Parts of this chapter have been reproduced from Wright et al.1 © 2022 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for Child and Adolescent Mental Health. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. The text below includes minor additions and formatting changes to the original text.
Parts of this chapter have been reproduced from Hayward et al.42 © 2022 Hayward et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The text below includes minor additions and formatting changes to the original text.
Parts of this chapter have been reproduced from Wang et al.2 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. The text below includes minor additions and formatting changes to the original text.
Summary of the main findings
Primary outcome
ASPECT sought to examine the non-inferiority of OST compared with routinely used CBT-based interventions for treating specific phobias in CYP in a multicentre study across England. The primary outcome measure was the BAT score at the 6-month follow-up, with a series of secondary measures completed alongside by both CYP and parents. The results demonstrated that using prospectively set non-inferiority limit OST is shown to be non-inferior in clinical effectiveness to multisession CBT. Where multiple imputation is used for missing values, this finding also applies. This signal pertains whether ITT or PP analysis is used. At the main outcome point, using the diagnostic measure (the ADIS), 27% of CYP are ‘no longer phobic’ using ITT analysis (same for OST and CBT) and 33% of CYP are ‘no longer phobic’ using PP analysis (34% in the OST group and 32% in the CBT group). In a review of the BAT outcome measure used with young people who have specific phobias, Castagna et al.40 suggested that an average of 60% step completion (6 out of 10 steps) post treatment is an ‘expected’ outcome when reviewing multiple studies and, therefore, our finding of over 70% means that step completion in both treatment groups post treatment shows similar or slightly better treatment success rates in this patient group. These findings were similar to those reported by Ollendick et al.,33 who found that those in the OST group achieved around 73% of BAT steps at the 6-month follow-up.
Secondary outcomes
Several secondary objectives were made a priori and were investigated by including an embedded health economic evaluation and qualitative interviews conducted with CYP, parents and clinicians. This work sought to examine:
- Corroboration for the primary outcome. All of the secondary outcome measures showed no differences between the two treatment groups. In particular, the other anxiety measures, including SUDS change scores, CAIS, goal-based outcome measures and RCADS anxiety scores, gave similar signals to the main outcome measure. When sensitivity analyses were examined, vomit phobia was the only phobia found to respond better to OST than multisession CBT, although subgroup numbers would have been too small to explore this definitively.
- The cost-effectiveness of OST compared with CBT and the relative impact of the interventions on CYP’s QoL, and school and social life. Compared with CBT, OST marginally decreased the mean service use costs while maintaining similar QALY improvement. Compared with baseline, it was observed that both treatments decreased the resource use over time, but the cost difference (in either NHS/PSS or societal perspectives) between the two treatment groups at the 6-month follow-up was marginal. The cost differences were driven by higher treatment costs and CAMHS-related service costs in the CBT group. The higher average cost of CBT was caused by the low unit cost but high-frequency sessions involved, whereas the higher CAMHS costs are likely to have been caused by a few high (but plausible) values (see Chapter 4, Service use and costs). The cost saving of OST within the 6-month time frame is expected to have been larger if CYP in the CBT group did not pause or delay their sessions as a result of the COVID-19 pandemic. A similar trend was also observed in the QALY outcomes. Although both the EQ-5D-Y and the CHU-9D utilities showed improvements over time in both treatment groups, the difference in QALY improvements between the two treatment groups was negligible. After adjusting for the imbalanced utility and other characteristics at baseline, the increase in QALYs remains small (0.002 QALYs). Both our primary and our sensitivity analyses suggest that OST is likely to save cost and maintain similar QALYs as CBT. The results of the health economic analysis demonstrated that, compared with CYP receiving CBT, on average, there is a £200–300 saving per person receiving OST. This is relatively small, but may be relevant for commissioners when considering resource utilisation if multiplied by the number of CYP with specific phobias in the UK.
- The acceptability of OST from the perspectives of CYP, parents/guardians and clinicians. The qualitative results show that the acceptability of and satisfaction with OST from the perspectives of CYP, parents/guardians and clinicians was good. When considering a range of acceptability parameters for OST, the core components that were consistently identified to influence treatment process and satisfaction were child readiness, therapist competency and collaboration. The main clinical challenges found in the study related to the flow of CYP through a complex health system, with many clinicians reporting that phobia was not deemed ‘core business’ by their managers. Implementation issues were also raised by clinicians, with emphasis on ensuring that adequate resources, time allocation, training and administrative support were in place to enable OST delivery, especially if rolling it out as a new therapeutic offering. It was felt that enhanced follow-up and improved handover may benefit parents, facilitate home-based skills practice and contribute to the maintenance of treatment effects.
Strengths and weaknesses
Strengths and weaknesses of the main trial
Strengths
Study aim and recruitment success
To the best of our knowledge, ASPECT was the first RCT to compare the clinical effectiveness and cost-effectiveness of OST with that of usual care (CBT) for CYP with specific phobias in England. It has been recommended that work be conducted to compare OST with multisession CBT in adults,99 which demonstrates the lack of evidence in the area and further justifies the importance of ASPECT. Although there were many difficulties in identifying CYP for study entry (as documented in Chapter 3), a sufficient number of CYP and their parent/guardians were recruited overall, demonstrating that it is possible to recruit to a study of this nature.
Widespread delivery of phobia treatments
Throughout the set-up and running of ASPECT, and as documented in more detail in Chapter 3, it became apparent that there was much variability in the services available for CYP with specific phobias across England. Although phobia treatments appear to be routinely delivered in some areas, other areas maintain high thresholds for service entry or do not offer such treatments at all. Through the completion of ASPECT, the importance of providing phobia treatments to CYP was widely disseminated to services across England. For this reason, many services started to identify and treat individuals with specific phobias who would otherwise have gone unnoticed. Furthermore, providing training to school-based services broadened the services available to CYP within this context.
Although ASPECT has now ended, it is hoped that a focus on delivering therapies for specific phobias will remain. As a result of the trial, approximately 170 clinicians have now received training in OST, which can continue to be delivered across services, as deemed appropriate. Anecdotal evidence from some of the participating services revealed that, despite not providing phobia treatment prior to ASPECT, provision of therapies for specific phobias is now being explored and expanded within services to ensure that treatment can be offered to CYP presenting with phobias. This was also evident in the findings obtained during the qualitative interviews with clinicians (see Chapter 4, Post-treatment reflections).
Pragmatic trial
Unlike other previous studies conducted in research settings,33,34,37,100 ASPECT was a pragmatic trial that allowed the effectiveness of OST and CBT to be compared in real-world usual clinical settings. Embedding the study in clinical teams was achieved without teams having to change their routine practice. This allowed us to compare OST against the usual delivery of CBT, which ensured a high level of ecological validity. Services using CBT were all following the usual practices of building a fear hierarchy, exposure and cognitive restructuring. Some services requested further information to ensure that their clinicians had clear guidelines and, for this reason, a CBT-as-usual guide was produced, which was co-written by several of the study principal investigators (BW and LG, 2018, personal communication) and used throughout the study.
Large number of UK centres
As discussed in Chapter 2, recruitment to ASPECT took place in the routine care pathways for phobia in CYP. This included 12 NHS trusts comprising 26 CAMHS sites, three third-sector/voluntary services and one university-based CYP well-being service in England. Recruiting from a diverse range of services allowed us to treat CYP in a variety of settings and positions in the care pathway.
Full range of phobia types included, unlike other studies
The design of ASPECT allowed CYP with a variety of phobias to take part in the research, with few exclusions. Providing that a child or young person could be in proximity with their phobic stimulus safely (either in reality or in simulation), they were able to enter the trial. As a result, across the sample, numerous phobias were identified that aligned to all of the subtypes described in the DSM-V criteria. All clinically presenting phobias were included, unlike some studies33 in which BII-type phobias were excluded.
Methodological strengths
ASPECT employed a number of methodological mechanisms to reduce bias. These included remote randomisation, blinding of researchers gathering outcome measures at all points, prospectively planned protocol and statistical analysis plans, and oversight of all procedures from both an independent TSC and a DMEC. The inclusion of the BAT alongside self-reported questionnaires allowed for a comprehensive and objective measure of phobia in CYP. We also measured treatment fidelity using both clinician reports and expert ratings of taped therapy sessions.
Weaknesses and challenges
Several challenges presented themselves to the study team and some weaknesses were apparent that affected both trial delivery and the overall results reported.
COVID-19
Most notable was the impact of the COVID-19 pandemic, which began in early 2020 and led to school closures, reduced service availability and changes to service delivery (e.g. from face to face to online). As documented in more detail in Chapter 3, several amendments were submitted to mitigate the impact of COVID-19 on this research trial, including remote data collection and remote therapy delivery where possible. However, despite these measures, the timing of the pandemic meant that recruitment to ASPECT remained unaffected, but therapy delivery for some CYP and the completion of the 6-month follow-up assessments were significantly impacted. Although only 4.5% (n = 3) of participants who withdrew from the trial cited COVID-19 as the reason, 37% (n = 98) of CYP completed their 6-month follow-up assessments having received no or incomplete treatment, some as a result of COVID-19 delays. Our estimates also suggest that approximately 17% (n = 7) of those in the OST group did not receive any treatment as a result of the pandemic, as well as 14% in the CBT group who received only one to three CBT sessions and 8% (n = 3) who received no sessions. Furthermore, measuring the primary outcome (the BAT) was not possible during periods in which social distancing restrictions and guidelines were in place because they prevented face-to-face contact with a researcher and/or exposure to a feared stimulus. This resulted in 48 (24.4%) CYP not completing this measure at follow-up (21.6% randomised to CBT and 27% randomised to OST). COVID-19 also may have affected treatment, such as continuing exposure work at home, because closing schools and reduced access to outside spaces may have reduced these opportunities.
Study withdrawals
In general, a large number of CYP withdraw from mental health treatment, with one meta-analysis reporting rates of between 25% and 75%.29 In our study, 31% of participants in the OST group and 27% of participants in the CBT group had received no treatment sessions by the 6-month follow-up point, which are both at the lower end of the figures reported by de Haan et al.’s29 review.
Large numbers of children in our study were withdrawn from phobia treatment by clinicians, which prompted us to look closely at reasons for this. First, inherent to phobia is an element of avoidance to the feared stimulus; however, treatment requires a child or young person to undergo exposure. This creates problems in two scenarios: first, where a parent/guardian is the motivation behind the treatment and the CYP is not motivated to engage, and, second, where the CYP is motivated but unable to overcome their fear to attend for treatment.
Given the pragmatic nature of ASPECT and its delivery in the real world, we expected some withdrawal of children by clinicians. Some of this was in response to the clinician not deeming the timing of the intervention to be right for the CYP (e.g. where a child had just entered foster care and was settling in). Similarly, at times, clinicians deemed that a child or young person with comorbidities needed assessment or treatment for a different condition before receiving phobia treatment (n = 4). This included young people with generalised anxiety disorder, post-traumatic stress disorder, generalised health anxiety and ASD. In some cases, the clinician believed that a more complex intervention was needed (n = 3), with two involving referrals to other trusts.
In two situations occurring in different sites, the clinician felt that the phobia was so severe that it would not be possible to treat it adequately using OST and, as a result, a child or young person who was allocated to OST received CBT as usual instead. Given that the clinicians had autonomy and were making decisions based on the best interests of the CYP, this made it important to consider the ITT analysis and the PP analysis together. This could be because some clinicians in a pragmatic trial involving large numbers of clinicians may not be in equipoise or because the child’s presentation may have changed. Indeed we had some instances in which the CYP had increased child mental health needs that had escalated since the initial assessment, perhaps related to long waiting times. One child or young person was referred to forensic CAMHS while they were on the phobia treatment waiting list. In another instance (not related to the phobia), a child or young person was admitted to hospital following a medication overdose.
Overall, 15 CYP had received a diagnosis for ASD or were currently awaiting an assessment (n = 4). Although we were able to treat those with ASD for phobia successfully, one of the clinicians described that the sensory fears of a young person with ASD were more related to this condition than to a phobia (as the initial clinical assessment had determined), and that they required an alternative therapeutic option for support.
Finally, on several occasions, triaging was completed by a clinician at the SPA of CAMHS and the assessment had recorded phobia as the main need for treatment; however, when they saw the clinician after waiting, the child was assessed as no longer having a phobia. It is not clear if these were spontaneous remissions or variations in clinician opinion. Previous research suggests that triage can rely on clinical experience to underpin decision-making rather than formal assessment tools.101 Our finding that many CAMHS did not regard phobia as severe enough to treat in their service may lead us to speculate that CAMHS clinicians in SPA triage centres may have limited information on which to base decision-making or, in some instances, limited experience of phobia. It also has implications, as discussed in Chapter 4 (parents’ and CYP’s experiences), because CYP entering the study tended to, therefore, be at the more severe end of the phobia spectrum, with higher impacts and cormobidities. This is evidenced by the high ADIS scores at study entry.
Sample size and statistical power
As described above, the number of withdrawals (from both treatment and follow-up) and the impact of COVID-19 meant that a smaller number of participants had contributed primary outcome data than originally estimated. We exceeded our target of 136 participants with primary outcome data given that 149 participants contributed to the ITT primary analysis. However, fewer (113 participants) contributed to the PP analysis. Despite this, the CI for PP non-inferiority assessment was consistent with ITT and wholly below the non-inferiority threshold and, thus, provided sufficient evidence to declare non-inferiority.
Generalisability
Although recruitment to ASPECT was geographically diverse and conducted within a variety of different service settings, limited ethnic diversity was evident in the overall sample recruited. Of the 274 CYP recruited to the study, 95.5% described their ethnicity as white English, Welsh, Scottish, Northern Irish or British, alongside 94.4% of their parents/guardians. This is below that expected from some of our localities, which had higher rates of ethnic minority populations. This may be because some families in ethnic minority groups are less likely to access CAMHS because of stigma in their community,102 fear of joining research studies,103 or a perception that specialist services are not accessible to minority cultural groups or lack of awareness of services.104,105 This may have also been compounded by a lack of awareness of treatment availability for specific phobias, as highlighted in the qualitative findings of the study.
Waiting times
Waiting times in CAMHS are currently long,106 which has been reported in a number of recent reports and papers.107,108 This created problems in this study in that some CYP had not started treatment (OST, n = 11; CBT, n = 12) or had started but not completed treatment (OST, n = 7; CBT, n = 32) prior to their main outcome research measurement point. This gives good justification for the use of both ITT and PP analysis. We did not research the impact on waiting times in ASPECT, but it is possible that OST (mean number of sessions 1.7 when including those attending and not attending treatment) could reduce waiting times compared with CBT as usual (mean number of sessions 4.9). This is not a straightforward comparison because, although OST involved fewer sessions and less overall therapy time, it may require more planning per session, but this could ultimately equate to less planning time overall. OST may also require better diary management than CBT as usual. Future research would be necessary to prospectively examine these subtle differences.
Defining one-session treatment
Training for OST has become available in some of the training programmes for child mental health practitioners within the CYP-IAPT landscape109 in the three settings of low-intensity work, school-based provision and specialist CAMHS. There were relatively small departures from the PP definition of OST. For example, some services provided a third session after OST, which was primarily to fulfil the service requirements for a follow-up session to gather outcome measurements; this was the main reason for an increased average session number in the OST group (2.2 sessions, among those who attended treatment), with 6% of participants receiving more sessions, or longer treatment, than the OST PP definition. The mean time in hours for each treatment group (OST group, 3.8 hours; CBT group, 5.4 hours) reflects the way that clinical time is organised in each treatment (i.e. concentrated into fewer sessions in OST), as well as time pressures facing CAMHS clinicians.
Outcome measurement challenges
Several weaknesses were also identified pertaining to the data collection methods employed within ASPECT.
It became clear early in the study that scoring the ADIS was complex, given that the scores recorded independently by CYP and parent/guardians were sometimes very different. The CSR ratings from the ADIS used for analysis were scored by trained research assistants (the majority of whom were mainly psychology graduates). The ability of the ADIS to allow clinical discretion in reaching a final ‘consensus’ score between CYP and parent/guardian has been used by some studies.33,37 A meeting of our clinical experts agreed that we should use the clear and strict process identified in a methodological paper by Evans et al.,110 as recommended by the ADIS instrument authors.111 This uses the higher of the scores given by CYP or parents/guardians, unless there is a strong reason to depart from this (e.g. clear evidence of unreliability of one score), in which case it is acceptable to apply judgement. Raters were trained in this difference in approach. The higher overall scores found in our study at baseline than those found by Ollendick et al.33 may have been because of this approach and/or may have been related to the higher threshold of mental health problems to enter CAMHS found in this study.
Unblinded assessments occurred in very similar amounts in both groups of the study and were unlikely to create bias to the main results. The study team did everything that it could to mitigate these by providing alternative blinded research assistants where possible. Future studies could consider how to achieve this more effectively when there are very diverse geographical sites and limited numbers of research assistants.
The BAT assessments appeared to work well where the stimulus was real. However, as discussed previously, the BAT was more problematic where the stimulus was simulated. There were two types of simulation. One was where the feared object was manufactured/feigned but not real (e.g. vomit). The second was where the situation was feigned (e.g. where the CYP with needle phobia watch a fake arm used for medical training receiving the injection but know that they themselves will not receive that injection) or videoed (e.g. a storm watched on a video when the weather outside is good). Some international clinical colleagues use live injections in the BAT (Dr T Davis III, personal communication) but this is possible only in centres where this is readily available. Some CYP were able to reach all 10 steps of the BAT. These CYP are distributed approximately equally across treatment groups (n = 22 CYP in CBT; n = 18 CYP in OST). It became apparent early in ASPECT that many CYP with a BII phobia were able to successfully reach step 10 of the BAT at baseline (n = 39); this was 58% of those with a BII phobia. We believe this to be related to the inability to create a realistic simulated BAT for this phobia. Given that 25% of the overall study sample had a BII phobia, we explored this with sensitivity analyses and found that the treatment difference between OST and multisession CBT was in keeping with the difference for the main analysis, as it was also between real and simulated stimuli used in the BAT (see Table 12). Comparability with other child phobia studies needs to consider that some studies do not include BII phobia (e.g. Ollendick et al.33). Difficulties simulating an actual blood draw/immunisation in therapies for adults with specific phobias have been described elsewhere (e.g. Hiermeier and Mofrad99).
Identification of subjects in complex systems
In some localities in which there was a SPA followed by referral into a service, there was a difference in opinion about whether or not a child required phobia treatment at that point. This was usually where the triaging SPA identified a child with a phobia and where a subsequent assessing clinician felt that treatment for a phobia at this point was not appropriate because of a comorbid mental health condition (e.g. depression or generalised anxiety). This could be both a weakness, as the child may have been randomised for treatment and then followed a different treatment path, or a strength, as it indicates that a child’s evolving mental health problems are given careful consideration to provide the most appropriate treatment. A total of 42.4% of those who withdrew from the trial did so in response to a therapist’s decision.
Strengths and weaknesses of the health economic analysis
To the best of our knowledge, this is the first study evaluating the cost-effectiveness of OST for CYP with specific phobias. The analyses account for the costs from a range of perspectives and the QALYs measured by different instruments. However, several weaknesses of the economic evaluation were apparent. First, and most importantly, cases that could not receive allocated interventions or where interventions were interrupted as a result of the COVID-19 pandemic were a concern. Although this could potentially introduce bias to our results, the sensitivity analysis for those who had interventions within the 6-month follow-up period showed that the results were largely the same as our primary analysis, thus supporting the robustness of the results. Sensitivity analyses imputing missing 6-month primary outcome data were consistent with the finding that OST was non-inferior from the primary analysis. Furthermore, the analysis of secondary outcomes (which included 48 participants who were unable to complete their primary outcome as a result of the COVID-19 pandemic but could complete secondary outcomes remotely) found the treatment groups to be similar, in keeping with the primary analysis results. Second, service use data were collected retrospectively and there may have been recall accuracy problems. However, this would not be likely to affect one group more than another. Hence, the study comparison results are likely to remain unchanged.
Finally, although the costs of DNA appointments to the NHS can be considerable,112 our study did not take these costs into account owing to the data constraints. This decision aligns with the NHS England’s 2020 national cost collection guidance for mental health, which advises that missed appointments/DNAs should not be included in the cost collection.113
Strengths and weaknesses of the qualitative study
Our nested qualitative studies explored the feasibility and acceptability of OST from the perspective of the clinicians delivering the intervention, and the CYP allocated to receive the intervention and their parents/guardians. Synthesising the views of these different stakeholders provides a multidimensional insight into the acceptability of delivering OST. Our qualitative findings help to elucidate the intrinsic and extrinsic factors that may have influenced intervention delivery, engagement and outcomes, and, in doing so, help to contextualise the effect sizes observed in ASPECT’s quantitative analysis.
The sampling at baseline was close to the overall sex ratio. Our qualitative studies are subject to some limitations. We explored, in depth, families’ and clinicians’ views of OST but owing to resource constraints did not explore the acceptability of standardised CBT in the trial. Although many of our findings were shaped by the format of OST, we are unable to discriminate with certainty the aspects of acceptability that are unique to this intervention from those that may also apply to alternative forms of provision. The pre-treatment preferences of clinicians and participants were quantitatively measured at trial baseline but this information was not updated at follow-up; therefore, we are unable to compare stakeholders’ reflections of the two treatments at trial completion. Where we examined preferences, there was a trend in the expected direction of preference for both therapist and participant, but these were not statistically significant. Both the specific and the transferable facets of treatment acceptability are important for implementation and add value to future health service development and refinement.
The recruitment pools for our parent and CYP interviews were set by the number of individuals who (1) were randomised to OST and (2) consented at baseline to participate in our qualitative studies. Obtaining consent to be contacted about the qualitative study at baseline guaranteed that recruitment occurred independently of OST engagement or outcome, but additional steps could have been taken to ensure maximum variability in our interview samples with respect to engaging and non-engaging participants. All interviewees had consented to participate in a RCT comparing treatments for specific phobias and, for this reason, may be argued to have displayed a levels of openness towards psychological therapies that was atypical of the general population. Demonstrable compromises in QoL were evident in our data set at both the individual and the family levels. It is, therefore, possible that participants’ reflections on the acceptability of OST may have been inflated by these past experiences or by the cognitive–emotional barriers that they had previously encountered in identifying and accessing appropriate care.
Data saturation was agreed between the researchers contributing to qualitative data analysis and was achieved in both our parent and our CYP samples. We continued to recruit after reaching this group decision and completed an additional three interviews in each sample to confirm our beliefs. Many participants articulated positive aspects of OST and a common appreciation for its personalised, child-centred approach emerged across all of our data sets. At the same time, these views were challenged by a strong narrative proposing that child referral and suitability guidance may need to be revisited, suggesting that our sampling approach and data collection procedures were sensitive to a variety of participant experiences and nuances in the views obtained.
The sample of CYP recruited to our qualitative interview study compared favourably to our trial population with respect to age, sex and phobia type. The inclusion of children in the acceptability study was dependent on the consent of their parent but was not reliant on a parent/carer–child dyad taking part. Children were interviewed separately from their parents wherever possible to maximise the likelihood of obtaining child-centred insights and accounts.
Multiple researchers were involved in conducting interviews, most of whom had previously collected baseline and follow-up measures from families as part of ASPECT. Therefore, for some participants, relationships may already have been built, facilitating interview style and dialogue. However, a clear distinction between service delivery teams and trial outcome assessments is not always apparent to research participants, and it is equally possible that some families may have been reticent to express unfavourable views. The mix of opinions evident in our data set goes some way to negating this concern.
All interviews were conducted after OST had been completed and the trial’s primary outcome measure had been collected. This design maximised trial validity but left our qualitative data vulnerable to recall bias. Children’s accounts, as with parents’ and clinicians’ accounts, may be subconsciously influenced by a post hoc reconstruction of events. Flexible and creative methods were employed to enable and support data collection with younger children, but the accounts of older children were noticeably more detailed, facilitating a deeper latent analysis. Although this may have biased our interpretative themes towards the views of older and more articulate children, our cross data set synthesis helps to ensure that a holistic and robust account of OST acceptability has emerged.
The recruitment pool for our clinician interviews was set by the number of health professionals who were trained in OST and participated in ASPECT. It may be argued that these individuals could be more open to service development than their peers; however, the decision for clinicians to participate was determined at the site level, by managerial agreement, rather than by practitioner self-selection. It is possible that the clinicians who participated in our qualitative interviews displayed higher levels of cognitive or emotional investment in OST than those who were more reticent in coming forward. However, a broad range of experiences were collated and the recruitment of participants from different trial sites and service backgrounds maximised heterogeneity in professional roles, service contexts and views.
By employing a purposive sampling approach, we recruited clinicians who had delivered OST to at least one child in ASPECT. Very few of these individuals had prior experience of treating specific phobias in CYP. Their assessments of their confidence and skills in delivering OST may not be too dissimilar to their broader practice community.
We did not set an a priori threshold for intervention acceptability. Our three qualitative studies provide a valuable insight into practitioner, patient and family experiences and, for this reason, enable the identification of early implementation barriers to OST uptake and delivery. Our analyses benefited from data coding that was undertaken independently by six researchers. Our data interpretations were also discussed with three PPI representatives, optimising the rigour of our synthesis. We separately undertook intervention fidelity assessments (reported in Chapter 4), but did not obtain observational data on intervention delivery during ASPECT.
Strengths and weaknesses of the fidelity assessment
It is difficult to draw conclusions regarding the fidelity in the trial owing to the small number of recordings available for fidelity assessment. Those that were assessed had moderate/high fidelity and were competent in the delivery of the treatments.
The main limitation of the fidelity assessment was the number of recorded sessions available for review by the team. This was because of young people not consenting to the recording, therapists choosing not to record sessions and difficulties in recording active treatment sessions that were highly mobile (e.g. with dogs). High clinical staff turnover (reported in Chapter 3) also created problems with keeping track of the location of recording devices. It is possible that this limited number of recordings was not representative of the delivery of the two treatments within the trial owing to the Hawthorne effect. However, the self-reports for the recorded sessions tend to report using fewer strategies than the fidelity assessors, suggesting that therapists were not recording their ‘best’ sessions only. We did aim to record as many sessions as possible with participant consent; however, owing to the geographical spread, the number of teams and therapists involved in the delivery, and a global pandemic we were able to record and retrieve only a small number.
ASPECT in the context of extant literature
In the phobia treatment RCT conducted by Ollendick et al.,33 49% of CYP were deemed phobia diagnosis free (as indicated by an ADIS CSR of < 4) in the OST treatment group at the similar time point of 6 months to our study. This apparent better recovery rate needs consideration. Their study and previous research in this area are different from our study in a number of key respects. First, Ollendick et al.33 used a consensus diagnosis procedure in which the highest ADIS CSR was not necessarily taken (often the midway point between parent and child was taken), whereas our study used the highest score from parent or child. Second, Ollendick et al.33 did not include BII phobias, which are a subtype of phobia that are difficult to treat. In ASPECT, 83.6% of CYP had one of three types of phobia: animal (29.9%), vomit (28.7%) and BII (25%). Higher rates of animal phobias have been reported in Ollendick et al.33 (56%) than the rates found in ASPECT and other similar RCTs (e.g. Öst et al.,34 50%; Flatt and King,114 43%). Although our rates have been comparable to some studies in relation to BII phobia (e.g. Öst et al.,34 23%), these were excluded from Ollendick et al.33 because they are regarded as complex to treat. The rates of vomit phobias were much higher in ASPECT than reported elsewhere (e.g. Ollendick et al.,33 1%). It should also be noted that many studies do not provide specific breakdowns of phobia types, instead grouping by subtype (e.g. situational); therefore, comparing rates of phobias, including BII or vomit, is difficult.
We had relatively few referrals for environmental phobia (e.g. storms) or dental phobias, although a number of isolated referrals were made for specific objects (e.g. buttons). Some of these carried high levels of impact (e.g. engaging in school activities in the context of a paper phobia). The large number of CYP with animal phobia is not surprising and these phobias were mainly dogs and spiders (both of which are very prevalent in society). Large numbers of BII phobias are also not unusual given CYP’s need for vaccination injections, particularly during our sample age range (e.g. human papillomavirus, MenACWY and three-in-one teenage booster), and for injections in the context of illness (e.g. blood tests and insulin in CYP diagnosed with diabetes mellitus). Vomit phobia has been relatively under-researched, but in most individuals the phobia starts before adulthood.115 It can affect them in multiple ways, leading to leaving or avoiding school, avoiding being with children for fear that they might vomit or avoiding certain foods or situations for fear that they themselves might vomit.115 These multifaceted impacts suggest that for some it can be a serious condition.
In addition, the study by Ollendick et al.33 used the ADIS CSR consensus rating score after clinician discussion of the parent and child scores, whereas we used the suggested methodology of taking the highest of the parent and child scores (i.e. ‘composite score’, Evans et al.110). Furthermore, our selection of recruits, mainly coming from CAMHS, meant that we had large numbers of children with severe, more complex anxiety disorders. This appears to be borne out by the finding that our pre-treatment ADIS phobic anxiety scores are higher than another large study in the USA that recruited through largely community mechanisms.37
One important factor is that earlier studies have been conducted in highly controlled research contexts 33,37 using trained psychology graduate therapists. Our study was a pragmatic trial in the real world involving existing clinicians from various child mental health treatment centres across England. It has been suggested that findings in research clinics or contexts may show larger differences than later real-world trials, one aspect of the research–practice gap in mental health research.116 This may be a factor here.
In summary, the finding that many CYP were still classified as having a specific phobia at follow-up (occurring in similar rates in both groups) may be related to a number of factors, including the COVID-19 pandemic affecting the study and causing treatment delays; the inclusion of phobias that are more difficult to treat; drawing largely from a CAMHS population in which CYP with phobias are more likely to have complex phobias or other cormobidities; a difference in the way that we measured phobia severity compared with other studies, which was more stringent; and the fact that the research in real-world services drew from clinicians with a range of backgrounds and skills (including trainees). Other factors may also have been at play.
Implications of ASPECT for future research
This study has shown that OST is as good as multisession CBT for CYP and that it is cost-saving. There are some challenges in organising therapy in busy services but it also carries some potential advantages, such as shorter treatment times.
A priority for future research would be to focus on whether some phobias respond better to OST or multisession CBT than others. For example, our subgroup analysis suggests that vomit phobia may respond better to OST than multisession CBT. Similarly, given some of the qualitative findings, it is possible that some phobias are less well suited to OST or that some CYP may do better with either OST or CBT (e.g. potentially based on age or comorbidities) and this could be explored in future research. On a similar note, there is some suggestion in the results that treatment response may be moderated by clinician experience, with more experienced clinicians associated with better OST outcomes. This would require further research because it may have implications for which therapies can be offered by different therapy teams (where the experience of clinicians may vary, for example, between school-based and clinical-based NHS child and adolescent mental health teams).
Further research would be helpful in identifying good outcome measures for phobia research and is also very important in clinical practice with CYP. The ADIS can be operationalised in different ways, as discussed above, and is influenced by the perspectives of raters. The BAT appears to be useful in animal phobias or those involving real exposure, but less useful in BII or other phobias for which the stimulus is simulated and ceiling effects appear.
Finally, given that a phobia is a life-long condition, more implementation science research could be conducted into how service delivery models and treatment options can enable more CYP to receive treatment and to improve the recovery rates. Any such research would need to be well integrated with commissioners and providers, and with strong PPI. The research has highlighted the need for future work in this area, particularly in developing quality standards and service specifications to develop training and care pathways for specific phobias in CYP.
Conclusions
We conclude that, in a UK context, OST for specific phobias in real-world current child mental health treatment centres is as clinically effective as the existing most common treatment of multisession CBT and is highly likely to be cost-saving. We also suggest that more work needs to be carried out in the development of service specifications, training and care pathways by commissioners and providers to ensure that treatments are readily available across the country for this treatable lifelong condition.