Journal of Anxiety Disorders 28 (2014) 787–794

Contents lists available at ScienceDirect

Journal of Anxiety Disorders

Developmental pathways of social avoidance across adolescence: The role of social anxiety and negative cognition夽 Anne C. Miers ∗ , Anke W. Blöte, David A. Heyne, P. Michiel Westenberg Faculty of Social and Behavioral Sciences, Institute of Psychology, Unit Developmental and Educational Psychology, Pieter de la Court Building, P.O. Box 9555, 2300 RB Leiden, The Netherlands

a r t i c l e

i n f o

Article history: Received 31 January 2014 Received in revised form 26 August 2014 Accepted 1 September 2014 Available online 17 September 2014 Keywords: Avoidance Adolescence Social anxiety Post-event rumination Life interference

a b s t r a c t It is argued that the adolescent onset of social anxiety disorder (SAD) may be partly attributable to an increase in avoidance of social situations across this period. The current cohort-sequential study investigated developmental pathways of social avoidance in adolescence and examined the explanatory role of social anxiety and negative cognitive processes. A community sample of youth (9–21 years, N = 331) participated in a four-wave study. Trajectory analyses revealed two pathways: an increased avoidance pathway and a low avoidance pathway. The pathways were hardly distinguishable at age 9 and they steadily diverged across adolescence. Logistic regression analyses showed that social anxiety and postevent rumination were significantly related to the increased avoidance pathway; anticipatory processing and self-focused attention were not. The findings suggest that adolescence is a key developmental period for the progression of social avoidance among youth who show relatively high levels of social anxiety and post-event rumination. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction Clinical and community studies of individuals with a DSM-IV (American Psychiatric Association, 1994) diagnosis of social anxiety disorder (SAD) indicate that the disorder typically has its onset in early adolescence (Knappe et al., 2011; Wittchen & Fehm, 2003). In a seminal review of the etiology of SAD, Rapee and Spence (2004) suggested that “the apparent onset of social phobia in early adolescence may perhaps have more to do with the increases in life interference caused by social anxiety [emphasis added] at this developmental stage than with increases in actual levels of social distress” (p. 741). Indeed, social anxiety does not appear to increase across adolescence. Recent studies of the developmental pathways of social anxiety reveal that individual differences in the levels of social anxiety appear in childhood and remain relatively stable across the adolescent period (Broeren, Muris, & Diamantopoulou,

夽 This research was supported by a grant from the Netherlands Organization for Scientific Research (056-34-014) awarded to the first and fourth authors. ∗ Corresponding author. Tel.: +31 071 527 3688; fax: +31 071 527 3619. E-mail addresses: [email protected] (A.C. Miers), [email protected] (A.W. Blöte), [email protected] (D.A. Heyne), [email protected] (P.M. Westenberg). http://dx.doi.org/10.1016/j.janxdis.2014.09.008 0887-6185/© 2014 Elsevier Ltd. All rights reserved.

2013; Marmorstein et al., 2010; Miers, Blöte, de Rooij, Bokhorst, & Westenberg, 2013). If the adolescent onset of SAD cannot be (fully) explained by social anxiety levels per se, might it be explained by an increase in life interference, as proposed by Rapee and Spence (2004)? We suggest that the adolescent onset of SAD may be partly attributable to an increase in the avoidance of feared social situations across this developmental period. Some adolescents with high social anxiety will come to avoid the social situations they are anxious about. Avoidance of social situations could then lead to life interference by limiting adolescents’ opportunities for developing and maintaining social relationships and thus building social competence, and by hindering school attendance and academic development. An increase in the avoidance of social situations across adolescence is likely an important factor contributing to life interference and, by extension, to the adolescent onset of SAD. There is preliminary evidence from cross-sectional studies that the avoidance of social situations increases from late childhood to adolescence. Sumter, Bokhorst, and Westenberg (2009) examined age patterns of self-reported avoidance in a community sample of normally developing adolescents aged between 9 and 17 years. The authors found evidence for an increase in the desire to avoid social situations, whereby adolescents 12 years and older reported a significantly stronger desire to avoid social situations than the group of 9–11 year olds. An age effect for avoidance of social situations

788

A.C. Miers et al. / Journal of Anxiety Disorders 28 (2014) 787–794

was also found in a clinical sample of children (aged 7–12 years) and adolescents (aged 13–17 years) with SAD (Rao et al., 2007). That is, a significantly higher percentage of adolescents reported avoidance of social situations as compared to children. In the current study we examined – longitudinally – the development of avoidance of social situations in a community sample of youth aged between 9 and 21 years. By using a non-clinical sample encompassing a wide range of social anxiety levels, we could measure potential growth in avoidance levels from late childhood to late adolescence. In particular, we investigated whether there is a distinct group of young people who follow a pathway of increasing avoidance of social situations from late childhood to late adolescence. In addition, we aimed to ascertain why some young people might show increased avoidance of social situations. One likely explanation for avoidance is the level of social anxiety. That is, an individual with high social anxiety is perhaps likely to avoid or try to avoid the situations he/she is anxious about. Indeed, moderate to high correlations between social distress and avoidance (.68–.76) have been reported in a non-clinical sample (Sumter et al., 2009). However, anxiety for social situations is not synonymous with avoidance of these situations. An individual with relatively mild anxiety may perceive the anxiety “as a major impediment to their quality of life” and start to avoid social situations, whereas an individual with a higher social anxiety level may not (Rapee & Spence, 2004, p. 739). This suggests that in addition to social anxiety, other factors may play a role in the development of avoidance. Research on clinical forms of social anxiety can shed light on those factors potentially associated with increased avoidance. The leading cognitive models of SAD (Clark, 2001; Clark & Wells, 1995; Heimberg, Brozovich, & Rapee, 2010; Rapee & Heimberg, 1997) suggest that increased avoidance of social situations is explained, in part, by negative cognitive processes that occur before, during, and after an encounter with a feared social situation. These cognitive processes are, respectively, anticipatory processing, self-focused attention, and post-event rumination. Anticipatory processing involves thinking about the upcoming situation, recalling past failures, and making predictions of poor performance or rejection (Clark, 2001). It is hypothesized that these negative anticipatory processes will increase anxiety for, and avoidance of, an upcoming social situation. The link between anticipatory processing and subsequent increased state anxiety has been tested and supported in adult samples (see Clark, 2001). However, to the best of our knowledge, the association between anticipatory processing and avoidance has not been tested in adult or youth samples. Self-focused attention is defined as “the process whereby attention is directed towards internal self-relevant stimuli” (Bögels & Mansell, 2004; p. 840). The stimuli can include behavior, arousal, thoughts, and emotions. The cognitive theories of SAD propose that individuals with SAD show heightened self-focus during a feared social situation (Clark & Wells, 1995; Rapee & Heimberg, 1997). In turn, this increases awareness of one’s potential flaws or errors in behavior and appearance, leading to a negative self-perception. In this way, self-focused attention may increase state anxiety levels and contribute to the avoidance of feared social situations. Whilst there is some support for the effect of heightened self-focused attention on increased state anxiety in adult and youth samples (Bögels & Mansell, 2004; Kley, Tuschen-Caffier, & Heinrichs, 2012), we are not aware of studies testing the relation between selffocused attention and avoidance. Post-event rumination is described as a post-mortem of a social situation, whereby the individual reviews the situation in detail and focuses on the negative self-perception experienced during the situation (Clark, 2001; Clark & Wells, 1995). In a similar way,

Heimberg et al. (2010) describe how a socially anxious person broods upon the specifics of the situation, including his or her actions and behavior and the reactions of other persons. The content of the rumination becomes more negative over time, and the process is said to strengthen the desire to avoid future social situations (Brozovich & Heimberg, 2011). In a sample of undergraduate students, Rachman, Grüter-Andrew, and Shafran (2000) found a positive correlation between post-event rumination and the selfreported wish to avoid future social interactions. In summary, we argue that the adolescent onset of SAD may be partly attributable to an increase in the avoidance of feared social situations across this developmental period. However, we do not yet know whether the avoidance of social situations does increase for some young people. Furthermore, if there is such an increase in the avoidance of social situations, we do not know to what extent it is explained by the level of social anxiety and by the aforementioned cognitive processes – anticipatory processing, self-focused attention, and post-event rumination. Hence, the present longitudinal study addressed two research questions. First, is it possible to identify a group of young people who show an increase in avoidance of social situations from late childhood to late adolescence? Second, is any such increase in avoidance related to social anxiety level and to negative cognitive processes occurring before, during, and after a social situation? To address the first research question we employed group-based trajectory modeling (Nagin, 2005) which is used to identify subgroups of individuals within a population that follow a distinct developmental pathway. We expected to find a developmental pathway showing an increase in the avoidance of social situations from late childhood (9 years) to late adolescence (21 years). Given the paucity of studies in this field, we did not formulate a hypothesis regarding the total number of developmental pathways that would be identified. In relation to the second research question, we hypothesized that higher levels of social anxiety would be positively associated with a pathway of increasing avoidance. We also expected that the three negative cognitive processes would be positively related to an increasing avoidance pathway. Due to the lack of direct evidence in the research literature it was not possible to make a prediction about the relative contribution of the three cognitive processes. Finally, Moulds, Kandris, Starr, and Wong (2007) showed a positive relationship between avoidance of social situations and depressive symptoms, independent of anxiety. Therefore, we took depressive symptoms into account to rule out the possibility that increased avoidance could be explained by inactivity related to depression.

2. Method 2.1. Participants The present study uses data from participants in the Social Anxiety and Normal Development study (SAND; Westenberg et al., 2009). At the start of the SAND study children and adolescents were recruited from one secondary school and two primary schools in an urban area of the Netherlands. This longitudinal study, with a cohort-sequential design, had four assessment waves and data from all four waves are used in the current study. Informed parental consent and participant assent was obtained in writing at each study wave. The SAND study was approved by the Leiden University Medical Ethical Committee. At the first wave (W1) the total sample comprised 331 participants (170 boys and 161 girls) aged between 9 and 17 years, with a mean age of 13.34 years (SD = 2.25). Eighty-two (81.6%) percent of participants lived with their biological parents, 5.7% with biological mother only, and 5.1% with biological mother and

A.C. Miers et al. / Journal of Anxiety Disorders 28 (2014) 787–794

stepfather. Ninety-two (91.5%) percent of participants were born in the Netherlands and 49.0% of biological mothers had completed tertiary education. The number of participants per wave was 331, 298, 248, and 236, respectively. The retention rate from wave to wave thus ranged between 83% and 95%. At W4 there were 121 boys and 115 girls with a mean age of 17.48 years (SD = 2.72). Due to the cohort-sequential design of the SAND study, each age cohort had a different number of participants with a value on our main variable, mean avoidance of social situations. The number of participants per age cohort with a value on mean avoidance was (these n’s include participants’ ages across all four waves): 9 years, n = 28; 10 years, n = 67; 11 years, n = 99; 12 years, n = 107; 13 years, n =131; 14 years, n = 134; 15 years, n = 152; 16 years, n = 124; 17 years, n = 110; 18 years, n = 65; 19 years, n = 39; 20 years, n = 33; 21 years, n =19. We examined missing data according to the number of missing values on the main variable across all four waves. No participant had a missing value at W1, hence the maximum number of missing values is three (i.e., missing at three waves). The number of participants with 0, 1, 2 or 3 missing value(s) was, respectively, 228, 19, 55, and 29. We conducted two regression analyses in order to test whether missing data on the main variable were independent of participant age at W1, avoidance at W1, gender, and the remaining study variables (social anxiety, depression, anticipatory processing, self-focused attention, and post-event rumination). In the first regression, a logistic regression, the dichotomous outcome variable was ‘value at W4 versus no value at W4’. In the second, a multiple regression, the continuous outcome variable was the total number of missing values across all waves. These analyses showed that missing data were independent of age and avoidance at W1, as well as gender and all other study variables (2 (8) = 4.13, ns; F(8) = 0.67, ns).

2.2. Procedure Waves one to three were conducted over three consecutive years and wave four took place between one and three years after W3. A complete description of the procedure employed in the four waves of the SAND study can be found in Miers et al. (2013). In short, participants attended the University for a PreLab Session (at W1, W2, W3, and W4) and again a week later for a Lab session (W1 and W3 only). During the Pre-Lab sessions participants completed a number of assessments. At W1 and W3 the Pre-Lab sessions also included instructions about the public speaking task (i.e., the Leiden Public Speaking Task; Leiden-PST, Westenberg et al., 2009) which would take place a week later during the Lab-session. The Leiden-PST involved participants giving a 5 min speech on the type of films they like and/or dislike and the reasoning why, using an example of a film to illustrate their reasoning. Participants spoke in front of a pre-recorded audience consisting of four boys and four girls (matched to the participant’s age) and a female teacher. The audience was filmed in a classroom setting. The recording began with an empty classroom and after 10 s the pupils and teacher walked into the room, took their seats and then looked into the camera. The audience was projected lifesize onto a screen, without a soundtrack. Participants were told that their performance would be evaluated by peers and teachers. For more details about the Leiden-PST see Westenberg et al. (2009). Data used for avoidance trajectory analyses were collected from participants at each assessment wave. Administration of the avoidance questionnaire always took place at the university during the Pre-Lab session, except at W2 when primary school participants completed the questionnaire at school during a supervised classroom session. The variables social anxiety, depression, anticipatory processing, self-focused attention, and post-event rumination, all

789

self-report, were measured at W1.1 Social anxiety and depression were measured during the Pre-Lab session. The three cognitive processes – anticipatory processing, self-focused attention and post-event rumination – were respectively measured one week before the Leiden-PST, immediately after the Leiden-PST, and one week following the Leiden-PST. The post-event rumination questionnaire was completed on a website designed specifically for SAND study participants. 2.3. Measures 2.3.1. Avoidance of social situations To measure avoidance of social situations we used the questionnaire published by Sumter et al. (2009). The questionnaire is based on the social phobia module from the Anxiety Disorders Interview Schedule for Children (ADIS-C; Silverman & Albano, 1996). The questionnaire begins with a standardized written instruction explaining that there are different ways to avoid situations (e.g., when a teacher asks a question in class a child looks downwards in the hope that he/she doesn’t have to answer the question). A situation is then presented, for example, ‘giving a speech in your class’, and the participant is asked to answer the question ‘Do you try to avoid this situation?’ on a nine-point Likert scale (0 = never, 4 = sometimes, 8 = always). Participants indicate whether they try to avoid twenty different social situations. Across the four waves of the SAND study the internal consistency (˛) of the 20 items in the avoidance questionnaire was excellent, ranging between .83 and .86. 2.3.2. Depression The Dutch translation (Timbremont & Braet, 2002) of the Children’s Depression Inventory (CDI; Kovacs, 1985) measured self-reported depression. The CDI includes 27 items that assess behavioral, affective, and cognitive signs of depression. For each item respondents are presented with three statements and asked to choose the one which best describes how they felt in the past two weeks. For example, “I do most things O.K.,” “I do many things wrong” and “I do everything wrong.” One item asking about suicide was removed from the questionnaire (see Wetter & El-Sheikh, 2012). In accordance with previous research we report sum scores (Roelofs et al., 2010) which range between 0 and 52. The Dutch version shows good reliability and validity (Roelofs et al., 2010) and ˛ for W1 participants in the current study was .80. 2.3.3. Social anxiety The Dutch translation (H. Koot & E. Utens, unpublished) of the Social Anxiety Scale for Adolescents (SAS-A; La Greca & Lopez, 1998) provided the measure of social anxiety. This 22-item instrument contains 18 descriptive self-statements about social anxiety symptoms (e.g., “I worry about what other kids think of me”) and four filler items. Respondents are asked to rate each item according to the degree to which the item “is true for you” (1 = not at all, 5 = all the time). Two of the SAS-A items have potential conceptual overlap with avoidance (i.e., item 5 “I only talk to people I know really well” and item 15 “I’m quiet when I’m with a group of people”). These two items were excluded when calculating the SAS-A scores used in the data analyses. The SAS-A has good internal consistency (La Greca & Lopez, 1998) and in the current study ˛ was .94 at W1.

1 In Sections 2.4 and 3 these variables (social anxiety, depression, anticipatory processing, self-focused attention, and post-event processing) are described as predictors because they are statistical predictors of trajectory membership. The interpretation of these variables in relation to trajectories of avoidance is addressed in Section 4.

790

A.C. Miers et al. / Journal of Anxiety Disorders 28 (2014) 787–794

2.3.4. Anticipatory processing To measure participants’ anticipatory processing before the Leiden-PST, we used a Dutch translation (see Miers, Blöte, Sumter, Kallen, & Westenberg, 2011) of the Expected Performance measure developed by Spence, Donovan, and Brechman-Toussaint (1999). This 7-item questionnaire measures expectation of the quality of performance during a speech, and predictions about how other people will judge the performance. Items are rated using a 5-point scale from negative to positive expectation. Internal consistency in the current study’s W1 sample was good (˛ = .76). Scores were re-scaled so that a higher score reflected negative anticipatory processing. 2.3.5. Self-focused attention The degree to which participants engaged in self-focused attention during the Leiden-PST was measured with the self-focus subscale from the Focus of Attention Questionnaire (FAQ; Woody, Chambless, & Glass, 1997). We used a 4-item version of the selffocus subscale (Miers et al., 2011). An example item is “I was focusing on my internal bodily reactions (for example, heart rate).” Each item is rated on a 5-point scale (1 = not at all, 5 = totally) according to how much the participant’s attention matched the item description. Previous studies have reported acceptable internal consistency in samples of adults (Woody et al., 1997) and youth (Hodson, McManus, Clark, & Doll, 2008). In the current W1 sample ˛ was .61. 2.3.6. Post-event rumination Rumination was measured with a Dutch translation of the Thoughts Questionnaire (TQ) developed by Edwards, Rapee, and Franklin (2003). The original version contains 29 items measuring positive and negative thoughts. For the SAND study some of the items were removed because of their difficulty for children. The shortened SAND version included 14 items measuring negative rumination, 9 items measuring positive rumination, and 1 neutral item. The present study used data from the negative rumination subscale only. Example items from the negative subscale are “How bad my speech was” and “What a failure I was”. The items are answered on a 5-point scale (0 = Never to 4 = Very Often) according to how often the participant had the thought in the week since the Leiden-PST. In the current W1 sample the internal consistency of the negative rumination subscale was very good, ˛ = .92. Ninety-four participants did not complete the rumination questionnaire. Missing data on the rumination questionnaire was unrelated to avoidance of social situations at W1 (t (329) = 0.77, ns). Of the 237 participants who completed this questionnaire, 42% completed it one week after the Leiden-PST, 46% between one and two weeks after the Leiden-PST, and 12% between two weeks and one and a half months after the Leiden-PST. Time to complete the rumination questionnaire was unrelated to avoidance of social situations at W1 (F(2, 234) = 1.04, ns). 2.4. Data analyses First, trajectories of social avoidance were identified using the SAS (version 9.2) PROC TRAJ program (Jones, Nagin, & Roeder, 2001). A censored normal distribution was used as the basis of model estimation. PROC TRAJ accommodates missing data (Nagin, 1999). We followed Nagin’s (2005) two stage model estimation procedure as described in Miers et al. (2013). Models with one to five cubic trajectory groups were estimated. The optimal number of trajectories was determined using the Bayesian information criterion (BIC) and the size of the trajectory groups (a minimum of 5% for the smallest group; Andruff, Carraro, Thompson, & Gaudreau, 2009). Next, the form of each trajectory was ascertained using backward removal of non-significant higher-order trends, removing cubic

trends first, then quadratic trends, and then, if appropriate, linear trends. The final model’s average posterior probabilities (PP) of group membership were examined and these should be greater than .70–.80 (Nagin, 1999). Second, logistic regression models were run in the Generalized Linear Models function in SPSS (version 21) to examine whether avoidance trajectories were statistically predicted by social anxiety and cognitive processes before, during, and after a social situation. In order to control for a relation between depression and avoidance trajectories, the first regression model tested depression as a predictor variable. In model 2 social anxiety was added as a predictor. In model 3 anticipatory processing, self-focused attention, and post-event rumination were added as predictors. Chi-square change was used to test whether the more complex model significantly improved the prediction of trajectory group membership.

3. Results 3.1. Trajectories of avoidance From the trajectory analyses a two group cubic model was identified as the best fit to the data with a BIC of −1233.53. A one group model had a BIC of −1373.11, and models with three, four, and five groups had BIC values of −1177.28, −1159.23 and −1170.29, respectively. Although the latter groups showed BIC values that were closer to zero, these models included groups with less than 5% of the sample. The initial model with two cubic groups was pared down, and the estimated avoidance trajectories are depicted in Fig. 1. The final two group model2 had a better fit to the data (BIC = −1228.36) than the initial model with cubic trends (BIC = −1233.53). The average posterior probability for both groups easily satisfied the 0.80 criterion. The first trajectory, increased avoidance (22.3%, n = 69, PP = .89, constant = 0.95, SE = 0.19, p < .001, linear slope = 0.33, SE = 0.06, p < .001, quadratic slope = −0.02, SE = 0.005, p < .001), comprised young people who began with low levels of social avoidance, showed an increase between 9 and 17 years of age, and then a slight decrease. The second trajectory, low avoidance (77.7%, n = 262, PP = .95, constant = 0.59, SE = 0.08, p < .001, linear slope = 0.08, SE = 0.03, p < .02, quadratic slope = −0.01, SE = 0.003, p < .02), comprised young people who began with low levels of avoidance and showed a minimal increase and decrease in the level of avoidance across adolescence (between 9 and 21 years). At 9 years the level of avoidance in the increased avoidance trajectory group did not significantly differ from that in the low avoidance trajectory group, 2 (1) = 3.03, ns, however, the linear and quadratic functions of the two trajectories were significantly different 2 (2) = 21.25, p < 001. Taken together, these results indicate that the trajectories are not parallel. The trajectory confidence intervals overlap at ages 9 and 10 years, but they cease to overlap from this point on, through to 21 years, indicating distinct trajectory groups in the adolescent period. The distribution of age cohort at W1 (2 (8) = 4.77, ns) and gender (2 (1) = 0.48, ns) was similar in each trajectory. In subsequent analyses, a participant was assigned to the trajectory group for which they had the highest posterior probability (Nagin, 1999).

2 The trajectory analyses were also conducted on the smaller sample used for the logistic regression analyses (n = 229). These analyses revealed the same two trajectories, a low avoidance pathway (79%, n = 185) and an increased avoidance pathway (20%, n = 44). Importantly, all participants assigned to the increased pathway based on the n = 229 sample were also assigned to the increased pathway as based on the total sample, n = 331.

A.C. Miers et al. / Journal of Anxiety Disorders 28 (2014) 787–794

791

4 3.5

Social avoidance

3 2.5 2 1.5 1 0.5 0 9

10

11

12

13

14

15

16

17

18

19

20

21

Age (years) Group 1 predicted

Group 2 predicted

Fig. 1. Trajectories of avoidance of social situations in solid lines and 95% confidence intervals in dashed lines. Group 1 = increased avoidance; Group 2 = low avoidance.

3.2. Predictors of avoidance trajectories In the following analyses we used data from 229 participants who had complete data on all predictor variables. There were 180 participants in the low avoidance group and 49 in the increased avoidance group. Table 1 presents the correlations among the predictor variables and between the predictor variables and avoidance trajectory group. Among the predictor variables the strongest correlations were between depression and social anxiety (r = .54, p < .01) and between self-focused attention and negative rumination (r = .49, p < .01). All five predictor variables were positively correlated with avoidance trajectory group. For the logistic regression analyses we chose the low trajectory group as the reference category because we were specifically interested in the group that shows an increase in avoidance across adolescence. In the first regression model (see Table 2) depression was a significant predictor of trajectory group membership (2 = 24.10 (1), p < .001, LL = −106.84, AIC = 217.68, BIC = 224.55). The odds ratio indicates that the probability of belonging to the increased group was greater for participants with higher levels of depression. The addition of social anxiety to the model including depression significantly improved the prediction of trajectory group membership, 2 (1) = 11.77, p < .001 (2 = 35.87 (2), p < .001, LL = −100.95, AIC = 207.91, BIC = 218.21). In this second model both depression and social anxiety contributed to the odds of belonging to the increased avoidance trajectory versus the low avoidance trajectory. However, social anxiety was the stronger of

the two predictors. The positive odds ratios indicate that high levels of depression and social anxiety increased the probability of belonging to the increased trajectory versus the low trajectory. The third model included the cognition variables and it showed a significant improvement in the prediction of trajectory group membership relative to model two, 2 (3) = 10.91, p < .025 (2 = 46.78 (5), p < .001, LL = −95.50, AIC = 203.00, BIC = 223.61). Following the addition of the cognition variables, depression no longer significantly differentiated between the trajectories, whereas social anxiety remained significant. Of the cognition variables, only postevent rumination showed a significant main effect. The positive odds ratio shows that the probability of belonging to the increased trajectory group relative to the low trajectory group was higher for participants with a tendency to engage in more negative post-event rumination after the speech task. 4. Discussion This is the first study to examine developmental pathways of the avoidance of social situations in adolescence and to test whether social anxiety and negative cognitive processes discriminate between these avoidance pathways. As expected, we identified a group of young people for whom the avoidance of social situations increased from late childhood to late adolescence (increased avoidance pathway). This supports the suggestion from cross-sectional research that the avoidance of social situations increases in adolescence (Rao et al., 2007; Sumter et al., 2009). The trajectory

Table 1 Correlations among predictor variables and avoidance trajectory group, and descriptive statistics of predictor variables (n = 229). Variable

1

1. Depression 2. Social anxiety 3. Anticipatory processing 4. Self-focused attention 5. Post-event rumination 6. Avoidance group (0, 1)a

– .54** .37** .24** .37** .47**

2 – .40** .38** .43** .51**

3

– .20** .28** .37**

4

– .49** .38**

Note: Boys and girls did not differ on any of the predictor variables. a Biserial correlations reported between avoidance trajectory group and predictors. 1 = increased avoidance group. ** p < .01 (1-tailed).

5

M(SD)

.48**

8.79 (5.30) 2.27 (0.75) 2.76 (0.46) 2.13 (0.73) 1.01 (0.73) –

Min–Max 0.00–31.00 1.00–4.63 1.33–4.33 1.00–4.75 0.00–3.64 –

792

A.C. Miers et al. / Journal of Anxiety Disorders 28 (2014) 787–794

Table 2 Logistic regression models predicting avoidance trajectory group membership (n = 229). Predictors

Model effects and parameter estimates

95% PL CI for OR

LR 2 (df)

B (SE)

Wald 2 (df)

OR

LL

UL

Model 1 W1 depression

24.10 (1)**

.15 (.03)

21.29 (1)

1.16**

1.09

1.24

Model 2 W1 depression W1 social anxiety

5.11 (1)* 11.77 (1)**

.08 (.04) .92 (.28)

5.10 (1) 10.65 (1)

1.09* 2.52**

1.01 1.47

1.17 4.49

2.28 (1) 4.41 (1)* 1.37 (1) 0.97 (1) 4.87 (1)*

.06 (.04) .64 (.31) .50 (.43) .27 (.28) .64 (.29)

2.28 (1) 4.26 (1) 1.34 (1) 0.96 (1) 4.92 (1)

1.06 1.89* 1.64 1.31 1.89*

0.98 1.04 0.72 0.77 1.08

1.15 3.52 3.90 2.27 3.34

Model 3 W1 depression W1 social anxiety W1 anticipatory processing W1 self-focused attention W1 post-event rumination

Note: Low avoidance group n = 180, increased avoidance group n = 49. CI, confidence interval; LR, likelihood ratio; OR, odds ratio; PL, profile likelihood. * p < .05. ** p < .01.

analyses also identified a group of young people with consistently low avoidance (low avoidance pathway). The two pathways were hardly distinguishable at age 9 and they steadily diverged across adolescence. Social anxiety and negative post-event rumination were both independently related to the increased avoidance pathway. Together, the study’s findings suggest that adolescence is a key developmental period for the progression of social avoidance among youth who show relatively high levels of social anxiety and post-event rumination. The increased avoidance pathway identified in our study is of particular relevance to Rapee and Spence (2004)’s suggestion that increased life interference caused by social anxiety may explain the adolescent onset of SAD. If, as proposed in the current study, increased life interference is indeed (partly) a result of an increase in avoidance then the increased avoidance pathway’s steady rise between 9 and 17 years suggests that adolescents following this pathway are likely to be at risk for developing SAD. To clearly determine the role of avoidance in the onset of SAD, it is necessary to prospectively track avoidance in a group of young people, and simultaneously screen for the presence of SAD by means of a clinical interview at the start and at the end of the study. As regards the relation between social anxiety and the increased avoidance pathway, our results show that the greater the anxiety for social situations, the more likely it is that these youth will increasingly try to avoid these social situations between late childhood to mid-adolescence. This finding is consistent with previous research yielding moderate to high correlations between self-reported social distress and avoidance in a clinical sample of adults with SAD (Heimberg et al., 1999) and a community sample of youth (Sumter et al., 2009). Based on the cognitive models of SAD (Clark, 2001; Heimberg et al., 2010) we also chose to examine three negative cognitive processing variables hypothesized to play a role in avoidance of social situations, namely anticipatory processing, self-focused attention, and post-event rumination. Of these cognitive variables only negative post-event rumination significantly discriminated between the increased and low avoidance pathways. This finding provides support for the post-event rumination component of the cognitive models of SAD (Clark, 2001; Heimberg et al., 2010). In the updated version of the Rapee and Heimberg model (Heimberg et al., 2010) specific attention is given to post-event rumination and the mechanisms by which this process may lead to avoidance. According to the authors, post-event rumination involves a process of the person “taking apart and putting together the elements of the situation and placing his or her own interpretations on them each time . . .” (p. 409). The authors propose that,

as a result of this iterative process, the individual’s view of the situation is likely to become increasingly negative over time. Further, when thinking about an upcoming social situation, the rumination that has taken place will connect previous social situations with the upcoming situation. It is likely, then, that such a rumination process will negatively influence young persons’ approach to a new social situation, whereby they will be inclined to try to avoid the situation. Further studies are required to shed more light on the (relative) role of the cognitive processes, anticipatory processing, self-focused attention, and post-event rumination, in the avoidance of social situations, particularly given the lower internal consistency of the self-focused attention measure used in the current study. In addition, other cognitive variables implicated in the maintenance of SAD may be related to the development of avoidance behavior in adolescence, such as interpretation bias and otherfocused attention. These cognitive factors also deserve attention in future research. In the present study, identification of social avoidance pathways was based on a predominantly white, middle-class sample. Future studies should include a more diverse sample in order to investigate potential differences in social avoidance by certain demographic variables like ethnicity and socio-economic status. In terms of gender, boys and girls were represented equally in each avoidance pathway. However, the relatively small size of the increased avoidance pathway group meant that we could not include gender as a variable in the regression analyses. It would be important for future studies to investigate avoidance pathways and their relation to social anxiety in boys and girls separately, particularly given the frequently found gender difference in social anxiety symptoms in non-clinical samples (Rapee & Spence, 2004). In our non-clinical sample the highest predicted avoidance score in the increased avoidance pathway was relatively low. It reflects the avoidance of social situations ‘every now and then’ according to the avoidance scale, falling short of the ‘sometimes’ point on the scale. It is worth keeping in mind that the pathways are based on the mean avoidance of twenty different social situations. Those adolescents who avoid social situations are not likely to avoid all types of social situations. Some situations may be avoided by more adolescents than other situations. Indeed, Blöte, Miers, Heyne, and Westenberg (2014) present data on the frequency with which low and high socially anxious youth avoid each of the same twenty different social situations. In particular, formal classroom situations (e.g., giving a speech in class) were found to be frequently avoided by high socially anxious youth. A higher avoidance score could therefore be expected given an avoidance pathway that is

A.C. Miers et al. / Journal of Anxiety Disorders 28 (2014) 787–794

represented by a subset of the social situations from our measure of avoidance. The present study has some limitations. In the SAND study’s cohort design, the measurement of the explanatory variables did not precede the start of the avoidance pathways. This means that the relations reported between the explanatory variables and the two developmental pathways do not demonstrate their actual temporal ordering. Second, our measure of avoidance asked respondents if they try to avoid a particular situation. We do not know, however, whether the respondents actually avoided the situations they reported trying to avoid. Because overt and covert avoidance may be relevant to the clinical diagnosis of SAD (Whiteside, Gryczkowski, Ale, Brown-Jacobsen, & McCarthy, 2013), it would be important to measure actual avoidance behavior in future studies. Third, we did not measure life interference per se. We measured avoidance, which may lead to interference in life functioning. It will be important for future studies to directly measure life interference caused by avoidance, fear, or anxiety, alongside the development of each of these, to better understand the complex relationship between these constructs. The study’s limitations notwithstanding, several practical implications are worth considering. Because youth who engage in post-event rumination appear to be at risk for increased avoidance across adolescence, which in turn may increase the risk for SAD, support should be provided to such youth. Selective prevention of the adolescent onset of SAD might be achieved by working with pre-adolescent youth who score high on a screening for rumination. Intervention might include youth-oriented psycho-education about the differentiation between adaptive and maladaptive forms of rumination (Nolen-Hoeksema, Wisco, & Lyubomirsky, 2008). Skills training might focus on the cessation of maladaptive postevent rumination (Clark & Wells, 1995) or the processing of social encounters in a concrete and constructive way (e.g., Makkar & Grisham, 2013). An alternative approach would be to help youth to detach from the negative cognition of post-event rumination via interventions such as mindfulness training, and to move toward value based goals (e.g., being part of a stable peer group at school; Murrell, Coyne, & Wilson, 2004). Finally, the present study’s findings suggest that the factors identified here as explaining avoidance of social situations, post-event rumination and social anxiety should be tackled at an age that precedes the rise in social avoidance. This should maximize the chance of preventing the avoidance of social situations, and ultimately, SAD.

References American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Andruff, H., Carraro, N., Thompson, A., & Gaudreau, P. (2009). Latent class growth modelling: a tutorial. Tutorials in Quantitative Methods for Psychology, 5, 11–24. Blöte, A. W., Miers, A. C., Heyne, D. A., & Westenberg, P. M. (2014). Social anxiety and the school environment of adolescents. (submitted for publication). Bögels, S. M., & Mansell, W. (2004). Attention processes in the maintenance and treatment of social phobia: hypervigilance, avoidance and self-focused attention. Clinical Psychology Review, 24, 827–856. http://dx.doi.org/10.1016/ j.cpr.2004.06.005 Broeren, S., Muris, P., & Diamantopoulou, S. (2013). The course of childhood anxiety symptoms: developmental trajectories and child-related factors in normal children. Journal of Abnormal Child Psychology, 41, 81–95. http://dx. doi.org/10.1007/s10802-012-9669-9 Brozovich, F., & Heimberg, R. G. (2011). The relationship of post-event processing to self-evaluation of performance in social anxiety. Behavior Therapy, 42, 224–235. http://dx.doi.org/10.1016/j.beth.2010.08.005 Clark, D. M. (2001). A cognitive perspective on social phobia. In: W. R. Crozier, & L. E. Alden (Eds.), International handbook of social anxiety: concepts, research and interventions relating to the self and shyness (pp. 405–430). Chichester: Wiley. Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. In: R. G. Heimberg, M. R. Liebowitz, D. A. Hope, & F. R. Schneier (Eds.), Social phobia: diagnosis, assessment, and treatment (pp. 69–93). New York: Guilford Press. Edwards, S. L., Rapee, R. M., & Franklin, J. (2003). Postevent rumination and recall bias for a social performance event in high and low socially anxious

793

individuals. Cognitive Therapy and Research, 27, 603–617. http://dx.doi.org/ 10.1023/a:1026395526858 Heimberg, R. G., Brozovich, F. A., & Rapee, R. M. (2010). A cognitive behavioral model of social anxiety disorder: update and extension. In: S. G. Hofmann, & P. M. DiBartolo (Eds.), Social anxiety. Clinical, developmental and social perspectives (pp. 395–422). New York: Elsevier. http://dx.doi.org/10.1016/b978-012-375096-9.00015-8 Heimberg, R. G., Horner, K. J., Juster, H. R., Safren, S. A., Brown, E. J., Schneier, F. R., et al. (1999). Psychometric properties of the Liebowitz Social Anxiety Scale. Psychological Medicine, 29, 199–212. http://dx.doi.org/10.1017/s0033291798007879 Hodson, K. J., McManus, F. V., Clark, D. M., & Doll, H. (2008). Can Clark and Wells’ (1995) cognitive model of social phobia be applied to young people? Behavioural and Cognitive Psychotherapy, 36, 449–461. http://dx.doi.org/ 10.1017/s1352465808004487 Jones, B., Nagin, D. S., & Roeder, K. (2001). A SAS procedure based on mixture models for estimating developmental trajectories. Sociological Methods and Research, 29, 374–393. http://dx.doi.org/10.1177/0049124101029003005 Kley, H., Tuschen-Caffier, B., & Heinrichs, N. (2012). Safety behaviors, self-focused attention and negative thinking in children with social anxiety disorder, socially anxious and non-anxious children. Journal of Behavior Therapy and Experimental Psychiatry, 43, 548–555. http://dx.doi.org/10.1016/j.jbtep.2011.07.008 Knappe, S., Beesdo-Baum, K., Fehm, L., Stein, M. B., Lieb, R., & Wittchen, H. (2011). Social fear and social phobia types among community youth: differential clinical features and vulnerability factors. Journal of Psychiatric Research, 45, 111–120. http://dx.doi.org/10.1016/j.jpsychires.2010.05.002 Kovacs, M. (1985). The Children’s Depression Inventory (CDI). Psychopharmacology Bulletin, 21, 995–998. La Greca, A. M., & Lopez, N. (1998). Social anxiety among adolescents: linkages with peer relations and friendships. Journal of Abnormal Child Psychology, 26, 83–94. http://dx.doi.org/10.1023/A:1022684520514 Makkar, S. R., & Grisham, J. R. (2013). The effect of post-event processing on mood, self-beliefs, and interpretive biases. Journal of Experimental Psychopathology, 4, 368–386. http://dx.doi.org/10.5127/jep.030912 Marmorstein, N. R., White, H., Chung, T., Hipwell, A., Stouthamer-Loeber, M., & Loeber, R. (2010). Associations between first use of substances and change in internalizing symptoms among girls: differences by symptom trajectory and substance use type. Journal of Clinical Child and Adolescent Psychology, 39, 545–558. http://dx.doi.org/10.1080/15374416.2010.486325 Miers, A. C., Blöte, A. W., de Rooij, M., Bokhorst, C. L., & Westenberg, P. M. (2013). Trajectories of social anxiety during adolescence and relations with cognition, social competence, and temperament. Journal of Abnormal Child Psychology, 41, 97–110. http://dx.doi.org/10.1007/s10802-012-9651-6 Miers, A. C., Blöte, A. W., Sumter, S. R., Kallen, V. L., & Westenberg, P. M. (2011). Subjective and objective arousal correspondence and the role of self-monitoring processes in high and low socially anxious youth. Journal of Experimental Psychopathology, 2, 531–550. http://dx.doi.org/10.5127/jep.019411 Moulds, M. L., Kandris, E., Starr, S., & Wong, A. C. M. (2007). The relationship between rumination, avoidance and depression in a non-clinical sample. Behaviour Research and Therapy, 45, 251–261. http://dx.doi.org/10.1016/j.brat.2006.03. 003 Murrell, A. R., Coyne, L. W., & Wilson, K. G. (2004). ACT with children, adolescents and their parents. In: S. C. Hayes, & K. D. Stroshal (Eds.), A practical guide to acceptance and commitment therapy (pp. 249–273). New York, NY: Springer. http://dx.doi.org/10.1007/978-0-387-23369-7 10 Nagin, D. S. (1999). Analyzing developmental trajectories: a semiparametric groupbased approach. Psychological Methods, 4, 139–157. http://dx.doi.org/10.1037/ 1082-989x.4.2.139 Nagin, D. S. (2005). Group-based modeling of development. Cambridge, MA: Harvard University Press. www.blackwell-synergy.com/doi/pdf/10.1111/j.17459125.2007.00078.x Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008). Rethinking rumination. Perspectives on Psychological Science, 3, 400–424. http://dx.doi.org/ 10.1111/j.1745-6924.2008.00088.x Rachman, S., Grüter-Andrew, J., & Shafran, R. (2000). Post-event processing in social anxiety. Behaviour Research and Therapy, 38, 611–617. http://dx.doi. org/10.1016/s0005-7967(99)00089-3 Rao, P. A., Beidel, D. C., Turner, S. M., Ammerman, R. T., Crosby, L. E., & Sallee, F. R. (2007). Social anxiety disorder in childhood and adolescence: descriptive psychopathology. Behaviour Research and Therapy, 45, 1181–1191. http://dx.doi.org/10.1016/j.brat.2006.07.015 Rapee, R. M., & Heimberg, R. G. (1997). A cognitive-behavioral model of anxiety in social phobia. Behaviour Research and Therapy, 35, 741–756. http://dx.doi.org/10.1016/s0005-7967(97)00022-3 Rapee, R. M., & Spence, S. H. (2004). The etiology of social phobia: empirical evidence and an initial model. Clinical Psychology Review, 24, 737–767. http://dx.doi.org/10.1016/j.cpr.2004.06.004 Roelofs, J., Braet, C., Rood, L., Timbremont, B., van Vlierberghe, L., Goossens, L., et al. (2010). Norms and screening utility of the Dutch version of the Children’s Depression Inventory in clinical and nonclinical youths. Psychological Assessment, 22, 866–877. http://dx.doi.org/10.1037/a0020593 Silverman, W. K., & Albano, A. M. (1996). Anxiety disorders interview schedule for DSM-IV child version, child interview schedule. San Antonio: The Psychological Corporation. Spence, S. H., Donovan, C., & Brechman-Toussaint, M. (1999). Social skills, social outcomes, and cognitive features of childhood social phobia. Journal of Abnormal Psychology, 108, 211–221. http://dx.doi.org/10.1037//0021-843x.108.2.211

794

A.C. Miers et al. / Journal of Anxiety Disorders 28 (2014) 787–794

Sumter, S. R., Bokhorst, C. L., & Westenberg, P. M. (2009). Social fears during adolescence: is there an increase in distress and avoidance? Journal of Anxiety Disorders, 23, 897–903. http://dx.doi.org/10.1016/j.janxdis.2009.05.004 Timbremont, B., & Braet, C. (2002). (Children’s Depression Inventory: Dutch version) Children’s Depression Inventory: Nederlandstalige versie. Lisse: Swets & Zeitlinger. Westenberg, P. M., Bokhorst, C. L., Miers, A. C., Sumter, S. R., Kallen, V. L., van Pelt, J., et al. (2009). A prepared speech in front of a pre-recorded audience: subjective, physiological, and neuroendocrine responses to the Leiden Public Speaking Task. Biological Psychology, 82, 116–124. http://dx.doi.org/10.1016/j. biopsycho.2009.06.005 Wetter, E. K., & El-Sheikh, M. (2012). Trajectories of children’s internalizing symptoms: the role of maternal internalizing symptoms, respiratory sinus

arrhythmia and child sex. Journal of Child Psychology and Psychiatry, 53, 168–177. http://dx.doi.org/10.1111/j.1469-7610.2011.02470.x Whiteside, S. P. H., Gryczkowski, M., Ale, C. M., Brown-Jacobsen, A. M., & McCarthy, D. M. (2013). Development of child- and parent-report measures of behavioral avoidance related to childhood anxiety disorders. Behavior Therapy, 44, 325–337. http://dx.doi.org/10.1016/j.beth.2013.02.006 Wittchen, H.-U., & Fehm, L. (2003). Epidemiology and natural course of social fears and social phobia. Acta Psychiatrica Scandinavica, 108, 4–18. http://dx.doi.org/10.1034/j.1600-0447.108.s417.1.x Woody, S. R., Chambless, D. L., & Glass, C. R. (1997). Self-focused attention in the treatment of social phobia. Behaviour Research and Therapy, 35, 117–129. http://dx.doi.org/10.1016/s0005-7967(96)00084-8