A scoping review investigating the use of exposure for the treatment and targeted prevention of anxiety and related disorders in young people

Abstract Background Cognitive Behavioural Therapy (CBT) is the gold standard intervention for anxiety and related mental health disorders among young people; however, the efficacy of individual elements of CBT (e.g., exposure to feared stimuli) have received little scrutiny. Aims This scoping review, informed by three stakeholder groups and a scientific advisory group, aimed to identify the nature and extent of the available research literature on the efficacy of exposure to feared stimuli, moderators of effectiveness in young people aged 14–24 years. Method Three international stakeholder groups composed of clinicians (N = 8), parents/carers (N = 5) and youth with lived experience of anxiety (N = 7) provided input into study design and results. Using the PRISMA extension for scoping reviews, a search of MEDLINE/Ovid, PsycINFO, PubMed, CINAHL, SCOPUS, EMBASE, ERIC, and Health Collection (informit) was conducted using terms related to anxiety, ages 14–24, and exposure. Results From 3508 unique abstracts, 64 papers were included for the review. While there was evidence for the efficacy of exposure as a treatment for youth anxiety disorders, fundamental gaps in knowledge of exposure in this age group were identified. Most studies examined post‐traumatic stress disorder, obsessive–compulsive disorder, and specific phobias with no randomised clinical trials uniquely evaluating exposure for the treatment of DSM‐5 anxiety disorders. Exposure was typically delivered accompanied by other anxiety management techniques. A multitude of optimisation strategies have been tested, yet only one of these effects (timing relative to sleep) showed preliminary evidence of replication. Conclusions A systematic and theoretically driven program of research investigating the efficacy of exposure in young people and factors that moderate its efficacy, along with methods to overcome barriers for delivery, is urgently needed.

most prevalent and chronic group of mental disorders in adolescents and young adults (Merikangas et al., 2010), it is vital that the most efficacious aspects of treatments are identified and we develop a greater understanding of the mechanisms for change (Bittner et al., 2007). A focus on this developmental stage is of particular importance because of the high rates of anxiety disorders during this period (NHS, 2019). Cognitive Behavioural Therapy (CBT) is recommended as the first line of treatment for anxiety and related disorders in young people in numerous treatment guidelines with approximately half to two-thirds of young people responding favourably to CBT (Higa-McMillan et al., 2016;James et al., 2020).
CBT with young people typically involves anxiety management strategies (e.g., cognitive restructuring, relaxation) followed by exposure (Whiteside et al., 2015), however, the efficacy of individual elements of CBT for youth has received little scrutiny (Whiteside et al., 2015) limiting efforts to improve both the efficacy and efficiency of treatments.
Exposure can be defined as prolonged and repeated confrontation with the feared object or situation in a systematic and controlled manner while preventing avoidance (Mobach et al., 2020) and is generally considered to be a critical component of CBT (Abramowitz et al., 2019). Exposure as a therapeutic technique was originally developed from the principles of associative learning, more specifically extinction learning (Marks, 1973). Exposure paradigms within CBT packages for anxiety disorders in young people have traditionally been based on a fear habituation model proposing that exposure allows the person to habituate to the emotion of fear (Foa & Kozak, 1986). More recently, the inhibitory learning model has proposed that exposure is successful because it establishes new memories about the feared object that compete with old memories (Craske et al., 2014). Based on this model, several strategies have been proposed to enhance the efficacy of exposure, such as violating expectancies about harm, occasional reinforced extinction, reducing safety behaviours, stimulus variability, enhancing memory consolidation, and affect labelling (Craske et al., 2014). Most empirical evaluation of these predictions has been limited to adult populations and our knowledge of the mechanisms of exposure in youth is less well-developed. This is a serious omission given that adolescents and young adults differ from older and mature adults in several important ways including cognitive maturity (Arain et al., 2013). Some evidence has indicated that both adolescent humans and rodents exhibit poorer extinction than younger and older groups (K. D. Baker et al., 2014). Therefore, it is vital to understand if exposure is an effective approach to reduce anxiety in an adolescent age-group.
Despite growing evidence informing the optimisation of exposure use with adults, there is a paucity of research examining strategies to improve the effects of exposure among young people. For example, treatment studies have applied exposure in many different forms such as in vivo (i.e., in real situations, not imaginal) or imaginal, graduated or flooding, one session treatment or across multiple treatment sessions, in session or out of session, and with or without coping strategies, with no comparisons of efficacy between various approaches. In most CBT packages for anxiety disorders in young people, exposure is graded and delivered both in session and out of session, and usually preceded by the introduction of coping strategies (Kendall & Peterman, 2015) despite recent evidence that these might not enhance outcomes, and may in fact impede positive treatment outcomes (Whiteside et al., 2020).
The objective of this scoping review was to identify the main sources of evidence regarding the use of exposure in the management of anxiety-related disorders among young people aged 14-24 years. This age range was selected by the agency funding this work, the Wellcome Trust, as part of a programme to identify the active ingredients in treatment of anxiety and/or depression in young people, given the unique needs of this age group and the high prevalence of anxiety and depression during these periods of development. This scoping review focused on the evidence for the efficacy of exposure and factors that increase or decrease its effectiveness. The scoping review included studies that specifically used exposure to treat or prevent anxiety disorders (clinical and subclinical) as included in the DSM-IV and DSM-5. This means that even though PTSD and OCD are no longer classified as anxiety disorders in DSM-5, they have been included, given exposure is a core component of treatment for these disorders and is commonly the primary ingredient.
There is also evidence to indicate that there are significant barriers to the delivery of CBT strategies, in particular exposure, due to a range of factors (such as clinician beliefs and confidence) Peer reviewed studies were included if they addressed questions on efficacy and/or moderators of exposure. We included studies in which the interventions or experimental conditions involved exposure as the primary ingredient (e.g., started in session 1 and comprised almost all sessions). These intervention studies needed to include a measure of anxiety-or fear-related outcomes that could potentially impact upon daily function. Studies examining interventions that included exposure as part of a broader treatment package (e.g., treatment involving cognitive restructuring, relaxation, and exposure) were also included if a measure of anxiety/impact was taken before and after the introduction of exposure so that the specific effect of exposure could be extracted. However, as exposure is rarely included entirely on its own and because we were interested in the potential augmenting effect of other concurrent strategies, studies were not excluded if young people were given additional strategies to use during exposure (e.g., relaxation, cognitive restructuring). In these cases, exposure still needed to comprise most sessions and be the primary ingredient. Studies were excluded if they were narrative reviews, meta-analyses, systematic reviews, or grey literature.

Study selection
After papers were extracted from a search of the electronic databases, each abstract was screened by two research team members using Covidence (Veritas Health Innovation, 2020). The abstracts of these studies were retrieved, and full texts screened for inclusion by two research team members. All conflicts regarding study eligibility were discussed and resolved between AT, LP, and MO.

Data synthesis
A coding framework was developed to extract data regarding details about the study, including date, location, sample characteristics, aim, design, details of exposure treatment, measures, and outcomes. An analysis of study quality is not typically included in a scoping review (Tricco et al., 2018).

Stakeholder engagement
The research team met with three stakeholder groups twice. The groups comprised clinicians delivering psychological treatments (N = 8), parents/carers of a young person with lived experience of anxiety (N = 5), and youth with lived experiences of anxiety disorders (N = 7). The focus groups were conducted in English (via Zoom) and included participants from Australia, UK, Japan, USA, Netherlands, Germany, India, Brazil, Thailand, and Malaysia. The research project was presented to each group in the first meeting and they provided feedback based on their experiences and expertise with exposure and/or psychological therapy for anxiety disorders. The search terms were then reviewed to ensure the experiences of the stakeholders were reflected. The research team met with each group again to present the review findings and to discuss how the results reflected their experiences. A scientific advisory group (N = 8; for membership see acknowledgements) was engaged on two occasions to provide feedback on the search strategy and the results. Figure 1 presents the flow diagram of articles selected for the scoping review. After screening 3508 abstracts and 885 full-text articles, 64 articles matched our criteria. The papers were diverse in methodologies, country of origin, and primary anxiety type (See Table 2). Table 3 outlines the details of the included studies such as author details, purpose of the study, and information on the sample and measures used. Table 4 specifies the types of exposure and additional elements used in each study (See Appendix S1). Across all the studies reviewed, 63% (n = 40) of the papers examined exposure using an in vivo format (91% of randomised controlled clinical trials (RCCTs; i.e., randomised controlled clinical trials where therapeutic doses of exposure were administered; n = 10) and 58% (n = 37) examined exposure using imaginal techniques (82% of randomised controlled clinical trials; n = 9). Most studies applied exposure using a gradual approach, that is, starting with less fearful situations (n = 39, TREATMENT AND TARGETED PREVENTION OF ANXIETY AND RELATED DISORDERS IN YOUNG PEOPLE -3 of 30 61%); with a high proportion of RCCTs using gradual exposure (n = 9; 82%). Studies less consistently examined exposure and included other skills during the program: psychoeducation (n = 20; 31%), relaxation (n = 16; 25%), or cognitive strategies (n = 3; 5%). Examining RCCTs specifically, 64% tested exposure with relaxation, 55% with psychoeducation about anxiety, and 9% with cognitive restructuring techniques. Only 16% (n = 10) of all studies reviewed used technology to assist exposures (0% of RCCTs). Only 9% of studies (n = 6) tested intensive exposure, that is exposure massed in a single session (at least 2 h) or a series of sessions (i.e., having more than one session a week).

RESULTS
Figure 2 provides a summary of the length and number of sessions across each study type. Typically, exposure sessions were fewer and shorter in the experimental studies, with the longest exposure sessions occurring in the randomised clinical trials and the greatest number of sessions occurring in the case series. We will provide a review of our main areas of focus: studies evaluating the efficacy of exposure, and studies identifying moderators of exposure. The studies will be organised according to the type of study. Within each section, we will first review the evidence from RCCTs, followed by experimental studies, quasi-experimental studies, cohort studies, case series and case studies.

Randomised controlled clinical trials (RCCT)
Eleven RCCTs directly examined the efficacy of exposure among 14-24-year-olds. All studies provided a comparison of exposure treatment compared to an active control condition, providing a conservative test of efficacy. Notably, no studies focused on youth with diagnosed DSM-5 anxiety disorders and instead focused on OCD, PTSD and subclinical presentations of anxiety. In many cases, the administration of exposure was integrated with other intervention strategies (e.g., relaxation/breathing exercises, psychoeducation, cognitive strategies, and eye movement desensitisation and reprocessing (EMDR)), preventing conclusions about the unique effects of exposure delivered as an ingredient on its own. If within and between group effect sizes were not provided they were calculated using Notes: When entering these terms into the database on July 13 th , 2020, the terms within the columns were combined with 'OR', and the terms between the columns were combined using 'AND'. For example, (anx* OR worr* OR panic* etc) AND (adolesc* OR student etc.) AND (cbt OR exposure therapy etc.) Where possible, filters were applied to target empirical studies, papers in English, and population ages between 14 and 25 years.
www.psychometrica.dl/effect_sizes.html (See Table 3). The average within-subject effect size for exposure therapies was very large at post-treatment (d = 1.97) and follow-up (d = 2.31). This means that overall there was a substantial change in anxiety/fear symptoms immediately following exposure-based therapies. This change appears to be maintained in the short to medium term. The average between-group effect sizes comparing exposure to an active control that did not specifically include exposure were moderate to large at post-treatment (d = 0.65) and follow-up (d = 0.66). This means that compared to other treatments for anxiety and related disorders, exposure-based therapies are superior (See Table 3  relaxation/breathing exercises (e.g., Gilboa-Schechtman et al., 2010) or through narrative exposure therapy (Ertl et al., 2011), or in combination with EMDR (Scheck et al., 1998). All of these approaches were found to be more efficacious in reducing PTSD severity scores or PTSD diagnoses in adolescents and young adults with PTSD when compared to an active control condition such as an academic tutoring program (Ertl et al., 2011), active listening (Scheck et al., 1998), or dynamic therapy (Gilboa-Schechtman et al., 2010). In a secondary analysis of another included trial (Foa et al., 2013), and one of the only studies to examine possible mechanisms of change, McLean et al. (2015) found that changes in negative trauma-related cognitions resulting from exposure therapy mediated changes in PTSD symptoms.
The remaining three RCCTs focused on treatment of social anxiety symptoms (Zaboski et al., 2019), fear of pain (Flack et al., 2018), and OCD (Neziroglu et al., 2000). Both Zaboski et al. (2019) and Neziroglu et al. (2000) reported large effect sizes for exposure treatment conditions compared to an active control condition at post-treatment and follow-up assessments (See Table 3).
For example, combining medication with graded, in vivo exposure with response prevention was more effective than medication alone in reducing OCD symptoms in adolescents with OCD (Neziroglu et al., 2000). Similarly, exposure significantly reduced social anxiety symptoms in university students (using group therapy with graded, in vivo exposure and response prevention) compared to a psychoeducation control (Zaboski et al., 2019). In contrast to these two studies observing large effect sizes, Flack et al. (2018) reported small, non-significant effect sizes at both post-treatment and follow-up favouring interoceptive exposure over relaxation for reducing fear of pain in a clinical sample of adolescents with chronic pain (Flack et al., 2018). Other (n = 17) Unable to obtain paper (n = 12) Outcome not relaƟng to anxiety or funcƟonal impairment (n = 8) Meta-analysis or review (n = 6) ParƟcipants not selected because of anxiety disorder (n = 5) Grey literature (n = 4) Studies included in scoping review (n = 64) Experimental studies Twenty-three experimental studies examined the efficacy of exposure, with most studies examining the effects on specific fears or phobias (n = 12; 52%) and in samples of university students (n = 19; 83%). There were few experimental findings where strong conclusions about the efficacy of exposure could be drawn, due to considerable variations in aims, limited methods (e.g., no random assignment) and lack of replication. Further, many of the studies were designed to test features which may facilitate exposure (these findings are reviewed in the next section).
Six of the experimental studies provided some information regarding the overall efficacy of exposure. In a sample of university students, both imaginal exposure with systematic desensitisation (which involved inducing deep muscle relaxation during imaginal graded exposure) and imaginal exposure with cue-controlled desensitisation (modified systematic desensitisation with self-administered cue-controlled relaxation) significantly reduced performance anxiety, compared to study skills alone (Lent & Russell, 1978). In this study, neither exposure condition differed from each other in terms of outcomes. Following an intervention involving imaginal exposure and EMDR in a sample of high-school students with PTSD, Therapist observation and client self-report 28 sessions (30-40 min per session) Saigh (1987), USA To describe the treatment of PTSD    Tang et al. (2015) showed significantly lower anxiety compared to a treatment as usual condition. There was also preliminary evidence for the efficacy of group systematic desensitisation, in a small sample of children (11-15 years old, n = 23) and university students (mean age 21.9 years n = 13) showing significantly decreased performance anxiety compared to autogenic training (involving exercises of relaxed breathing, progressive muscle relaxation) and imaginal exposure alone (Kondaš, 1967).
In vivo exposure was found to be associated with reductions in anxiety/fear in several experimental studies using university students. In a single session, in vivo exposure was associated with significantly lower spider fear and avoidance than viewing neutral images (Müller et al., 2011). In vivo exposure was also associated with lower avoidance and anxiety for young people with a diagnosis of Specific Phobia (heights) when delivered weekly using virtual reality (VR) compared to a waitlist condition (Rothbaum et al., 1995a).
Blocked and constant exposure (exposure that focused on repeating one step in each session using a gradual hierarchy) did not differ significantly from random and variable exposure in reducing contamination-related fears for university students, immediately and 2 weeks later (Kircanski et al., 2012). This preliminary evidence suggests that in vivo exposure is effective in treating simple/specific fears and that the delivery may not necessarily need to be gradual or use a typical exposure hierarchy.
Two small experiments found exposure treatments to be less or no more efficacious than an alternative treatment (Kim, 2008;Possis et al., 2013). In university students with elevated social anxiety (n = 61), Possis et al. (2013) found that compared to a psychoeducation control, one session of cognitive restructuring (without exposure) was more efficacious at reducing social anxiety than one behavioural experiment (involving exposure). Further, there was no significant difference between improvisation-assisted desensitisation (involving graded imaginal exposure) or music-assisted progressive muscle relaxation (6 weekly sessions of 30 min) in reducing performance anxiety in young pianists (n = 30; Kim, 2008).

Quasi-experimental studies
There were four quasi-experimental studies identified in this scoping review, with three out of four studies testing exposure in university students with performance or specific fears and one clinical study of patients with OCD. All four studies showed significant reductions in anxiety symptoms following exposure. For example, in vivo exposure was effective in significantly reducing performance anxiety compared to students allocated to a control condition involving written assignments (Finn et al., 2009). Similarly, Longo and Vom Saal (1984) showed that graded imaginal exposure with or without breathing exercises showed greater reductions in performance anxiety in university students compared to a waitlist condition. In a third study, university students who reported high dog fears also demonstrated reductions in fears following exposure but still had significantly higher fear than a low dog fear control group (Hoffmann & Odendaal, 2001). Finally in an open clinical trial with a sample of children and adolescents with a diagnosis of OCD, exposure in either intensive or weekly treatment was associated with a significant reduction in OCD severity (Franklin et al., 1998).
Cohort studies. The scoping review identified eight cohort studies, one of which provided information about moderators of efficacy and is discussed in the next section. Three cohort studies provided information on the potential benefits of using virtual reality to deliver exposure to treat social/performance anxiety (Kahlon et al., 2019;Morina et al., 2015;Stupar-Rutenfrans et al., 2017). Two cohort studies examined the efficacy of exposure in clinical adolescent populations with OCD (Fischer et al., 1998;Riise et al., 2018). Significant reductions in OCD symptoms were observed following group graded in vivo exposure (with response prevention) delivered weekly (Fischer et al., 1998)  Case series and case studies. 11 case reports and seven case series reported favourable effects when using exposure to treat specific fears/phobias (Buchanan & Houlihan, 2008;Burton et al., 2017;Chok et al., 2010;Newman & Adams, 2004;Rothbaum et al., 1995b), performance anxiety (Culver et al., 2012), OCD (Abramowitz, 2002;Farrell et al., 2016;Iniesta-Sepúlveda et al., 2018;Marr, 2012;Whiteside & Abramowitz, 2006;Whiteside et al., 2008;Woods et al., 2000;Wu & Storch, 2016), and PTSD (Frye & Spates, 2012;Hendriks et al., 2017;Kitchiner, 2000;Saigh, 1987). Of these studies, Focusing on specific details. Using a randomised design, asking university students (n = 49) with elevated social anxiety to focus on the distinctive features of the stressful personal experience during imaginal exposure (e.g., participants were instructed to recall all the specific features of the target event, in detail) was significantly more effective at decreasing distress than focusing on generic elements (Vrielynck & Philippot, 2009).
Challenging exposures. The findings of one randomised controlled experiment of university students with height fears indicated that adding actions to make exposure more challenging (such as running towards the railing) was significantly more effective than exposure that did not include this feature in reducing reported fear (Wolitzky & Telch, 2009).
Inadequate use. In adolescents with OCD who had received a non-effective course of CBT and medication, Krebs et al. (2015) showed that 95.5% of the previous courses of CBT were rated as inadequate, primarily because there had not been a sufficient focus on exposure in the treatment.
Multiple contexts. Olatunji (2017) conducted a randomised controlled experiment in a community sample of young adults with Specific Phobia (Snakes; n = 108). Exposure delivered in multiple contexts resulted in significantly better outcomes in terms of avoidance and fear than when young people were exposed to videos of the feared object in single contexts.
Evoking disgust: Olatunji et al., 2012 showed that activating emotions of disgust prior to repeated exposure did not significantly enhance reduction of fear, avoidance or disgust in a university sample of students (n = 44) with high injection fears.
Drug enhancement. Powers et al. (2008) randomly allocated university students and community volunteers (n = 95) with elevated fears of enclosed spaces to take either yohimbine hydrochloride (a selective competitive alpha2-adrenergic receptor antagonist that appears to enhance extinction learning) or placebo prior to exposure.
Although there were no immediate differences in reduction of fear, with both groups showing significant fear reduction, participants in the drug condition were significantly less likely to experience a return of fear at the 1-week follow-up assessment.
Safety behaviours. Three randomised controlled experiments investigated the impact of safety behaviours on the efficacy of exposures in high fearful university samples. In a non-clinical sample, Goetz and Lee (2015) found that allowing safety behaviours (e.g., hygienic wipes) after exposures to treat contamination fears was associated with significantly more rapid reductions in fear and behavioural avoidance compared to participants using safety behaviours prior to the exposure. Rachman et al. (2011) showed that safety behaviours used after exposure for contamination fears did not lead to significantly enhanced outcomes compared to when participants were not given the opportunity to use safety behaviours.
Similarly, there was no significant difference in outcomes for participants with high claustrophobic fear when given access to safety aids during exposure compared to exposure without safety aids (Deacon et al., 2010).

Cue reminders.
Four experimental studies have examined the impact of providing cues during the exposure process (as a reminder of an object/element associated with the feared event). For example, one randomised controlled experiment on university students with elevated performance anxiety conducted by Shin and Newman (2018) presented all participants with specific cues during exposure sessions and then randomly allocated participants to receive or not receive the cues during a behavioural avoidance test 1 week later. The reminder cues prevented a return of fear on two of three outcomes. In contrast, in another pseudo-randomised experiment of university students with high spider fears, reminders of cues presented during the exposure did not significantly attenuate fear renewal in a different exposure context (Dibbets et al., 2013).
Stimuli Information. Two studies examining different types of information provision were included in the review. The first study examined safety information and the second study examined the degree to which the participant was told the extinction stimuli were like other stimuli. First, Johnson and Casey (2015) found that safety information presented 10 min prior to extinction attenuated the recovery of fear in both teenagers (n = 36) and young adults (n = 38). Participants who were not provided with safety information (during post retrieval extinction) displayed a robust recovery of the fear compared to those participants who were given the safety information. In the second experiment with university students (n = 69), Scheveneels et al. (2017) found that providing information about the object (artificial animal like objects) as 'typical' of feared stimuli promoted significantly greater generalisation of extinction, measured by lower shock expectancies, compared to atypical explanations. Although there were no significant differences between the groups in terms of arousal or subjective distress, those with greater control were significantly more successful in approaching a spider at the end of the session and less avoidant at follow-up.
Relaxation. Few experimental studies that examined the effect of relaxation techniques on exposure outcomes. A quasi-experiment with university students with elevated speech anxiety found that incorporating breathing exercises with imaginal exposure was significantly more effective than exposure without breathing exercises in reducing self-reported and observed (but not physiological) measures of speech anxiety (Longo & Vom Saal, 1984). Similarly, one case study of a young adult with PTSD reported that incorporating mindfulness exercises before and after exposure was helpful as measured by verbal client feedback and overall reductions in selfreport measures of fear and anxiety (Frye & Spates, 2012).

Stakeholder engagement
Efficacy. Stakeholders were unsurprised by the evidence supporting the efficacy of exposure, as those who had experience with the technique had found it helpful. One clinician believed that anxiety was secondary to other underlying issues so was trained not to target the anxiety specifically and hence had not used the technique. At the second stakeholder meeting, this same clinician had begun adopting the technique successfully.
Moderators. Based on their own experience, two of the moderators identified in the review were identified as important by stakeholders: timing of the exposures and control. Both youth and carers believed that exposure was effective when it was graded with "very small clear steps" that enabled the young person to feel "in control." Both groups also voiced the importance of youth autonomy. Second, the review suggested that exposures closer to the time a young person goes to sleep may be more effective, yet in contrast, one youth found that engaging in exposures in the mornings was more helpful, because if she had exposure sessions in the afternoon she would spend the day catastrophising about it.
The remaining moderators identified in the review were not considered to be as important to stakeholders. However, stakeholders highlighted several other factors that they believed were important to the delivery and efficacy of exposure. For example, clinicians believed exposure was successful when youth had external support (e.g., from family or friends), felt understood, comprehended the treatment rationale, were willing and motivated to engage in treatment, and trusted the therapist.
Youth and carers identified the importance of carers being given clearly defined roles in the exposure process. Youth and carers often felt unsupported, with carers being unsure of their role in homework tasks and how much assistance to provide. Both groups also talked about the importance of therapeutic alliance (with outcomes from exposure being more favourable when the therapist was trusted and there was a positive relationship). Parents and carers emphasised the importance of using empathetic language, psychoeducation, rewards, preventing unhelpful reassurance giving, and having more than one family member or having friends engaged in supporting the young person.
Some of the strategies young people employed to help with exposure practice included realistic thinking, breathing exercises, self-talk strategies, and mindfulness. Young people reported engaging in exposure exercises with a therapist or informally on their own or with peers. One young person received CBT strategies as a child and so found it "childish" to return to the same program again as an adolescent.

Barriers to delivery
Many clinicians expressed difficulties motivating clients to engage in exposure and reported anxieties about their client's wellbeing which often led them to avoid using exposure despite its potential benefits. Clinicians emphasised that preserving the relationship with the client was a very important factor for them in deciding whether to conduct exposure and they often avoided implementing the strategy with teenagers due to fear of jeopardising the client relationship and engagement. For some clinicians, their own anxiety about and lack of confidence in conducting exposure impeded exposure delivery.
Both parent and youth stakeholder groups expressed reluctance to engage in group exposure therapy, perceiving it to be less effective than one-on-one exposure sessions as sessions were not personalised. While shared experiences may enhance group therapy, the clinicians expressed how it could be "very overwhelming for both clients and therapists" and difficult to personalise treatment.
The primary barrier to exposure faced by the parents was difficulty motivating the young person to take ownership of their treatment and complete homework exercises. A common factor that was brought up by carers was the difficulty they experienced supporting the young person through the exposure homework. Moreover, parents voiced feeling unsupported and untrained in helping their child engage in exposures. Although parents were more involved with treatment when their child was younger, as the children got older young people reported they typically wanted their parents to be less us only including treatment studies where exposure was the key treatment ingredient and was introduced from the start of treatment, exposure was rarely delivered on its own and was typically accompanied by other anxiety management techniques. This is an important limitation as the addition of these techniques may have positive or negative effects. For example, a recent meta-analysis examining CBT for a younger cohort (5-15 years) identified stronger intervention effects when CBT with exposure was delivered without the addition of relaxation (Whiteside et al., 2020). Furthermore, while a broad range of optimisation strategies have been evaluated within the identified body of research, the timing of exposures relative to going to sleep (where an exposure immediately prior to sleep appeared to augment the reduction of spider fears), was the only one that had been replicated in a second study. Together these issues highlight the need for a systematic program of robust and reproducible research that examines the efficacy of exposure and the impact of additional anxiety management strategies (such as relaxation, cognitive restructuring) on the efficacy related to exposure in 14-24-year-olds is an essential next step for research.
A further limitation of the available literature is the extent to which it reflects stakeholders' experiences and preferences. Stakeholder discussions highlighted the importance of a number of potential optimisation strategies that have yet to be evaluated in this population such as the benefits of informal exposures with peers, completion of out of session exposures (homework), and parental role definition and support. These provide clear directions for future research. There were a couple of notable exceptions and these present interesting dilemmas for clinical practice given the potential mismatch between expectations or experiences and research evidence. Clinicians reported sometimes being reluctant to engage in exposure due to the risk that it might rupture the therapeutic relationship, however, in one randomised controlled trial, client-reported therapeutic alliance was stronger in exposure treatments for PTSD compared to client centred therapy (Capaldi et al., 2016). Contrary to some clinicians' expectations, this preliminary evidence suggests there is potential for exposure to augment rather than disrupt the therapeutic alliance. This suggestion is consistent with findings from a recent meta-analysis that attrition from exposure with response prevention is generally lower than from other therapies among youth (under 17-year-olds) with OCD (Johnco et al., 2020). Finally, stakeholders highlighted that a graded approach is important to promote engagement and a sense of control among young people. While this has not been directly tested, Kircanski et al. (2012) showed in an experimental study that a graded approach to exposure delivered similar outcomes to exposures of random difficulty among young people with OCD. Whether the graded approach increases young people's readiness to engage in exposure within clinical settings remains to be seen. Indeed, the stakeholder discussions emphasised the need for more developmentally sensitive research that identifies how best to help young people, parents and clinicians overcome concerns about conducting exposure within treatment.
It was clear from our stakeholder groups that exposure was not always offered to young people seeking help for anxiety problems or not always delivered in an adequate manner. In support of this, one cohort study in our review found that client experiences of previous CBT were inadequate at treating OCD due to insufficient use of exposure (Krebs et al., 2015). The authors suggested that clinicians may avoid behavioural aspects of the treatment due to fear of triggering distress in the client. As mentioned, this idea was supported by our clinician stakeholders, some of whom expressed reluctance to use exposure with their clients. This is also consistent with several Consistent with a scoping review approach (Tricco et al., 2018), methodological quality was not appraised and, given the wide variation in study research questions and approaches, we took a descriptive approach to clarifying the identified research. We hope 26 of 30 - that with a growing evidence-base future systematic reviews will include both meta-analytic approaches to synthesise findings and systematic appraisal of research quality.

CONCLUSION
Exposure is typically seen as the central component in anxiety treatments for young people. Yet there are fundamental and substantial gaps in our knowledge of exposure-based therapies in treating anxiety-related conditions specifically in adolescents. Few robust conclusions can be drawn about the outcomes from exposure and potential ways to optimise it given the lack of replicated research using rigorous methods. It is unclear whether exposure is effective as a strategy on its own and whether adjunctive anxiety management strategies help or hinder. Given the high prevalence of anxiety disorders among adolescents and the long-term consequences of ineffective treatment, it is paramount for a systematic program of highquality research to be undertaken to address the fundamental gaps identified in this review to make clinical recommendations that are developmentally sensitive to the needs of adolescents and their families.