- Email: [email protected]
Relationships between social anxiety, depressive symptoms, and antisocial behaviors: Evidence from a prospective study of adolescent boys
Journal of Anxiety Disorders 23 (2009) 718–724
Contents lists available at ScienceDirect
Journal of Anxiety Disorders
Relationships between social anxiety, depressive symptoms, and antisocial behaviors: Evidence from a prospective study of adolescent boys Maria Tillfors a,*, Bassam El-Khouri b, Murray B. Stein c, Kari Trost a,d Department of Behavioral, Social and Legal Sciences, O¨rebro University, SE-701 82 O¨rebro, Sweden Department of Public Health, Karolinska Institute, Stockholm, Sweden c Department of Psychiatry, University of California San Diego, La Jolla, CA, USA d Department of Psychology, Stockholm University, Stockholm, Sweden a
b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 31 March 2008 Received in revised form 12 February 2009 Accepted 18 February 2009
Little is understood about generalized and non-generalized social anxiety disorder (SAD) and their associations with co-occurring internalizing and externalizing problems in adolescence. In the present study, we investigated adolescent boys with SAD symptoms and considered depressive symptoms as well as antisocial behaviors when looking for patterns during two developmental time periods; junior high and high school. Participants in the analyses were part of a longitudinal study. No patterns were found linking antisocial problems and non-generalized SAD in either junior high or high school. Furthermore, it was uncommon for youths in the non-generalized SAD subgroup to develop comorbidity over time. The generalized SAD subgroup of boys, however, was likely to develop comorbidity either with depressive symptoms only or with depressive symptoms and antisocial tendencies. Our findings suggest that developmental pathways for SAD subgroups may differ. ß 2009 Elsevier Ltd. All rights reserved.
Keywords: Social anxiety disorder Depressive symptoms Antisocial tendencies Adolescent boys Prospective study
Social anxiety disorder (SAD), also known as social phobia, is one of the most prevalent mental disorders. SAD is characterized by intense anxiety and avoidant behavior during performance and/or social interactional situations. Approximately 7–13% of individuals in westernized populations will suffer from social phobia during their lifetime and women are more likely to be affected (Furmark, 2002; Rapee & Spence, 2004). Although most research to date has focused on social phobia in adults, it is clear that SAD symptoms often begin in childhood or adolescence, typically around the age of 13 (Rapee & Spence, 2004). However, considering prevalence and seriousness of this disorder (Fehm, Pelissolo, Furmark, & Wittchen, 2005; Schneier, 2006; Stein & Stein, 2008) it is surprising how little is understood about its characteristics and course early on in its development. In terms of the general pathology of SAD, there are indications that SAD usually belong to one of two subgroups (i.e., generalized or non-generalized; APA, 2000). In turn, these subgroups are associated with different patterns of co-occurring internalizing problems like depressive symptoms (Chartier, Walker, & Stein, 2003; Wittchen, Stein, & Kessler, 1999), and externalizing problems like antisocial behaviors (Sareen, Stein, Cox, & Hassard, 2004). However, only a few studies have considered prospective information gathered during adolescence. In the present study, we will (a) identify patterns based on two subgroups within SAD with consideration to depressive
* Corresponding author. Tel.: +46 19 303959; fax: +46 19 303484. E-mail address: [email protected] (M. Tillfors). 0887-6185/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.janxdis.2009.02.011
symptoms and behavioral problems during two time points during adolescence, and (b) study the developmental pathway of those individuals with pronounced SAD symptoms. The generalized form of SAD is defined in the DSM system as a fear of ‘‘most social situations’’ (APA, 2000). The literature further refers to another SAD subgroup described varyingly as ‘‘discrete,’’ ‘‘circumscribed,’’ ‘‘limited,’’ ‘‘performance,’’ or ‘‘non-generalized,’’ although not explicitly defined in the DSM system (Hofmann, Heinrichs, & Moscowitch, 2004). This latter subgroup in most cases seems to include those whose social fears are constrained to one or more performance situations. In a recent cross-sectional study among youths, Marmorstein (2006) found that generalized SAD among females aged 15–17 was associated with major depression. Among males however, only the performance-focused SAD was found to be associated with dysthymia as well as conduct disorder, an instance of severe externalizing behavior. The differential patterning of symptoms associated within the two SAD subgroups suggest that the two dimensions of internalizing and externalizing problems should be simultaneously considered when evaluating these subgroups and in settings where both time and gender are taken into account. 1. Understanding the link between depressive symptoms and SAD subgroups Social avoidant behavior is a key aspect of SAD and also one of the known risk factors for depression. It is therefore not surprising
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
that prior research has found a link between early behavioral inhibition and later development of SAD and major depression (Hirshfeld-Becker et al., 2007; Neal, Edelmann, & Glachan, 2002). Inhibited children typically react with fearfulness, avoidant behavior, and reticence when confronted with unfamiliar or novel people and places (Kagan, Snidman, & Arcus, 1993). However, previous studies have only found a weak to moderate association between childhood behavioral inhibition and SAD. One reason may be that behavioral inhibition, like SAD, consists of different subgroups, where one subgroup of children reacts with social worry and another with high physiological reactivity. The subgroup typified by social worry may be characterized by stability over time as well as being associated with later development of generalized SAD. This and the finding that an association between social worry and adult depressive symptoms exists suggest that generalized SAD and depression share a common vulnerability factor (Rapee & Spence, 2004). Hence, this could in turn explain the high level of co-occurrence observed between generalized SAD, in comparison with non-generalized SAD, and depression in retrospective studies (Hofmann et al., 2004) and prospective ones (e.g., Beesdo et al., 2007; Wittchen et al., 1999). 2. Understanding the link between antisocial behaviors and SAD subgroups It has been suggested that high levels of social inhibition, characterizing generalized SAD, may function as a protective mechanism against antisocial behaviors. This could be congruent with the observation that adolescents with generalized SAD may be too socially inhibited to associate with delinquent peers. For instance, Kerr and colleagues (1997) report that behaviorally inhibited boys in late childhood, defined by those who were too shy to make friends, seem to be less delinquent than their counterparts with low levels of behavioral inhibition. Thus, although social inhibition may be a risk factor for developing internalizing problems, it may protect against developing externalizing problems. Repeatedly, however, studies have found an association of SAD with antisocial behaviors, as for example the earlier mentioned association between non-generalized SAD and conduct disorder observed among youths (Marmorstein, 2006). Sareen and colleagues (2004) found in two large cross-sectional community studies, this time involving adults, an association between generalized SAD and antisocial behavior. Hence, an association between the two SAD subgroups and externalizing problems seems to emerge but within different age groups. This could indicate that the SAD subgroups in relation to externalizing problems connect in different ways as a function of time, i.e., the pattern of cooccurrence changes during the transition to adulthood. For the last few decades, it has been noted that youths who are violent and persistent over the life course with their problem behaviors start early and tend to lack guilt over time (Moffitt, 1993; Moffitt & Caspi, 2001). Unlike these antisocial individuals, however, SAD individuals seem not to lack the ability to feel empathy or guilt. Because of this presumed retention of empathic ability, it is unlikely that most SAD individuals will display violent problem behaviors. In the present study, we looked specifically at boys with SAD symptoms and considered problem behaviors as well as depressive symptoms when looking for patterns during two developmental time periods in adolescence: junior high school and high school. We first studied possible patterns of SAD, depressive symptoms and antisocial behaviors separately at these two different time periods in order to understand possible cross-sectional configurations. Second, in order to understand
719
developmental pathways of SAD in terms of depressive symptoms and antisocial behaviors, we studied individual developmental pathways for boys with pronounced SAD symptoms. In accordance with prior research (Beesdo et al., 2007; Chartier et al., 2003; Wittchen et al., 1999) we expected more co-occurrence of depressive symptoms to be observed among generalized SAD adolescents and we expected this relationship to become stronger over time, in comparison to non-generalized SAD adolescents. Since previous research shows conflicting findings (Marmorstein, 2006; Sareen et al., 2004) when it comes to the association between SAD subgroups and antisocial behaviors, the a priori hypotheses for this relationship were less clear. However, we expected that none of the adolescents in either SAD subgroup would engage in violent antisocial behaviors. 3. Analytical approach Both variable and person-oriented approaches were used in the present study. The variable oriented approach is useful in analyzing and understanding what characteristics co-aggregate in a group of individuals (Block, 1971). For the purpose of the present study, we used the variable oriented approach to look at the characteristics that are most similar in order to find possible underlying SAD factors. The person-oriented approach is useful in finding configurations and patterns of longitudinal trajectories (Bergman, Magnusson, & El-Khouri, 2003; Bergman & Trost, 2006). We used the person-oriented approach to find and understand individual patterns and pathways of social anxiety symptoms and their relationship to depressive symptoms and antisocial behaviors. 4. Method Participants were part of a prospective longitudinal program of research. Approximately 3000 children and adolescents between the ages of 10 and 18 took part in a 6-year longitudinal survey in an average sized community in mid-Sweden. The present study focuses on boys at two age periods specifically during junior high school and high school. These age periods were chosen specifically since they represented two socially relevant developmental time points. Research on prevalence, etiology, and treatment of psychological disorders is primarily based on categorical diagnoses and restricted to arbitrary cut-offs based on extreme levels of symptoms (DSM-IV-TR; APA, 2000; Harvey, Watkins, Mansell, & Shafran, 2004) which has limited the comprehensive portrayal of disorders in the literature. In the present study, the dimensional perspective is implemented. By taking the dimensional view, no distinct difference is made between, for example, sub-threshold social anxiety and the diagnostic category SAD except in the frequency and/or severity of experienced symptoms (Kollman, Brown, Liverant, & Hofmann, 2006). As part of the larger survey, adolescents filled out a depression scale, a social anxiety scale, and questions about problem behaviors. Adolescents’ reports were used since it has been suggested that adolescents may provide more accurate information about themselves than teachers or parents (Steinberg, Lamborn, Darling, Mounts, & Dornbusch, 1994; Trost, Biesecker, Stattin, & Kerr, 2007). When adolescents are asked to report normbreaking behaviors like their own use of tobacco products, their reporting have been found to be highly reliable (Post et al., 2005). Furthermore, it has been reported that adolescents with certain psychological disorders may be able to give a unique and reliable description of their own health (Smith, Pelham, Gnagy, Molina, & Evans, 2000).
720
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
4.1. Symptoms of depression The Center for Epidemiologic Studies Depression Scale for Children (CES-DC; Fendrich, Weissman, & Warner, 1990; Olsson & vonKnorring, 1997) was used in the present study. It is a self-rating scale and contains of 20 items. In a validation study using a diagnostic interview as a reference the specificity for depression measured by CES-DC was about .70. Further, the CES-DC correlated highly with Beck’s Depression Inventory (BDI; r = .81), indicating adequate concurrent validity (Olsson & vonKnorring, 1997). Although the instrument is used as a screening for depressive symptoms, it has also been said to adequately assess clinical depression in youths (Roberts, Lewinsohn, & Seeley, 1991). In more detail, if the CES-DC is used with high cut off scores the scale has been found to be specific enough to detect clinical depression (Olsson & vonKnorring, 1997). The response scale is based on 1, not at all, to 4, often (Min/Max = 20/80; Cronbach’s alpha = .89). 4.2. Symptoms of SAD The social anxiety criteria were based on the first out of two parts of the Social Phobia Screening Questionnaire for Children (SPSQ-C; Green-Landell et al., 2009), a modified version of the Social Phobia Screening Questionnaire (SPSQ; Furmark et al., 1999). The first part consisted of eight items describing situations that tend to elicit social anxiety. The response scale was based on 1, no fear, to 3, severe fear. In a validation study using a diagnostic interview as a reference the sensitivity for SAD measured by SPSQC was .71 and the specificity was .86. The test–retest reliability over 3 weeks was .60 (Green-Landell et al., under review). In order to be able to study two different potential forms of SAD, a factor analysis was conducted. Based on a principal component analysis on social anxiety symptoms at grade 8 and 9 (junior high school), two components appeared. The Kaiser–Meyer–Oklin correlation of .72 exceeded the recommended value of .60, and the Bartlett’s test was significant (Pallant, 2005). As seen in Table 1, symptoms of SAD were found to load on two separate components. The variance explained for factor 1 was 28% and for factor 2 was 24%. Fears that are characteristic of an individual with nongeneralized SAD were represented by a composite score on the items ‘‘To speak in front of the class,’’ ‘‘To raise your hand during a lesson,’’ ‘‘To phone someone unfamiliar,’’ and ‘‘To initiate a conversation with someone unfamiliar’’ (Min/Max = 4/12; Cronbach’s alpha = .64). Fears that are more characteristic of an individual with generalized SAD were represented by a composite score on the items ‘‘To be together with others during the breaks,’’ ‘‘To go to a party,’’ ‘‘To be eating together with others during the lunch-break,’’ and ‘‘To look someone in the eyes during a conversation’’ (Min/Max = 4/12; Cronbach’s alpha = .66). 4.3. Antisocial behaviors Antisocial behaviors were used as a marker for antisocial tendencies. The response scale was based on 1, it has never
happened to 5, more than ten times. For violent antisocial behaviors, we used composite scores of six items, e.g., ‘‘Have you threatened or forced someone to give you money, cigarettes, or anything else?’’ or ‘‘Have you been part of beating someone so that you believe or know that he or she needed to be treated at the hospital?’’ (Min/Max = 6/30; Cronbach’s alpha = .77). For nonviolent antisocial behaviors, we used composite scores created from eleven items. For example ‘‘Have you taken things from a store, stand, or shop without paying?’’ or ‘‘Have you taken a bicycle without permission?’’ (Min/Max = 11/55; Cronbach’s alpha = .83). 4.4. Cluster analysis All available data with respect to the five factors from totally three measurement occasions (with a gap of 1 year) were used in the cluster analyses. At each measurement occasion, data were collected for junior high students as well as for high school students. For the junior high cluster analysis, 630 unique individuals, spread over the three measurement occasions, contributed in total 933 data vectors. For the high school cluster analysis, 553 unique individuals, spread over the three measurement occasions, contributed in total 783 data vectors. Because each time period, e.g., junior high, consists of more than 1 school year, e.g., grade 8 and grade 9, some individuals contributed data vectors more than once. In all, 239 individuals were eligible for longitudinal analyses taking into account those who entered the study already in high school. 5. Results 5.1. Preparatory and classification analyses Tables 2 and 3 shows the strength of associations as reflected by pairwise correlations between all five factors cross-sectionally at each time period. Person-oriented methods were used by performing the three steps of the LICUR (Linking of ClUsters after removal of Residue; Bergman, 1998) method with SLEIPNER 2.1 (Bergman & El-Khouri, 2002; Bergman et al., 2003). No imputation was performed on the data. In the first step, residue analyses were conducted to find patterns of values across the cluster variables that matched no one else in the sample. No such cases were found and in turn, no residual patterns were excluded from the analyses. In the second step, separate cluster analyses using Ward’s method and squared Euclidean distances were conducted to find homogenous subgroups at each time point. A 7-cluster solution was chosen for both time points based on homogeneity of cluster solutions, explained error sum of squares (EESS), and theoretical meaningfulness of the cluster solutions (see Fig. 1). In Table 4, characteristics of the cluster solutions including standardized means and standard deviations are presented. At both time points, the clusters were reasonably homogenous for majority of clusters and EESS values were relatively high (64% for both time points). At the first time point, one well adjusted pattern and one depressive symptom only pattern were observed. Two
Table 1 Varimax rotation of two factor solution of social anxiety symptoms based on grades 8 and 9. Items
Factor 1 non-generalized SAD
To To To To To To To To
.76 .71 .67 .53 .49
phone someone unfamiliar initiate a conversation with someone unfamiliar speak in front of the class raise your hand during a lesson go to a party be together with others during the breaks be eating together with others during the lunch-break look someone in the eyes during a conversation
.43
Factor generalized SAD
.46 .88 .78 .50
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
721
Table 2 Correlations and autocorrelations between the variables measured at grade 8 and 9 (n = 933). Non-violent Non-violent Violent Depressive Non-generalized SAD
Violent
Depressive
.83
.31 .25
Non-generalized SAD .03 ns .23 .23
Generalized SAD .11 .10 .28 .46
Note: All correlations are significant at the 0.01 level unless specified. ns = not significant.
Table 3 Correlations and autocorrelations between the variables measured at grade 10 and 11 (n = 783). Non-violent Non-violent Violent Depressive Non-generalized SAD
Violent
Depressive
.79
.29 .23
Non-generalized SAD .02 ns .01 ns .22
Generalized SAD .15 .21 .25 .46
Note: All correlations are significant at the 0.01 level unless specified. ns = not significant.
Table 4 Standardized means (M), standard deviations (S.D.), homogeneity (HC) and cluster total (N) based on 7-cluster solutions at both time points.
Grades 8 and 9 Cluster 1 Cluster 2 Cluster 3 Cluster 4 Cluster 5 Cluster 6 Cluster 7
Non-violent behaviors M (S.D.)
Violent behaviors M (S.D.)
Depressive symptoms M (S.D.)
.49 .13 2.12 .38 .37 .79 5.21
.19 .26 2.25 .30 .30 .64 5.55
.88 1.54 .09 .58 .20 .51 1.59
(.72) (.55) (1.05) (.28) (.29) (1.21) (.94)
(.70) (.30) (.88) (.29) (.29) (1.24) (1.46)
(.90) (.92) (.87) (.51) (.69) (1.06) (1.31)
Non-generalized SAD M (S.D.)
.40 1.25 .66 .64 .87 .48 .15
(.79) (.84) (.76) (.57) (.54) (1.37) (1.32)
Generalized SAD M (S.D.)
HC
N
.35 1.52 .49 .46 .15 3.49 .86
1.03 1.11 1.32 .33 .53 3.44 4.08
151 67 38 376 260 27 14
2.00
933
.91 .37 .24 .71 1.48 2.84 4.53
162 264 197 57 51 38 14
2.00
783
(.37) (.91) (.30) (.27) (.63) (1.62) (1.94)
All Grades 10 and 11 Cluster 1 .00 (.71) Cluster 2 .30 (.38) Cluster 3 .36 (.32) Cluster 4 .04 (.56) Cluster 5 1.40 (1.18) Cluster 6 .23 (.94) Cluster 7 4.98 (1.25)
.14 .28 .30 .25 1.47 .50 5.41
(.44) (.18) (.20) (.31) (1.04) (1.28) (1.75)
.20 .35 .63 1.98 .28 1.08 1.13
(.74) (.52) (.54) (.59) (.87) (1.43) (1.33)
All
patterns were found with heightened antisocial behavioral tendencies. Two patterns were found with higher non-generalized SAD symptoms, one with only heightened non-generalized SAD and one pattern with high depressive, non-generalized SAD, and generalized SAD symptoms. No patterns were found with antisocial problems and non-generalized SAD. One pattern was found with high generalized SAD and non-violent antisocial problems. At the second time point, two well adjusted patterns were found and one depressive symptom only pattern was found. Once again a pattern with high depressive, non-generalized SAD, and generalized SAD symptoms was found, one pattern with high non-generalized SAD only was found, and two patterns with antisocial behavioral tendencies were found. Once again, no pattern was found with indicating both non-generalized SAD and antisocial problems. Structural stability of the cluster solutions were studied by using the CENTROID method where the Average Squared Euclidean Distance (ASED) indicates the degree to which two paired profiles are the same shape (Bergman et al., 2003). The correspondence between the solutions found between junior high school ages and high school ages were fairly high (ASED = .038–1.583). In the final step, in order to study expected and unexpected movements between cluster memberships over time, the EXACON
1.18 .14 1.06 .08 .93 .87 .05
(.66) (.46) (.34) (.88) (.64) (.80) (1.14)
.49 .26 .52 .22 .53 2.94 1.29
(.77) (.50) (.22) (.50) (.20) (1.38) (1.90)
procedure was performed (Bergman & El-Khouri, 1987) on those who had complete data at time one (8th to 9th grade) and at time two (10th to 11th grade) which is approximately 2 years later. This resulted in studying a total of 239 individuals. Individuals who contributed more than one data vector to the analysis of junior high data were assigned the cluster membership at junior high to which the first of their data vectors belonged. Correspondingly, individuals who contributed more than one data vector to the analysis of high school data were assigned the cluster membership at high school to which the last of their data vectors belonged. If significantly more individuals than would be expected by chance alone were observed belonging to any two clusters each pertaining to different time point then the link between these two clusters is denoted as a typical (longitudinal) pathway. These pathways thus found are shown in Fig. 2 depicted with an arrow. Those high on social anxiety symptoms, particularly non-generalized, seemed to have high individual stability independent of co-occurring problems. It was typical for an individual to move from any cluster with non-generalized SAD to another cluster with nongeneralized SAD symptoms. Those high on generalized SAD symptoms seemed to either reappear in the pattern reflecting depressive symptoms or the pattern reflecting antisocial behaviors, depressive symptoms, and generalized SAD symptoms.
722
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
Fig. 2. Descriptive representation of clusters at each time point. The first row represents clusters in grade 8 and 9. The second row represents clusters in grade 10 and 11. Arrows show typical significant cluster pathways (p < .05).
We also studied atypical individual changes (significant unlikely movements of individuals from one cluster to another). As expected, it was unlikely for those with multiple problems at grade 8 or 9 to appear in any of the well adjusted patterns 2 years later and it was unlikely for those in any pattern with no or low problems to appear in the non-generalized SAD pattern 2 years later. More interestingly, those from any pattern with high nongeneralized SAD were unlikely to appear in any of the well adjusted patterns or in the pattern with high antisocial tendencies and low non-generalized SAD 2 years later. 6. Discussion Fig. 1. Graphical representation of clusters at each time point. In the first row, each graph represents plotted cluster means for those in grade 8 and 9. In the second row, each graph represents plotted cluster means for those in grade 10 and 11. NV = nonviolent behaviors, V = violent behaviors, D = depressive symptoms, NG = nongeneralized social anxiety symptoms, G = generalized social anxiety symptoms. Arrows show typical significant cluster pathways (p < .05).
Based on the EXACON analyses, as expected and although not depicted in Fig. 1, it was typical (2–3 times more likely to occur than expected by chance for an individual to move from a cluster with no problems to another cluster with no problems 2 years later.
The present study extends prior research by prospectively following adolescent boys with SAD symptoms, taking into account both problem behaviors and depressive symptoms. To gain a comprehensive view of this interrelationship it is essential to use longitudinal data from the general population. We were interested in understanding cross-sectional configurations as well as individual developmental pathways among youths with nongeneralized and generalized SAD symptoms, respectively. No patterns were found linking antisocial problems and nongeneralized SAD in either junior high or high school. This finding is in line with Sareen and colleagues’ study (2004) which, using two
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
independent population-based datasets in adults, found a significant association between any DSM-III-R anxiety disorder diagnosis (including SAD with multiple social fears, which would be representative of the generalized subgroup) and any DSM-III-R antisocial diagnosis. But the finding is in contradiction to the study of Marmorstein (2006) who found that severe externalizing behavior (oppositional defiant disorder and conduct disorder) was only associated with performance-focused (i.e., non-generalized) SAD in a community-based sample of youth (aged 9–17). Reasons for the disparate findings in these two studies are unclear, but the differences in methodology and samples – notably their very different age ranges – could be operative. It also seems that once one has developed non-generalized SAD symptoms, these symptoms persist over time. This may be contrary to some studies, but not others (Vriends, Becker, Meyer, Williams, et al., 2007), which have found relatively little chronicity in nongeneralized SAD (e.g., Massion et al., 2002), but prior work has mostly focused on clinical, rather than general population samples. It is important to note that the longitudinal period in the present study was short and it is possible that additional remission of symptoms might be observed with longer follow-up. The present findings are consistent, however, with other findings that social anxiety disorder in adolescents does tend to be relatively stable over a several-year period of observation (Ferdinand, Dieleman, Orme, & Verhulst, 2007). It was atypical for youths in the non-generalized SAD cluster to develop comorbidity (both regarding depressive symptoms as well as antisocial tendencies) over time. These observations are highly consistent with most prior research showing that it is generalized SAD (or, conceptualized differently, SAD with multiple social fears) that carries the highest burden of mental health comorbidity (Beesdo et al., 2007; Chartier et al., 2003; Ruscio et al., 2008). Unlike the youths in the non-generalized SAD cluster, the generalized SAD cluster boys in junior high school were likely to develop comorbidity (either with depressive symptoms only or with depressive symptoms and antisocial tendencies) 2 years later. This is in line with previous literature that those with generalized SAD are associated with a higher degree of comorbidity, especially with depression, compared to persons with non-generalized SAD (Chartier et al., 2003; Wittchen et al., 1999). These observations further strengthen the idea that generalized SAD and depression share common vulnerability (e.g., genetic) factor(s) (Hettema, Neale, Myers, Prescott, & Kendler, 2006). The association between generalized SAD and antisocial tendencies is also in line with findings of Sareen et al. (2004) but not with those of Marmorstein (2006). In addition, it does not seem to be the case that inhibited behavior in older youths (grade 10 and 11) is protective against problem behaviors as it seems to be in late childhood (Kerr, Tremblay, Pagani, & Vitaro, 1997). A tentative interpretation of this finding is that abusing alcohol could be a gateway leading to antisocial behaviors for those with generalized SAD in high school. Those with generalized SAD symptoms may use alcohol as self-medication. Such an interpretation is in accordance with previous research showing that SAD is a predictor of alcohol problems among youths (Zimmerman et al., 2003). It is possible that the drinking behavior in some youths with generalized SAD symptoms help them associate with socially deviant peers since they do not fit in with their adjusted peers. Contrary to our expectation, however, youth with generalized SAD symptoms showed co-occurrence with violent antisocial tendencies in spite of their presumed emphatic abilities. One possible explanation of this divergent finding could be found in the above stated alcohol abuse hypothesis in combination with the mechanism of group pressure. Hence, this explanation should be interpreted cautiously, and further research is needed to empirically test the stated alcohol abuse hypothesis in relation to youths with generalized SAD.
723
Our results may also be interpreted in light of Moffitt’s (1993) theoretical model which includes two differing developmental paths for antisocial behavior reflecting two subgroups: life course persistent and adolescence limited. Life-course-persistent adolescents whose early onset of problem behaviors starts in childhood is hallmarked by physical aggression that persists into midlife (Moffitt & Caspi, 2001). The other and larger subgroup is described as adolescence-limited since according to Moffitt (1993), these youth have temporary antisocial behaviors during adolescence and their behaviors are more normative than pathological. The generalized SAD cluster of the present study may fit in well with Moffitt’s adolescence-limited subgroup. Additional longitudinal observation will be needed to confirm this possibility. The finding of high comorbidity in those with generalized SAD symptoms could be conceptualized as interpersonal fears spreading into other interpersonally sensitive domains such as depressive symptoms, antisocial tendencies or also into more performance fears. Hence, the present observations of longitudinal discontinuity from interactional to performance fears may reflect non-generalized and generalized SAD as simply part of a continuum of social anxiety, rather than qualitatively distinct subgroups (Ruscio et al., 2008; Vriends, Becker, Meyer, Michael, & Margraf, 2007). Noteworthy however, is that performance fears do not seem to spread or morph into more interpersonal fears. Further research is needed, however, to determine if these are indeed two qualitatively different subgroups. Some limitations should be noted. First, we used exclusively self-report measures. However, previous research has demonstrated that youth’s own self-report measures of internalizing problems are likely accurate for subjective experiences (Trost et al., 2007). Furthermore, if youths are assured of confidentiality, selfreport measures of externalizing problems have shown good validity (Marini, Dane, Bosacki & YCL-CURA, 2006). Second, we used dimensional measures of symptoms. Although, the dimensional view is argued to better measure elevated symptomatology that later proceeds on to clinical levels, this approach could yield different results than a diagnostic view. Strengths of our approach that should be highlighted are first that we have a very high response rate (over 90%) and that the longitudinal data are from a normative general population sample. Second, we use both variable oriented and person-oriented approaches. Hence, both approaches are legitimate and should be seen as complementary when trying to capture the complexity in the social reality where we human beings are living and interacting (Bergman et al., 2003). Moreover, the person-oriented approach is needed when information about the process characteristics of the developmental course is of interest (Bergman et al., 2003). 7. Conclusion The present study indicates that we should be aware of and take symptoms of both the SAD subgroups seriously but for different reasons. Junior high school youths who show symptoms of nongeneralized SAD (i.e., predominant performance fears) and are untreated should be taken seriously since these symptoms seem to be persistent over time. On the other hand, these symptoms seem unlikely to spread into other kinds of symptoms (interactional fears) or lead to comorbidity with depressive symptoms or disruptive problem behaviors. In contrast, generalized SAD (i.e., predominant interactional social fears) is more likely to develop into co-morbid symptoms such as depressive symptoms and/or disruptive problem behaviors. Both SAD subgroups appear to have in common a level of chronicity (at least during the 6-year period of observation) that seems unlikely to spontaneously dissipate. Whether or not early identification and intervention can reliably
724
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
and cost-effectively change this course of symptoms is a question for further empirical research. In particular, the pathway(s) from interpersonal (generalized) social fears to antisocial behaviors deserve further study in this regard. Acknowledgment This study was supported by the Swedish Council for Working Life and Social Research. References American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: American Psychiatric Association. Text Revision. Beesdo, K., Bittner, A., Pine, D. S., Stein, M. B., Ho¨fler, M., Lieb, R., et al. (2007). Incidence of social anxiety disorder and the consistent risk for secondary depression in the first three decades of life. Archives of General Psychiatry, 64, 903–912. Bergman, L. R. (1998). A pattern-oriented approach to studying individual development: snapshots and processes. In: R. B. Cairns, L. R. Bergman, & J. Kagan (Eds.), Methods and models for studying the individual (pp. 83–122). Thousand Oaks, CA: Sage. Bergman, L. R., & El-Khouri, B. M. (1987). EXACON—A Fortran 77 program for the exact analysis of single cells in a contingency table. Educational and Psychological Measurement, 47, 155–166. Bergman, L. R., & El-Khouri, B. M. (2002). SLEIPNER 2.1-A statistical package for patternoriented analyses. Stockholm: Stockholm University, Department of Psychology. Bergman, L. R., & Trost, K. (2006). The person-oriented versus the variable-oriented approach: are they complementary, opposites, or exploring different worlds? Merrill-Palmer Quarterly, 52, 601–632. Bergman, L. R., Magnusson, D., & El-Khouri, B. (2003). Studying individual development in an interindividual context: a person oriented approach. Mahwah, NJ: Lawrence Erlbaum. Block, J. (1971). On further conjectures regarding acquiescence. Psychological Bulletin, 76, 205–210. Chartier, M. J., Walker, J. R., & Stein, M. B. (2003). Considering comorbidity in social phobia. Social Psychiatry and Psychiatric Epidemiology, 38, 728–734. Fehm, L., Pelissolo, A., Furmark, T., & Wittchen, H.-U. (2005). Size and burden of social phobia in Europe. European Neuropsychopharmacology, 15, 453–462. Fendrich, M., Weissman, M. M., & Warner, V. (1990). Screening for depressive disorder in children and adolescents: validating the Center for Epidemiologic Studies Depression Scale for Children. American Journal of Epidemiology, 131, 538–551. Ferdinand, R. F., Dieleman, G., Orme, J., & Verhulst, F. C. (2007). Homotypic versus heterotypic continuity of anxiety symptoms in young adolescents: evidence for distinctions between DSM-IV subtypes. Journal of Abnormal Child Psychology, 35, 325–333. Furmark, T. (2002). Social phobia: overview of community surveys. Acta Psychiatrica Scandinavica, 105, 84–93. Furmark, T., Tillfors, M., Everz, P.-O., Marteinsdottir, I., Gefvert, O., & Fredrikson, M. (1999). Social phobia in the general population: prevalence and sociodemographic profile. Social Psychiatry and Psychiatric Epidemiology, 34, 416–424. Green-Landell, M., Tillfors, M., Furmark, G., Bohlin, G., Andersson, G., & Svedin, C. G. (2009). Social phobia in Swedish adolescents: prevalence and gender differences. Social Psychiatry and Psychiatric Epidemiology, 44, 1–7. Green-Landell, M., Bjo¨rklind, A., Tillfors, M., Furmark, T., Svedin, C. G., & Andersson, G. (under review). Psychometric evaluation of the Social Phobia Screening Questionnaire for Children (SPSQ-C) in a community sample of Swedish adolescents. European Child and Adolescent Psychiatry. Harvey, A., Watkins, E., Mansell, W., & Shafran, R. (2004). Cognitive behavioral processes across psychological disorders: a transdiagnostic approach to research and treatment. New York: Oxford University Press Inc. Hettema, J. J., Neale, M. C., Myers, J. M., Prescott, C. A., & Kendler, K. S. (2006). A population-based twin study of the relationship between neuroticism and internalizing disorders. American Journal of Psychiatry, 163, 857–864. Hirshfeld-Becker, D. R., Biederman, J., Henin, A., Faraone, S. V., Davis, S., Harrington, K., et al. (2007). Behavioral inhibition in preschool children at risk is a specific
predictor of middle childhood social anxiety: a five-year follow-up. Journal of Developmental and Behavioral Pediatrics, 28, 225–233. Hofmann, S. G., Heinrichs, N., & Moscowitch, D. A. (2004). The nature and expression of social phobia: toward a new classification. Clinical Psychology Review, 24, 769–797. Kagan, J., Snidman, N., & Arcus, D. (1993). On the temperamental categories of inhibited and unhibited children. In: K. H. Rubin & J. Asendorf (Eds.), Social withdrawal, inhibition, and shyness in childhood (pp. 19–28). New Jersey: Lawrence Erlbaum. Kerr, M., Tremblay, R. E., Pagani, L., & Vitaro, F. (1997). Boy’s behavioral inhibition and the risk of later delinquency. Archives of General Psychiatry, 54, 809–816. Kollman, D. M., Brown, T. A., Liverant, G. I., & Hofmann, S. G. (2006). A taxometric investigation of the latent structure of social anxiety disorder in outpatients with anxiety and mood disorders. Depression and Anxiety, 23, 190–199. Marini, Z. A., Dane, A. V., & Bosacki, S. L. YCL-CURA. (2006). Direct and indirect bullyvictims: differential psychosocial risk factors associated with adolescents involved in bullying and victimization. Aggressive Behavior, 32, 551–569. Marmorstein, N. R. (2006). Generalized versus performance-focused social phobia: patterns of comorbidity among youth. Journal of Anxiety Disorders, 20, 778–793. Massion, A. O., Dyck, I. R., Shea, M. T., Phillips, K. A., Warshaw, M. G., & Keller, M. B. (2002). Personality disorder and time to remission in generalized anxiety disorder, social phobia, and panic disorder. Archives of General Psychiatry, 59, 434–440. Moffitt, T. E. (1993). Adolescence-limited and life-course-persistent antisocial behavior: a developmental taxonomy. Psychological Review, 100, 674–701. Moffitt, T. E., & Caspi, A. (2001). Childhood predictors differentiate life-course persistent and adolescence-limited antisocial pathways among males and females. Development and Psychopathology, 13, 355–375. Neal, J. A., Edelmann, R. J., & Glachan, M. (2002). Behavioural inhibition and symptoms of anxiety and depression: is there a specific relationship with social phobia? British Journal of Clinical Psychology, 41, 361–374. Olsson, G., & vonKnorring, A. L. (1997). Depression among Swedish adolescents measured by the self-rating scale Center for Epidemiolgy Studies-Depression Child (CES-DC). European Journal of Child and Adolescent Psychiatry, 6, 81–87. Pallant, J. (2005). SPSS survival manual (2nd ed.). New York: Open university Press. Post, A., Gilljam, H., Rosendahl, I., Meurling, L., Bremberg, S., & Galanti, M. R. (2005). Validity of self reports in a cohort of Swedish adolescent smokers and smokeless tobacco (snus) users. Tobacco Control, 14, 114–117. Rapee, R. M., & Spence, S. H. (2004). The etiology of social phobia: empirical evidence and an initial model. Clinical Psychology Review, 24, 737–767. Roberts, R. E., Lewinsohn, P. M., & Seeley, J. R. (1991). Screening for adolescent depression: a comparison of the CES-D and BDI. Journal of American Academy of Child and Adolescent Psychiatry, 30, 58–66. Ruscio, A. M., Brown, T. A., Chiu, W. T., Sareen, J., Stein, M. B., & Kessler, R. C. (2008). Social fears and social phobia in the USA: results from the National Comorbidity Survey Replication. Psychological Medicine, 38, 15–28. Sareen, J., Stein, M. B., Cox, B. J., & Hassard, S. T. (2004). Understanding comorbidity of anxiety disorders with antisocial behavior. Journal of Nervous and Mental Disease, 192, 178–186. Schneier, F. R. (2006). Clinical practice. Social anxiety disorder. New England Journal of Medicine, 355, 1029–1036. Smith, B. H., Pelham, W. E., Gnagy, E., Molina, B., & Evans, S. (2000). The reliability, validity, and unique contributions of self-report by adolescents receiving treatment for attention-deficit/hyperactivity disorder. Journal of Consulting and Clinical Psychology, 68, 489–499. Stein, M. B., & Stein, D. J. (2008). Social anxiety disorder. Lancet, 371, 1115–1125. Steinberg, L., Lamborn, S. D., Darling, N., Mounts, N. S., & Dornbusch, S. M. (1994). Overtime changes in adjustment and competence among adolescents from authoritative, authoritarian, indulgent, and neglectful families. Child Development, 65, 754–770. Trost, K., Biesecker, G., Stattin, H., & Kerr, M. (2007). Not wanting parents’ involvement: sign of autonomy or sign of problems? European Journal of Developmental Psychology, 4, 314–331. Vriends, N., Becker, E. S., Meyer, A., Williams, S. L., Lutz, R., & Margraf, J. (2007a). Recovery from social phobia in the community and its predictors: data from a longitudinal epidemiological study. Journal of Anxiety Disorders, 21, 320–337. Vriends, N., Becker, E. S., Meyer, A., Michael, T., & Margraf, J. (2007b). Subtypes of social phobia: are they of any use? Journal of Anxiety Disorders, 21, 59–75. Wittchen, H.-U., Stein, M. B., & Kessler, R. C. (1999). Social fears and social phobia in a community sample of adolescents and young adults: prevalence, risk factors and co-morbidity. Psychological Medicine, 29, 309–323. Zimmerman, P., Wittchen, H.-U., Ho¨fler, M., Pfister, H., Kessler, R. C., & Lieb, R. (2003). Primary anxiety disorders and the development of subsequent alcohol use disorders: a 4-year community study of adolescents and young adults. Psychological Medicine, 33, 121.
Contents lists available at ScienceDirect
Journal of Anxiety Disorders
Relationships between social anxiety, depressive symptoms, and antisocial behaviors: Evidence from a prospective study of adolescent boys Maria Tillfors a,*, Bassam El-Khouri b, Murray B. Stein c, Kari Trost a,d Department of Behavioral, Social and Legal Sciences, O¨rebro University, SE-701 82 O¨rebro, Sweden Department of Public Health, Karolinska Institute, Stockholm, Sweden c Department of Psychiatry, University of California San Diego, La Jolla, CA, USA d Department of Psychology, Stockholm University, Stockholm, Sweden a
b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 31 March 2008 Received in revised form 12 February 2009 Accepted 18 February 2009
Little is understood about generalized and non-generalized social anxiety disorder (SAD) and their associations with co-occurring internalizing and externalizing problems in adolescence. In the present study, we investigated adolescent boys with SAD symptoms and considered depressive symptoms as well as antisocial behaviors when looking for patterns during two developmental time periods; junior high and high school. Participants in the analyses were part of a longitudinal study. No patterns were found linking antisocial problems and non-generalized SAD in either junior high or high school. Furthermore, it was uncommon for youths in the non-generalized SAD subgroup to develop comorbidity over time. The generalized SAD subgroup of boys, however, was likely to develop comorbidity either with depressive symptoms only or with depressive symptoms and antisocial tendencies. Our findings suggest that developmental pathways for SAD subgroups may differ. ß 2009 Elsevier Ltd. All rights reserved.
Keywords: Social anxiety disorder Depressive symptoms Antisocial tendencies Adolescent boys Prospective study
Social anxiety disorder (SAD), also known as social phobia, is one of the most prevalent mental disorders. SAD is characterized by intense anxiety and avoidant behavior during performance and/or social interactional situations. Approximately 7–13% of individuals in westernized populations will suffer from social phobia during their lifetime and women are more likely to be affected (Furmark, 2002; Rapee & Spence, 2004). Although most research to date has focused on social phobia in adults, it is clear that SAD symptoms often begin in childhood or adolescence, typically around the age of 13 (Rapee & Spence, 2004). However, considering prevalence and seriousness of this disorder (Fehm, Pelissolo, Furmark, & Wittchen, 2005; Schneier, 2006; Stein & Stein, 2008) it is surprising how little is understood about its characteristics and course early on in its development. In terms of the general pathology of SAD, there are indications that SAD usually belong to one of two subgroups (i.e., generalized or non-generalized; APA, 2000). In turn, these subgroups are associated with different patterns of co-occurring internalizing problems like depressive symptoms (Chartier, Walker, & Stein, 2003; Wittchen, Stein, & Kessler, 1999), and externalizing problems like antisocial behaviors (Sareen, Stein, Cox, & Hassard, 2004). However, only a few studies have considered prospective information gathered during adolescence. In the present study, we will (a) identify patterns based on two subgroups within SAD with consideration to depressive
* Corresponding author. Tel.: +46 19 303959; fax: +46 19 303484. E-mail address: [email protected] (M. Tillfors). 0887-6185/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.janxdis.2009.02.011
symptoms and behavioral problems during two time points during adolescence, and (b) study the developmental pathway of those individuals with pronounced SAD symptoms. The generalized form of SAD is defined in the DSM system as a fear of ‘‘most social situations’’ (APA, 2000). The literature further refers to another SAD subgroup described varyingly as ‘‘discrete,’’ ‘‘circumscribed,’’ ‘‘limited,’’ ‘‘performance,’’ or ‘‘non-generalized,’’ although not explicitly defined in the DSM system (Hofmann, Heinrichs, & Moscowitch, 2004). This latter subgroup in most cases seems to include those whose social fears are constrained to one or more performance situations. In a recent cross-sectional study among youths, Marmorstein (2006) found that generalized SAD among females aged 15–17 was associated with major depression. Among males however, only the performance-focused SAD was found to be associated with dysthymia as well as conduct disorder, an instance of severe externalizing behavior. The differential patterning of symptoms associated within the two SAD subgroups suggest that the two dimensions of internalizing and externalizing problems should be simultaneously considered when evaluating these subgroups and in settings where both time and gender are taken into account. 1. Understanding the link between depressive symptoms and SAD subgroups Social avoidant behavior is a key aspect of SAD and also one of the known risk factors for depression. It is therefore not surprising
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
that prior research has found a link between early behavioral inhibition and later development of SAD and major depression (Hirshfeld-Becker et al., 2007; Neal, Edelmann, & Glachan, 2002). Inhibited children typically react with fearfulness, avoidant behavior, and reticence when confronted with unfamiliar or novel people and places (Kagan, Snidman, & Arcus, 1993). However, previous studies have only found a weak to moderate association between childhood behavioral inhibition and SAD. One reason may be that behavioral inhibition, like SAD, consists of different subgroups, where one subgroup of children reacts with social worry and another with high physiological reactivity. The subgroup typified by social worry may be characterized by stability over time as well as being associated with later development of generalized SAD. This and the finding that an association between social worry and adult depressive symptoms exists suggest that generalized SAD and depression share a common vulnerability factor (Rapee & Spence, 2004). Hence, this could in turn explain the high level of co-occurrence observed between generalized SAD, in comparison with non-generalized SAD, and depression in retrospective studies (Hofmann et al., 2004) and prospective ones (e.g., Beesdo et al., 2007; Wittchen et al., 1999). 2. Understanding the link between antisocial behaviors and SAD subgroups It has been suggested that high levels of social inhibition, characterizing generalized SAD, may function as a protective mechanism against antisocial behaviors. This could be congruent with the observation that adolescents with generalized SAD may be too socially inhibited to associate with delinquent peers. For instance, Kerr and colleagues (1997) report that behaviorally inhibited boys in late childhood, defined by those who were too shy to make friends, seem to be less delinquent than their counterparts with low levels of behavioral inhibition. Thus, although social inhibition may be a risk factor for developing internalizing problems, it may protect against developing externalizing problems. Repeatedly, however, studies have found an association of SAD with antisocial behaviors, as for example the earlier mentioned association between non-generalized SAD and conduct disorder observed among youths (Marmorstein, 2006). Sareen and colleagues (2004) found in two large cross-sectional community studies, this time involving adults, an association between generalized SAD and antisocial behavior. Hence, an association between the two SAD subgroups and externalizing problems seems to emerge but within different age groups. This could indicate that the SAD subgroups in relation to externalizing problems connect in different ways as a function of time, i.e., the pattern of cooccurrence changes during the transition to adulthood. For the last few decades, it has been noted that youths who are violent and persistent over the life course with their problem behaviors start early and tend to lack guilt over time (Moffitt, 1993; Moffitt & Caspi, 2001). Unlike these antisocial individuals, however, SAD individuals seem not to lack the ability to feel empathy or guilt. Because of this presumed retention of empathic ability, it is unlikely that most SAD individuals will display violent problem behaviors. In the present study, we looked specifically at boys with SAD symptoms and considered problem behaviors as well as depressive symptoms when looking for patterns during two developmental time periods in adolescence: junior high school and high school. We first studied possible patterns of SAD, depressive symptoms and antisocial behaviors separately at these two different time periods in order to understand possible cross-sectional configurations. Second, in order to understand
719
developmental pathways of SAD in terms of depressive symptoms and antisocial behaviors, we studied individual developmental pathways for boys with pronounced SAD symptoms. In accordance with prior research (Beesdo et al., 2007; Chartier et al., 2003; Wittchen et al., 1999) we expected more co-occurrence of depressive symptoms to be observed among generalized SAD adolescents and we expected this relationship to become stronger over time, in comparison to non-generalized SAD adolescents. Since previous research shows conflicting findings (Marmorstein, 2006; Sareen et al., 2004) when it comes to the association between SAD subgroups and antisocial behaviors, the a priori hypotheses for this relationship were less clear. However, we expected that none of the adolescents in either SAD subgroup would engage in violent antisocial behaviors. 3. Analytical approach Both variable and person-oriented approaches were used in the present study. The variable oriented approach is useful in analyzing and understanding what characteristics co-aggregate in a group of individuals (Block, 1971). For the purpose of the present study, we used the variable oriented approach to look at the characteristics that are most similar in order to find possible underlying SAD factors. The person-oriented approach is useful in finding configurations and patterns of longitudinal trajectories (Bergman, Magnusson, & El-Khouri, 2003; Bergman & Trost, 2006). We used the person-oriented approach to find and understand individual patterns and pathways of social anxiety symptoms and their relationship to depressive symptoms and antisocial behaviors. 4. Method Participants were part of a prospective longitudinal program of research. Approximately 3000 children and adolescents between the ages of 10 and 18 took part in a 6-year longitudinal survey in an average sized community in mid-Sweden. The present study focuses on boys at two age periods specifically during junior high school and high school. These age periods were chosen specifically since they represented two socially relevant developmental time points. Research on prevalence, etiology, and treatment of psychological disorders is primarily based on categorical diagnoses and restricted to arbitrary cut-offs based on extreme levels of symptoms (DSM-IV-TR; APA, 2000; Harvey, Watkins, Mansell, & Shafran, 2004) which has limited the comprehensive portrayal of disorders in the literature. In the present study, the dimensional perspective is implemented. By taking the dimensional view, no distinct difference is made between, for example, sub-threshold social anxiety and the diagnostic category SAD except in the frequency and/or severity of experienced symptoms (Kollman, Brown, Liverant, & Hofmann, 2006). As part of the larger survey, adolescents filled out a depression scale, a social anxiety scale, and questions about problem behaviors. Adolescents’ reports were used since it has been suggested that adolescents may provide more accurate information about themselves than teachers or parents (Steinberg, Lamborn, Darling, Mounts, & Dornbusch, 1994; Trost, Biesecker, Stattin, & Kerr, 2007). When adolescents are asked to report normbreaking behaviors like their own use of tobacco products, their reporting have been found to be highly reliable (Post et al., 2005). Furthermore, it has been reported that adolescents with certain psychological disorders may be able to give a unique and reliable description of their own health (Smith, Pelham, Gnagy, Molina, & Evans, 2000).
720
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
4.1. Symptoms of depression The Center for Epidemiologic Studies Depression Scale for Children (CES-DC; Fendrich, Weissman, & Warner, 1990; Olsson & vonKnorring, 1997) was used in the present study. It is a self-rating scale and contains of 20 items. In a validation study using a diagnostic interview as a reference the specificity for depression measured by CES-DC was about .70. Further, the CES-DC correlated highly with Beck’s Depression Inventory (BDI; r = .81), indicating adequate concurrent validity (Olsson & vonKnorring, 1997). Although the instrument is used as a screening for depressive symptoms, it has also been said to adequately assess clinical depression in youths (Roberts, Lewinsohn, & Seeley, 1991). In more detail, if the CES-DC is used with high cut off scores the scale has been found to be specific enough to detect clinical depression (Olsson & vonKnorring, 1997). The response scale is based on 1, not at all, to 4, often (Min/Max = 20/80; Cronbach’s alpha = .89). 4.2. Symptoms of SAD The social anxiety criteria were based on the first out of two parts of the Social Phobia Screening Questionnaire for Children (SPSQ-C; Green-Landell et al., 2009), a modified version of the Social Phobia Screening Questionnaire (SPSQ; Furmark et al., 1999). The first part consisted of eight items describing situations that tend to elicit social anxiety. The response scale was based on 1, no fear, to 3, severe fear. In a validation study using a diagnostic interview as a reference the sensitivity for SAD measured by SPSQC was .71 and the specificity was .86. The test–retest reliability over 3 weeks was .60 (Green-Landell et al., under review). In order to be able to study two different potential forms of SAD, a factor analysis was conducted. Based on a principal component analysis on social anxiety symptoms at grade 8 and 9 (junior high school), two components appeared. The Kaiser–Meyer–Oklin correlation of .72 exceeded the recommended value of .60, and the Bartlett’s test was significant (Pallant, 2005). As seen in Table 1, symptoms of SAD were found to load on two separate components. The variance explained for factor 1 was 28% and for factor 2 was 24%. Fears that are characteristic of an individual with nongeneralized SAD were represented by a composite score on the items ‘‘To speak in front of the class,’’ ‘‘To raise your hand during a lesson,’’ ‘‘To phone someone unfamiliar,’’ and ‘‘To initiate a conversation with someone unfamiliar’’ (Min/Max = 4/12; Cronbach’s alpha = .64). Fears that are more characteristic of an individual with generalized SAD were represented by a composite score on the items ‘‘To be together with others during the breaks,’’ ‘‘To go to a party,’’ ‘‘To be eating together with others during the lunch-break,’’ and ‘‘To look someone in the eyes during a conversation’’ (Min/Max = 4/12; Cronbach’s alpha = .66). 4.3. Antisocial behaviors Antisocial behaviors were used as a marker for antisocial tendencies. The response scale was based on 1, it has never
happened to 5, more than ten times. For violent antisocial behaviors, we used composite scores of six items, e.g., ‘‘Have you threatened or forced someone to give you money, cigarettes, or anything else?’’ or ‘‘Have you been part of beating someone so that you believe or know that he or she needed to be treated at the hospital?’’ (Min/Max = 6/30; Cronbach’s alpha = .77). For nonviolent antisocial behaviors, we used composite scores created from eleven items. For example ‘‘Have you taken things from a store, stand, or shop without paying?’’ or ‘‘Have you taken a bicycle without permission?’’ (Min/Max = 11/55; Cronbach’s alpha = .83). 4.4. Cluster analysis All available data with respect to the five factors from totally three measurement occasions (with a gap of 1 year) were used in the cluster analyses. At each measurement occasion, data were collected for junior high students as well as for high school students. For the junior high cluster analysis, 630 unique individuals, spread over the three measurement occasions, contributed in total 933 data vectors. For the high school cluster analysis, 553 unique individuals, spread over the three measurement occasions, contributed in total 783 data vectors. Because each time period, e.g., junior high, consists of more than 1 school year, e.g., grade 8 and grade 9, some individuals contributed data vectors more than once. In all, 239 individuals were eligible for longitudinal analyses taking into account those who entered the study already in high school. 5. Results 5.1. Preparatory and classification analyses Tables 2 and 3 shows the strength of associations as reflected by pairwise correlations between all five factors cross-sectionally at each time period. Person-oriented methods were used by performing the three steps of the LICUR (Linking of ClUsters after removal of Residue; Bergman, 1998) method with SLEIPNER 2.1 (Bergman & El-Khouri, 2002; Bergman et al., 2003). No imputation was performed on the data. In the first step, residue analyses were conducted to find patterns of values across the cluster variables that matched no one else in the sample. No such cases were found and in turn, no residual patterns were excluded from the analyses. In the second step, separate cluster analyses using Ward’s method and squared Euclidean distances were conducted to find homogenous subgroups at each time point. A 7-cluster solution was chosen for both time points based on homogeneity of cluster solutions, explained error sum of squares (EESS), and theoretical meaningfulness of the cluster solutions (see Fig. 1). In Table 4, characteristics of the cluster solutions including standardized means and standard deviations are presented. At both time points, the clusters were reasonably homogenous for majority of clusters and EESS values were relatively high (64% for both time points). At the first time point, one well adjusted pattern and one depressive symptom only pattern were observed. Two
Table 1 Varimax rotation of two factor solution of social anxiety symptoms based on grades 8 and 9. Items
Factor 1 non-generalized SAD
To To To To To To To To
.76 .71 .67 .53 .49
phone someone unfamiliar initiate a conversation with someone unfamiliar speak in front of the class raise your hand during a lesson go to a party be together with others during the breaks be eating together with others during the lunch-break look someone in the eyes during a conversation
.43
Factor generalized SAD
.46 .88 .78 .50
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
721
Table 2 Correlations and autocorrelations between the variables measured at grade 8 and 9 (n = 933). Non-violent Non-violent Violent Depressive Non-generalized SAD
Violent
Depressive
.83
.31 .25
Non-generalized SAD .03 ns .23 .23
Generalized SAD .11 .10 .28 .46
Note: All correlations are significant at the 0.01 level unless specified. ns = not significant.
Table 3 Correlations and autocorrelations between the variables measured at grade 10 and 11 (n = 783). Non-violent Non-violent Violent Depressive Non-generalized SAD
Violent
Depressive
.79
.29 .23
Non-generalized SAD .02 ns .01 ns .22
Generalized SAD .15 .21 .25 .46
Note: All correlations are significant at the 0.01 level unless specified. ns = not significant.
Table 4 Standardized means (M), standard deviations (S.D.), homogeneity (HC) and cluster total (N) based on 7-cluster solutions at both time points.
Grades 8 and 9 Cluster 1 Cluster 2 Cluster 3 Cluster 4 Cluster 5 Cluster 6 Cluster 7
Non-violent behaviors M (S.D.)
Violent behaviors M (S.D.)
Depressive symptoms M (S.D.)
.49 .13 2.12 .38 .37 .79 5.21
.19 .26 2.25 .30 .30 .64 5.55
.88 1.54 .09 .58 .20 .51 1.59
(.72) (.55) (1.05) (.28) (.29) (1.21) (.94)
(.70) (.30) (.88) (.29) (.29) (1.24) (1.46)
(.90) (.92) (.87) (.51) (.69) (1.06) (1.31)
Non-generalized SAD M (S.D.)
.40 1.25 .66 .64 .87 .48 .15
(.79) (.84) (.76) (.57) (.54) (1.37) (1.32)
Generalized SAD M (S.D.)
HC
N
.35 1.52 .49 .46 .15 3.49 .86
1.03 1.11 1.32 .33 .53 3.44 4.08
151 67 38 376 260 27 14
2.00
933
.91 .37 .24 .71 1.48 2.84 4.53
162 264 197 57 51 38 14
2.00
783
(.37) (.91) (.30) (.27) (.63) (1.62) (1.94)
All Grades 10 and 11 Cluster 1 .00 (.71) Cluster 2 .30 (.38) Cluster 3 .36 (.32) Cluster 4 .04 (.56) Cluster 5 1.40 (1.18) Cluster 6 .23 (.94) Cluster 7 4.98 (1.25)
.14 .28 .30 .25 1.47 .50 5.41
(.44) (.18) (.20) (.31) (1.04) (1.28) (1.75)
.20 .35 .63 1.98 .28 1.08 1.13
(.74) (.52) (.54) (.59) (.87) (1.43) (1.33)
All
patterns were found with heightened antisocial behavioral tendencies. Two patterns were found with higher non-generalized SAD symptoms, one with only heightened non-generalized SAD and one pattern with high depressive, non-generalized SAD, and generalized SAD symptoms. No patterns were found with antisocial problems and non-generalized SAD. One pattern was found with high generalized SAD and non-violent antisocial problems. At the second time point, two well adjusted patterns were found and one depressive symptom only pattern was found. Once again a pattern with high depressive, non-generalized SAD, and generalized SAD symptoms was found, one pattern with high non-generalized SAD only was found, and two patterns with antisocial behavioral tendencies were found. Once again, no pattern was found with indicating both non-generalized SAD and antisocial problems. Structural stability of the cluster solutions were studied by using the CENTROID method where the Average Squared Euclidean Distance (ASED) indicates the degree to which two paired profiles are the same shape (Bergman et al., 2003). The correspondence between the solutions found between junior high school ages and high school ages were fairly high (ASED = .038–1.583). In the final step, in order to study expected and unexpected movements between cluster memberships over time, the EXACON
1.18 .14 1.06 .08 .93 .87 .05
(.66) (.46) (.34) (.88) (.64) (.80) (1.14)
.49 .26 .52 .22 .53 2.94 1.29
(.77) (.50) (.22) (.50) (.20) (1.38) (1.90)
procedure was performed (Bergman & El-Khouri, 1987) on those who had complete data at time one (8th to 9th grade) and at time two (10th to 11th grade) which is approximately 2 years later. This resulted in studying a total of 239 individuals. Individuals who contributed more than one data vector to the analysis of junior high data were assigned the cluster membership at junior high to which the first of their data vectors belonged. Correspondingly, individuals who contributed more than one data vector to the analysis of high school data were assigned the cluster membership at high school to which the last of their data vectors belonged. If significantly more individuals than would be expected by chance alone were observed belonging to any two clusters each pertaining to different time point then the link between these two clusters is denoted as a typical (longitudinal) pathway. These pathways thus found are shown in Fig. 2 depicted with an arrow. Those high on social anxiety symptoms, particularly non-generalized, seemed to have high individual stability independent of co-occurring problems. It was typical for an individual to move from any cluster with non-generalized SAD to another cluster with nongeneralized SAD symptoms. Those high on generalized SAD symptoms seemed to either reappear in the pattern reflecting depressive symptoms or the pattern reflecting antisocial behaviors, depressive symptoms, and generalized SAD symptoms.
722
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
Fig. 2. Descriptive representation of clusters at each time point. The first row represents clusters in grade 8 and 9. The second row represents clusters in grade 10 and 11. Arrows show typical significant cluster pathways (p < .05).
We also studied atypical individual changes (significant unlikely movements of individuals from one cluster to another). As expected, it was unlikely for those with multiple problems at grade 8 or 9 to appear in any of the well adjusted patterns 2 years later and it was unlikely for those in any pattern with no or low problems to appear in the non-generalized SAD pattern 2 years later. More interestingly, those from any pattern with high nongeneralized SAD were unlikely to appear in any of the well adjusted patterns or in the pattern with high antisocial tendencies and low non-generalized SAD 2 years later. 6. Discussion Fig. 1. Graphical representation of clusters at each time point. In the first row, each graph represents plotted cluster means for those in grade 8 and 9. In the second row, each graph represents plotted cluster means for those in grade 10 and 11. NV = nonviolent behaviors, V = violent behaviors, D = depressive symptoms, NG = nongeneralized social anxiety symptoms, G = generalized social anxiety symptoms. Arrows show typical significant cluster pathways (p < .05).
Based on the EXACON analyses, as expected and although not depicted in Fig. 1, it was typical (2–3 times more likely to occur than expected by chance for an individual to move from a cluster with no problems to another cluster with no problems 2 years later.
The present study extends prior research by prospectively following adolescent boys with SAD symptoms, taking into account both problem behaviors and depressive symptoms. To gain a comprehensive view of this interrelationship it is essential to use longitudinal data from the general population. We were interested in understanding cross-sectional configurations as well as individual developmental pathways among youths with nongeneralized and generalized SAD symptoms, respectively. No patterns were found linking antisocial problems and nongeneralized SAD in either junior high or high school. This finding is in line with Sareen and colleagues’ study (2004) which, using two
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
independent population-based datasets in adults, found a significant association between any DSM-III-R anxiety disorder diagnosis (including SAD with multiple social fears, which would be representative of the generalized subgroup) and any DSM-III-R antisocial diagnosis. But the finding is in contradiction to the study of Marmorstein (2006) who found that severe externalizing behavior (oppositional defiant disorder and conduct disorder) was only associated with performance-focused (i.e., non-generalized) SAD in a community-based sample of youth (aged 9–17). Reasons for the disparate findings in these two studies are unclear, but the differences in methodology and samples – notably their very different age ranges – could be operative. It also seems that once one has developed non-generalized SAD symptoms, these symptoms persist over time. This may be contrary to some studies, but not others (Vriends, Becker, Meyer, Williams, et al., 2007), which have found relatively little chronicity in nongeneralized SAD (e.g., Massion et al., 2002), but prior work has mostly focused on clinical, rather than general population samples. It is important to note that the longitudinal period in the present study was short and it is possible that additional remission of symptoms might be observed with longer follow-up. The present findings are consistent, however, with other findings that social anxiety disorder in adolescents does tend to be relatively stable over a several-year period of observation (Ferdinand, Dieleman, Orme, & Verhulst, 2007). It was atypical for youths in the non-generalized SAD cluster to develop comorbidity (both regarding depressive symptoms as well as antisocial tendencies) over time. These observations are highly consistent with most prior research showing that it is generalized SAD (or, conceptualized differently, SAD with multiple social fears) that carries the highest burden of mental health comorbidity (Beesdo et al., 2007; Chartier et al., 2003; Ruscio et al., 2008). Unlike the youths in the non-generalized SAD cluster, the generalized SAD cluster boys in junior high school were likely to develop comorbidity (either with depressive symptoms only or with depressive symptoms and antisocial tendencies) 2 years later. This is in line with previous literature that those with generalized SAD are associated with a higher degree of comorbidity, especially with depression, compared to persons with non-generalized SAD (Chartier et al., 2003; Wittchen et al., 1999). These observations further strengthen the idea that generalized SAD and depression share common vulnerability (e.g., genetic) factor(s) (Hettema, Neale, Myers, Prescott, & Kendler, 2006). The association between generalized SAD and antisocial tendencies is also in line with findings of Sareen et al. (2004) but not with those of Marmorstein (2006). In addition, it does not seem to be the case that inhibited behavior in older youths (grade 10 and 11) is protective against problem behaviors as it seems to be in late childhood (Kerr, Tremblay, Pagani, & Vitaro, 1997). A tentative interpretation of this finding is that abusing alcohol could be a gateway leading to antisocial behaviors for those with generalized SAD in high school. Those with generalized SAD symptoms may use alcohol as self-medication. Such an interpretation is in accordance with previous research showing that SAD is a predictor of alcohol problems among youths (Zimmerman et al., 2003). It is possible that the drinking behavior in some youths with generalized SAD symptoms help them associate with socially deviant peers since they do not fit in with their adjusted peers. Contrary to our expectation, however, youth with generalized SAD symptoms showed co-occurrence with violent antisocial tendencies in spite of their presumed emphatic abilities. One possible explanation of this divergent finding could be found in the above stated alcohol abuse hypothesis in combination with the mechanism of group pressure. Hence, this explanation should be interpreted cautiously, and further research is needed to empirically test the stated alcohol abuse hypothesis in relation to youths with generalized SAD.
723
Our results may also be interpreted in light of Moffitt’s (1993) theoretical model which includes two differing developmental paths for antisocial behavior reflecting two subgroups: life course persistent and adolescence limited. Life-course-persistent adolescents whose early onset of problem behaviors starts in childhood is hallmarked by physical aggression that persists into midlife (Moffitt & Caspi, 2001). The other and larger subgroup is described as adolescence-limited since according to Moffitt (1993), these youth have temporary antisocial behaviors during adolescence and their behaviors are more normative than pathological. The generalized SAD cluster of the present study may fit in well with Moffitt’s adolescence-limited subgroup. Additional longitudinal observation will be needed to confirm this possibility. The finding of high comorbidity in those with generalized SAD symptoms could be conceptualized as interpersonal fears spreading into other interpersonally sensitive domains such as depressive symptoms, antisocial tendencies or also into more performance fears. Hence, the present observations of longitudinal discontinuity from interactional to performance fears may reflect non-generalized and generalized SAD as simply part of a continuum of social anxiety, rather than qualitatively distinct subgroups (Ruscio et al., 2008; Vriends, Becker, Meyer, Michael, & Margraf, 2007). Noteworthy however, is that performance fears do not seem to spread or morph into more interpersonal fears. Further research is needed, however, to determine if these are indeed two qualitatively different subgroups. Some limitations should be noted. First, we used exclusively self-report measures. However, previous research has demonstrated that youth’s own self-report measures of internalizing problems are likely accurate for subjective experiences (Trost et al., 2007). Furthermore, if youths are assured of confidentiality, selfreport measures of externalizing problems have shown good validity (Marini, Dane, Bosacki & YCL-CURA, 2006). Second, we used dimensional measures of symptoms. Although, the dimensional view is argued to better measure elevated symptomatology that later proceeds on to clinical levels, this approach could yield different results than a diagnostic view. Strengths of our approach that should be highlighted are first that we have a very high response rate (over 90%) and that the longitudinal data are from a normative general population sample. Second, we use both variable oriented and person-oriented approaches. Hence, both approaches are legitimate and should be seen as complementary when trying to capture the complexity in the social reality where we human beings are living and interacting (Bergman et al., 2003). Moreover, the person-oriented approach is needed when information about the process characteristics of the developmental course is of interest (Bergman et al., 2003). 7. Conclusion The present study indicates that we should be aware of and take symptoms of both the SAD subgroups seriously but for different reasons. Junior high school youths who show symptoms of nongeneralized SAD (i.e., predominant performance fears) and are untreated should be taken seriously since these symptoms seem to be persistent over time. On the other hand, these symptoms seem unlikely to spread into other kinds of symptoms (interactional fears) or lead to comorbidity with depressive symptoms or disruptive problem behaviors. In contrast, generalized SAD (i.e., predominant interactional social fears) is more likely to develop into co-morbid symptoms such as depressive symptoms and/or disruptive problem behaviors. Both SAD subgroups appear to have in common a level of chronicity (at least during the 6-year period of observation) that seems unlikely to spontaneously dissipate. Whether or not early identification and intervention can reliably
724
M. Tillfors et al. / Journal of Anxiety Disorders 23 (2009) 718–724
and cost-effectively change this course of symptoms is a question for further empirical research. In particular, the pathway(s) from interpersonal (generalized) social fears to antisocial behaviors deserve further study in this regard. Acknowledgment This study was supported by the Swedish Council for Working Life and Social Research. References American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: American Psychiatric Association. Text Revision. Beesdo, K., Bittner, A., Pine, D. S., Stein, M. B., Ho¨fler, M., Lieb, R., et al. (2007). Incidence of social anxiety disorder and the consistent risk for secondary depression in the first three decades of life. Archives of General Psychiatry, 64, 903–912. Bergman, L. R. (1998). A pattern-oriented approach to studying individual development: snapshots and processes. In: R. B. Cairns, L. R. Bergman, & J. Kagan (Eds.), Methods and models for studying the individual (pp. 83–122). Thousand Oaks, CA: Sage. Bergman, L. R., & El-Khouri, B. M. (1987). EXACON—A Fortran 77 program for the exact analysis of single cells in a contingency table. Educational and Psychological Measurement, 47, 155–166. Bergman, L. R., & El-Khouri, B. M. (2002). SLEIPNER 2.1-A statistical package for patternoriented analyses. Stockholm: Stockholm University, Department of Psychology. Bergman, L. R., & Trost, K. (2006). The person-oriented versus the variable-oriented approach: are they complementary, opposites, or exploring different worlds? Merrill-Palmer Quarterly, 52, 601–632. Bergman, L. R., Magnusson, D., & El-Khouri, B. (2003). Studying individual development in an interindividual context: a person oriented approach. Mahwah, NJ: Lawrence Erlbaum. Block, J. (1971). On further conjectures regarding acquiescence. Psychological Bulletin, 76, 205–210. Chartier, M. J., Walker, J. R., & Stein, M. B. (2003). Considering comorbidity in social phobia. Social Psychiatry and Psychiatric Epidemiology, 38, 728–734. Fehm, L., Pelissolo, A., Furmark, T., & Wittchen, H.-U. (2005). Size and burden of social phobia in Europe. European Neuropsychopharmacology, 15, 453–462. Fendrich, M., Weissman, M. M., & Warner, V. (1990). Screening for depressive disorder in children and adolescents: validating the Center for Epidemiologic Studies Depression Scale for Children. American Journal of Epidemiology, 131, 538–551. Ferdinand, R. F., Dieleman, G., Orme, J., & Verhulst, F. C. (2007). Homotypic versus heterotypic continuity of anxiety symptoms in young adolescents: evidence for distinctions between DSM-IV subtypes. Journal of Abnormal Child Psychology, 35, 325–333. Furmark, T. (2002). Social phobia: overview of community surveys. Acta Psychiatrica Scandinavica, 105, 84–93. Furmark, T., Tillfors, M., Everz, P.-O., Marteinsdottir, I., Gefvert, O., & Fredrikson, M. (1999). Social phobia in the general population: prevalence and sociodemographic profile. Social Psychiatry and Psychiatric Epidemiology, 34, 416–424. Green-Landell, M., Tillfors, M., Furmark, G., Bohlin, G., Andersson, G., & Svedin, C. G. (2009). Social phobia in Swedish adolescents: prevalence and gender differences. Social Psychiatry and Psychiatric Epidemiology, 44, 1–7. Green-Landell, M., Bjo¨rklind, A., Tillfors, M., Furmark, T., Svedin, C. G., & Andersson, G. (under review). Psychometric evaluation of the Social Phobia Screening Questionnaire for Children (SPSQ-C) in a community sample of Swedish adolescents. European Child and Adolescent Psychiatry. Harvey, A., Watkins, E., Mansell, W., & Shafran, R. (2004). Cognitive behavioral processes across psychological disorders: a transdiagnostic approach to research and treatment. New York: Oxford University Press Inc. Hettema, J. J., Neale, M. C., Myers, J. M., Prescott, C. A., & Kendler, K. S. (2006). A population-based twin study of the relationship between neuroticism and internalizing disorders. American Journal of Psychiatry, 163, 857–864. Hirshfeld-Becker, D. R., Biederman, J., Henin, A., Faraone, S. V., Davis, S., Harrington, K., et al. (2007). Behavioral inhibition in preschool children at risk is a specific
predictor of middle childhood social anxiety: a five-year follow-up. Journal of Developmental and Behavioral Pediatrics, 28, 225–233. Hofmann, S. G., Heinrichs, N., & Moscowitch, D. A. (2004). The nature and expression of social phobia: toward a new classification. Clinical Psychology Review, 24, 769–797. Kagan, J., Snidman, N., & Arcus, D. (1993). On the temperamental categories of inhibited and unhibited children. In: K. H. Rubin & J. Asendorf (Eds.), Social withdrawal, inhibition, and shyness in childhood (pp. 19–28). New Jersey: Lawrence Erlbaum. Kerr, M., Tremblay, R. E., Pagani, L., & Vitaro, F. (1997). Boy’s behavioral inhibition and the risk of later delinquency. Archives of General Psychiatry, 54, 809–816. Kollman, D. M., Brown, T. A., Liverant, G. I., & Hofmann, S. G. (2006). A taxometric investigation of the latent structure of social anxiety disorder in outpatients with anxiety and mood disorders. Depression and Anxiety, 23, 190–199. Marini, Z. A., Dane, A. V., & Bosacki, S. L. YCL-CURA. (2006). Direct and indirect bullyvictims: differential psychosocial risk factors associated with adolescents involved in bullying and victimization. Aggressive Behavior, 32, 551–569. Marmorstein, N. R. (2006). Generalized versus performance-focused social phobia: patterns of comorbidity among youth. Journal of Anxiety Disorders, 20, 778–793. Massion, A. O., Dyck, I. R., Shea, M. T., Phillips, K. A., Warshaw, M. G., & Keller, M. B. (2002). Personality disorder and time to remission in generalized anxiety disorder, social phobia, and panic disorder. Archives of General Psychiatry, 59, 434–440. Moffitt, T. E. (1993). Adolescence-limited and life-course-persistent antisocial behavior: a developmental taxonomy. Psychological Review, 100, 674–701. Moffitt, T. E., & Caspi, A. (2001). Childhood predictors differentiate life-course persistent and adolescence-limited antisocial pathways among males and females. Development and Psychopathology, 13, 355–375. Neal, J. A., Edelmann, R. J., & Glachan, M. (2002). Behavioural inhibition and symptoms of anxiety and depression: is there a specific relationship with social phobia? British Journal of Clinical Psychology, 41, 361–374. Olsson, G., & vonKnorring, A. L. (1997). Depression among Swedish adolescents measured by the self-rating scale Center for Epidemiolgy Studies-Depression Child (CES-DC). European Journal of Child and Adolescent Psychiatry, 6, 81–87. Pallant, J. (2005). SPSS survival manual (2nd ed.). New York: Open university Press. Post, A., Gilljam, H., Rosendahl, I., Meurling, L., Bremberg, S., & Galanti, M. R. (2005). Validity of self reports in a cohort of Swedish adolescent smokers and smokeless tobacco (snus) users. Tobacco Control, 14, 114–117. Rapee, R. M., & Spence, S. H. (2004). The etiology of social phobia: empirical evidence and an initial model. Clinical Psychology Review, 24, 737–767. Roberts, R. E., Lewinsohn, P. M., & Seeley, J. R. (1991). Screening for adolescent depression: a comparison of the CES-D and BDI. Journal of American Academy of Child and Adolescent Psychiatry, 30, 58–66. Ruscio, A. M., Brown, T. A., Chiu, W. T., Sareen, J., Stein, M. B., & Kessler, R. C. (2008). Social fears and social phobia in the USA: results from the National Comorbidity Survey Replication. Psychological Medicine, 38, 15–28. Sareen, J., Stein, M. B., Cox, B. J., & Hassard, S. T. (2004). Understanding comorbidity of anxiety disorders with antisocial behavior. Journal of Nervous and Mental Disease, 192, 178–186. Schneier, F. R. (2006). Clinical practice. Social anxiety disorder. New England Journal of Medicine, 355, 1029–1036. Smith, B. H., Pelham, W. E., Gnagy, E., Molina, B., & Evans, S. (2000). The reliability, validity, and unique contributions of self-report by adolescents receiving treatment for attention-deficit/hyperactivity disorder. Journal of Consulting and Clinical Psychology, 68, 489–499. Stein, M. B., & Stein, D. J. (2008). Social anxiety disorder. Lancet, 371, 1115–1125. Steinberg, L., Lamborn, S. D., Darling, N., Mounts, N. S., & Dornbusch, S. M. (1994). Overtime changes in adjustment and competence among adolescents from authoritative, authoritarian, indulgent, and neglectful families. Child Development, 65, 754–770. Trost, K., Biesecker, G., Stattin, H., & Kerr, M. (2007). Not wanting parents’ involvement: sign of autonomy or sign of problems? European Journal of Developmental Psychology, 4, 314–331. Vriends, N., Becker, E. S., Meyer, A., Williams, S. L., Lutz, R., & Margraf, J. (2007a). Recovery from social phobia in the community and its predictors: data from a longitudinal epidemiological study. Journal of Anxiety Disorders, 21, 320–337. Vriends, N., Becker, E. S., Meyer, A., Michael, T., & Margraf, J. (2007b). Subtypes of social phobia: are they of any use? Journal of Anxiety Disorders, 21, 59–75. Wittchen, H.-U., Stein, M. B., & Kessler, R. C. (1999). Social fears and social phobia in a community sample of adolescents and young adults: prevalence, risk factors and co-morbidity. Psychological Medicine, 29, 309–323. Zimmerman, P., Wittchen, H.-U., Ho¨fler, M., Pfister, H., Kessler, R. C., & Lieb, R. (2003). Primary anxiety disorders and the development of subsequent alcohol use disorders: a 4-year community study of adolescents and young adults. Psychological Medicine, 33, 121.