Future Directions for Research on Early Intervention for Young Children at Risk for Social Anxiety

Abstract

Anxiety disorders are common among young children, with earlier onset typically associated with greater severity and persistence. A stable behaviorally inhibited (BI) temperament and subsequent shyness and social withdrawal (SW) place children at increased risk of developing anxiety disorders, particularly social anxiety. In this Future Directions article, we briefly review developmental and clinical research and theory that point to parenting and peer interactions as key moderators of both the stability of BI/SW and risk for later anxiety, and we describe existing interventions that address early BI/SW and/or anxiety disorders in young children. We recommend that future research on early intervention to disrupt the trajectory of anxiety in children at risk (a) be informed by both developmental science and clinical research, (b) incorporate multiple levels of analysis (including both individual and contextual factors), (c) examine mediators that move us closer to understanding how and why treatments work, (d) be developed with the end goal of dissemination, (e) examine moderators of outcome toward the goal of treatment efficiency, (f) consider transdiagnostic or modular approaches, (g) integrate technology, and (h) consider cultural norms regarding BI/SW/anxiety and parenting.

Anxiety disorders are among the most common disorders of childhood, with 50% of cases appearing before age 6 (Dougherty et al., 2013; Egger & Angold, 2006; Merikangas et al., 2010). Youth with anxiety disorders are at risk for persistent anxiety and comorbid disorders (e.g., depression, substance use) into adolescence and adulthood (Davies et al., 2016; Frenkel et al., 2015; Hussong, Jones, Stein, Baucom, & Boeding, 2011), with early onset of anxiety often indicating a more chronic and severe course (Ramsawh, Weisberg, Dyck, Stout, & Keller, 2011). Given these findings, efforts to identify risk factors present very early in development can have significant impact in reducing individual and societal burden.

Behavioral inhibition (BI) is a temperamental predisposition to experience negative affect and withdraw in the face of novel situations, objects, and people. Extreme BI can be reliably identified as early as 4 months of age in approximately 15% of infants (Fox, Henderson, Marshall, Nichols, Ghera, 2005; Kagan, 1997). In infancy and toddlerhood, BI is typically measured in the laboratory by observing the child’s reactivity to novel nonsocial and unfamiliar social stimuli (e.g., unfamiliar room, mechanical robot, adult stranger; Fox, Henderson, Rubin, Calkins, & Schmidt, 2001; Kagan, Reznick, & Snidman, 1987). Early assessments of infant reactivity and toddler BI are suggested to be antecedents of later childhood social withdrawal (SW) in familiar social contexts (Rubin, Coplan, & Bowker, 2009). The related term shyness has been used to refer to inhibition in response to novel social situations, more specifically. In infancy and early childhood, shyness is elicited by feelings of distress when confronted by unfamiliar people. To some, such behavior serves an adaptive purpose in that it removes children from situations they perceive as discomforting and dangerous (Schmidt & Buss, 2010); however, the extreme response of some children to novel situations far exceeds any potential danger and often leads to SW. The term social withdrawal refers to consistent (across time and situations) display of solitary behavior when encountering both familiar and/or unfamiliar peers (Rubin et al., 2009). The developmental cascade from BI to shyness and SW, in combination with maladaptive interactions with important others in these children’s social worlds (e.g., parents and peers), can place them on a developmental trajectory leading to anxiety disorders, particularly social anxiety.

Across studies within the developmental and clinical literatures, BI is considered a relatively stable trait (Degnan & Fox, 2007). However, evidence also suggests that environ-mental factors can impact the stability and links among BI and SW over time. Experiences such as being in nonparental care (i.e., daycare) and having a supportive best friendship are associated with the discontinuity of BI/SW; on the other hand, exposure to particular parenting behaviors (e.g., low warmth and autonomy granting, overprotection) predicts the continuity of BI/SW (Hastings, Rubin, Smith, & Wagner, in press; Rubin, Bowker, Barstead, & Coplan, in press).

Important to note, stable BI/SW has been found in many longitudinal studies to predict the development of anxiety disorders, particularly social anxiety disorder (e.g., Chronis-Tuscano et al., 2009). Fortunately, as noted, not all children who display BI/SW develop anxiety disorders. For instance, in our research, only one third of children who demonstrated stable high BI across infancy and toddlerhood had a current anxiety disorder in adolescence, and slightly more than 50% met criteria for a lifetime anxiety disorder (Chronis-Tuscano et al., 2009). Thus, at best, half of children with this robust risk factor go on to develop an anxiety disorder. This is good news! Elucidating predictors of the stability of BI/SW, as well as moderators of the relations between BI/SW and later anxiety, can inform intervention efforts to interrupt the progression from BI/SW to anxiety in at-risk children.

In this Future Directions article, we summarize developmental science and clinical research on BI/SW and anxiety disorders in young children, describe existing interventions, and discuss what we view as key directions for future research on the topic of early interventions to prevent the development of anxiety among at-risk children and/or reduce burden or disability in young children already meeting diagnostic criteria for anxiety disorders.

THEORETICAL FRAMEWORK

Rubin and colleagues’ (2009) theoretical model of the development of BI, SW, and internalizing problems pro-vides a guiding framework for the transactional relations between BI/SW/shyness, parenting, and peer relations (see Figure 1 for an adaptation of this model). Within this model, early childhood BI is reinforced and exacerbated by children’s reciprocal interactions with their parents. Children high in BI “pull for” their parents to rescue them when they are in distress upon encountering new situations or people (referred to in the clinical literature as “parental accommodation”; Lebowitz et al., 2013). Parents instinctually wish to alleviate their child’s distress; this tendency holds evolutionary value when true danger is present. However, over time, parents of BI/SW children come to perceive their children as vulnerable in almost any new situation or upon encountering unfamiliar adults and peers (Coplan, Reichel, & Rowan, 2009) and, as a result, respond to them in an overly protective, directive, and controlling manner (Degnan, Henderson, Fox, & Rubin, 2008; Hane, Cheah, Rubin, & Fox, 2008). They may “step in” and either refrain from encouraging their children to engage in developmentally important social opportunities (e.g., birthday parties, extracurricular activities) or speak/act for their child in uncomfortable social situations (e.g., introducing oneself, ordering in a restaurant) in an effort to protect their child from distress. Such avoidance is main-tained by negative reinforcement processes in which the child’s anxiety decreases, strengthening and maintaining the child’s anxiety and avoidance; seeing the child’s dis-tress alleviated is also negatively reinforcing for the parent, who may be quicker to accommodate the child’s anxiety by allowing avoidance the next time as a result. Over time, these children become increasingly dependent on their parents in a developmentally aberrant manner and feel unable of handling anxiety-provoking social situations on their own (Gazelle & Ladd, 2003).

FIGURE 1.

Developmental-transactional model of child behavioral inhibition and anxiety. Adapted from Rubin et al. (2009).

In line with this theoretical model, parenting is one modifiable factor that has been found in many studies to both predict the stability of and growth in BI/SW across development (e.g., Booth-LaForce et al., 2012) and to moderate risk for later anxiety among children displaying stable BI/SW (Lewis-Morrarty et al., 2012). Indeed, mothers of socially anxious children are more protective, directive, and intrusive than mothers of typical children (Hudson & Rapee, 2004; Moore, Whaley, & Sigman, 2004), and this style of parenting prospectively predicts later anxiety (Hudson & Rapee, 2000; McLeod, Wood, & Weisz, 2007). Moreover, parental accommodation, in which parents adjust their behavior to help their child minimize distress when faced with feared stimuli or situations (Lebowitz et al., 2013), is associated with more severe child anxiety symptoms and functional impairment in cross-sectional studies (Storch et al., 2015). Meta-analytic findings suggest a medium effect (d = .58) for the association between maternal overcontrol and childhood anxiety, and a larger effect (d = .76) for social anxiety specifically (Van Der Bruggen, Stams, & Bögels, 2008).

On the other hand, when parents respond to BI/SW toddlers and preschoolers with appropriate warmth and sensitivity, their children may be placed on a healthier developmental trajectory (Coplan, Arbeau, & Armer, 2008). For instance, Hane and colleagues (2008) reported that preschoolers’ socially reticent behavior remained stable until 7 years of age if their mothers were observed as lacking in warm, supportive behavior, whereas maternal warmth was associated with discontinuity in socially reticent behavior from ages 4–7 years. At the same time, too much warmth and/or sensitivity is also not adaptive in that children benefit most from gentle encouragement from parents to approach new situations and people (Kiel, Premo, & Buss, 2016).

Further contributing to this clinical picture, approximately one third to two thirds of children with elevated BI have a parent with an anxiety or mood disorder (e.g., Hirshfeld, Biederman, Brody, Faraone, & Rosenbaum, 1997), and having a parent with an anxiety disorder increases the risk of children with moderate to high levels of BI developing an anxiety disorder themselves (Wichstrøm, Belsky, & Berg-Nielsen, 2013). Anxiogenic (i.e., restrictive and low autonomy grant-ing) parenting behaviors are thought to be mechanisms by which high-BI young children of parents with anxiety disor-ders are at increased risk of developing anxiety disorders (Murray, Creswell, & Cooper, 2009). Although meta-analytic findings suggest a weak association between parent anxiety and parental control overall (d = .08), larger effects have been found for studies in which both parental control and autonomy granting were examined and for studies with higher measurement quality (d = .25–.49; Van Der Bruggen et al., 2008).

Rubin et al.’s (2009) transactional model holds that childhood BI/SW is also reinforced by children’s reciprocal interactions with their peers (Figure 1). The socially avoidant behaviors of shy, withdrawn young children often do not allow the child to develop age-appropriate social skills and social self-efficacy that derive from repeated exposure to peer situations (Bohlin, Hagekull, & Andersson, 2005; Gazelle & Druhen, 2009; Rubin et al., in press). For instance, in comparison to their typically developing peers, preschool-age children who exhibit inhibited and shy/withdrawn behavior display less social competence when interacting with their peers (Bohlin et al., 2005; Chen, DeSouza, Chen, & Wang, 2006; Nelson, Rubin, & Fox, 2005). Moreover, shy-withdrawn children rarely initiate contact with peers and take longer to initiate conversations with their typically developing counterparts (Coplan et al., 2008; Crozier Perkins, 2002). These social deficits persist with age, as BI/SW at 4 years of age has been shown to negatively predict social competence at school age (Bohlin et al., 2005). Withdrawn kindergarteners and early elementary-school-age children also employ fewer social problem-solving strategies and are less successful in their attempts to initiate social interactions (Nelson et al., 2005; Stewart & Rubin, 1995). This lack of social skill and success serves to maintain and/or exacerbate the child’s anxiety and social discomfort over time.

Moreover, when these children exhibit poor social skills while attempting to interact with their peers (or avoid interacting with peers altogether), they are more likely to be perceived by peers as potentially “easy targets” for victimization (Rubin, Wojslawowicz, Rose-Krasnor, Booth-LaForce, & Burgess, 2006) and thus may be rejected and excluded by them (Ladd, Kochenderfer-Ladd, Eggum, Kochel, & McConnell, 2011). Subsequently, these children feel poorly about themselves and withdraw more from their peers over the course of elementary school (Bukowski, Laursen, & Hoza, 2010; Gazelle & Ladd, 2003; Rubin et al., in press). This negative self-regard further contributes to the emergence of later anxiety and internalizing problems more broadly and ultimately reinforces anxious social behaviors associated with BI/SW (Nelson et al., 2005; Rubin et al., in press). Thus, encouraging productive social inter-actions among peers early in development for children classified as BI/SW may prevent the maladaptive cascade of peer difficulties initiated by social withdrawal and avoidance, and in turn reduce risk for the development of anxiety and other internalizing problems.

TREATMENT IMPLICATIONS

These transactional influences of child BI, parent, and peer interactions on the development, maintenance, and exacerbation or amelioration of children’s inhibited behaviors provide a compelling rationale for intervening with both parents and children in treatment to redirect children high in BI/SW off the trajectory to anxiety (Burgess, Rubin, Cheah, & Nelson, 2005; Hirshfeld et al., 1997; Hudson & Rapee, 2000; Lewis-Morrarty et al., 2012). Moreover, we argue that treatment for the child should be conducted within the peer context because this is precisely where children high in BI/SW are most affected and because (as reviewed herein) unsuccessful peer interactions have been shown to contribute to the worsening of anxiety, depression, and self-regard among children with BI/SW (Coplan, Schneider, Ooi, & Hipson, 2018). Consistent with this approach, Beidel and Turner incorporate anxious and non-anxious peers in Social Effectiveness Therapy for Children and Adolescents (Beidel, Turner, & Morris, 2000), suggesting that peers can be utilized in group interventions for social anxiety across the developmental spectrum. In light of the established link between early, stable BI/SW (mea-sured via observations as early as 4 months of age; Kagan, 1997) and later anxiety, we also argue that parenting interventions emphasizing gentle encouragement of approach behaviors in anxiety-provoking situations should begin as early as possible to interrupt the stability of BI and prevent associated social and emotional consequences.

Although there have been some efforts to develop and test prevention/early intervention programs for young children with high BI/SW and/or anxiety disorders, most intervene with parents only (e.g., Cool Little Kids, Rapee & Jacobs, 2002; Rapee, Kennedy, Ingram, Edwards, & Sweeney, 2010; BRAVE-ONLINE, Donovan & March, 2014) or with individual parent–child dyads (e.g., Being Brave, Hirshfeld-Becker et al., 2010; Parent–Child Interaction Therapy for Separation Anxiety [PCIT for SAD], Pincus, Eyberg, & Choate, 2005; Coaching Approach Behavior & Leading by Modeling, Puliafico, Comer, & Pincus, 2012). Several of these programs share a similar theoretical orientation (i.e., behavioral or cognitive-behavioral), content, and sequencing, and some are adapted from the same or similar programs (e.g., Coping Cat, Kendall, 1994; Parent–Child Interaction Therapy, Eyberg et al., 2001).

Among the most promising of programs involving both parents and children is Hirshfeld-Becker et al.’s (2008) Being Brave, loosely modeled after Coping Cat. Being Brave operates by targeting maladaptive parenting behaviors understood to contribute to child anxiety, such as actively encouraging avoidance and modeling unhelpful coping behaviors. Being Brave has been found to reduce child anxiety disorders com-pared to a waitlist control group (WLC); however, child BI moderated treatment effects such that children classified with elevated BI did not benefit from Being Brave as much as children with lower BI (Hirshfeld-Becker et al., 2010). Moreover, although Being Brave has numerous strengths and strong empirical support, it does not involve peers in treatment and is composed of up to 20 sessions, making it much lengthier than other programs for young children with BI/anxiety.

PCIT for Separation Anxiety Disorder (SAD) (Pincus et al., 2005) focuses on restructuring parent–child interaction patterns to increase warmth, decrease overcontrol, and decrease avoidance of separation situations. Results from a pilot study indicated that 73% of participants in the intervention group did not meet diagnostic criteria for SAD at posttreatment com-pared to WLC, in which all children retained their clinical diagnoses (Carpenter, Puliafico, Kurtz, Pincus, & Comer, 2014). Impressively, the majority of participants receiving PCIT for SAD maintained treatment gains at 3-month follow-up. However, PCIT for SAD does not target anxiety in nonseparation situations, nor does it involve in vivo coaching during exposure practice. CALM (Puliafico et al., 2012), a promising later adaptation of Pincus’s intervention, incorporates in vivo exposures with parent coaching (though not in the peer context) but awaits testing in an RCT.

Cool Little Kids (Rapee & Jacobs, 2002) is a six-session parent-only program that was designed to prevent anxiety in young children classified as BI.¹ In addition to providing psychoeducation about anxiety and targeting anxiogenic parenting behaviors, Cool Little Kids specifically addresses strategies to manage parental anxiety. Cool Little Kids has been shown to have long-term effects, including fewer child anxiety disorders and parent-reported anxiety symptoms at 3 years posttreatment compared to a no-treatment control group (Rapee et al., 2010). Yet a sizeable proportion of children retained or developed new anxiety disorders.

Other existing programs focus exclusively on the child by intervening with shy/withdrawn young children in their peer group. One such example is the Social Skills Facilitated Play program (SSFP; Coplan, Schneider, Matheson, & Graham, 2010). SSFP targets peer interactions in vivo, teaching specific social skills in a developmentally appropriate manner and scaffolding peer interactions during free play and organized activities. SSFP has been shown to increase observed social competence and reduce anxious and inhibited behaviors compared to a WLC group, but no group differences have been found on teacher reports of social behavior. More recent quasi-experimental research combining SSFP with Cool Little Kids showed greater improvements in clinician-rated anxiety outcomes for combined SSFP–Cool Little Kids but no differences between the combined condition and Cool Little Kids alone on maternal-reported anxiety outcomes (Lau, Rapee, & Coplan, 2017).

The Turtle Program: PCIT for Young Children Displaying Behavioral Inhibition is a brief yet comprehensive early intervention consisting of eight weekly concurrent parent and child group sessions adapted from PCIT, Being Brave, and SSFP (Danko, O’Brien, Rubin, & Chronis-Tuscano, in press). A unique feature of the Turtle Program is that the child group allows for in vivo coaching of parents with their child within the very peer contexts in which these children display inhibition (e.g., show-and-tell, parties). This developmental approach stands in contrast to other interventions for young children, some of which attempt to apply a downward extension of cognitive-behavioral therapy (CBT), which has an established evidence base for older children, adolescents, and adults with anxiety. Thus, in line with our transactional model (Figure 1), this intervention simultaneously addresses child BI/SW and parent–child and peer interactions to influence the developmental course of anxiety in this high-risk group.

The parent component of the Turtle Program was modeled after PCIT for SAD (Puliafico et al., 2012), with two primary exceptions: (a) exposures occur within treatment so that parents can be coached in vivo and (b) the group format. The Turtle Program child group is modeled after SSFP and involves a brief “circle time” in which social skills training is delivered, and there is free play and group activities, during which group leaders encourage social interactions among peers and practice of these newly acquired social skills (Coplan et al., 2010). In the pilot investigation of the Turtle Program, children randomized to the treatment condition demonstrated significant improvements on maternal-reported BI and anxiety symptoms, teacher-rated school anxiety symptoms, and observed maternal positive affect/sensitivity compared to WLC (Chronis-Tuscano et al., 2015). Important to note, treatment effects generalized to the classroom setting as evidenced by an increased frequency of observed classroom social interactions with, and initiations toward, peers for children who received the Turtle Program (Barstead et al., 2017).

FUTURE DIRECTIONS

Fortunately, in recent years there has been increasing research on some very promising early interventions for young children presenting with anxiety disorders and those who are at risk for anxiety by virtue of being behaviorally inhibited and socially withdrawn. Moving forward, the following recommendations can be made for how to continue to advance the field with the ultimate goal of reducing the individual and societal burden associated with anxiety disorders and their developmental sequelae.

Integrate Developmental Science and Clinical Psychology Literatures

First, intervention development must take into account both the developmental and clinical child psychology literatures. In line with a developmental psychopathology framework (Hayden & Mash, 2014), investigators need to incorporate foundational knowledge regarding developmental norms and competencies, the importance of parents versus peers at various stages of development, situations that are most relevant at a particular stage of development and for children with a certain clinical presentation (e.g., for preschool-age children with BI/SW, “show-and-tell,” birthday parties) and the cognitive capabilities of children of a certain age to ensure that our treatments are developmentally grounded (Holmbeck, Devine, & Bruno, 2010). For example, Kendall and colleagues have written about developmental considerations in adapting Coping Cat for younger children and adolescents (Beidas, Benjamin, Puleo, Edmunds, & Kendall, 2010).

However, we need to go well beyond this to understand the correlates, concomitants, and consequences of the phenomenon we are studying, which may not be referred to in the same terms across disciplines. Research on young children who have the tendency to experience distress and withdraw in the face of both novel and familiar social situations has been referred to across the developmental and clinical literatures as “behaviorally inhibited,” “shy,” “socially withdrawn,” “socially reticent,” and “socially anxious.” As noted in our introductory comments, these terms are not synonymous (e.g., some of these constructs are examined in unfamiliar, laboratory-based contexts, whereas others are measured specifically in familiar social contexts) but are most certainly overlapping (Rubin et al., in press). The same is true for research on parenting characterized by “overcontrol,” “intrusiveness,” “oversolicitiousness,” “restrictiveness,” “low autonomy granting,” “accommodation,” or “anxiogenic parenting.” We must consider research on all of these related constructs in order to achieve a comprehensive read of this literature.

Clarifying “intervention” versus “prevention” terminology when describing programs and reporting outcomes is also needed in the scientific literature. Prevention studies may report a treatment effect (i.e., child symptoms were reduced compared to a control condition) rather than a prevention effect (i.e., child symptoms did not worsen over time for the prevention group relative to a control group). Programs that aim to both treat and prevent disorders and symptoms may be more accurately described as “early intervention programs.”

Taking a developmental science approach also means that we must integrate that which is known about the course of BI/SW and anxiety over time when designing our treatment studies. We know that (despite popular belief) BI/SW is not always stable over time but that when it is, children are at increased risk for later anxiety and other forms of psychopathology (e.g., depression, substance use; e.g., Caspi, Moffitt, Newman, & Silva, 1996; Frenkel et al., 2015). In addition, we know that early onset of anxiety is associated with a more chronic and severe course (Ramsawh et al., 2011). Thus, we would be remiss to design treatment studies to simply examine outcomes at posttreatment or after a brief follow-up period. Longer follow-up periods have been examined in several seminal treatment and prevention studies for elementary-age children with anxiety disorders (e.g., Barrett, Farrell, Ollendick, & Dadds, 2006; Ginsburg et al., 2014; Kendall, Safford, Flannery-Schroeder, Webb, 2004) but are clearly also needed in early intervention studies for young children at risk for or with anxiety.

As a brilliant example of such examination of treatment effects across development, Rapee (2013) conducted a follow-up of Cool Little Kids 11 years postintervention when children reached middle adolescence and found that girls in the Cool Little Kids condition demonstrated fewer internalizing disorders and maternal-reported anxiety symptoms compared to the control group, but there were no significant differences for boys. Examining longer term outcomes of early intervention programs for BI/anxiety clarifies whether participants are protected from developing internalizing disorders in the future (“sleeper effects”) and can thus elucidate the impact of mediators of intervention effects on develop-mental trajectories.

Examine Multiple Levels of Analysis

Second, although a multimethod assessment approach is always recommended in studies of children’s mental health (De Los Reyes et al., 2015), it is not always the case that studies involve gathering information in the settings that are “at the heart of” a given phenomenon. For early intervention for children with BI/SW, examination of the social (i.e., peer) context is critical to understanding targets for intervention and to assessing whether intervention is having the desired effects on real-world functioning (Coplan et al., 2018). Surprisingly few of the existing studies reviewed herein involved the collection of teacher ratings and/or observational measures assessing social behavior at school (see Chronis-Tuscano et al., 2015; Coplan et al., 2010, for exceptions). Such data collection is hypothesis generating and absolutely necessary to assess the functional impact of intervention.

Another aspect of the child’s social context, parental psychopathology (in this case, anxiety disorders; Murray et al., 2009), is an important contextual consideration in future intervention studies for BI/SW and anxiety in young children. Although the literature on parental anxiety as a predictor or moderator of intervention effects for children with anxiety disorders is mixed (e.g., Hirshfeld-Becker et al., 2010; Kendall, Hudson, Gosch, Flannery-Schroeder, Suveg, 2008), researchers who have studied preschool-age children have not examined the effect of specific parental anxiety disorders (e.g., social anxiety) on child treatment outcomes. Indeed, many parents with anxiety struggle to parent effectively in contexts specific to their anxiety (e.g., Murray et al., 2012); thus, these parents may need a more comprehensive, “hands-on” intervention approach to overcome their parenting difficulties in situations that provoke anxiety in both the parent and child. It also may be that parent anxiety impacts longer term intervention outcomes for young children, affecting maintenance of treatment gains or the development of psychopathology over time.

Although many studies of parental psychopathology have focused on mothers, more recent research has examined the impact of paternal anxiety and paternal parenting on child anxiety. Evidence suggests that, compared to mothers, paternal anxiety may be related to different anxiogenic parenting behaviors and may play a significant role in the development of child anxiety and treatment outcomes (Liber et al., 2008; Teetsel, Ginsburg, & Drake, 2014). These findings underscore the importance of examining the unique impact of paternal psychopathology and parenting on early intervention outcomes.

Multimethod assessment should also consider the vast literature on biomarkers associated with stable BI/SW, including right frontal EEG asymmetry, greater autonomic reactivity, elevated morning cortisol levels, heightened startle responses, and more vigilant attention styles (see Fox et al., 2005). For instance, child elevated heart rate reactivity has been associated with BI/SW in the developmental literature (e.g., Schmidt, Fox, Schulkin, & Gold, 1999), and lower vagal tone has been shown to predict maternal over-control (Kennedy, Rubin, Hastings, & Maisel, 2004; Root, Hastings, & Rubin, 2016); thus, child and/or parent physio-logical reactivity and regulation may predict or moderate treatment effectiveness. For example, in a study of anxious school-age children treated with CBT, researchers found that baseline child skin conductance level moderated child anxiety symptom decline at 1-year follow-up (i.e., higher skin conductance was associated with less anxiety symptom improvement; Dieleman, Huizink, Tulen, Utens, & Tiemeier, 2016). The extent to which parents become physiologically aroused during a child stressor task may be related to their tendency to be overprotective outside the laboratory (Kennedy et al., 2004; Root et al., 2016). Assessing parent, in addition to child, physiology is in line with a developmental-transactional model (Figure 1), whereby parental distress has a bidirectional relationship with child distress. Moreover, collection of these measures at multiple time points can allow for the examination of whether change in child or parent physiology may result from treatment. For example, in the externalizing literature, researchers have found that increased use of PCIT skills was significantly related to improved child emotion regulation (measured via respiratory sinus arrhythmia suppression) at posttreatment (Graziano, Bagner, Sheinkopf, Vohr, & Lester, 2012).

Examine Mediators of Intervention Effects

The majority if not the entire literature on interventions for young children with BI/SW and/or anxiety has involved comparing the investigational intervention to a no-treatment or WLC group. Thus, if intervention effects are found, we cannot know for sure if any treatment is simply better than no treatment; we are unable to isolate specific aspects of treatment (or treatment components) that result in change. In other words, we cannot know for sure if our treatments are working in the way we theorize that they are working. Going forward, as a first step, treatments for young children with BI/anxiety should compare the investigational inter-vention to an active control group to be sure that there is something specific to the intervention that has led to change.

The next step in understanding how our treatments for young children with BI/anxiety work is to conduct formal tests of mediation (Carper, Makover, & Kendall, 2017). To do so, potential mediators should be measured during (not simply before and after) treatment to establish temporal precedence (Kazdin, 2007). In line with the National Institute of Mental Health’s emphasis on “experimental therapeutics,” it will be important to show that the investigational treatment is “moving the target mechanism” (e.g., parental overcontrol, social participation, physiological indices of emotion regulation) and that moving the target mechanism results in clinical benefit (e.g., alleviating impairment/distress for the child, improving developmental competencies and participation in age-appropriate social activities; Gordon, 2017). Such research can also advance developmental theories about the processes or mechanisms leading from BI/SW to anxiety (Kazdin, 2007).

Keep Treatment Simple with the End Goal of Dissemination

To begin to disseminate and implement interventions in community settings, delivery by individuals in these settings needs to be sustained over time without the support of federal grant funding and a large research team. Rapee’s Cool Little Kids is an excellent example of a brief program that has been found to reduce child anxiety disorders and parent-reported anxiety symptoms and has demonstrated effects up to 11 years after treatment ended (Rapee, 2013; Rapee, Kennedy, Ingram, Edwards, & Sweeney, 2005). Yet this intervention is exceedingly simple, requiring only basic training in CBT and space needed to run a parent group (e.g., in a community center, library, or school), in comparison to the Turtle Program or other PCIT adaptations, which require specialized training, concurrent parent and child group leaders, technology, and the appropriate space. SSFP (Coplan et al., 2010) is another intervention that is not too resource intensive and can be implemented by trained educational staff, allowing it to be disseminated easily. A future research agenda is to elucidate for whom these simpler, efficient, and easily disseminated interventions are effective, and in contrast which children and families require more intensive approaches delivered by more skilled clinicians (see the upcoming Point 5). Stepped care (e.g., Salloum et al., 2016) or Sequential Multiple Assignment Randomization Trial (Almirall & Chronis-Tuscano, 2016) designs can inform a more personalized and efficient intervention approach.

As we have argued herein, early intervention for children who are inhibited or anxious should be delivered within the most relevant social context(s). For this reason, (pre)school is an ideal setting in which to work on approaching peers and exposure to a range of social interactions (Coplan et al., 2018; Ginsburg & Drake, 2002). Adapting early intervention programs for children with BI and/or anxiety for implementation in the school setting will pose a unique set of challenges that researchers will need to address, however. For instance, although evidence demonstrates that teachers view shy behavior in the classroom as problematic (e.g., Coplan, Hughes, Bosacki, & Rose-Krasnor, 2011; Coplan Rudasill, 2016), teachers and administrators may consider this a lower priority than disruptive behavior problems. Considering organizational and individual-level factors in dissemination efforts to school settings will be essential to yield the greatest public health impact (Aarons, Hurlburt, & Horwitz, 2011).

Examine Moderators

Building on the last point, the question of for whom we need more complex interventions versus simpler approaches is very important to consider. Existing evidence-based interventions for young children with BI or anxiety range in length from six sessions (Cool Little Kids) to 20 sessions (Being Brave). In the spirit of treatment efficiency, examination of who can benefit from simple versus more complex or lengthy interventions can extend the number of children who are able to receive quality services for problems related to BI/SW/anxiety.

Multiple types of measures (across modalities) may be considered as predictors or moderators of treatment efficacy. For instance, it may be the case that children with certain biomarkers (e.g., patterns of EEG asymmetry, physiological response to novel situations in the laboratory; Fox et al., 2005) or attention bias to threat will require a more intensive treatment approach; it is also possible that families in which one or both parents struggle with their own anxiety may require more comprehensive treatment. Other children may show a healthier developmental trajectory if they sim-ply gain exposure to peers in a daycare or preschool setting without any formal therapeutic intervention (Degnan & Fox, 2007). Knowledge of predictors and moderators of developmental trajectories can help to inform which children require intervention and which are more likely to adapt to social demands on their own. Examination of moderators will allow us to personalize treatment in a manner that is efficient and most effective for a given child in a given context with certain clinical characteristics (Paul, 1967), again with the goal of improving developmental outcomes for the largest number of at-risk children using easily deployed approaches.

Consider a Transdiagnostic or Modular Approach

Transdiagnostic programs “present a consolidated set of interventions aimed to efficiently and effectively treat multiple … problem sets simultaneously” (Chu, 2012, p. 1). Taking a transdiagnostic approach in early intervention programs targeting internalizing disorders with high-risk children may be more efficient and effective from a dissemination perspective. As previously described, early childhood BI has been shown to predict later depression and substance use disorders, in addition to anxiety disorders, in adulthood (e.g., Caspi et al., 1996; Frenkel et al., 2015). A transdiagnostic program for young children with elevated BI could thus target mechanisms in the development of anxiety and depression more broadly. Transdiagnostic programs that treat or aim to prevent internalizing disorders with elementary-school-age children have demonstrated promising results (Bilek & Ehrenreich-May, 2012; Martinsen, Kendall, Stark, & Neumer, 2016), but none have been tested with preschoolers. How a transdiagnostic early intervention would compare to a more specific approach targeting risk factors for the development of social anxiety remains to be seen.

Modular intervention approaches take components of evidence-based treatments and provide a set of modules and treatment algorithms for providers to treat patients, who often present with comorbid disorders and changing symptomatology in “real-world” settings (e.g., Weisz et al., 2012). This approach has demonstrated promising results in comparison to usual care and traditional evidence-based treatment conditions with school-age children and adolescents (Chorpita et al., 2013) but has yet to be tested with younger children and their parents. A modular intervention for young children with BI/SW/anxiety would need to consider which intervention components are necessary (e.g., exposure) and whether more resource-intensive intervention components (e.g., in vivo parent coaching, child social skills group) could be add-on modules.

Integrate Technology

Adapting and developing early interventions that incorporate technology also have the potential to improve intervention outcomes and disseminability (Comer, 2015). Work on adapting early interventions for internet delivery has demonstrated promising results (e.g., Morgan et al., 2017), and future work will need to examine for whom in-person delivery is still needed. Interventions for children with anxiety have incorporated technology such as virtual reality (e.g., Wong Sarver, Beidel, & Spitalnick, 2014) and video-teleconferencing (e.g., Cooper-Vince, Chou, Furr, Puliafico, & Comer, 2016) and demonstrated feasibility and acceptability. Determining whether this is feasible for early intervention programs for BI/SW is an important next step in this research. In addition, other technology such as apps for smartphones and tablets have been utilized in behavioral parent training for disruptive behavior disorders (Jones et al., 2014) and hold promise for parenting interventions for BI/SW/anxiety.

Integrating technology into the dissemination and implementation of early intervention programs, such as using telemedicine technology to provide remote, internet-delivered practitioner supervision (e.g., Funderburk, Ware, Altshuler, Chaffin, 2008), can improve treatment fidelity and training cost-effectiveness. Technology can also improve intervention scalability and enhance sustainability. For example, newly trained practitioners using a web-based program in behavioral parent training sessions reported significantly less time spent on session preparation and follow-up activities than those in the usual delivery condition (Self-Brown, Osborne, Rostad, & Feil, 2016). When intervention implementation is less burdensome on newly trained providers and more cost-effective for organizations, dissemination and implementation outcomes can improve. Incorporating these considerations during the initial phases of intervention design will be important to improve the real-world “reach” of early intervention programs.

Consider Culture

Finally, the vast majority of research on BI/SW, its outcomes, and related interventions comes from studies conducted in Western cultures (Rubin et al., 2009). However, the meaning and implications of BI, SW, shyness, social anxiety, and related constructs may vary across cultures (e.g., Chen, 2010; Schreier et al., 2010). For example, studies conducted in mainland China in the 1990s indicated that behaviorally inhibited and shy behaviors were viewed quite positively (as signs of humbleness, respect, and obedience) and tended to be associated with adaptive social and academic outcomes (Chen, Rubin, & Li, 1995). Indeed, in this early cross-cultural work, parents of inhibited Chinese children were more likely to be warm and supportive, whereas North American parents were more intrusive and overly protective (Chen et al., 1998). In contrast, more recent studies indicate a rather stunning reversal, with shyness now associated with negative outcomes such as peer rejection and depressive symptoms (Coplan, Liu, Cao, Chen, & Li, 2017). Chen (2010) suggested that shyness is no longer an adaptive trait in contemporary urban China because ongoing societal and economic shifts have increased the importance of competition in Chinese schools and workplaces.

Of critical importance is the need to incorporate relevant cultural aspects into intervention and treatment protocols (Hofmann, Asnaani, & Hinton, 2010). For example, as compared to their Western counterparts, Asians and individuals of Asian heritage are particularly unwilling to seek treatment for anxiety (Hsu & Alden, 2008; Lee, Lee, & Kwok, 2005). Further, within the treatment context, socially anxious individuals of Asian descent may be hesitant to express their thoughts and emotions out of concern for causing distress to others (Zhu et al., 2014). Notwithstanding, as a promising first step, Li et al. (2016) recently adapted SSFP (Coplan et al., 2010) for use with extremely inhibited kindergarten children in Shanghai. Among their results, as compared to high-BI controls, high-BI children who participated in the intervention demonstrated significant increases postintervention in observed peer interactions and prosocial behaviors at kindergarten, and these gains were maintained after 2 months.

Finally, within the United States there are large populations who have yet to be studied vis-à-vis the developmental origins, concomitants, and consequences of BI/SW. Specifically, there are virtually no studies of BI/SW young children who are growing up in stressful, dangerous community and family settings. However, there is evidence suggesting that similar to European American youth, early-onset anxiety symptoms in African American children are associated with both concurrent and long-term academic, social, and psychological difficulties (Grover, Ginsburg, & Ialongo, 2007) and that CBT delivered in high schools with African American youth with anxiety disorders results in similar improvements in anxiety symptoms, impairment, and clinical diagnoses as seen in studies conducted with White, middle-class children and adolescents (Ginsburg & Drake, 2002). However, there are few if any BI/SW studies that focus specifically on young African American children who reside in poverty. To the extent that living in a dangerous neighborhood may be associated with a more protective parenting style (Leventhal Brooks-Gunn, 2000), one might ask about the developmental course for those young children who are inhibited and withdrawn – does parental protectiveness serve a different developmental function for these children than it does for the vast majority of children who have participated in the extant developmental research on BI/SW? If it does, then clearly, our early interventions will have to be mindful of cultural and community norms and values.

CONCLUSIONS

In line with our developmental-transactional model we have presented herein (Figure 1), compelling evidence showing that an early-emerging temperamental characteristic, BI, and the subsequent shyness and social withdrawal experienced by a subset of children classified as BI can set forth a developmental cascade that contributes to the emergence of impairing anxiety disorders (particularly social anxiety). This developmental model points to interactions with parents and peers as key targets for intervention, as they both influence the stability of BI/SW and moderate risk for anxiety.²

In this Future Directions article, we recommend that future research on early intervention for young children with BI/SW/anxiety (a) be informed by developmental science and clinical research, (b) incorporate multiple levels of analysis (including both individual and contextual factors), (c) examine mediators and mechanisms that explain how and why treatments work, (d) be developed with an eye toward dissemination, (e) examine moderators of outcome toward the goal of treatment efficiency, consider a transdiagnostic or modular approach, (g) integrate technology, and (h) consider cultural norms regarding BI/SW/anxiety. Fortunately, many of these recommendations fall in line with current National Institute of Mental Health funding priorities (e.g., examination of constructs at multiple levels of analysis, personalized medicine, experimental therapeutics), which should help to make such investigations possible.

Our hope is that future research in line with our recommendations will serve to interrupt the developmental progression from early BI/SW to later anxiety, making efficient treatments more readily available in the contexts in which at-risk children are most impaired (e.g., preschool) and reserving more intensive treatments for those children and families who need them most. Attention to identified risk and protective factors, such as behavioral inhibition and social withdrawal, early in a child’s life can ultimately reduce the substantial burden of anxiety disorders on an individual and societal level.