Social phobia: A comparison of specific and generalized subtypes and avoidant personality disorder

Journal of Abnormal Psychology 1992, Vol. 101, No. 2, 326-331

Copyright 1992 by the American Psychological Association Inc 002I-843X/92/S3.00

Social Phobia: A Comparison of Specific and Generalized Subtypes and Avoidant Personality Disorder Samuel M. Turner, Deborah C. Beidel, and Ruth M. Townsley Western Psychiatric Institute and Clinic, Department of Psychiatry, University of Pittsburgh School of Medicine The validity of the social phobia subtype distinction was examined in a large sample of carefully diagnosed social phobics (N= 89). Generalized and specific subtypes were diagnosed reliably, and the generalized subtype showed a consistent pattern of greater symptom severity than did the specific subtype. In addition, generalized social phobics with and without avoidant personality disorder were compared, and a difference was found for only 1 of 4 parameters. The results are discussed in terms of the validity of subtyping in social phobia and the diagnostic boundary between social phobia and avoidant personality disorder.

With the publication of the Diagnostic and Statistical Manual of Mental Disorders (3rd ed.; DSM-III; American Psychiatric Association, 1980), social phobia was introduced formally into the diagnostic nomenclature. Social phobia is currently defined as "a persistent fear of one or more situations in which the person is exposed to possible scrutiny by others and fears that he or she may do something or act in a way that will be humiliating or embarrassing" (p. 243, American Psychiatric Association, 1987). Although social phobia is conceptualized as a specific phobia, those afflicted with the disorder frequently suffer from pervasive social inhibition, as well as significant impairment in social and occupational functioning and increased rates of alcohol and drug use (Amies, Gelder, & Shaw, 1983; Liebowitz, Gorman, Fyer, & Klein, 1985; Sanderson, Rapee, & Barlow, 1987; Turner & Beidel, 1989; Turner, Beidel, Dancu, & Keys, 1986). A number of changes in the diagnostic criteria for social phobia were introduced in the revised DSM-III (DSM-IH-R; American Psychiatric Association, 1987). One major change was the creation of the generalized subtype. The DSM-HI-R instructs clinicians to "specify a generalized subtype if the phobic situation includes most social situations" (p. 243), but it gives no information about a subtype specification for those patients who do not fear most social situations. However, fearing only specific situations (such as public speaking) has been referred to as the specific subtype. Under the guidelines provided in the DSM-III-R, the distinction between the specific and generalized subtypes is unclear and can be interpreted in a number of different ways. For example, most situations may mean most types of situations, although it is unclear what number qualifies as most. Alternatively, some researchers have proposed a subtype distinction between those who have fears of the most commonly occurring social situations (e.g., conversations and social gatherings or parties) and those who have difficulty

with less common, performance-oriented situations (e.g., speeches and meetings; Schlenker & Leary, 1982). However, the utility of these two methods (i.e., number of types of feared situations vs. frequency with which the situations occur) for distinguishing between the subtypes has yet to be examined empirically. Furthermore, it is unclear whether the subtype distinction provides useful information in terms of symptomatology, etiology, prognosis, and response to treatment. Only one published study has attempted to determine whether social phobics who had either the generalized or specific subtype differed in terms of symptomatology. In this study, Heimberg, Hope, Dodge, and Becker (1990) retrospectively reviewed the records of 57 social phobics in order to assign a subtype diagnosis. Patients with fears in only circumscribed situations (i.e., public speaking) were assigned the specific subtype, and patients with fears in most or all social situations (no set number was specified) were assigned the generalized subtype. On a battery of self-report instruments, the subtypes differed on indexes of depression and cognitions related to social situations but did not differ significantly on measures of social phobic anxiety or anxiety in public speaking situations. When differences were found, the generalized subtype patients were more severely impaired. In addition, patients' subjective anxiety, cognitions, heart rate, and objectively rated performance and anxiety were recorded during a feared behavioral task. The specific social phobics gave a speech and the generalized social phobics had a conversation with a confederate. Although some differences were found between the subtype groups (i.e., the specific social phobics had greater heart rate response and more positive thoughts during the task and reported less anxiety after the task), the fact that each group had individualized tasks makes it difficult to determine if the differences were due to the subtype distinction or to variation in the tasks. A second major change in the DSM-III-R relates to the diagnosis of avoidant personality disorder (APD). On the basis of the criteria in the DSM-III, a patient who received the diagnosis of APD could not be assigned the diagnosis of social phobia. The central components of APD according to the DSM-IH were hypersensitivity to rejection, unwillingness to enter into

This project was supported in part by National Institute of Mental Health Grant MH41852. Correspondence concerning this article should be addressed to Samuel M. Turner, Western Psychiatric Institute and Clinic, 3811 O'Hara Street, Pittsburgh, Pennsylvania 15213. 326

327

SOCIAL PHOBIA

relationships, a strong desire for relationships, low self-esteem, and social withdrawal. In contrast, the DSM-III-R allows a patient to receive both diagnoses of social phobia and APD. Furthermore, the criteria for APD have changed. Currently, the central aspect of the revised APD criteria is a fear of negative evaluation and discomfort in social situations. When APD is denned in this fashion, the similarity between it and social phobia is obvious. Thus, the boundary between the two disorders has become blurred, particularly with respect to the generalized subtype. Essentially, there is significant overlap between the diagnostic criteria for social phobia and three of the seven APD criteria: (a) avoids social or occupational activities that involve significant interpersonal contact, (b) is reticent in social situations because of a fear of saying something inappropriate or foolish, and (c) fears being embarrassed by blushing, crying, or showing signs of anxiety in front of other people. Only one additional criterion would be required for the APD diagnosis. In fact, in a sample of 71 £>SA/-///-.R-diagnosed social phobics, 75% had either a DSM-III-R APD diagnosis or traits of APD (Turner, Beidel, Borden, Stanley, & Jacob, 1991). However, no study has examined empirically the relation between APD and social phobia as they are defined in the DSMIII-R. The primary purposes of this study are to examine the validity of the social phobia subtypes by determining if there are symptomatic differences between the two and to clarify the relation between the generalized subtype of social phobia and APD. Method Subjects The subjects were 89 patients (47 women and 41 men) consecutively admitted to the Anxiety Disorders Clinic at Western Psychiatric Institute and Clinic with a DSM-III-R diagnosis of social phobia. Data from 71 of these subjects were used in a study in which the relation of social phobia to Axis I and Axis II disorders was examined (Turner et al, 1991). Diagnoses were assigned on the basis of the results of interviews conducted by experienced clinicians. First, a semistructured psychiatric interview schedule, the Initial Evaluation Form (IEF; Mezzich, Dow, Rich, Costello, & Himmelhoch, 1981) was conducted by a trained clinician. The patient was then clinically interviewed conjointly by the clinician and a staff psychiatrist, and a preliminary diagnosis was derived by consensus. The patient was then referred to the Anxiety Disorders Clinic. In order to confirm the diagnosis of social phobia, the Anxiety Disorders Interview Schedule-Revised (ADIS-R; DiNardo et al., 1986), which follows DSM-III-R criteria, was administered by a clinician experienced with anxiety disorders. Twenty-five percent of the ADIS-R interviews were videotaped and rated by a second clinician who was unaware of the initial diagnosis. The second clinician assigned the diagnosis of social phobia in all cases. A third clinician administered the full interview version of the Structured Clinical Interview for DSM-III-R (SCID-II; Spitzer & Williams, 1986) in order to establish personality disorder diagnoses. Among the 89 subjects, 15 met criteria for APD. Twenty-five percent of the interviews were videotaped and rated by a different clinician who, again, was unaware of the initial diagnosis. Interrater reliability for the diagnosis of APD, as estimated by the kappa coefficient, was .65. Patients were excluded from the study if they met criteria for another primary Axis I or Axis II disorder other than APD. Thirty-eight percent of the sample met criteria for at least one other

Axis I disorder, and 24% met criteria for an Axis II disorder (including APD). The frequencies of each of the comorbid diagnoses are shown in Table 1. The average age of the subjects was 36.70 years (SD = 8.69). Eighty-four of the subjects were Caucasian, 2 were African American, 2 were Asian, and 2 were Arabic.

Assignment of Specific or Generalized Subtype Diagnosis The DSM-III-R criteria do not clearly specify an operationalized definition of most social situations. Therefore, in order to assign subtype diagnoses, operationalized criteria were formulated. Our clinical experience with these patients suggested that defining most social situations to mean fears of the most commonly occurring social situations (e.g., conversations), as opposed to only performance-oriented situations (e.g., speeches), had the most clinical significance. Therefore, the criteria used in this study were defined as follows: Patients were assigned a generalized subtype diagnosis if they feared parties (social gatherings), initiating conversations, or maintaining conversations, and patients were given a specific subtype diagnosis if they feared only circumscribed situations, such as giving speeches, speaking in meetings, eating or writing in public, or using public restrooms. Subjects who feared multiple specific types of situations (e.g., speeches and writing in public) but did not fear parties or conversations received the specific subtype diagnosis. In order to meet criterion for a given type of situation, subjects were required to rate that situation as at least moderately fear provoking (i.e., 2 on a 0-4 Likert scale). These determinations were made by a review of the ADIS-R interview schedules in which patients were questioned about each of these types of situations. In order to evaluate interrater agreement, two independent raters determined the subtype diagnosis for each patient. One subject's subtype diagnosis was determined from IEF information because ADIS-R data were not available. Two raters independently agreed on the diagnosis for this subject. Twenty-eight subjects met criteria for the specific subtype, and 61 subjects met criteria for the generalized subtype. Table 2 shows the number of subjects who feared each of the situations that were listed in the ADIS-R and used to determine subtype diagnosis. In the small number of cases in which the specific-subtype patients feared parties or conversations, they did so only in circumscribed situations (e.g., related to their job) and, therefore, were not assigned the generalizedsubtype diagnosis.

Instruments Subjects who took part in this study were participants in a study of treatment outcome, for which they underwent a comprehensive assess-

Table 1 Frequencies of Concurrent DSM-III-R Axis I and Axis II Disorders Disorder

Frequency Axis I

Generalized anxiety disorder Simple phobia Dysthymia Alcohol abuse

26 8 3 1

29 9 3 1

15 6 2 1

17 7 2 1

Axis II Avoidant Obsessive-compulsive Dependent Antisocial

328

S. TURNER, D. BEIDEL, AND R. TOWNSLEY

Table 2 Feared Situations Used to Determine Social Phobia Subtype Diagnosis Generalized

Specific (n = 27) Situation

Frequency

Speeches Meetings Eating Writing Restrooms Parties Initiating conversations Maintaining conversations

27 19 3 3 2 1 2 2

Frequency 100 69 11 11 7 4 7 7

60 52 23 15 3 50 41 35

98 85 38 25 5 82 67 57

two unstructured interpersonal interactions (one with an opposite-sex confederate and the other with a same-sex confederate). Confederates (research associates) were trained to remain neutral during the interaction and were instructed to respond in a matter-of-fact manner without asking questions; this left the burden of the conversation to the patient. The heterosocial scene required the patient to interact with the confederate as if he or she was on a first date; for the same-sex task, the patient was asked to interact with the confederate as if he or she was a new roommate or neighbor. Each interaction lasted 3 min. The interactions were videotaped and subsequently rated for facial gaze, voice tone, number of verbal initiations, length of verbalizations, and overall skill by a trained research associate. The raters and the confederates were unaware of the patients' subtype diagnosis. The average interrater reliability of the ratings was .75 (range, .70-.82) as estimated by the Pearson product-moment coefficient. Data for social skills were not available for 18 of the patients.

Results ment of their disorder and functioning before treatment. The following subset of measures obtained during this assessment were selected to address the specific questions examined in this study. General distress. As a measure of general distress, the Hamilton Rating Scale for Anxiety (HRSA; Hamilton, 1959), the Symptom Checklist-90-Revised, General Symptom Index and Anxiety subscales (SCL-90-R GSI and SCL-90-R Anxiety; Derogatis, 1983), and the Beck Depression Inventory (BDI; Beck & Beamsderfer, 1974) were used. Social anxiety and social functioning. Social anxiety and functioning were assessed with the Social Phobia and Anxiety Inventory (SPAI; Turner, Beidel, Dancu, & Stanley, 1989), the SCL-90-R Interpersonal Sensitivity subscale (SCL-90-R IS; Derogatis, 1983), independent evaluator ratings of symptom severity on the Clinical Global Impressions Severity Scale (CGI; Guy, 1976), and the Social Adjustment Scale-SelfReport (SAS-SR; Weissman & Bothwell, 1976). Responses during a behavioral-performance task. Patients were asked to make a 10-min impromptu speech during which physiological, self-report, and behavioral data were obtained. Although speech anxiety was not the primary fear of every patient, this speech task has been found to be a good behavioral task for anyone with maladaptive social anxiety (Beidel, Turner, Jacob, & Cooley, 1989). For the purpose of the speech, patients were provided with a number of possible topics and were given 3 min to prepare a speech on up to three of the topics. Three research assistants made up the audience. Before the assignment of the task, the patient was requested to rest quietly for 10 min while baseline physiological activity was monitored. After the baseline period, patients were requested to speak for 10 min. However, they were told that if they felt they could not continue, they could stop (i.e., escape from) the speech when a light went off to indicate that they had spoken for 3 min. All but six spoke for at least 3 min. The length of each speech was recorded. In addition, heart rate was assessed at 2-min intervals with an automated blood pressure and pulse rate monitor (Model SD-700A, IBS Corp., Waltham, MA). Because patients spoke for varying amounts of time, either the average of the first two recorded heart rates or a single heart rate, in cases in which they spoke for less than 4 min, was used for data analysis. Heart-rate data were not available for 3 patients who spoke for less than 2 min and for two subjects because of an inability to obtain accurate readings. Immediately after the speech, patients were asked to rate their level of anxiety during the speech on a 9-point Subjective Units of Distress Scale. Patients also reported their thoughts during the speech by completing the Social Interaction Self-Statement Test (SISST; Glass, Merluzzi, Biever, & Larsen, 1982). For this study, only the Negative Thoughts subscale was used. Ratings of social skills. Interpersonal skill was examined within

Reliability of Subtype Diagnoses Interrater agreement was estimated with the kappa coefficient. Agreement between the two independent ratings of social phobia subtypes was .97.

Design and Analyses Conceptually, there were three primary questions of interest in this study. First, what differences, if any, exist between specific and generalized social phobics, regardless of whether or not they have APD? Second, if patients with APD are excluded from receiving a social phobia diagnosis (as in DSM-III), are the remaining generalized and specific social phobics different? Finally, do generalized social phobics with APD differ from those without APD? These questions were addressed by using four sets of dependent variables specifically selected because of their potential to reveal relevant similarities or differences among these groups. The four domains included (a) level of general distress, (b) social anxiety and social functioning, (c) behavior, affect, and psychophysiological response during the speech task, and (d) ratings of social skill during role play tasks. Given the questions of interest, the analyses were conducted in two steps. In order to examine the first question, the specific and generalized subtypes were subjected to a two-group multivariate analysis of variance (MANOVA) on each of the four sets of variables. Next, the second and third questions were addressed in one-way MANOVAs to compare the following three groups of subjects: those with the specific subtype, those with the generalized subtype without APD, and those with the generalized subtype and APD. (There were no subjects with both the specific subtype and APD.) Significant differences were examined further with univariate tests on each variable within the appropriate set. Although these two sets of analyses are not orthogonal, they are necessary to address the critical question of how specific social phobics compare with generalized social phobics in the absence of APD.

Specific Versus Generalized Social Phobics Regardless of the Presence ofAvoidant Personality Disorder There was a significant multivariate difference between the specific and generalized social phobics on general distress, Ho-

329

SOCIAL PHOBIA telling's 7^(85) = 16.96, Wilks's X = 0.8337, p < .01. The univariate tests revealed that the specific-subtype subjects had less overall distress than did the generalized-subtype subjects on all four variables in this set (see Table 3): HRSA, F(1,85) = 5.90, p < .05; SCL-90-R GSI, F(l, 85) = 8.86, p < .01; SCL-90-R Anxiety, F(l, 85) = 5.74, p < .05; and BDI, F(l,85) = 14.59, p < .001. In addition, there was a significant multivariate difference between the groups on social anxiety and social functioning, Hotelling's H86) = 34.55, Wilks's X = 0.7134, p < .0001. Univariate differences showed that the specific subtype was less severe with respect to problematic social anxiety and social functioning on all four indexes: SPAI, F(l, 86) = 29.62, p < .0001; SCL-90-R IS, F(l, 86) = 13.60, p < .001;CGI, F(l, 86) = 9.55, p < .01; and SAS-SR, F(l, 86) = 14.34, p < .001. There were no significant differences between the two groups with respect to the third and fourth parameters, the responses during the speech and the social skills tasks.

Social Phobia Subtypes and the Presence ofAvoidant Personality Disorder This analysis was conducted in order to make comparisons between the specific social phobics and the generalized social phobics without APD (Question 2) and between the generalized social phobics without APD and generalized social phobics with APD (Question 3). General distress. The omnibus MANOVA across the three groups was significant, F(8,162) = 2.76, Wilks's X = 0.7747, p < .01. When analyzed with univariate tests, the groups differed with respect to all four indexes: HRSA, F(2,84) = 3.60, p < .05; SCL-90-R GSI, F(2, 84) = 5.02, p < .01; SCL-90-R Anxiety, F(2, 84) = 3.36, p < .05, and BDI, F(2, 84) = 10.78, p < .0001. Post hoc tests conducted with the Tukey honestly significant difference procedure revealed that the specific subtype had a lower level of distress than the generalized subtype without APD on both the SCL-90-R GSI and the BDI. The generalized subtype without APD differed from those with APD on only one variable, the BDI, on which the latter group was more

depressed. The specific subtype had lower distress than the APD group on all four indexes. Means for the three groups on the variables related to general distress are presented in Table 4. Social anxiety and social functioning. The MANOVA was significant, F(8, 164) = 5.87, Wilks's X = 0.6043, p < .0001. Univariate tests revealed that there were differences between the groups on all four variables: SPAI, F(2, 85) = 15.72, p < .0001; SCL-90-R IS, F(2,85) = 10.54, p < .0001; CGI, F(2,85) = 6.05, p