Journal of Psychiatric Research

Transcription

1 Journal of Psychiatric Research 45 (2011) 111e120 Contents lists available at ScienceDirect Journal of Psychiatric Research journal homepage: Social fear and social phobia types among community youth: Differential clinical features and vulnerability factors Susanne Knappe a, *, Katja Beesdo-Baum a, Lydia Fehm b, Murray B. Stein c, Roselind Lieb d,e, Hans-Ulrich Wittchen a,e a Institute of Clinical Psychology and Psychotherapy, Technische Universitaet Dresden, Dresden, Germany b Department of Psychology, Humboldt University Berlin, Germany c Department of Psychiatry, University of California San Diego, USA d Institute of Psychology, Epidemiology and Health Psychology, University of Basel, Switzerland e Max Planck Institute of Psychiatry Munich, Germany article info abstract Article history: Received 5 November 2009 Received in revised form 30 April 2010 Accepted 5 May 2010 Keywords: Social phobia Subtypes Community Risk factor DSM-V Objective: To compare different social fears and social phobia subtypes with regard to clinical (age of onset, avoidance, impairment, comorbidities) and vulnerability factors (behavioural inhibition (BI), parental psychopathology and parental rearing) among community youth. Methods: Fears of 6 social situations and Social Phobia (SP), along with their clinical features, were assessed using the Munich-Composite International Diagnostic Interview (DIA-X/M-CIDI) in a population-based sample of N ¼ e24 year olds that were followed up for 10 years. BI and parental rearing were assessed using self-report questionnaires. Parental psychopathology was assessed directly in parents via DIA-X/M-CIDI, supplemented by offsprings family history reports. Results: In the total sample, 20.0%, 11.6%, 11.7% reported fear of 1, 2, 3 or more social situations, respectively; rates were 24.2%, 18.7%, and 57.1% in SP-cases (6.6% of the total sample). Exploring the factorial structure indicated rather unidimensionality of social fears than mutual distinction of social fears by interaction vs. performance situations. Except for fear of taking tests and public speaking, social fears rarely occurred in isolation. Social fears of both interaction and performance situations were associated with severe avoidance (vs. fear of either situation; Odds Ratios, OR ¼ 1.5, 95%CI: 1.1e1.9) and impairment (OR ¼ 3.6, 95%CI: 2.6e4.9), and more comorbid anxiety and depressive disorders (OR range 3.2e5.8, p >.001). Fear of interaction situations was associated with higher BI (vs. performance-related fears, OR range 1.2e2.1, p <.05). Associations with parental psychopathology and unfavourable parental rearing were less consistent, albeit stronger for fear of interaction situations (vs. performance-related fears). Interactions with time indicated an earlier onset of SP for higher BI, but not for parental psychopathology or unfavourable parental rearing. Conclusions: Interaction-related social fears differ in their clinical and vulnerability factors from performance-related social fears. The current DSM-IV specifier of generalized SP may fall short of adequately denoting these differences. Fear of taking tests appears to be conceptually and, possibly, etiologically distinct from other social fears, and may be better placed in another category (e.g., as a type of specific phobia). Ó 2010 Elsevier Ltd. All rights reserved. 1. Objectives With lifetime prevalences up to 13% (Beesdo et al., 2007; Kessler et al., 1994), social phobia (SP) is one of the most common mental disorders. SP can encompass a heterogenous spectrum of social fears, which may affect one (isolated) or multiple social situations including * Corresponding author. Tel.: þ ; fax: þ address: knappe@psychologie.tu-dresden.de (S. Knappe). fears of performance situations (e.g., speaking in front of others, taking tests or exams, or writing) as well as fears of interaction situations (e.g., going to a party or fear of talking to others) (Heimberg et al., 1993; Piqueras et al., 2008; Stein and Deutsch, 2003; Stemberger et al., 1995; Turner et al., 1992). In light of the upcoming revisions of our classificatory systems, it is timely to examine qualitative and/or quantitative differences between different social fears, or among the often suggested dichotomization of social fears according to interaction vs. performance situations, and to consider alternative specifiers for SP subtypes (Bögels et al., 2010) /$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi: /j.jpsychires

2 112 S. Knappe et al. / Journal of Psychiatric Research 45 (2011) 111e120 Large-scale epidemiologic studies, using standardized and concise assessment strategies report on the frequency of social fears and SP, their clinical features and associations with SP-related risk factors (e.g., Acarturk et al., 2008; DeWit et al., 2005; Stein et al., 2000; Vriends et al., 2007; Wells et al., 1994). In addition, factor analytic (Cox et al., 2003; Perugi et al., 2001; Safren et al., 1999; Vriends et al., 2007), latent-class (Kessler et al., 1998; Ruscio et al., 2008), and cluster analyses (Eng et al., 2000; Furmark et al., 2000; Stein and Deutsch, 2003) were conducted to explore the structure of social fears in population-based samples. Findings yielded homogeneous, albeit diverging classes of social fears based on temperamental or personality factors (Hofmann et al., 2004), descriptive characteristics (e.g., Furmark et al., 2000) or number of social fears (Ruscio et al., 2008; Vriends et al., 2007). Accordingly, individuals with isolated fear of performance situations may be qualitatively distinct from those with only interactional fears and from those with fear of interaction plus performance situations. For example, fear of public speaking was found to be the most common social fear (Faravelli et al., 2000; Kessler et al., 1994; Magee et al., 1996; Stein et al., 1994). In its isolated form, fear of speaking was noted to be less persistent, less impairing, and less comorbid than in combination with other social fears (Kessler et al., 1998). Among SP-cases, those with isolated fear of performance situations reported less distress, less social anxiety, and higher ratings on social functioning than SP-cases with interaction fears (Turner et al., 1992). Analyses were however unable to dissect whether fear of performance vs. interaction situations load on separate dimensions, and whether fear of performance situations represents a more heterogenous group than fear of interaction situations (Perugi et al., 2001; Ruscio et al., 2008; Safren et al., 1999). Likewise, similarities and differences between these types of fears and, indeed, their validity as categories (e.g., interaction vs. performance-related social fears) have been rarely examined. Heterogeneity among social fears may be driven by differences in their clinical features and developmental roots (Blöte et al., 2009; Lieb et al., 2000b; Mannuzza et al., 1995; Stein et al., 1998). Associations with common vulnerability factors such as behavioural inhibition (BI; Beesdo et al., 2010; Biederman et al., 1993; Hirshfeld et al., 1992; Prior et al., 2000; Rosenbaum et al., 1991; Schwartz et al., 1999), parental psychopathology (e.g., Knappe et al., 2009b,c) and unfavourable parental rearing (e.g., Elizabeth et al., 2006; Knappe et al., 2009c; Rapee and Spence, 2004) have been widely examined for SP, but rarely for different social fears or SP subtypes. BI refers to a temperament-based disposition to show restraint and withdrawal in unfamiliar settings or novel situations (Kagan et al., 1990). The two dimensions of BI, i.e., social/school and fear/illness (Reznick et al., 1992; Van Ameringen et al., 1998) may differ between fears within the SP spectrum, with the social fear component of BI to be more consistently linked to fear of interactional situations than fairly isolated fear of public speaking or other performance situations. SP strongly aggregates in families, and some authors suggest that BI serves as a genetic liability to SP and other anxiety disorders (Biederman et al., 1990; Rosenbaum et al., 1991). Similarly, parental psychopathology (parental SP, other parental anxiety, depressive, alcohol use disorders) is associated with an increased risk for social fears and SP in offspring (e.g., Knappe et al., 2009b,c; Wittchen et al., 2000). The genetic heritability of SP is, however, estimated to be moderate (Kendler et al., 1999), and environmental factors and geneeenvironment interactions are likely to contribute to SP (Elizabeth et al., 2006). For example, unfavourable parental rearing behaviour (e.g., parental overprotection, rejection, lack of emotional warmth, disturbed family functioning) is linked to onset and persistence of SP-conditions, independent of and in combination with parental psychopathology (Knappe et al., 2009a,b). Family environment may promote offspring SP by parental modelling of anxious or avoidance behaviour (DeRosney et al., 2006; Gerull and Rapee, 2002; Gruener et al., 1999; Muris et al., 1996; Murray et al., 2007) or attitudes and actions (Bögels et al., 2003; Whaley et al., 1999) by which parents actively manage their offsprings behaviour (Chorpita et al., 1996; Murray et al., 2008; Rapee, 1997). Such developmental pathways may particularly contribute to fear of interaction situations. Finally, with regard to clinical features such as age of onset patterns for specific social fears on the one hand, and their relationships with putative SP-related risk factors on the other hand, their interactive effects may further explain homogeneities and heterogeneities among the SP spectrum. First onset of SP is usually located in late childhood and early adolescence; onset of new cases after 20 years of age is rare (Beesdo et al., 2009, 2007; Wittchen and Fehm, 2003). There is little evidence that individuals with an early onset of SP (e.g., during childhood) differ from those with a later onset of the disorder (e.g., during adolescence or adulthood) in terms of clinical features (Davidson et al., 1993; de Menezes et al., 2005). Vulnerability factors though may promote an earlier manifestation of social fears. That is, social fears may occur earlier in the presence of vulnerability factors (higher levels of BI, parental psychopathology, unfavourable parental rearing), perhaps resulting in higher persistence of SP-symptoms and co-occurrence/incidence of comorbid disorders. In light of the ongoing revision process of future classificatory systems, we want to explore commonalities among social fears of interaction and performance situations, and to contrast differences between fears of interaction vs. performance situations. Using data from a representative community sample of adolescents and young adults, we examine 1) clinical features (age of onset, avoidance behaviour, impairment, comorbid disorders) and 2) vulnerability factors (BI, parental psychopathology, unfavourable parental rearing) as validators for social fears, and criteria that may discriminate between fears of interaction vs. performance situations. 2. Materials and methods 2.1. Sampling and data collection Data were collected as part of the prospective longitudinal Early Developmental Stages of Psychopathology (EDSP) e Study. Detailed descriptions of the EDSP design and field procedures are reported elsewhere (Lieb et al., 2000a; Wittchen et al., 1998b,c). Briefly, the study consists of a baseline survey conducted in 1995 (T0) with N ¼ 3021 individuals (response rate 70.9%) of a younger (N ¼ 1395, aged 14e17 years at baseline) and an older study cohort (N ¼ 1626, 18e24 years at baseline). The first follow-up (T1) (mean interval 1.64 years, SD ¼ 0.19) was conducted only for the younger cohort (N ¼ 1228; response rate: 88.0%), whereas the second (T2) (N ¼ 2548, mean interval after T0 3.5 years, SD ¼ 0.3, response rate: 84.3%) and third follow-up (T3) (N ¼ 2210, mean interval after T0 8.4 years, SD ¼ 0.7; response rate: 73.2%) was conducted for both cohorts. There was no selective drop out (attrition) from baseline (T0: N ¼ 3021) to 10-year follow-up (T3: N ¼ 2210) for SP (OR ¼ 1.1, 95%CI: 0.8e1.6). All participants (in cases of aged 18 or younger their parents) provided written informed consent. The EDSP project and its family genetic supplement have been approved by the Ethics Committee of the Medical Faculty of the Technische Universitaet Dresden (No: EK-13811) Diagnostic assessment DIA-X/DSM-IV diagnoses, along with information on age of onset, avoidance behaviour and degree of impairment were directly

3 S. Knappe et al. / Journal of Psychiatric Research 45 (2011) 111e assessed at each assessment wave using the computer-assisted version of the standardized Munich-Composite International Diagnostic Interview (DIA-X/M-CIDI) (Wittchen and Pfister, 1997). Diagnostic information was aggregated across all available assessment waves for cumulative lifetime status. Reliability and validity is moderate to good for all the disorders covered by the DIA-X/M- CIDI. Kappa for diagnostic testeretest reliability was 0.75 for SP stem items, 0.72 for DSM-IV SP, and 0.81 for any anxiety disorder (Wittchen et al., 1998a). Testeretest reliability for the SP module is acceptable (kappa ¼ 0.57) (Lachner et al., 1998; Reed et al., 1998). In contrast to an earlier study (Beesdo et al., 2007), but in line with more recent contributions (Knappe et al., 2009a,b Beesdo et al., 2010) criterion E was required when respondents were 18 years or older, because estimation of impairment due to SAD may be of limited reliability in respondents of younger age (Wittchen et al., 1999). Positive affirmation of the DIA-X/M-CIDI stem question Have you ever had an unusually strong fear or avoidance of doing things in front of others or of being the centre of attention? For example (have you ever had an unusually strong fear of). indicated the presence of social fears. To increase validity (Wittchen et al., 2001), a list of 6 social fear situations (fear of eating or drinking, writing, going to a party or meeting, taking tests or exams, speaking in front of others, and talking to others) and a list of social fear cognitions (something embarrassing or shameful could happen, being regarded as dumb or weak, being regarded as crazy, to experience an anxiety (panic) attack, to be confused, to be ashamed, to throw up, to lose control over intestinal organs, or to turn red whenever one was in a social situation, thought about it, or was about to enter such a situation) were used. From T1 on, other social fears (e.g. using public restrooms, working while being observed) were additionally assessed. When only one out of the situations was reported, we refer to it as an isolated social fear. Age of onset information on social fears was aggregated from available assessment waves by using the minimum age of onset reported. Consistent with prior reports (Beesdo et al., 2007), age of onset for social fears and SP was based on the minimum age of onset reported by the respondent at any of the assessment waves. This definition agreed well (95%) with other age of onset aggregation methods (e.g., using the first or mean of reported ages-of-onset), i.e. the absolute value of differences were at most 1 year. Degree of avoidance refers to the frequency at which social situations were avoided because of social anxiety (never, rarely, frequently, or always; maximum score across all completed waves). Degree of impairment indicates how much anxiety of social situations or avoidance interfered with life and daily activities (not at all, a little, much, or very much; maximum score across all completed waves). The maximum report of impairment and avoidance was used for analyses indicating the most severe expression for the participant throughout the study. Ratings of avoidance and impairment were aggregated into no/low and moderate/severe. Behavioural Inhibition (BI) was assessed at baseline using the Retrospective Self-Report of Inhibition (RSRI, Reznick et al., 1992). Extensive research demonstrated moderate stability and predictive validity of BI (i.e., moderate correlation of r ¼ 0.56 between the RSRI and the Adult Self-Report of Inhibition) (Reznick et al., 1992; Rohrbacher et al., 2008). Internal consistency and validity for the German version of the RSRI including its two dimensional subscores on social/school fear and fear/illness was found to be good (Rohrbacher et al., 2008). Parental psychopathology (lifetime diagnoses in either mother or father: SP; any other anxiety disorder including specific phobia, generalized anxiety disorder, panic disorder, agoraphobia; any depressive disorder including major depressive disorder or dysthymia, any alcohol use disorder including alcohol abuse or dependence) was derived by aggregation of diagnostic information from direct DIA-X/M-CIDI interviews in parents (at T1, T3; N ¼ 1152 mothers, N ¼ 211 fathers, overall: N ¼ 1189) and indirect family history information using the respondents as informants (at T0, T2, T3; overall: N ¼ 3021). Indirect parental diagnoses were derived from family history data collected with offspring as informants using the M-CIDI family history module using a modified version of the Family History Research Diagnostic Criteria (Merikangas et al., 1998). At baseline, offspring were asked M-CIDI-questions to assess the key symptoms of parental DSM-IV disorders and whether their parent sought professional help because of his or her respective symptoms. At T2 and T3, an extended version of the family history module was used, containing fully structured sections covering M-CIDI/DSM-IV criteria. A priority hierarchy to aggregate information on parental psychopathology was derived after examination of agreement patterns between family history reports and available parent interview data (Beesdo et al., 2010). That is, parents direct information was considered most reliable and was used when available. When direct information was not available, T3 family history information was used, followed by T2 and T0, respectively, because greater agreement was found for respondents at higher ages (T3: 21e34 years, T2: 17e28 years, T0: 14e17 years). Perceived parental rearing with regard to parental rejection, emotional warmth, and overprotection was assessed using the German Version of the Questionnaire of Recalled Parental Rearing Behaviour administered to offspring (FEE, Schumacher et al., 1999) at T1 (N ¼ 1100/1395) Statistical analyses Results (%, means, standard deviations, coefficients) are weighted by age, gender, and geographic location at baseline to match the distribution of the original sampling frame (Lieb et al., 2000a), frequencies (Ns) are reported unweighted. The Stata Software package 11.0 (StataCorp, 2009) was used to compute robust variances, confidence intervals, and p-values (by applying the HubereWhite sandwich matrix) which is required when analyses are based on weighted data (Royall, 1986). To explore the structure of social fears, we used exploratory (principal) factor analyses (EFA) based on tetrachorical correlations to account for binary items and rotated the factors with orthogonal oblimin rotation. Logistic regressions provided odds ratios (ORs) for associations between clinical (age of onset, moderate/severe vs. no/low avoidance behaviour, moderate/severe vs. no/low impairment, comorbidities) and vulnerability factors (BI, presence vs. absence of parental psychopathology, parental rearing) with respondent s social fears and SP. Notably no adjustment for multiple testing was applied, because the individual tests were related to individual hypotheses and adjustment would treat them as reflecting a global hypothesisdwhich is questionable in substantive terms (Savitz and Olshan, 1995). Scores on the BI and FEE scales were standardized such that the OR reveals the factors by which the likelihood for the outcome increases for each increase of 1 standard deviation (SD) in the independent variable. Cox regressions were applied to assess overall differences in the risk of developing social fears of interaction and performance situations among respondents with DSM-IV SP, respectively over time by age of onset of social fears. We allowed for different curves according to age and gender ( stratified Cox regression, Therneau and Grambsch, 2000). To assess whether respondents with the risk factor covariate (e.g., higher levels of BI, presence of parental psychopathology, higher levels of unfavourable parental rearing styles) were at increased risk for earlier onset of social fears, the proportional hazard assumption was tested using Schoenfeld

4 114 S. Knappe et al. / Journal of Psychiatric Research 45 (2011) 111e120 residuals (Therneau et al., 2000). When the assumption was violated, the interaction term covariate age in years was added to the model in order to improve the model fit and to assess how strongly the hazard ratios (HR) depended on time. Here, the modelbased time-dependent hazard ratio equals HR (main effect of covariate) HR (interaction effect of covariate) age in years. Again the proportional hazard assumption was tested using Schoenfeld residuals. Interaction with time is interpreted as follows: (a) HR (main effect) >1 and HR (interaction effect) <1: Social fear occurs earlier in respondents with an unfavourable risk factor (e.g., higher levels of BI, presence of parental psychopathology, dysfunctional parental rearing) under the condition that SP occurs. (b) HR (main effect) <1 and HR (interaction effect) <1: Social fears occur later in respondents with an unfavourable risk factor under the condition that SP occurs. (c) HR (main effect) <1 and HR (interaction effect) >1: Social fears occur earlier in respondents with an unfavourable risk factor under the condition that SP occurs. (d) HR (main effect) >1 and HR (interaction effect) >1: Social fears occur later in respondents with an unfavourable risk factor under the condition that SP occurs. All analyses are based on 3021 respondents, using the LOCF method, except for associations on parental rearing using data from 1100 respondents of the younger study cohort (14e17 years at baseline). 3. Results 3.1. Factor structure of social fears Exploratory (principal) factor analyses (EFA) indicated that the 6 social situations loaded onto a single latent factor in the total sample (N ¼ 3201), and similarly in the SP subsample (N ¼ 209/ 3021, 6.6% weighted), where fear of taking tests or exams was negatively intercorrelated with other social fears. Logistic regressions to examine the associations between feared social situations and catastrophic anxiety cognitions yielded strong positive associations for the majority of social fears (table available upon request), except for fear of taking tests, which was negatively associated with something embarrassing or shameful could happen, to be ashamed, and to turn red (OR range 0.63e0.76, p <.05). In the following analyses, we examine clinical and vulnerability factors for different types of social fears. Given the current discussion on (content-related) subtypes for SP and following procedures of Cox et al. (2003) and Stein and Deutsch (2003), we also explored associations for fear of interaction (fear of going to a party or meeting, talking to others) and performance situations (fear of eating or drinking, writing, taking tests, public speaking), although this differentiation is not clearly indicated in our data based on EFA and by associations with anxiety cognitions Descriptives and clinical features of social fears in the total and SP subsample Frequencies of social fears in the total and SP subsample are presented in Table 1. Almost half of the total sample (N ¼ 1359/ 3021, 43.3%) had at least one social fear with 20.0% (N ¼ 642/3021) reporting one feared social situation, and 11.6% (N ¼ 347/3021) reporting two, and 11.7% (N ¼ 370/3021) three or more feared social situations. Fear of taking tests or exams, speaking in front of others, and fear of talking with others were the most prevalent social fears in both males and females (Range 9.8e32.3%). Among respondents with DSM-IV SP (N ¼ 209/3021, 6.6%), rates for all types of social fears were higher than in the total sample, ranging from 15.1% to 75.2%. Overall, rates for isolated forms of social fears were lower than 5% both in the total and the SP subsample. One notable exception is that isolated fear of taking tests was present in 11.2% of the total sample and in 14.3% of respondents with SP. Fig. 1A shows the cumulative age of onset distributions for isolated social fears up to age 34 in the total sample. Onset was located primarily in childhood for isolated fear of eating (mean 11.4 years, 5.7) and isolated fear of writing (mean 11.8 years, 6.9) and in adolescence for isolated fear of talking to others (mean 13.8 years, 5.6), isolated fear of speaking in front of others (mean 14.2 years, 4.5) and isolated fear of taking tests (mean 14.7 years, 4.5). Fig. 1B shows the age of onset patterns separately for fear of interaction and performance situations and DSM-IV SP up to age 34. Mean age of onset for fear of interaction situations with or without contemporaneously emerging performance-related social fears was 11.6 years (4.8) and 11.4 years (4.7), respectively. Mean age of onset for isolated fear of performance situations was 12.9 years (4.7). Age of onset for DSM-IV SP was 13.1 years (5.4). Among those with at least one social fear (N ¼ 1359/3201), all social situations were associated with significant avoidance (moderate/severe vs. no/low: OR range 5.3e18.3, p <.05). Regarding isolated social fears, isolated fear of taking tests (OR ¼ 6.7, 95%CI: 5.1e8.6, p <.001) and isolated fear of speaking in front of others (OR ¼ 3.6, 95%CI: 2.6e5.0, p <.001), isolated fear of interaction situations (OR ¼ 1.8, 95%CI: 1.1e3.0, p ¼.030) and their co-occurrence with fear of performance situations (OR ¼ 1.5, 95%CI: 1.1e1.9, p ¼.009) were associated with significant avoidance. Similarly, all social fear situations (except fear of speaking in front of others) were associated with impairment (moderate/severe vs. no/low: OR range 1.7e3.5, p <.05), with strongest associations for fear of both interaction and performance situations. Again, no associations with impairment emerged for most isolated social fears. One notable exception was that respondents with isolated fear of speaking in front of others more frequently reported no/low levels of impairment (OR ¼ 0.1, 95%CI: 0.1e0.3, p <.001) than respondents with fear of other social situations. Regarding the presence of comorbid lifetime anxiety (i.e., other than SP), depressive or substance use disorders, strong positive associations were found for all social fear situations (all p-values <0.01). Considering isolated social fears, the only significant finding was that isolated fear of going to a party was associated with a higher risk of anxiety disorders (OR ¼ 2.9, 95%CI: 1.3e6.7, p ¼.008, vs. other social fears). Isolated fear of any interaction situation was associated with other anxiety (OR ¼ 4.2, 95%CI: 2.4e7.3, p <.001) and depressive disorders (OR ¼ 1.8, 95%CI: 1.5e2.3, p <.001), while isolated fear of any performance situation was associated with all comorbid disorders considered (OR range 1.7e2.0, p <.05). The highest odds for comorbid disorders was found in individuals with fear of both interaction and performance situations (OR range 3.2e5.8, p <.001, vs. fear of either interaction or performance situations) Vulnerability factors for social fears in the SP subsample Behavioural Inhibition Among SP-cases, social fears of all situations except for fear of taking tests were associated with higher levels of BI (upper part of Table 2). Interestingly, isolated fear of taking tests (vs. all other social fears) was associated with lower levels of the total BI scale and the BI subscale social/school fear, whereas isolated fear of interaction situations was associated with higher levels of BI (vs. isolated fear of performance situations). The co-occurrence of social fear in both interaction and performance situations was related to higher levels of BI (vs. fear of either interaction or performance situations) (lower part of Table 2).

5 S. Knappe et al. / Journal of Psychiatric Research 45 (2011) 111e Table 1 Frequency of social fears in adolescents and young adults (cumulated across T0eT3, locf). In the total sample Among those with DSM-IV social phobia Total N ¼ 3021 Males N ¼ 1533 Females N ¼ 1488 Total N ¼ 209 Males N ¼ 70 Females N ¼ 139 N %w N %w N %w N %w N %w N %w Number of social fears (cumulated T0eT3) c e e e Type of feared social situation Eating or drinking Writing Taking tests or exams Speaking in front of others Going to party or meeting Talking with others Fears in other social situations Isolated eating or drinking Isolated writing isolated test/exam Isolated speaking in front of others Isolated going to party or meeting Isolated talking with others Isolated performance fear a,* Isolated interaction fear b,* Combined performance and interaction* Any fear c Any performance fear a Any interaction fear b N, unweighted number; %w, weighted percentages (Note: percentages do not add up to 100% due to rounding). a Includes fear of eating/drinking, writing, tests/exam, speaking in front of others. b Includes fear of going to a party or meeting, talking with others. c Includes fear of eating or drinking, writing, going to a party or meeting, tests, speaking in front of others, talking with others, and fears in other social situations. * 17 cases in the total sample (1 case in the SP subsample) reported isolated 'other social fears', and where therefore not allocated to either the performance or interaction social fear subtype. Among respondents with SP, survival analyses tested whether those with higher levels of BI had an earlier age of onset of SP, respectively the interaction and performance social fear type, than respondents with lower levels of BI (Table available upon request). Compared to SP-cases with lower BI, SP-cases with higher levels on BI scales (total, social/school, fear/illness) had an overall increased risk to develop SP (HR range 7.8e46.6, p <.05), fear of performance situations (HR range 2.9e7.8, p <.05) and fear of interaction situations (HR range 2.7e8.0, p <.05). An interaction with time was found, indicating that respondents with higher levels on the BI fear/illness scale have an earlier onset of performance-related social fears than respondents with lower levels on the BI fear/illness scale (HR main effect ¼ 3.1, 95%CI: 1.5e6.5, p ¼.003, HR interaction effect ¼ 0.9, 95%CI: 0.9e1.0, p ¼.048); no such interaction was found for the BI social scale. Adjusting analyses on performance situations for fear of interaction situations and vice versa did not change the findings Parental psychopathology Among SP-cases, different social fears did not differ with regard to parental psychopathology, with two notable exceptions: First, SP-cases reporting fear of going to a party or meeting had higher rates of parents with SP (69.2% vs. 42.5%; OR ¼ 3.6, 95%CI: 1.3e9.6, p ¼.011) and parents with other anxiety disorders (54.1% vs. 37.6%; OR ¼ 2.1, 95%CI: 1.1e3.9, p ¼.023) than SP-cases without these social fears. Second, SP-cases with fear of talking to others had higher rates of parents with alcohol use disorders than SP-cases without this fear (69.4% vs. 46.7%; OR ¼ 2.8, 95%CI: 1.2e6.5, p ¼.017). SP-cases with fear of any interaction situation had higher rates of parental SP (86.1% vs. 62.3%; OR ¼ 4.4, 95%CI: 1.1e17.7, p ¼.039) and alcohol use disorders (78.8% vs. 61.0%; OR ¼ 2.5, 95%CI: 1.1e5.9, p ¼.036) than SP-cases without such fears. Having any performance-related social fear (vs. no performance fears) was positively associated with other parental anxiety disorders (99.5% vs. 93.5%, OR ¼ 12.9, 95%CI: 1.6e109.2, p ¼.018). Regarding isolated social fear types, SP-cases with isolated fear of any performance situation had substantially lower rates of parents with SP (11.4% vs. 37.8%; OR ¼ 0.2, 95%CI: 0.04e0.9, p ¼.034) and parents with alcohol use disorders (38.7% vs. 21.2%, OR ¼ 0.4, 95%CI: 0.2e0.9, p ¼.039) than SP-cases without isolated fear of any performance situation. Albeit SP-cases reporting fear of both interaction and performance situations had higher rates of parental disorders (range 68.4e88.3%) than SP-cases with fear of either interaction or performance situations (range 55.0e58.3%), these differences however were not statistical significant. Again, survival analyses tested whether SP-cases with parental psychopathology had an earlier onset of SP than their counterparts without parental psychopathology (Table available upon request).

6 116 S. Knappe et al. / Journal of Psychiatric Research 45 (2011) 111e120 A distribution onset Cumulative age of B distribution onset Cumulative age of Patterns of age of onset of isolated social fears in the total sample isolated fear of eating isolated fear of going to a party Age in years isolated fear of speaking in front of others isolated fear of writing isolated fear of taking tests isolated fear of talking to others Patterns of age of onset for interaction and performance fears in the total sample, and for DSM-IV social phobia isolated interaction fears Age in years combined interaction and performance fears isolated performance fears DSM-IV Social Phobia Fig. 1. (A) Patterns of age of onset for isolated social fears in the total sample (N ¼ 3021). (B) Patterns of age of onset for interaction and performance fears and for DSM-IV social phobia in the total sample (N ¼ 3021). All parental disorders considered increased the risk for offspring SP (HR range 1.8e2.3, p <.05) but without any interaction with time. Separate analyses on fear of performance and interaction situations, respectively, did not reveal significant findings Parental rearing Parental rearing styles were not associated with different types of social situations among SP-cases with few exceptions (Table 3). Fear of talking to others was associated with lower levels of recalled emotional warmth (vs. other social fears). Fear of any performance situation was related to higher levels of parental overprotection (vs. other social fears). Of note, fear of both interaction and performance situations (vs. fear of either interaction or performance situations) was not associated with higher levels of unfavourable parental rearing. Survival analyses tested whether SP-cases reporting unfavourable parental rearing (e.g., higher levels of rejection, overprotection, lower levels of emotional warmth) had an earlier onset of SP than SP-respondents reporting more favourable levels of parental rearing (Table available upon request). Overall, more negative parental rearing (high rejection, high overprotection, low emotional warmth) was associated with increased risk for offspring SP at a given year (HR range 0.9e1.5, p <.05), yet no interactions with time occurred. For fear of interaction or performance situations no associations were found, probably due to limited statistical power in this subsample analyses. 4. Discussion Examining clinical and vulnerability factors of 6 different social fears and SP in a community sample of adolescents and young adults up to age 34 yielded three major results: First, different social fears and types differed in terms of clinical features (comorbidities, degree of avoidance and impairment). Second, among SP-cases, interaction-related social fears were particularly associated with higher levels of BI. Associations with parental psychopathology and unfavourable parental rearing were less consistent, albeit stronger for fear of interaction situations (vs. performance-related fears). Third, interactions with time indicated an earlier onset of social fears in SP-cases with higher BI, but not for presence of parental psychopathology or unfavourable parental rearing. Table 2 Associations between Behavioural Inhibition and social fears among respondents with DSM-IV Social Phobia (N ¼ 209). c Type of feared social situation BI total BI social fear BI illness fear OR 95%CI p OR 95%CI p OR 95%CI p Eating or drinking vs. no eating or drinking < < Writing vs. no writing < Test vs. no test < Speaking vs. no speaking < Party vs. no party < < Talking vs. no talking < < Any performance fear a vs. no performance fear Any interaction fear b vs. no interaction fear < < Isolated test vs. other < Isolated speaking vs. other Isolated interaction b vs. isolated performance a Combined interaction b and performance a vs. interaction a or performance a < < Other: all other fears than the respective isolated fear. OR, from logistic regressions, controlled for age and gender; CI, Confidence Interval. p.05, bold prints indicate statistical significance. a Includes fear of eating/drinking, writing, tests/exam, speaking in front of others. b Includes fear of going to a party or meeting, talking with others. c N ¼ 209/3021 respondents with social phobia.

7 S. Knappe et al. / Journal of Psychiatric Research 45 (2011) 111e Table 3 Associations between parental rearing behaviour and social fears among respondents with DSM-IV Social Phobia (N ¼ 106) c. Type of feared social situation Parental rejection Parental overprotection Parental emotional warmth OR 95%CI p OR 95%CI p OR 95%CI p Eating or drinking vs. no eating or drinking Writing vs. no writing Party vs. no party Test vs. no test Speaking vs. no speaking Talking vs. no talking Any performance fear a vs. no performance fear Any interaction fear b vs. no interaction fear Isolated test vs. other Isolated speaking vs. other Isolated interaction b vs. isolated performance a Combined interaction b and performance a vs. interaction b or performance a Other: all other fears than the respective isolated fear. M, mean; SD, standard deviation; OR, odds ratio from logistic regression, controlled for age and gender; CI Confidence interval. p ¼.05, bold prints indicate statistical significance. a Includes fear of eating/drinking, writing, tests/exam, speaking in front of others. b Includes fear of going to a party or meeting, talking with others. c N ¼ 106/1100 respondents with Social Phobia, for whom also data on perceived parental rearing were available. Up to age 34 and cumulated across all assessments, a substantial proportion of respondents feared only one (20%) or two (11.6%) social situations; the number of SP-cases affected by 3 or more social fears was similarly high (11.7%). Findings demonstrate that many individuals in the general population report mild or moderate social fears without meeting criteria for a SP diagnosis. Concordant with reviews on social fears and SP subtypes (Bögels et al., 2010; Hofmann et al., 1995; Hook and Valentiner, 2002), our findings suggest that interaction-related social fears qualitatively differ from other (e.g., performance-related) social fears: Fears of interaction situations rarely occurred in isolation, whereas fears of performance situations (such as taking tests, speaking in front of others) emerged as independent from interaction-related social fears. Isolated performance-related fears were less impairing and less comorbid than isolated interaction-related fears. Based on these findings, a differentiation between interaction and performance fears appears indicated, namely that individuals with only performance-related social fears apparently differ from those with fear of interaction situations (with or without fear of performance situations) on a number of validating factors. However, in the vast majority of individuals, interaction- and performance-related social fears co-occur, challenging this conclusion. Compared to isolated fear of either interaction or performance situations, co-occurrence was associated with greater severity (avoidance, impairment), and higher rates of comorbid anxiety and depressive disorders. Thus, overall, findings do not suggest mutual subtyping of social fears by interaction vs. performance situation. In addition, the interaction and performance fear differentiation was not supported by our results of the exploratory factor analysis. Hence, a specifier in DSM-V indicating the presence of combined interaction and performance fears may be more useful. The current specifier of generalized SP may be problematic as it does only indicate the number of problematic situations, but not the type or content of social anxiety (although it can be inferred that the requirement that it involve most social situations would refer to a combination of performance and interactional situations). Fear of taking tests was the most common social fear in both the total sample and in the SP subsample, and even exceeds the proportion of individuals with fear of public speaking, particularly in its isolated form. This may be attributable to the young age of our sample because adolescents and young adults are often confronted with such situations during their academic activities. Previous research has focused on the position of fear of public speaking within the SP spectrum, given its differences from other social fears in terms lower persistence, impairment, comorbidity rates and physiological measures (e.g., Blöte et al., 2009; Kessler et al., 1998; Stein and Deutsch, 2003). It has therefore been tentatively suggested to categorize fear of public speaking as a specific phobia (Eng et al., 2000). Our findings suggest that it will also be important to carefully rethink the position of fear of taking tests as residing within the SP spectrum. With regard to the ongoing revision of future diagnostic systems, fear of taking tests may be considered for inclusion in the SP spectrum as a form of SP, for example by specifying test anxiety only. Although prior studies have found fear of negative evaluation to be a core feature of test anxiety (Beidel and Turner, 1988; McDonald, 2001), we found fear of taking tests unlike other social fears to be negatively associated with characteristic social-phobic cognitions such as that something embarrassing could happen or to be ashamed. In addition, isolated fear of taking tests was found to be not related to moderate/severe impairment and comorbid disorders (similar to fear of public speaking); yet it was associated with significant avoidance behaviour and (among SP-cases) with lower levels of BI (different than fear of public speaking). Hence (isolated) performance fears like public speaking and taking tests may be considered as two different entities. This conclusion is consistent with our findings on the factor structure of fears in the SP subsample. Similarly, Perugi et al. (2001) found that fear of taking tests did not load on any of the five social anxiety factors derived by principal component factor analysis of the Liebowitz Social Anxiety Scale, while fear of public speaking was allocated to the formal speaking anxiety -factor, along with fear of speaking up to a meeting, giving a report to a group, or acting, performing or giving a talk in front of an audience. Hence, fear of taking tests may also be considered as belonging better to the category of specific phobias. Bögels et al. (2010) sensibly point to the need for rigorous impairment criteria at a sufficiently high diagnostic threshold (e.g., avoidance of higher educational opportunities, failure on all types examinations despite adequate preparation) to preclude too many individuals with normal test anxiety from being diagnosed as having a disorder. In summary, our findings favor consideration of fear of taking tests as a syndrome distinct from social phobia. However, its relationship to other (anxiety) disorders, for example specific phobia, is yet to be determined.

8 118 S. Knappe et al. / Journal of Psychiatric Research 45 (2011) 111e120 Our study is not without limitations. Only one age of onset was reported relating to onset of any social fear, and age of onset information may therefore be subject to recall bias. Because many individuals reported fears in two or more social situations, we restricted comparative onset analyses to subjects with isolated fears. Since respondents were assessed up to 4 times while passing through the core high risk phase for SP onset (Beesdo et al., 2007; Knappe et al., 2009c), age of onset information likely reflects rather unbiased age of onset patterns of social fears and SP. Analyses were limited to the 6 social situations proposed in the CIDI respond list (plus a heterogenous group of other social fears). Another limitation is the restriction of analyses on parental rearing to the younger study cohort, which may cut statistical power. However, recall of rearing experiences referred to contemporary living situations of most respondents at home with their parents, likely attenuating recall bias at time of assessment. With these limitations in mind, our findings demonstrate that interaction-related social fears differ from performance-related social fears in their clinical and vulnerability factors. The current DSM-IV specifier of generalized SP may fall short of adequately denoting differences between social fears. Because most social phobics fear both interaction and performance situations, our study supports an alternative specifier for combined interaction/performance fears or alternatively performance fears alone for SP-cases with rather isolated performance-related social fears such as public speaking. Importantly, our findings also suggest that the categorization of test taking anxiety within social phobia should be strongly reconsidered, given the lack of association with typical social phobia cognitions, and the low concordance with other social fears. Role of funding source This work is part of the Early Developmental Stages of Psychopathology (EDSP) Study and is funded by the German Federal Ministry of Education and Research (BMBF) project no. 01EB9405/6, 01EB 9901/6, EB , 01EB0140, and 01EB0440. Part of the fieldwork and analyses were also additionally supported by grants of the Deutsche Forschungsgemeinschaft (DFG) LA1148/1-1, WI2246/1-1, WI 709/7-1, and WI 709/8-1. Contributions The manuscript presents original material and has not been considered for publication elsewhere. All of the authors have made substantial contributions to this work, and approved this manuscript for submission. Susanne Knappe prepared the manuscript and performed the statistical analyses. Katja Beesdo-Baum and Lydia Fehm helped to draft the manuscript. Murray B. Stein, Roselind Lieb and Hans-Ulrich Wittchen critically revised the manuscript for important intellectual content. Principal investigators of the EDSP-study are Drs Lieb and Wittchen, who take responsibility for the integrity of the study data. All authors and co-authors had full access to the data. Conflict of interest statement Susanne Knappe and Lydia Fehm declare to have no conflict of interest. Katja Beesdo-Baum receives or has in the past three years received speaking honoraria from Eli Lilly. Murray B. Stein receives or has in the past three years received research support from Eli Lilly and Company and GlaxoSmithKline, and was a consultant for AstraZeneca, Avera Pharmaceuticals, BrainCells Inc., Bristol-Myers Squibb, Eli Lilly and Company, EPI-Q Inc., Forest Laboratories, Hoffmann-La Roche Pharmaceuticals, Integral Health Decisions Inc., Jazz Pharmaceuticals, Johnson & Johnson, Mindsite, Sanofi-Aventis, Transcept Pharmaceuticals Inc., and Virtual Reality Medical Center. Roselind Lieb receives or has in the past three years received speaking honoraria from Wyeth. Hans-Ulrich Wittchen receives or has in the past three years received research support from Eli Lilly and Company, Novartis, Pfizer, Schering-Plough. He is currently or in the past three years has been a consult for Eli Lilly, GlaxoSmithKline Pharmaceuticals, Hoffmann-La Roche Pharmaceuticals, Novartis, Pfizer; Wyeth. He receives or has in the past three years received speaking honoraria from Novartis, Schering-Plough, Pfizer; Wyeth. Acknowledgement Principal investigators are Dr. Hans-Ulrich Wittchen and Dr. Roselind Lieb. Core staff members of the EDSP group are: Dr. Kirsten von Sydow, Dr. Gabriele Lachner, Dr. Axel Perkonigg, Dr. Peter Schuster, Dr. Michael Höfler, Dipl.-Psych. Holger Sonntag, Dr. Tanja Brückl, Dipl.-Psych. Elzbieta Garczynski, Dr. Barbara Isensee, Dipl.-Psych. Agnes Nocon, Dr. Chris Nelson, Dipl.-Inf. Hildegard Pfister, Dr. Victoria Reed, Dipl.-Soz. Barbara Spiegel, Dr. Andrea Schreier, Dr. Ursula Wunderlich, Dr. Petra Zimmermann, Dr. Katja Beesdo-Baum, Dr. Antje Bittner, Dr. Silke Behrendt and Dipl.-Psych. Susanne Knappe. Scientific advisors are Dr. Jules Angst (Zurich), Dr. Jürgen Margraf (Basel), Dr. Günther Esser (Potsdam), Dr. Kathleen Merikangas (NIMH, Bethesda), Dr. Ron Kessler (Harvard, Boston) and Dr. Jim van Os (Maastricht). Appendix. Supplementary data Supplementary data associated with this article can be found in online version at doi: /j.jpsychires References Acarturk C, De Graaf R, van Straten A, Ten Have M, Cuijpers P. Social phobia and number of social fears, and their association with comorbidity, health-related quality of life and help seeking. A population-based study. Social Psychiatry and Psychiatric Epidemiology 2008;43:273e9. Beesdo K, Bittner A, Pine DS, Stein MB, Höfler M, Lieb R, et al. Incidence of social anxiety disorder and the consistent risk for secondary depression in the first three decades of life. Archives of General Psychiatry 2007;64:903e12. Beesdo K, Knappe S, Pine DS. Anxiety and anxiety disorders in children and adolescents: developmental issues. Psychiatric Clinics of North America 2009;32:483e524. Beesdo K, Pine DS, Lieb R, Wittchen HU. Incidence and risk patterns of anxiety and depressive disorders and categorization of generalized anxiety disorder. Archives of General Psychiatry 2010;67:47e57. Beidel DC, Turner SM. Comorbidity of test anxiety and other anxiety disorders in children. Journal of Abnormal Child Psychology 1988;16:275e87. Biederman J, Rosenbaum JF, Balduc-Murphy EA, Faraone SV, Chaloff J, Hirshfeld DR, et al. A 3 year follow-up of children with and without behavioral inhibition. Journal of the American Academy of Child and Adolescent Psychiatry 1993;32:814e21. Biederman J, Rosenbaum JF, Hirshfeld DR, Faraone SV, Bolduc EA, Gersten M, et al. Psychiatric correlates of behavioral inhibition in young children of parents with and without psychiatric disorders. Archives of General Psychiatry 1990;47:21e6. Blöte AW, Kint MJW, Miers AC, Westenberg PM. The relation between public speaking anxiety and social anxiety: a review. Journal of Anxiety Disorders 2009;23:305e13. Bögels SM, Stein MB, Beidel DC, Clark LA, Pine DS, Voncken M. Social anxiety disorder: questions and answers for the DSM-V. Depression and Anxiety 2010;27:168e89. Bögels SM, van Dongen L, Muris PU. Family influences on dysfunctional thinking in anxious children. Infant and Child Development 2003;12:243e52. Chorpita B, Albano A, Barlow D. Child anxiety sensitivity index: considerations for children with anxiety disorders. Journal of Clinical Child Psychology 1996;25:77e82.