Anxious Solitude/Withdrawal and Anxiety Disorders: Conceptualization, Co-occurrence, and Peer Processes Leading Toward and Away from Disorder in Childhood

Abstract

This chapter contains (1) an analysis of commonalities and differences in anxious solitude and social anxiety disorder, and a review of empirical investigations examining (2) correspondence among childhood anxious solitude and anxiety and mood diagnoses and (3) the relation between peer difficulties and temporal stability of anxious solitude and depressive symptoms. Findings support a diathesis-stress model in which anxious solitude forecasts symptoms of psychopathology primarily in the context of interpersonal stress. Additionally, evidence for individual and environmental factors which moderate risk for peer difficulties among anxious solitary children is reviewed. Implications for intervention are discussed.

Whereas the preceding chapters examine the relation between the temperamental construct of behavioral inhibition (BI) and social anxiety disorder (SAD) in childhood, this chapter examines relations between the affective–behavioral profile of anxious solitude/withdrawal and SAD, as well as additional anxiety and mood disorders. This focus on connections between anxious solitude/withdrawal and SAD is important because, consistent with the principle of equifinality (see Chapter 1), it is likely that there are multiple pathways to SAD. If BI is used as the exclusive means of identifying children as at risk for SAD, then other pathways to this disorder will remain unknown. This is not to say that BI and anxious solitude/withdrawal constitute mutually exclusive pathways to SAD; indeed, BI is considered to be a risk factor for anxious solitude/withdrawal (Calkins & Fox, 2002). However, anxious solitude/withdrawal should be considered a risk factor for SAD in its own right, because some children display this profile in middle childhood without having demonstrated BI in early childhood (Asendorpf, 1990b).

Multiple sources of evidence suggest that BI contributes to the etiology of SAD in only a portion of cases. First, evidence from multiple longitudinal investigations of BI indicate that approximately half of individuals with lifetime diagnoses of SAD in middle childhood and adolescence (M age = 6–15 years) did not display BI in early childhood (Biederman et al., 1990, 1993, 2001; Hirshfeld-Becker et al., 2007; McDermott et al., 2009; Schwartz, Snidman, & Kagan, 1999). Although rates of SAD are approximately double in BI versus non-BI children, non-BI children make up a larger proportion of the population. Thus, although a lower percentage of non-BI than BI children demonstrate SAD, in extant research about equal numbers of non-BI and BI children receive SAD diagnoses. However, the majority of this literature stems from studies in which children of parents with psychopathology were overselected, so rates may differ in the general population. Second, as noted in Chapters 2 and 4, heritability for SAD is estimated at 30 to 50% (Kendler, Karkowski, & Prescott, 1999; Kendler, Neale, Kessler, Heath, & Eaves, 1992), indicating that environmental factors are also vital to understanding the etiology of SAD. Insofar as BI is assumed to contribute to genetically influenced routes to SAD (see Chapter 2), this suggests that environmental factors contribute approximately half or more of the variance in SAD. (However, it should also be noted that early environmental factors may also contribute to BI itself. One hint supporting this possibility is that only a portion of children who demonstrate BI also exhibit the expected physiological profile. For instance, of those children who initially demonstrated BI at 9 months, approximately half changed to non-BI at 2–4 years, and this group, compared to a continuously inhibited group, demonstrated both less electroencephalogram (EEG) asymmetry at 9 months and more experience with nonparental care (Fox, Henderson, Rubin, Calkins, & Schmidt, 2001). Finally, because recent epidemiological data indicate that onset of SAD frequently occurs during childhood (for a review see Rapee, Schniering, & Hudson, 2009), it is important to understand environmental contributions to SAD throughout childhood. These environmental factors are likely to exacerbate social anxiety in children with preexisting anxious tendencies. They may also trigger new social anxiety in children without preexisting tendencies, although research is needed into this possibility. Importantly, preexisting anxious tendencies are likely to be multifaceted in nature (BI, insecure attachment, goodness-of-fit between temperament and parenting, parental overcontrol, parental modeling of anxiety, etc.) and include early environmental factors (e.g., parenting factors, see Chapter 6).

Conceptual Relations Between Anxious Solitude and Social Anxiety Disorder

Anxious solitude/withdrawal was introduced in Chapter 1 as an affective–behavioral profile in which children engage in elevated rates of solitary behavior among familiar peers due to social anxiety. Specifically, anxious solitude is defined by elevated rates of solitary onlooking behavior (watching peers’ play without joining), shyness, social hesitancy, and verbal inhibition when children are among familiar peers (Coplan, Rubin, Fox, Calkins, & Stewart, 1994; Gazelle & Ladd, 2003). These behaviors and interaction patterns are correlated with other manifestations of anxiety in young children (automanipulatives such a thumb sucking, lip biting, and hair pulling), suggesting that they are indicative of social anxiety (Coplan et al., 1994), the core state conceptualized as motivating anxious solitude/withdrawal. Anxious solitary children are conceptualized as wanting to interact with their peers (having normative levels of social approach motivation), but being blocked by worry that they will perform poorly and/or be poorly received by peers (having heightened social avoidance motivation spurred by social evaluative concerns; Asendorpf, 1990a). Anxious solitude/withdrawal is typically identified by either peer nominations from classmates (e.g., Gazelle, 2008) or behavioral observations of free play (Rubin, 2001) among familiar peers at school, with teacher or childcare provider ratings being the next-best option (Spangler & Gazelle, 2009). Anxious solitude/withdrawal can be identified as early as 2 years of age (e.g., Gazelle & Spangler, 2007), but peer nominations are typically not used for this purpose until 8 years of age.

Diagnosis of SAD in childhood, as defined in previous chapters, requires that (a) when in a social situation, the child is either afraid or worries that she or he will do something embarrassing or that others will think is stupid or will laugh at; (b) at least one specific social situation is either avoided or endured with distress; and (c) that this problem interferes with the child’s life (e.g., friendships, activities at school or home, keeps child from doing things she or he wants to do) at least some of the time (American Psychiatric Association, 2000; Silverman & Albano, 1996). Diagnostic criteria also specify that the child must be anxious about interactions with other children, not just adults. The core state is conceptualized as marked and persistent fear of social situations. Diagnoses are typically made via clinical interview of the child (if the child is at least 8 years old) and his or her parents. Diagnoses of SAD can be made as early as 2 years of age based on parental-report on the Preschool Age Psychiatric Assessment (Egger & Angold, 2004; Egger, Ascher, & Angold, 1999).

Although anxious solitude and SAD share the core emotional state of social anxiety and are both identifiable by age 2, additional similarities and differences between the two constructs are relatively nuanced. The construct of anxious solitude/withdrawal emphasizes behavior. Both of the preferred methods of identifying anxious solitary/withdrawn children—peer nominations and behavioral observation—require that a child demonstrate observable manifestations of social anxiety in their behavior (e.g., onlooking solitary behavior, verbal inhibition, hesitancy, etc.). The hypothetical child who feels anxious with familiar peers but does not manifest behavioral signs would not qualify as anxious solitary/withdrawn. Notably, self-reports of anxious solitude/withdrawal are rarely used for purposes of identification in the developmental literature, in part because although self-reports of anxious solitude/withdrawal are significantly related to information from other informants, the strength of this association is modest (Spangler & Gazelle, 2009). In contrast, clinical diagnoses of SAD rely, in part, on subjective distress as reported by the child and/or their parent. Therefore, although anxious solitude/withdrawal and SAD converge in identifying children who experience social anxiety among peers, each construct is defined in ways that privilege certain aspects of the phenomenology of social anxiety (i.e., behavior versus subjective distress). In a similar vein, different informants are likely to yield the best information about different aspects of the phenomenology of social anxiety—peers and research observers have direct access to children’s interaction with their familiar classmates, whereas only the child themselves can directly assess subjective distress (but parents may serve as confidants). Thus, differences in emphasis on particular aspects of the phenomenology of social anxiety are often confounded by reliance on different informants (but for a multitrait, multimethod analysis of anxious solitude see Spangler & Gazelle, 2009).

Despite these differences in emphasis, there are nonetheless commonalities in the assessment of both constructs. For instance, behaviors characteristic of anxious solitude/withdrawal (e.g., solitary onlooking behavior) could be captured in the “avoidance of social situations” or “interference” criteria used in clinical diagnosis. Because children are often not at liberty to avoid social situations connected with schooling altogether, such behavior can be seen as avoidance demonstrated within the confines of an inescapable social situation. However, it is also possible that some anxious solitary/withdrawn children demonstrate the behaviors that qualify them for this developmental classification without meeting the interference criteria necessary for clinical diagnosis (Gazelle, Workman, & Allan, in press). In particular, anxious solitary/withdrawn children differ in the degree of impairment they experience with peer relations, with some children experiencing chronic peer mistreatment whereas others do not experience mistreatment and enjoy normative levels of friendship and peer acceptance (Gazelle, 2008; Gazelle & Ladd, 2003).

Empirical Evidence of Elevated Social Anxiety Disorder Among Anxious Solitary Children

Until recently, rates of SAD among anxious solitude/withdrawal children were unknown. Because anxious solitude/withdrawal and SAD share the core emotional state of social fear or anxiety, many researchers have assumed that anxious solitary/withdrawn children ought to be at risk for SAD (e.g., Rubin & Burgess, 2001). However, it is only recently that anxious solitude/withdrawal and SAD have been assessed within a single study (Gazelle et al., in press) because the appropriate methodologies are typically used by researchers from separate research traditions.

In a recent investigation (Gazelle et al., in press), 30% of anxious solitude/withdrawal versus 12.5% of comparison children received SAD diagnoses (a significant difference; anxious solitude/withdrawal was assessed via peer behavioral nominations in the fall of fourth grade) based on parental reports on the Anxiety Disorder Interview Schedule (Silverman & Albano, 1996) during the following spring and summer. These children also demonstrated elevation in other anxiety disorders (15.0% vs. 3.6% specific phobia—a significant effect; 15% vs. 5.4% generalized anxiety disorder, 5.0% vs. 0.0% posttraumatic stress disorder—both trends) and anxiety-related disorders (5.0% vs. 0.0% selective mutism, a significant effect). Interestingly, these SAD rates are comparable to those found in studies of BI versus non-BI children (Biederman et al., 1993; Hirshfeld-Becker et al., 2007). This is particularly striking because participants in Gazelle et al. (in press) were drawn from a community sample of children attending public school, whereas the majority of child participants recruited into the cited BI studies were children of parents with panic and depressive disorders. Factors that may have contributed to concordance among anxious solitude/withdrawal and SAD in Gazelle et al. (in press) relative to the BI studies include conducting assessments of behavioral development and psychopathology closer in time and conducting developmental assessments with somewhat older children. Nonetheless, this study suggests that a substantial minority of children who demonstrate anxious social behavior at school experience clinically significant levels of social anxiety. Because approximately 12% of children from community samples are typically identified as anxious solitary/withdrawn at any given assessment point, and 30% of these children can be assumed to experience SAD according to this investigation, approximately 3.6% (12% × 30%) of children from community samples would be estimated to have SAD (note, however, that some comparison children also received SAD diagnoses). This is a rough estimate, as this study was not designed to yield accurate population estimates. Nonetheless, it is within the estimates from epidemiological studies of the prevalence of SAD in childhood (ranging from 1.5 to 5%, see also Chapters 2 and 4, Costello & Angold, 1995; Rapee et al., 2009; Shaffer, Fisher, Dulcan, & Davies, 1996). Most important, evidence for elevated SAD among children with anxious solitude/withdrawal suggests that assessment of anxious solitude/withdrawal would be useful both in targeting children with socially anxious tendencies for preventative intervention prior to the onset of clinically significant disorder, as well as a means of identifying children from screening samples as in need of further assessment for the presence of current clinically significant disorder and early intervention.

Results of Gazelle et al. (in press) also indicate that children who demonstrated any history of anxious solitude/withdrawal during four assessment points in third and fourth grade not only evidenced elevated SAD, but also elevated major depression (10.0% vs. 0.0%, a significant effect) and dysthymia (5.0% vs. 0.0%, a marginally significant effect). This is consistent with evidence from other childhood samples indicating that anxiety typically precedes the onset of depression (Cole, Peeke, Martin, Truglio, & Seroczynski, 1998). This suggests that targeting anxious solitary/withdrawn children for early intervention could potentially prevent not only SAD, but also depressive disorders. It is not only important to prevent these disorders in their own right, but also to prevent the developmental and psychiatric consequences of such disorders. For instance, once a child experiences an episode of major depression, they are at increased risk for future episodes, and anxiety and mood disorders are known to interfere with normative social and emotional development (Rudolph, Flynn, & Abaied, 2008). Below, evidence is reviewed in support of a conceptual model of the relation between anxiety and depression and the role that peer relations processes may play in triggering depression among anxious youth.

Peer Relations and the Diathesis-Stress Model of Psychopathology in Anxious Solitary Children

Now that an empirical linkage between child anxious solitude/withdrawal and SAD has been established, investigators from both subdisciplines should be better able to benefit from research available in both developmental and clinical research traditions. Developmental research offers detailed information about anxious solitary/withdrawn children’s peer relations and the influence of peer relations on trajectories of anxiety and other symptoms of psychopathology over the course of middle childhood and early adolescence.

I have proposed a diathesis-stress model in which peer relations difficulties reinforce social anxiety and trigger symptoms of depression among anxious solitary children (Gazelle & Ladd, 2003; Gazelle & Rudolph, 2004). According to this model, anxious solitary/withdrawn children feel socially helpless. That is, they worry that negative social events such as having no one to play with at recess will occur, and they doubt their ability to cope effectively with such events. These worries or social evaluative concerns will not necessarily support the persistence of anxious solitude/withdrawal or trigger depressive symptoms over time. However, if an anxious solitary/withdrawn child experiences peer relations difficulties, particularly peer exclusion, this interpersonal stress is hypothesized to confirm and reinforce initial social concerns and trigger the transformation of the initial helpless orientation into a hopeless orientation characteristic of depression (Abramson, Alloy, & Metalsky, 1988; Alloy, Hartlage, & Abramson, 1988; Alloy, Kelly, Mineka, & Clements, 1990). Specifically, an anxious solitary/withdrawn child who is frequently excluded may not just worry that he or she may have negative social interactions with peers, but come to feel certain that he or she will have negative peer interactions and continue to feel unable to change this situation. In support of this model, anxious solitary/withdrawn children who encounter peer exclusion in the early years of elementary school subsequently demonstrate more stable elevation in anxious solitude/withdrawal over the course of middle childhood according to peer reports (Gazelle & Ladd, 2003), and heightened depressive symptoms in both middle childhood and early adolescence according to both self- and teacher-reports (Gazelle & Ladd, 2003; Gazelle & Rudolph, 2004).

The nature of peer relationships and peer interaction for socially anxious children in these developmental periods has received little attention in the clinical literature, yet peer relations appear to influence continuity and discontinuity in social anxiety. Developmental evidence suggests that peer relations can alter, for better or worse, the anxiety and depressive symptom trajectories of anxious children over the course of multiple years. Children who succeed in developing and maintaining normative peer relationships despite anxious tendencies appear to demonstrate healthy psychological adjustment, whereas anxious solitary/withdrawn children who encounter persistent peer relations problems appear to demonstrate increasing or more stable social anxiety and depressive symptoms (Gazelle & Ladd, 2003; Gazelle & Rudolph, 2004). Although there had been a long-standing debate in the peer relations literature about whether such peer effects truly indicate a unique influence on children’s adjustment, or simply reflect children’s preexisting vulnerabilities, recent analysis supports the first possibility—that peer relations make unique contributions to children’s adjustment trajectories over time (Ladd, 2006). Therefore, comprehensive models of social anxiety in childhood must include peer adversity. Although all forms of peer adversity (e.g., peer exclusion and victimization) are detrimental to children in general, and anxious solitude/withdrawal children in particular, peer exclusion (being passively ignored by peers or actively refused participation in peer activities) is an especially salient form of peer adversity for anxious solitary/withdrawn children. Research indicates a stronger relation between anxious solitude/withdrawal and exclusion than rejection (peer dislike) (Gazelle & Ladd, 2003). This is probably because peers exclude children who are vulnerable (e.g., unlikely to defend themselves) beyond the extent to which they might dislike a child (Gazelle, 2008). Similarly, naturalistic observation of anxious solitary/withdrawn children has revealed that they encounter peer exclusion (especially passive ignoring) more frequently than peer victimization (Druhen & Gazelle, 2009, April). This may occur because it is simply very easy for peers to ignore a child who infrequently approaches and maintains a distance from them. Despite the passive nature of being ignored by one’s peers, this social phenomenon appears to impact children’s psychosocial adjustment in important ways. This is perhaps because the message conveyed by exclusion is clear: “No one wants to play with me.”

Given the impact of peer relations on anxious children’s adjustment, it is critical to understand why some anxious children encounter persistent problems with peer adversity, whereas others do not. Recent research has uncovered multilevel factors that moderate anxious solitary/withdrawn children’s risk for peer adversity.

In regard to individual factors, social behaviors that children display in addition to anxious solitude/withdrawal moderate their risk for peer adversity (Gazelle, 2008). In comparison to other anxious solitary/withdrawn children, anxious solitary/withdrawn children who display attention-seeking behavior are more excluded by peers, and aggressive anxious solitary/withdrawn children are more victimized by peers. However, it is not the case that these other social behaviors are solely responsible for peer mistreatment because when anxious solitary/withdrawn versus nonanxious solitary/withdrawn children display the same behavior (e.g., attention seeking, aggression), children who display the combination of anxious solitude/withdrawal and aversive behavior encounter more peer adversity. Thus, it seems the combination of behavioral vulnerability (anxious solitude/withdrawal) and aversive behavior (attention seeking, aggression) is most likely to be met with peer adversity. This may be due to the “perfect storm” of peer dislike and being viewed by peers as unlikely to defend oneself (or to have friends who might defend oneself). This work has also indicated that anxious solitary/withdrawn children who display highly agreeable characteristics enjoy normative peer relations and participation in friendships. Peers also perceived these children to be highly intelligent. These individual characteristics were strongly associated with concurrent peer relations, and developmental analysis would also suggest that they may have roots in early experience. For instance, in another investigation, children identified as anxious solitary/withdrawn at 2 to 4½ years who encountered high versus low maternal sensitivity, subsequently were more responsive social partners during observed interaction with a friend at 4½ years, and less likely to be rejected by peers in first grade (Gazelle & Spangler, 2007). Thus, it would seem that young children with anxious tendencies can nonetheless learn social responsivity through synchronous interactions with their mother, which later translates into healthy peer relationships.

Child sex has also been found to moderate anxious solitary children’s risk for peer adversity, but the nature of this moderation differs by developmental period and outcome of interest. In early middle childhood, anxious solitary boys have been found to be more excluded than their female counterparts (Gazelle & Ladd, 2003). However, in early adolescence, no sex difference in exclusion among anxious solitary youth emerged, and anxious solitary girls versus boys demonstrate earlier elevation in depressive symptoms in response to peer exclusion (Gazelle & Rudolph, 2004).

In addition to the influence of early maternal sensitivity, it also appears that broader contextual factors can moderate anxious solitude/withdrawn children’s risk for peer adversity. For instance, children identified as anxious solitary/withdrawn by childcare providers at 2 to 4^½ years of age were significantly more likely to encounter peer rejection in first grade if the concurrent classroom climate was observed to be low in emotional support (Gazelle, 2006). Examination of the nonfamilial proximal environment on the social and emotional adjustment of anxious solitary/withdrawn children is rare and constitutes an area ripe for further investigation. Perhaps the most important implication of this research is that interventions for anxious solitary/withdrawn children should not only be aimed at the level of the individual child, but also endeavor to assess and modify the child’s environment as well.

Implications for Intervention

By linking the literatures on childhood SAD and anxious solitude/withdrawal, present findings provide a view of the social behavior and peer relations of children with SAD at school. An important implication that can be drawn from this literature is that because some children with SAD may have entrenched negative reputations among their peers that are resistant to change, it is insufficient to expect a child to overcome their social anxiety without actually directly addressing the child’s peer environment (not just how the child thinks about the environment or what interpersonal skills the child brings into the environment). Consequently, it would be recommended that clinicians consider not only treating the individual child, but also assessing their peer context and making attempts to modify this context if it is found to be aversive.

Assessments of peer context may either be conducted by using peer behavior nominations such as those employed in many of the studies reviewed above, observing the child’s peer interactions at school (particularly at recess or lunchtime when children’s interactions are not structured by the teacher), or at least asking the child’s teacher for their assessment of the child’s peer relations (teachers typically have more opportunity to observe this than do parents). Clinicians should be especially attentive to the degree that a child is excluded or victimized by his or her peers, as these experiences are known to contribute to symptoms of psychopathology (Gazelle & Ladd, 2003; Gazelle & Rudolph, 2004).

In regard to modifying an adverse peer context, several approaches are possible. First, the clinician could time individual intervention so that it capitalizes on annual school break and the reshuffling of children upon entry to the next grade. This naturalistic reorganization of children’s social structure may provide an increased opportunity for a fresh start despite some familiarity among classmates. Another potentially complimentary strategy is to work with classroom teachers to create a classroom climate that promotes social inclusion. A thoughtful case study of this sort of classroom-level intervention in kindergarten is available in the book, You Can’t Say You Can’t Play (Paley, 1992), although empirical work is needed to evaluate such programs. Such a universal intervention ought not only benefit an anxious child with peer difficulties in particular, but also benefit all children in the class. Finally, is important to assess the degree to which a child’s peer relations may have changed during and following such intervention efforts by conducting follow-up assessments. Overall, a developmental science framework in which childhood social anxiety is conceptualized as the product of dynamic interaction among multilevel factors within the child and in their interpersonal environment implies that intervention should also be directed at multiple levels within the child and their interpersonal environment.