, in which not one of approximately 300 patients developed agoraphobia without first experiencing a panic attack. Several questions arise that pertain to the face validity of the hypothesis that panic is a significant factor in the development of agoraphobia avoidance, and the corollary hypothesis that panic occurs upon exposure to feared situations. First, do patients with extensive avoidance have a history of more frequent panic attacks than patients with minimal avoidance? Second, does extensive avoidance successfully eliminate panic? The present study attempted to address these questions by examining patients who reported varying amounts of avoidance. Of interest was the frequency of panic in the past and at present, and its relationship to situational fear and avoidance. Of additional interest were the relationships between duration since first panic and extent of avoidance, and the role of fear of panic in avoidance behavior. METHOD
Subjects were 57 patients assessed at the Phobia and Anxiety Disorders Clinic, State University of New York at Albany. All received a primary diagnosis of panic disorder with agoraphobia (DSM-III-R). Exclusionary criteria included current alcohol or drug abuse, psychotic disorder, and a primary diagnosis of major depression. Assessment was conducted through administration of a structured interview schedule, the Anxiety Disorders Interview Schedule-Revised, which yields reliable diagnostic judgments (DiNardo, O’Brien, Barlow, Waddell, & Blanchard, 1983). Fear and reported avoidance of 23 specific situations were assessed within the structured interview. These included the following: driving, riding in a car, grocery stores, malls. crowds, buses, planes, taxis. waiting in line, walking, elevators, being at home, movies, restaurants, theaters, auditoriums, church, tunnels, small rooms, parks, squares. work, and other. Clients were asked how much fear they experienced in each situation “because you might be unable to leave in case you felt faint or panicky or ill.” That is, fear was rated as the discomfort felt upon anticipation of panic, using a scale of 0 (no fear) to 4 (very severe fear). Avoidance of each situation also was rated from 0 (no avoidance) to 4 (never enters the situation, even with a safe person). Ratings were as-
PANIC,
FEAR,
AND AVOIDANCE
155
signed by clinicians and were analyzed in terms of: (a) total sum of fear and avoidance ratings across all situations, (b) mean fear and avoidance, and (c) the frequency of situations feared and avoided. The patients were grouped into one of three categories: mild avoidance (n = 35), moderate avoidance (n = 16) and severe avoidance (n = 6). Groupings were established on the basis of global clinical judgement, in accordance with descriptions provided in DSM III-R: mild = some avoidance (or endurance with distress) but relatively normal lifestyle; moderate = avoidance results in constricted lifestyle; and severe = avoidance results in being nearly or completely housebound or unable to leave the house unaccompanied. Specific criteria were not established with respect to the amount of fear and avoidance reported for the 23 situations. The three “avoidance” groups were compared on the following variables (all obtained from the structured interview): duration since first panic attack. frequency of panic at the time of initial assessment, the greatest frequency with which panic occurred per week since onset, and the pressure of fear of panic. RESULTS Reliabilit]
Avoidance was rated by two independent raters for 9 patients who received a diagnosis of panic disorder with agoraphobia, based on separate administrations of the ADIS-R. Correlation coefficients were as follows: total fear, I’ = .98. and total avoidance, r = .98. Subject
Characteristics
The sample consisted of 39 females and 18 males, whose ages ranged from 19 to 66 years (mean age = 35.2 years). Sixty-nine percent of the mild group, 56 percent of the moderate group, and 100 percent of the severe group were female. The mean ages of the groups did not differ: 35.7, 32.8, and 35.7 years, respectively. The groups did not differ with respect to the duration of their disorder, as measured by the mean number of months since the first panic attack. The overall average was 102 months, or 8% years, from the first panic until initial assessment. Six of the moderate group (38%) and 9 of the mild group (26%) reported that their panic began less than 29 months before initial assessment; 29 months was the minimum number of months reported by the severe group. Fear and Alvidance
Fear and avoidance ratings for the 23 situations were compared across the three groups through the use of one-way analyses of variance. The
156
M. G. CRASKEETAL.
group effect was significant for each measure: total scores for fear and avoidance, mean scores, and number of situations feared and avoided. Means, standard deviations, and F values are presented in Table 1. Significant effects were followed by paired comparisons through the use of Scheffe’s test (alpha = .05). All groups differed significantly from each other with respect to total fear, total avoidance, mean fear, and mean avoidance. The moderate and severe avoidance groups were not discriminable with regard to the frequency with which situations were feared or avoided. Fear and avoidance ratings correlated very highly: total, r = .94, p < .OOl; mean, r = .94, p < .OOl; and frequency, r = .87, p < .OOl. However, it is noteworthy that, while avoidance tended to increase as fear increased, absolute values were not equivalent. An interesting pattern emerged when the concordance between absolute fear and avoidance was examined. Percentage of concordance was calculated at each level of fear. For example, concordance occurred when a fear rating of 2 was matched by an avoidance rating of 2 for a given situation. Discordance was signified for all other avoidance ratings (0, 1, 3 and 4) at that fear level. The percentage of concordance increased linearly as the fear level increased: when fear = 1, fear and avoidance ratings agreed in 29% of the cases; fear = 2, agreement = 55%; fear = 3, agreement = 63%; and fear = 4, agreement = 84%. TABLE 1 FEARANDAVOIDANCEMEASURESFOREACHDIAGNOSTIC Avoidance Variable
Groups
Mean
GROUP Standard Deviation
F
Total Fear
Mild Moderate Severe
16.39 37.00 57.83
8.91 13.15 13.01
46.29*
Total Avoid
Mild Moderate Severe
10.36 29.69 54.00
7.92 12.04 14.57
54.93*
Mean
Mild Moderate Severe
0.71 1.79 2.68
0.43 0.55 0.67
4X.69+
Mild Moderate Severe
0.50
5s. IO”
2.54
0.37 0.53 0.77
Number of Situations Feared
Mild Moderate Severe
9.03 15.13 18.50
4.91 4.26 2.59
16.60*
Number of Situations Avoided
Mild Moderate Severe
6.00 13.88 18.00
4.38 4.26 2.83
31.44*
Fear
Mean Avoid
lp<.w1
1.44
PANIC. FEAR.
157
AND AVOIDANCE
TABLE
2
PANICFREQUENCYIN EACHDIAGNOSTIC GROUP*
Panic frequency Greatest
frequency
Mild Avoidance (N = 35)
Moderate Avoidance (N = 16)
Severe Avoidance (N = 6)
9.74 (14.58) 7.15 (5.30)
15.46 (16.83) 7.06 (6.06)
17.40 (16.85) 6.67 (2.25)
over past month of panic
per week
Panic
Twenty-four patients were not asked “Have you been afraid that you might have more attacks?,” due to the structure of the interview. All except two of the remaining 33 patients stated that they were afraid of the recurrence of panic. Both people who reportedly did not anticipate panic with dread belonged to the mild avoidance group. Panic frequency measures for each diagnostic group are presented in Table 2. The trend for a linear increase in panic frequency (for the month prior to assessment) across the three avoidance groups was not significant due to the large degree of within-group variability. The different avoidance groups reported average estimates of the greatest frequency with which panic had ever occurred (per week) that were very similar (see Table 2). Estimates of current panic frequency and greatest panic frequency correlated significantly: r = .45, p < .Ol. Nonparametric correlations were performed between panic frequency (current and greatest frequency) and the various fear and avoidance measures. Correlation coefficients are presented in Table 3. Estimates of the grsarestfrequenc_y of panic did not correlate significantly with any measure of fear or avoidance. Estimates of the current frequency of panic correlated significantly with total fear and mean fear, and correlated moderately with the frequency of situations feared @ < .06). Estimates of the TABLE
3
CORRELATIONSBETWEENPANICFREQUENCYANDFEARANDAVOIDANCE Current Frequency of Panic Total fear Total avoid Mean fear Mean avoid Number of situations Number of situations
feared avoided
.36* .27 .36* .25 .29 .20
Greatest Frequency of Panic .I8 .07 .18 .04 .15 .02
158
M. G. CRASKE ET AL.
current frequency of panic did not correlate sure of avoidance.
significantly
with any mea-
DISCUSSION The primary purpose of this investigation was to examine the relationship between estimates of panic and of avoidance behavior. Several points of interest from the analyses will be noted before commenting upon the implications of these findings. Analyses that compared measures of fear and avoidance among the three groups indicated that the clinicians’ global judgement of diagnostic level of avoidance was closely related to the amount of self-reported fear and avoidance of the 23 situations. The groups differed markedly in terms of total amount and average amount of fear and avoidance. It is noteworthy for the purpose of assessment of phobic anxiety that frequerlcy of the situations feared and avoided was a less discriminating measure. In general, the more avoidance subjects reported, the more fear they reported also. Hence, correlations between fear and avoidance measures were very high. Of interest, was the finding that concordance between the absolute levels of fear and avoidance was increasingly evident as fear intensified. At milder levels of fear, avoidance ratings varied considerably. These findings tend to support the hypothesis of Rachman and Hodgson (1974) that discordance between the subjective and behavioral response systems of anxiety is most likely to occur at mild fear intensities, whereas concordance is most likely to occur at high fear intensities. The frequently made clinical assumption that clients who present with extensive avoidance have experienced anxiety over a longer duration than those presenting with minimal avoidance requires further testing. The avoidance groups in this study did not differ significantly with respect to duration since their first panic attack. There was a trend for extensive avoiders to report a longer duration of their disorder in comparison to the other groups, but several minimal avoiders reported that their first panic occurred many years prior to our assessment. The chronological relationship between panic and avoidance was not assessed adequately by the procedures herein, given the inadequacy of the recall of an initial panic attack as a marker for the onset of a current anxiety disorder. However, the data suggest that it cannot be assumed that severe avoidance is the inevitable consequence of a long history of panic attacks. nor that it always develops after a pattern of minimal avoidance. The finding that 94% of the subjects anticipated the recurrence of panic lends credence to recent revisions proposed for the DSM-III which require the presence of only one panic attack for a diagnosis of panic disorder given that it is accompanied by substantive fear of the recurrence of panic. The results suggest that fear of panic may not determine extensiveness of avoidance, as even mild avoiders feared the recurrence
PANIC,
FEAR,
AND AVOIDANCE
1.59
of panic. However, the dichotomous nature of the data concerning anticipatory fear (“yes or no”) reduces the sensitivity of this measure. A measure of intensity of anticipation may, on the other hand, correlate with the extent of avoidance and differentiate the three avoidance groups. A definite but nonsignificant trend was present for severe avoiders to report more panic attacks in the past month than mild avoiders. Certainly, they did not experience fewer panics than mild avoiders, a finding which pertains to the initial question of the success with which avoidance eliminates panic. The report of repeated panic despite avoidance of situations in which panic is considered likely to occur tends to reinforce the notion that panic in agoraphobia is not only situationally cued, but may occur in response to nonsituational, and perhaps interoceptive, cues (Barlow, in press). Imperfect coupling was observed between panic and avoidance behavior. Avoidance of situations was, reportedly, a means of avoiding the experience of panic, and, to that extent, the results support recent behavioral conceptualizations of agoraphobia. However, given the same frequency of panic, different people avoid to different extents. Moreover, the failure of the avoidant strategy to extinguish in those patients for whom avoidance seemingly did not alleviate the experience of panic is surprising. It suggests that panicfrequency is not the only determinant of avoidance behavior, especially in view of the finding that previous history of panic frequency seemed to have little bearing on current presentation. Possibly, cognitive parameters such as the feared consequences of panic, or the expectancy that panic will occur, or behavioral parameters such as how one has learned to cope with anxiety (especially given the absence of males in the severe avoidance group), may be important determinants of avoidant behavior. Indeed, panic frequency related much more strongly to fear than to avoidance behavior. Rachman and Levitts’ (1985) experimental analysis of the consequences of panic demonstrated that the occurrence of an unexpected panic attack is followed by an increase in predicted fear (i.e., the anticipation of panic). In a similar fashion, the current data demonstrates that higher frequencies of panic attacks tend to be accompanied by higher levels of fear. A major difference between the two sets of results is that the 1985 finding is restricted to unexpected panic. An interesting speculation emerges: panickers who avoid extensively may prevent the occurrence of panic attacks that are expected to occur in the situations they avoid, but continue to experience unexpected panics (e.g.. when at home relaxing). Unexpected attacks may be cued interoceptively. as suggested by Barlow (in press). Hence, in accordance with Rachman and Levitts’ findings, anticipatory fear is elevated, which, in turn. provides a strong motivation for avoidance of situations that have been in some way associated with panic. Of course, this also implies that situational avoidance is a successful means of preventing some (expected) panics and is maintained through negative reinforcement. Further investigation is deemed necessary to fully answer these questions, with
160
M.
‘3.
CRASKE
ET AL.
special attention to the situational and nonsituational, expected and unexpected dimensions of panic attacks in different avoidance groups, and with a larger sample of severe avoiders than was obtained herein. REFERENCES American
Psychiatric
Association. (1980). Diagnosric stntisrical rn~~r~d of menr~i/ disDC: Author. Association. (1986). DSM-IIIR in development 8/l/86. Washington.
orders (3rd ed.). Washington,
American Psychiatric DC: Author. Barlow, D. H. (1986). In defense of panic disorder with agoraphobia and the behavioral treatment of panic: A comment on Kleiner. The Behavior Therclpist. 9, 99- 100. Barlow, D. H. (in press). Panic, anxiety and the anxiety disorders. Guilford Press. New York. DiNardo, P. A., O’Brien, G. T., Barlow, D. H., Waddell. M. T., & Blanchard. E. B. (1983). Reliability of DSM-III anxiety disorder categories using a new structured interview. Archives
ofGeneral Psychiarry. 40, 1070-1074.
Goldstein, A. J., KcChambless.
D. (1978). A reanalysis of agoraphobia. Erhcn?ov Thercrpv
9, 47-59.
Klein, D. E (1981). Anxiety reconceptualized. New research
and changing
concepts.
In D. F. Klein & J. Rabkin (Eds.). An.riery.
New York: Raven.
Rachman, S.. & Hodgson, R. (1974). Synchrony and desynchrony Behavior Research
and Therapy,
Rachman. S., & Levitt, K. (1985). Panics and their consequences. Therupy,
in fear and avoidance.
12, 3 I 1-3 18. BehrtrYor Resetrrch cmd
23(5), 585-600.
Spitzer, R. L.. & Williams, J. B. W. (1985). Proposed revisions in the DSM-III classification of anxiety disorders based on research and clinical experience. In A. H. Tuma & J. D. Maser (Eds.), Anxiefy and rhe anxiety disorders. New Jersey: Erlbaum. Swinson, R. P. (1986). Reply to Kleiner. The Behavior Therapist. 9(6). 110-128. Uhde, T. W.. Roy-Byrne, P. P.. Boulenger, J.. Vittone, B., & Post, R. M. (1985). Phenomenology and neurobiology of panic disorder. in A. H. Tuma &J. D. Maser (Eds.). Anrief? and the anxiety disorders. New Jersey: Erlbaum.