Current and Future Approaches to Pediatric Anxiety Disorder Treatment

Abstract

This overview critically appraises literature on the treatment of pediatric anxiety disorders. The two established treatments for these conditions comprise cognitive behavioral therapy (CBT) and antidepressant medications. Many youth receiving these treatments fail to achieve remission, which creates a need for new treatments. After summarizing the literature for CBT and current medications, the manuscript describes research that lays a foundation for improvements in the treatment of pediatric anxiety disorders. This foundation leverages neuroscientific investigations also described in the overview, which provide insights on mechanisms of successful treatment.

1-. Introduction

This article describes research on three pediatric anxiety disorders: generalized anxiety (GAD), separation anxiety (SA), and social anxiety disorders (SAD). The article considers these disorders as a group, because most treatment studies target them together, given their high prevalence and frequent comorbidity. The article omits specific phobia, the fourth common anxiety disorder, which manifests lower rates of persistence and impairment as well as better long-term outcomes than the three conditions covered in the article (1). A fifth diagnosis, selective mutism, often occurs comorbid with SAD. Because large clinical trials only include children with selective mutism when they also manifest with SAD, data on treatment of SAD largely apply to the treatment of selective mutism. Similarly, panic disorder is rare before early adulthood, and large clinical trials in youth only include this condition when it occurs alongside GAD, SA, or SAD. Hence, data on treatment for these disorders also likely apply to panic disorder.

These three pediatric anxiety disorders are highly prevalent and may cause long-term impairment (2). On one hand, effective treatments exist for the disorders, and many patients manifest long-term remissions (1, 3). Longitudinal studies find that between one-third to one-half of youth with these anxiety disorders are free of psychopathology when evaluated over the next 10 years (1, 4). This provides evidence of a better prognosis than for most other forms of pediatric psychopathology, including autism spectrum disorders, attention deficit hyperactivity disorder, and major depressive disorder (5). On the other hand, even with adequate treatment, many patients remain symptomatic; other youth relapse after robust initial responses; and most adults with persistent mood or anxiety disorders previously had manifested a pediatric anxiety disorder. Thus, clinicians confront the duality of a relatively good prognosis with adequate treatment for one set of youth (1, 4), existing alongside a high risk for chronicity in sizable groups of other youth. Clinicians approaching the treatment of pediatric anxiety disorders can rely on two established treatments: antidepressants and cognitive behavioral therapy (CBT). The first two sections of this overview summarize research on these established treatments and novel approaches for improving them that may make them more widely applicable. The overview closes by briefly outlining a path for finding other innovative treatments that will likely take longer to be widely applied.

The discussion of novel approaches for both relatively immediate and longer-term improvements in treatment leverages a robust literature on the pathophysiology of anxiety disorders. As reviewed elsewhere (6–12), this literature delineates the unique opportunities for therapeutic discovery from the melding of preclinical and clinical research, given strong cross-species conservation in circuitry deployed by mammals when they confront threats. Research on pediatric anxiety disorders finds many signs of dysfunction in this threat-responsive neural circuitry. Signs of such dysfunction manifest at many levels. This includes blood-based biomarkers such as cortisol, C-reactive protein, and interleukins that are sensitive to threat exposure (13), neuropsychological and psychophysiological indicators of information processing functions that mammals deploy in the presence of threats (9), as well as measures of brain structure and function implicated in the mammalian threat response (10, 12). Rather than providing a comprehensive review, the current paper narrowly describes research on pathophysiology that provides promising targets for novel treatments. Such research relates modifiable biomarkers to the course of anxiety disorders and describes preliminary evidence that attempts to alter these biomarkers for therapeutic purposes to enhance treatment responses in pediatric anxiety disorders.

2-. Established treatments

The current treatment of anxious youth is mostly conducted in traditional clinical settings where, after the diagnosis of a disorder the current treatment consists of one of three modalities: SSRI, CBT or combination of SSRI and CBT (see Figure 1). Table 1 summarizes relevant meta-analyses published in the past five years on these modalities, and this section highlights findings for each treatment.

Figure 1:

Current treatment for individuals diagnosed with an anxiety disorder (red) consists of either medication, CBT, or a combination of both. Near term improvements in outcomes might follow by increasing availability of these treatments and by offering psychological treatment to people at-risk (‘purple’) for anxiety disorders. Current research seeks novel treatments based on their capacity to influence mammalian neural circuitry that is engaged by threats. The figure illustrates two behaviors and associated circuitry. One behavior concerns the capture of attention by a threat, such as the snake in the picture. This behavior engages the lateral prefrontal cortex, as also shown in the figure in green, and may be targeted by cognitive training as also illustrated in the figure. The other behavior appears in the bottom of the figure, where a mouse first learns that a sound predicts a shock before the mouse next extinguishes this learning. Retention of this subsequent extinction learning is represented by the associated sound that appears without the shock. Such learning requires the medial prefrontal cortex, shown in blue. This circuit could be targeted by brain stimulation, as also shown in the figure.

Table 1:

Published meta-analysis of anxiety treatment from 2019 – 2023.

Meta-Analysis	Study	N Studies	N Subjects	Age	Comparison	Effect Size	CI
CBT	James et al. 2020 (28)	87	5,964	2–19	WL/no treatment, TAU, attention control, alternative treatment, and medication	5.45c	3.90 to 7.60
	Baker et al. 2021 (29)	16	766	11–18	Active vs. passive control	0.454*b	0.22 to 0.69
CBT - Group Therapy	Sigurvinsdóttir et al. 2020 (101)	81	3,386	Under 18	WL	8.96**c	4.03 to 19.90
SSRI	Dobson et al. 2020 (18)	20	2,623	5–17	Placebo	5.2b	2.8 to 8.8***
SNRI	Dobson et al. 2020 (18)	20	2,623	5–17	Placebo	2.5b	−0.1 to 5.1
Placebo Response	Motta et al. 2023 (102)	135	12,583	N/A	SSRI/SNRI	−1.11b	−1.22 to −1.00

^*:reduction of symptoms

^**:no significant differences compared to individual-CBT

^***:credible interval. TAU: treatment as usual, WL: waitlist.

^a:Hedge’s g

^b:standard mean difference

^c:odds ratio

^d:Cohen’s d.

2.1. Pharmacology

Pharmacological treatment represents one of two established therapies for pediatric anxiety disorders. Meta-analyses provide robust evidence of efficacy for virtually all selective serotonin reuptake inhibitors (SSRIs) and selective norepinephrine reuptake inhibitors (SSNIs) in the treatment of generalized anxiety disorder, social anxiety, and separation anxiety disorders. Nevertheless, approval from the Food and Drug Administration (FDA) only exists for two medications, duloxetine and escitalopram, both for generalized anxiety disorder.

Strong evidence derives both from meta-analyses summarized in Table 1 and from notable individual randomized controlled trials (RCTs). Perhaps the most influential trial is the Child/Adolescents Anxiety Multimodal Study (CAMS) (14). CAMS included 488 children and adolescents randomly assigned to one of four groups: monotherapy with placebo, sertraline, or CBT, as well as combined CBT/sertraline therapy. Treatment with sertraline used a flexible dosing design, with a maximal dose of 200 mg/day. For the sertraline monotherapy group, the effect size was 0.45 (95% CI, 0.17 to 0.74), and the number necessary to treat (NNT) was 3.2 (95% CI, 3.2 to 3.5). These results are comparable with results for other RCTs for SSRIs (15, 16) and SNRIs (17). Meta-analytic evidence summarized in Table 1 documents larger effect sizes for SSRIs than SNRIs (3, 18), with no clear superiority for any drug within a class. Across many pharmacological studies, worse treatment response relates to higher baseline symptom severity (19) and a diagnosis of social anxiety disorder (20). Nevertheless, despite these notable beneficial effects, fewer than half of patients receiving antidepressants manifest long-term remissions. In the CAMS trial, only approximately 20% of patients receiving any treatment displayed consistently low levels of anxiety over a four-year follow-up period, with no differences among treatments, and, in the group randomized to sertraline, only between 40 and 52% of patients manifested remission in any of the four yearly follow-up assessments (4). Thus, while medications reduce symptoms in the short term, a need exists for new therapies that might produce long-term remission.

The use of pharmacological treatment carries the risk of adverse effects. Behavioral activation is one of the most common, appearing more common in SSRIs than SNRIs. It typically occurs at rates below 15%. Antidepressants carry a black box warning in their label due to an increased risk of suicide in indivduals younger than 25. This warning stems from a meta-analysis of industry sponsored clinical trials that found a moderate effect size for suicidality (21). The reported risk in the FDA analysis was higher for treatment with paroxetine or venlafaxine, as compared to other antidepressants. Other data suggest that paroxetine may exhibit a relatively high risk for thoughts about suicide (18). Despite potential differences among particular medications, clinicians should still carefully monitor suicidality in patients receiving any SSRI or SNRI.

Other classes of medication are commonly used to treat anxiety in adults, and some such medications have been investigated in children and adolescents. However, these medications should be considered second or third-line options. Tricyclic antidepressants appear less effective than SSRIs while having more concerning side-effects (18). Moreover, the medications are more lethal in overdose than SSRIs and SNRIs, making them third-line options. Similarly, benzodiazepines are prescribed due to their rapid onset of action and wealth of data in adults. Nevertheless, only three small RCTs exist in youth (22). These studies included small samples, with the largest RCT including only 30 patients treated with alprazolam (23). Additionally, meta-analyses provide no evidence of efficacy, showing a small effect size and a wide confidence interval (CI) (Odds Ratio [OR]: 1.4 95%CI, [0.3–6.1]) (18). Benzodiazepines also include problematic side effects among youth, including a risk for disinhibition and harmful impacts on cognition. Nevertheless, given the underpowered nature of existing studies and notable effects in adults, it is reasonable to consider treatment with benzodiazepines, particularly in acute scenarios where short-term use is needed.

Finally, buspirone and guanfacine represent two other medications studied in RCTs with more than 50 patients. For buspirone, one RCT with 227 patients and another RCT with 341 patients examined efficacy (24), whereas for guanfacine, one RCT with 83 patients examined efficacy (25). Across the three trials, primary efficacy analyses show no advantage for either medication. Nevertheless, both medications are well-tolerated in youth and widely used for indications other than pediatric anxiety disorders. Moreover, secondary analyses provide some suggestive evidence of efficacy for both medications. Hence, both medications are reasonable second-line options for patients who fail to respond to multiple SSRIs or SNRIs.

2.2. CBT

Cognitive Behavioral Therapy (CBT) is an established first-line treatment for pediatric anxiety disorders, manifesting effect sizes comparable to those for antidepressants (26). CBT for anxiety disorders usually targets avoidance through gradual exposure to threatening stimuli combined with psychoeducation and the teaching of coping strategies. Many RCTs examine the efficacy of manual-based CBT, usually delivered across 12 to 20 sessions in various formats, such as group or parent-focused CBT (27). The effect size for CBT alone in the CAMS trial was 0.31 (95% CI: 0.02–0.59) (14), while meta-analytic evidence finds larger effect sizes favoring CBT over waitlist or treatment as usual (table 1). Even with moderate effect sizes, between half (28) and two thirds of subjects will not achieve full remission (29). As in pharmacological treatment, higher baseline symptoms predict worse treatment response, while parental psychopathology may also predict outcome in some settings (30). Notably CBT involves lower risk for adverse effects than medication, with no serious adverse effects reported in most RCTs. As a result, it is universally viewed as at least one of two possible first-line therapies.

CBT delivered to different age and diagnostic groups involves many similarities. The treatment often begins with the creation of a fear hierarchy, where the patient lists mildly threatening scenarios on the bottom and very frightening scenarios on the top. Next, the therapist teaches the patient coping strategies and other approaches that help the patient reduce their distress when exposed to feared situations. Finally, patients undergo guided exposures across the typical two-to-three-month course of therapy, where each session involves confronting scenarios on successively higher hierarchy levels. Beyond such similarities, notable differences exist in approaches to CBT among patients with distinct clinical presentations. With separation anxiety disorder, therapists often must work closely with families to craft experiences that allow graded separations from parents, whereas in social anxiety disorder, therapists might venture out of the office with patients, to assist patients performing frightening social tasks, such as purchasing a snack in a crowded store. Unlike separation anxiety and social anxiety disorders, with GAD, patients often fear abstract scenarios, which may require the therapist to creatively devise effective exposure sessions. For example, in patients who fear failure, therapists might work with patients to purposely make minor errors in a homework assignment or perform an after-school activity below their level of proficiency. Thus, these differences in treatment reflect unique clinical scenarios.

2.3. Combination

More research is needed comparing SSRI and CBT monotherapy to each other and to combined SSRI/CBT therapy. On their own, effect sizes over control therapies across distinct studies appear comparable for SSRIs and CBT. However, such comparisons can be misleading, since medication and psychotherapy trials can recruit different types of patients and rely on different control therapies. Given comparable efficacy and more concern with adverse effects with SSRIs, some European guidelines (31) consider CBT to be the only first-line therapy, unlike some United States guidelines, which consider the two treatments to be equivalent first-line options (32). Guidelines uniformly recommend combined treatment in patients who fail to respond to monotherapy.

In terms of combined treatment, only two notable RCTs exist. First, CAMS found the combination of CBT with medication to be superior to either modality alone (14). However, patients who received combination treatment were told that they were randomized to active medication, and no patient groups received CBT with placebo augmentation. Only patients randomized to either sertraline or placebo monotherapy received blinded therapy. Hence, combined treatment’s superiority could reflect expectancy as opposed to particular effects of SSRIs. Second, consistent with this possibility, another study compared the combination of CBT with SSRI augmentation to CBT with placebo augmentation, in an attempt to equilibrate between-condition expectancy. Unlike CAMS, in this study, the outcome for youth receiving CBT with an SSRI was no better than for youth receiving CBT with placebo (26).

As a result stronger effects in CAMS could reflect design issues or differences in sample size, and more research that controls for expectancy with combination treatment is needed. Nevertheless, the effect size in the more recent study equating expectancy was small, suggesting the need for very large sample sizes in future studies (26). Regardless, when expectancies of improvement are equalized, data suggest that SSRI addition to CBT may provide no benefit over CBT. Hence, many clinicians begin treatment with CBT, and augment the treatment with medication in cases where there is no adequate clinical response after a period of approximately eight-to-12 weeks.

3-. Near-term improvements in treatment outcomes

Figure 1 highlights current treatments and approaches for improving them using approaches summarized in Table 2. We discuss three promising avenues for generating near-term improvements in the treatment of anxiety disorders. One avenue extends the reach of available treatments. A second applies precision-medicine approaches, and the final avenue leverages research on mechanisms to be targeted by novel therapies.

Table 2:

Novel treatments/approaches table. Results from meta-analysis 2019 – 2023

Meta-Analysis & RCT Trials	Study	N Studies	N Subjects	Age	Effect Size	CI
Improving treatment Access	Wakefield et al. 2021 (103)	47	636,734	Over 18	0.88d	0.79 to 0.97
ABMT	Fodor et al. 2020 (71)	65	3,879	Over 18	−0.55b	−0.91 to −0.19*
School based	Hugh-Jones et al. 2021 (104)	18	2,076	Children and Adolescents	−0.28a	−0.50 to −0.05*
	Gee et al 2020 (105)	45	1,075**	10–19	−.49*b	−0.79 to −0.19
Telehealth	Venturo-Conerly et al. 2022 (106)	37	2,054***	M = 9.38 years (SD = 4.19 years)	.62a	.16 to 1.07
Computer-Based CBT	Grist et al. 2019 (107)	34	3,113	6–18	0.66a	0.42 to 0.90
Mindfulness Based Stress Reduction	Zhou et al. 2020 (108)	14	1,489	12–25	−0.14b	−0.24 to −0.04
Parental CBT	Yin et al. 2021 (27)	6	407	Children and Adolescents	−0.72b	−1.41 to −0.03*
Low/Middle Income Countries CBT	Venturo-Conerly et al. 2023 (109)	34	4,176	M = 13.38 years (SD = 2.66 years)	0.76****a	−0.13 to 1.65

^*:outcome was the mean change, reduction of anxiety symptoms

^**:subset of anxious children

^***:total N not reported in the main manuscript; Derived from suppl Table 1.

^****:effect size of anxiety (included PTSD and OCD studies) as moderator of treatment response

^a:Hedge’s g

^b:standard mean difference

^c:odds ratio

^d:Cohen’s d.

3.1. Extending Available Treatments

Established approaches exist for maximizing the impact of established treatments. These include techniques for leveraging technology and for reducing the burden faced by primary care providers and educators. As discussed below, a final approach focuses on prevention to interrupt, early in the course of a developmental cascade, processes that can generate chronic emotional problems.

Technology provides unique opportunities for addressing the needs of youth with anxiety disorders. Considerable research suggests that technology can increase the availability of CBT and other forms of therapy. Meta-analyses of RCTs for internet-delivered therapy are summarized in Table 2; the table considers treatment of various conditions given the paucity of RCTs focused specifically on pediatric anxiety disorders. Studies summarized in Table 2 generate two conclusions (33). First, comparable levels of efficacy might be achieved by using technology to deliver CBT in patients’ homes, as compared to the clinic. Second, some level of regular therapist contact appears necessary to achieve such efficacy.

If results in future studies confirm existing research, the standard of care may one day involve having patients conduct some facets of CBT while being directly monitored by a therapist and conduct other facets on their own with the aid of technology. Extending this approach from research settings to the clinic could rapidly increase therapist availability, if other limitations in existing work could be addressed. For example, current studies often use a control group of “treatment as usual” or waitlist. Although data in adults indicate that internet delivered CBT may be as effective as face-to-face (34), more large, well designed studies are needed in children and adolescents that directly compare CBT delivered via telehealth to face-to-face therapy.

Beyond increasing availability of established treatments, technology provides other opportunities for improvement. Studies in adults suggest that therapeutic processes can be monitored using technology in ways that identify barriers to successful CBT outcomes (35). Studies in youth find that better clinical outcomes occur following specific types of therapeutic encounters, such as exposure sessions involving sustained threat confrontation (36). Future research might integratively extend these two observations by using technology to monitor therapeutic processes related to clinical outcomes in pediatric anxiety disorders. With such technology, patients, families, and clinicians might monitor the engagement of processes that predict remission. Finally, mobile technology provides other opportunities through real-time, multi-day monitoring of multiple parameters. Later sections of this review describe therapeutic research that targets mechanisms. Wearable devices may facilitate such work by providing abundant data on physiology, sleep patterns, as well as avoidance and other facets of physical activity that may reflect a treatment’s impact on a targeted mechanism.

Independent of technology, another opportunity for advancing treatment outcomes leverages patients’ regular contact with pediatricians and educators. Considerable research confirms pediatricians’ ability to identify anxiety disorders, which supports recent recommendations for in-office anxiety screening during routine well-child visits (37). Enhancing the capacity for in-office treatment delivery could improve outcomes. Among adults, CBT delivered in primary care settings, relative to a control intervention, produces greater reductions in anxiety symptoms that persists for at least 18 months (9, 38). Some research begins to extend this approach to pediatric anxiety disorders (37, 39, 40). However, large-scale RCTs are needed to test whether the embedding of services within primary care settings produces long-term reductions in pediatric anxiety as it does in adult anxiety disorders. Much like pediatricians, educators also can identify pediatric anxiety disorders, and RCTs already document the efficacy of CBT delivered in school settings (41, 42). While more work is needed to confirm efficacy comparable to traditional forms of CBT (43), offering CBT in schools might help participants access treatment while lowering the associated cost and burden. School-based treatment also could represent one component of stepped-care, where youth who fail to respond or who manifest severe problems are referred to specialized services (40).

A final approach for extending the impact of available treatments utilizes prevention. Pediatric anxiety disorders represent an intermediate step on a developmental cascade that involves early-life risks related to temperament and late-stage consequences that extend beyond anxiety into other emotional realms (44). For example, pediatric anxiety disorders strongly predict risk for mood disorders, which are more likely to persist and more difficult to treat than pediatric anxiety disorders. Hence, treatments delivered early in the course of this developmental cascade may be more effective than those delivered at later points. From this perspective, enhancing treatment for pediatric anxiety disorders can be viewed as a form of prevention for mood disorders. For universal and selective forms of prevention, this logic supports other approaches for bolstering resilience and assisting high-risk subjects.

Universal prevention targets population-level risk. To approach this goal, some studies deliver brief psychological interventions in schools or through technology by adapting features implemented in CBT and related therapies. Recent RCTs suggest that effect sizes are not large for these interventions, but a small beneficial effect applied to a large population could be important. As such, the scalable nature of these treatments may support broad delivery in ways that produce clinical impact (45–47). Other universal prevention research suggests the importance of targeting poverty, deprivation, and social stressors that accompany these community-level risks in populations of children where anxiety disorders are particularly common (48). Nevertheless, while programs of cash transfer or universal minimal income appear to benefit many aspects of mental health, weaker or less consistent effects exist for anxiety than symptoms of behavior disorders and problems with substance use or school failure (2, 49–51).

Figure 1 depicts facets of selective prevention focused on youth at-risk for anxiety. In terms of factors amenable to a selective-prevention approach, anxiety disorders have one clear targetable risk marker, that is behavioral inhibition (BI). BI manifests in the first few years of life as a tendency to react with hesitancy when confronting novel stimuli, scenarios, or people. BI strongly predicts risk for multiple anxiety disorders; associations with social anxiety disorder appear particularly robust and well-replicated (44, 52). Early work suggests that children with BI benefit from targeted intervention that reduces or prevents the worsening of anxiety and other problems (53).

3.2. Precision medicine

The availability of CBT, antidepressants, and their combination creates situations well-suited for precision-medicine approaches. Practitioners rely on impressionistic information or consensus guidelines that vary across countries to choose from among these three options. Precision medicine approaches might generate useful data-driven alternatives to guide in a more uniform manner practitioners around the world who face similar choices about treatment options (54).

Longitudinal research and investigations of response moderators in RCTs provide some guidance about predictors of pediatric anxiety-disorder outcome. Thus, high levels of severity, comorbidity, temperamental factors, and social anxiety symptoms all predict relatively poor outcomes in multiple studies (1). These and other factors identified through meta-analysis could be combined through risk calculators to create precision-medicine tools, as has been done for other mental illnesses (55). Moreover, emerging strategies of non-supervised machine learning can inform these efforts by leveraging research to cluster subjects according to high dimensional data. Cluster membership can then inform creation of risk calculators that predict treatment response to CBT or medications. Particularly useful tools would go beyond prediction of overall outcomes by differentially predicting response to CBT, antidepressants, or their combination. However, this will require additional work, since available data from large RCTs fail to identify factors that predict differential responses (56).

Multivariate analysis of brain imaging data also may inform the creation of risk calculators. This approach could leverage resting state functional magnetic resonance imaging (rsfMRI) to relate brain function to treatment response. One recent report (57) utilizes rsfMRI in adults with major depressive disorder (MDD) randomized to receive either CBT or antidepressant therapy. The report finds that distinct rsfMRI patterns predict response to these two therapies. Attempts to extend the approach to anxiety must contend with inconsistent findings for rsfMRI research in this population (58). However, multivariate analysis does relate rsfMRI to symptoms of pediatric anxiety disorders with some consistency, which sets the stage for future development (59, 60). Of note, these methods require large samples to minimize the risk of overfitting. Existing reports, such as from the ABCD study (61), show rsfMRI to be sufficiently scalable to support this work. Moreover, mega-analyses can pool individuals from multiple samples to generate data sets with greater numbers of patients than in single cohorts, even ones as large as the ABCD study (62–64). Findings from studies adopting these approaches suggest that rsfMRI, as compared to structural MRI data, may provide more accurate targets for precision medicine. These approaches suggest that rsfMRI might be used to identify children and adolescents with anxiety disorders most likely to benefit from SSRIs, CBT, or their combination. When embedded in a large RCT comparing these treatment arms, future imaging studies might convincingly relate patterns of functional connectivity to treatment outcomes. If outcomes relate to particular treatment choices, as in preliminary work on MDD (57), this could reveal how clustering according to certain models renders superior results to treatment as usual.

Beyond rsfMRI, another useful approach exploits similarities between CBT and extinction, a process where neural circuitry has been mapped in many species (12). The bottom section of Figure 1 depicts an extinction experiment conducted with a rodent. A first experimental phase involves exposure to a threat, a sound paired with a shock. Extinction occurs in the next experimental phase, when the animal learns to treat the sound as neutral, after it no longer predicts the shock. Processes engaged during exposure therapy in CBT resemble processes that allow rodents to reduce defensive response to an extinguished threat, such as the sound in the bottom of Figure 1 (65).

Research extending experiments on extinction to humans incorporates psychophysiology, electrophysiology, and fMRI markers (66). This work links pediatric anxiety disorders to heightened responding to neutral stimuli misperceived as threatening (12, 40). Recent research on extinction in large pediatric samples confirms findings from smaller samples. For example, two studies (67, 68) report positive relations between levels of physiologic responding to extinguished threats and levels of pediatric anxiety disorder symptoms, whereas a third study (69) uses fMRI to show that extinguished threats evoke stronger amygdala-ventro-medial-prefrontal-cortex (vmPFC) connectivity in youth with anxiety disorders than healthy youth. This latter study also reports age differences in amygdala-vmPFC connectivity that resemble age differences reported in other research on extinction, both in rodents and people (12). Finally, few studies of extinction examine relations with treatment (40). As in rsfMRI research, large scale RCTs are needed that incorporate measures of extinction as potential predictors of treatment response. Such a focus on extinction also informs other strategies discussed in the next section for creating novel therapies.

3.3. Targeting Mechanisms

Research on mechanisms contributing to pediatric anxiety disorders provides insights on a multi-phased approach involving scenarios depicted in Figure 1. This approach begins in cross-sectional studies relating pediatric anxiety symptoms to correlates of mechanisms presumed to initiate or sustain anxiety disorders. This is followed by longitudinal research linking future risk for anxiety to such mechanistic correlates, which then also can be targeted in open trials that alter these correlates and examine changes in symptoms. Finally, RCTs evaluate the role of such mechanisms in treatment. The availability of antidepressants and CBT, two effective therapies, informs this approach. Given children’s vulnerable state, novel treatments either must possess advantages over these therapies, be associated with relatively low risk of adverse effects, or be prioritized for use in treatment-refractory patients.

A recent review (11) discusses the need to balance the availability of effective treatment with the need to find new ones for pediatric anxiety disorders. As an example, it highlights the low risk of adverse effects as an important feature of cognitive bias modification, where the approach utilizes attention bias modification therapy (ABMT) to target threat-related attention bias. This type of attention bias occurs when threats, as compared to neutral stimuli, produce greater disruptions in attention. Children expressing high risk for anxiety or ongoing anxiety disorders manifest greater threat-related attention bias than healthy children (11, 40). Various schemes exist for reducing this bias, but the most consistent results arise for ABMT that trains children to shift their attention away from threat. While ABMT exerts replicable, statistically significant effects on anxiety, meta-analyses of stand-alone ABMT in both adults and children find modest effect sizes, typically below d=0.60 (70, 71). In addition, modest evidence of efficacy exists across three studies that added ABMT to CBT for pediatric anxiety disorders, here with small-to-medium effect sizes (6, 72, 73). Thus, evidence of efficacy exists, but increasing the magnitude of ABMT efficacy will be important.

The next section of this review more fully describes research on brain stimulation, but brain stimulation is mentioned here, as preliminary evidence suggests that it could increase the effect size for ABMT. Imaging research implicates dysfunction in the dorsolateral prefrontal cortex (dlPFC) in threat-related attention bias that occurs in pediatric anxiety disorders, as highlighted in the middle section of Figure 1 (74). By hindering patients’ abilities to maintain neural representations of their goals, dlPFC dysfunction may enable attention bias by allowing mild threats to inappropriately capture attention (75). Data suggest that CBT and ABMT partially alleviate dlPFC dysfunction in pediatric anxiety disorders (74). This therapeutic effect could be enhanced through direct dlPFC stimulation designed to further correct dlPFC dysfunction associated with pediatric anxiety disorders. Consistent with this possibility, a study in adults (75) shows that active relative to sham dlPFC stimulation during ABMT produces greater changes in attention.

While examined less extensively than threat-related attention bias, other forms of cognitive dysfunction also correlate with pediatric anxiety symptoms in ways that provide additional targets for novel therapies. For example, children with anxiety disorders manifest elevated activity in the ventrolateral prefrontal cortex and associated interconnected areas that comprise the brain’s so-called ventral attention network. This anxiety-related enhancement in brain activity occurs when generic, non-emotional, salient stimuli involuntarily capture attention (11). Thus, pediatric anxiety disorders involve a hypersensitivity to distractors on both neural and behavioral grounds that could be reduced through cognitive training designed to alleviate this hypersensitivity. Similarly, Craske and colleagues (76) suggest that deficient inhibitory control, which occurs in pediatric anxiety disorders (77, 78), impedes CBT response by hindering the extinction of conditioned fear. Preliminary work in adults suggests that cognitive training regimens capable of enhancing inhibitory control could improve response to CBT (79).

Research in adults uses many approaches to target mechanisms and improve treatment responses. This research informs studies of pediatric anxiety disorders, since more work in adults than in youth uses pharmacotherapy to target mechanisms. One recent review (80) describes compounds used in this manner, including oxytocin, cortisol and other modulators of hypothalamic-pituitary-adrenal-axis function, as well as d-cycloserine and other medications that modulate glutamate. Like the targeting of inhibitory control deficits with cognitive training, many such compounds target the circuitry supporting extinction of conditioned fear. As discussed above, imaging studies implicate this circuitry in pediatric anxiety disorders, suggesting the relevance of studies in adults for youth. Unfortunately, limited research in youth employs pharmacotherapy to target extinction, possibly due to concerns about adverse effects. The best-developed approach uses d-cycloserine to augment CBT in obsessive compulsive disorder (OCD), a condition related to pediatric anxiety disorders. However, results from an RCT in 142 pediatric patients with OCD suggest the medication fails to augment response to CBT (81).

Examination of sleep provides a final example where research on mechanisms informs approaches to novel treatment in pediatric anxiety disorders. Problems with sleep correlate with ongoing and future risk for pediatric anxiety disorders (82). At a mechanistic level, such problems may disrupt memory consolidation in ways that hinder extinction of fear. Evidence (83) suggests that novel techniques for consolidating memory during sleep could potentially be adapted to facilitate response to CBT through consolidation of extinction. Similarly, other work generates a model (82) that links disrupted sleep to pediatric anxiety disorders through effects on functioning of the brain’s default mode network. Novel treatments (84) apply this model through sleep enhancement programs targeting multiple processes. These include distressing thoughts and associated rumination at bedtime in addition to alterations in motivation, deficient savoring of positive experiences, and problematic sleep hygiene. These programs attempt to increase response to CBT in pediatric anxiety disorders by targeting sleep’s associated disruptions in brain function.

4-. Treatment Improvements into the Future

The above-noted studies of attention bias and extinction illustrate how research on mechanisms provides a useful path for therapeutic discovery involving techniques with low risk of adverse effects. Of note, cognitive training influences brain function in less direct ways than other techniques, such as medications and some forms of brain stimulation, which carry greater risks for adverse effects. This provides avenues for synergistic work that combines research on cognitive training, a technique with low risk of adverse effects, with research on other therapies carrying greater risk for adverse effects. By providing complementary perspectives on mechanisms, such synergistic work could generate saltatory long-term improvements in treatment.

While minimal research examines brain stimulation in any pediatric mental disorder, let alone pediatric anxiety disorders, research on ABMT summarized in the previous section illustrates the promise of melding studies of cognitive training and brain stimulation. Accordingly, this final section highlights some facets of brain stimulation. Some techniques, such as transcranial direct current stimulation (tDCS), carry relatively few risks and hold some promise, where stimulation of the prefrontal cortex might influence extinction in ways that could facilitate response to CBT. However, one study in healthy adults found unexpected effects of tDCS on extinction: stimulation parameters expected to facilitate extinction of conditioned fear generated the opposite effect (85, 86). Unexpected results could reflect a less clear understanding of tDCS’s neural effects than for another promising treatment, transcranial magnetic stimulation (TMS).

TMS has some risk to induce seizures. While no TMS data exist in pediatric anxiety disorders, overall seizure risk with TMS in people of all ages is well below 1% (87). Work relevant to pediatric anxiety disorders includes an RCT of TMS in 103 pediatric patients with MDD, where no seizures occurred among the 48 patients who received as many as 30 treatment sessions (88). As described in the section above for ABMT, studies of TMS and cognitive training could inform each other, where TMS could be appropriately delivered to children with refractory disorders. Of note, such work could progress slowly, given findings among adults, where this approach has yet to generate the expected therapeutic benefits for anxiety, beyond successes in the treatment of OCD. For example, reminiscent of the above-noted unexpected tDCS effects on extinction (86), results from a study in 125 adults with posttraumatic stress disorder suggest that TMS stimulation intended to facilitate extinction could evoke the opposite effect, slowing CBT-related extinction (89).

Increasing the cross-talk between basic and clinical researchers is important. Available work links theories of anxiety disorder pathogenesis to neuroscience research on mammalian defensive behavior (7, 8, 10, 44, 90–95). Moreover, this work specifically highlights the relevance of research in non-human primates, given the evolutionary relatedness of nonhuman primates to humans, the prolonged development of the primate brain, and the well-developed prefrontal cortex of primates. This work supports neurodevelopmental theories of anxiety, identifying neural circuits and altered gene expression within the amygdala and prefrontal cortex that appear related to risk for anxiety disorders (10, 91–93). As with research on prevention, this work suggests that more durable therapeutic effects may result from earlier treatment, and it also delineates many pathways for clinical research.

Cross-species similarities allow neuroscience to inform a long-term agenda for novel treatment discovery in pediatric anxiety disorders. As reviewed elsewhere (80, 96–98), new treatments could involve novel pharmacological agents, including small molecules capable of targeting circuits or mechanisms related to defensive behavior, compounds that enhance neural plasticity, and medications showing some initial success in research on adult mood and anxiety disorders. However, two factors suggest that brain stimulation provides unique advantages. First, in children and adolescents, risk for adverse effects constrains studies of novel therapeutics (11). Adverse effects initiated by ingested substances can be difficult to terminate and can last for hours or even days; many effects generated by brain stimulation more rapidly abate, disappearing with cessation of stimulation. Thus, risk for enduring adverse effects may be lower with brain stimulation than novel medications. Second, studies of brain stimulation for mood disorders use imaging to leverage cross-species research on neural circuitry function (11). A comparable approach in anxiety could allow imaging studies in youth and experiments in other species to identify neural-circuit targets associated with avoidance and other dysfunctional behaviors evoked by threats. These circuits might be targeted through brain stimulation in pediatric patients with refractory anxiety disorders.

Whether through pharmacology, brain stimulation, or other means, long-term treatment advances likely require changes in practice involving closer basic-clinical integration than exists in current guidelines (31, 32). Closer integration is needed due to the degenerate nature of the nervous system (99), which allows multiple circuits to support many specific behaviors. From this perspective, any codification of specific behaviors remains wanting until the code connects these behaviors to particular circuits. In this instance, degeneracy would require dysfunctional behaviors to be classified based on the combination of phenomenology existing alongside specific neural circuitry correlates. This departs from current nosologies based only on phenomenology.

Applied to defensive behavior and anxiety, degeneracy allows approach-avoidance conflicts to evoke biologically distinct forms of avoidance that appear superficially similar on behavioral, phenomenological grounds (7, 65). In these instances, avoidance in some individuals may reflect the brain’s intrinsic tendency to overvalue avoidance of threats; in other individuals, avoidance may reflect the brain’s intrinsic tendency to undervalue procurement of rewards, and, in a final group of individuals, avoidance may reflect problems in the brain’s ability to adapt effectively following prior exposures to threats and rewards. Thus, these three groups of individuals might manifest superficially similar problems with approach-avoidance conflicts that arise from distinct source-mechanisms. Such groups could require different treatments each tailored to particular mechanisms (7, 100). In the long term, such degeneracy requiring distinct mechanism-guided treatments ultimately could create scenarios where appropriate treatment of mental disorders requires behavior to be classified, diagnoses to be assigned, and treatments to be selected based on assessments of brain function.

5-. Conclusion

This overview summarizes evidence on CBT and SSRI treatment for pediatric anxiety disorders. This is followed by a discussion of three approaches for improving treatment outcomes over the coming decade. Finally, the review describes an approach that combines research on cognitive training that targets mechanisms causing or sustaining anxiety disorders with studies of brain stimulation. This approach leverages leads from basic research on the targeting of circuits that support defensive behavior in mammals.