Feasibility randomised controlled trial – results

The project was planned to be conducted over 24 months, beginning on 1 October 2019. Training of identified staff members to deliver IPC-A was planned to commence before the project began and be completed during months 1–3. Recruitment was planned to take place in months 4–15. Follow-up assessments were planned to be completed in month 21. Months 22–24 were to be dedicated to data cleaning, analysis and dissemination.

Although recruitment started as planned in January 2020, the study was suspended in March 2020 due to the first lockdown of the COVID-19 pandemic, which temporarily suspended National Institute for Health and Care Research (NIHR) research activities. The study recommenced in July 2020, with refresher training being offered for therapists. Recruitment then recommenced in September 2020, creating a 7-month delay. By February 2021, however, due to reported impacts on services from continued lockdowns and other COVID-19-related complications, the number of referrals was significantly impacted. Sites had to prioritise urgent cases only or focus only on safeguarding issues. Teams also experienced a change in types of referrals received, reducing the number of trial-eligible young people and lowering the capacity of staff to take on ICALM cases. IPC-A-trained staff also withdrew, citing new pressures resulting from the increased demand and reduced capacity, reducing the number of staff able to carry out the intervention. This difficulty in recruitment prompted changes such as the addition of new delivery sites, offering additional training sessions to train more staff and creating alternative referral pathways (see Recruitment and retention).

In July 2021, due to the ongoing difficulties resulting from the COVID-19 pandemic, and the delay caused by the suspension of the study at the start of the pandemic, an extension was requested and granted by NIHR. According to the revised timeline, recruitment and follow-ups were due to end between December 2021 and February 2022, with March 2022–October 2022 being dedicated to data cleaning, analysis and dissemination. Recruiting sites felt they had begun to normalise and were more positive about their ability to recruit, although the ongoing pressures on the health system limited their ability to prioritise the research trial.

Recruitment and retention

Recruitment of delivery sites

In September 2019, eight teams across Norfolk and Suffolk were approached about taking part in the study. Seven teams (Site-01_Early Help and NEET Team, Site-02_Early Help and NEET Team, Site-03_Early Help and NEET Team, Site-05_Charity, Site-07_Charity, Site-11_Early Help and NEET Team, Site-14_School Nursing) gave their agreement to take part. Six teams were based in Suffolk, and one team was based in Norfolk. The team that declined to take part had previously participated in the single-arm study of IPC-A and stated the reason for declining was that they felt IPC-A did not fit well due to the complexity of their clients.

Site-05_Charity suspended their involvement in December 2020, citing competing pressures and planned to restart their involvement once they had greater capacity; however, this did not happen, and they withdrew from the study in April 2021.

In response to COVID-19, changing team responsibilities and no longer having any teams referring to the study based in Norfolk, it became necessary to recruit further teams to support the study. In May 2021, six additional teams gave their agreement to take part (Site-04_Wellbeing Service, Site-06_Charity, Site-08_Community NHS Trust, Site-09_Family Support Team, Site-12_Family Support Team, Site-13_Family Support Team).

The Family Support Teams provide support through the Family Support Process and child-in-need (CIN) care plans. They work with children and families where there are concerns about the children’s well-being. It is unusual for these teams to get referrals where the only concern is adolescent mental health; there is a much more complex picture, often including a level of relational dysfunction. However, mental health concerns are the norm within their service users. Unlike mental health services, assessments will nearly always include all family members and will most likely be carried out using a Signs of Safety framework.

Site-04_Wellbeing Service offer Tier 2 mental health services to young people aged 4–18. The team consists of children’s wellbeing practitioners (CWPs) (without a core professional mental health qualification) and PMHWs (with a core professional mental health qualification) and provides brief interventions to young people with mild/moderate difficulties. Referrals are via the Site-10_Central referral point, the same referral point as Suffolk Early Help teams. As per trial protocol, staff without a core professional mental health qualification delivered IPC-A as part of the trial.

Site-06_Charity sees young people with mild to moderate mental health difficulties aged between 0 and 21 and works closely with Site-05_Charity. The service has a variety of staff without core mental health qualifications, including CWPs and educational mental health practitioners. Referrals to Site-06_Charity come via either professional or self-referral. Site-06_Charity have a track record of actively participating in research and have a keen awareness of what this involves. Site-06_Charity withdrew in September 2021, citing staff turnover and lack of capacity to train more staff as the reasons for withdrawing.

Site-08_Community NHS Trust provide health advice and support for young people at a universal services level. Part of this role involves delivering time-limited interventions to young people aged 12 and above with new and emerging mental health difficulties. Such interventions are offered by resilience and emotional health practitioners (REHPs), who do not have a core mental health qualification. Most referrals to this service are either from the young person themselves, a parent or a school representative.

A diagram of the sites can be seen in Figure 5.

Of the 13 teams who were sites for the feasibility trial, 7 made referrals to the study, 3 (Site-02_Early Help and NEET Team, Site-06_Charity and Site-08_Community NHS Trust) delivered IPC-A/TAU of which 2 teams provided recordings (Site-08_Community NHS Trust and Site-02_Early Help and NEET Team).

Changes to referral pathways

Although the majority of teams had agreed to offer both IPC-A and TAU, due to difficulties with retention of trained IPC-A staff, clinical capacity, lack of suitable cases and stagnant waiting lists, there were a number of changes to referral pathways as well as the type of intervention teams were able to offer.

Site-04_Wellbeing Service/Site-14_School Nursing

With difficulties due to many young people being ‘stuck’ on the Site-10_Central referral point waiting list, preventing possible eligible referrals, new pathways were sought. There were difficulties with Site-14_School Nursing having limited capacity to deliver the trial, but having eligible young people, and Site-04_Wellbeing Service who had IPC-A trainees but few eligible young people. It was suggested and agreed with locality managers and commissioners to set up a new pathway to allow referrals to the Site-04_Wellbeing Service from the Site-14_School Nursing. Site-14_School Nursing team checked for eligibility and willingness to take part in the study and then referred to the research team for consent and eligibility checks. Site-14_School Nursing would then transfer the case from Site-14_School Nursing to Site-04_Wellbeing Service. If ineligible, the case remained with Site-14_School Nursing. Only at the point of assessment at the Site-04_Wellbeing Service, if a case withdraws from the study or is no longer suitable, would the case stay with Site-04_Wellbeing Service to finish any treatment, as the case would unlikely be suitable for Site-14_School Nursing. If suitable, the research team would be informed, and the case randomised to Site-04_Wellbeing Service for the intervention and informed if the case is TAU or IPC-A. The intervention and the follow-ups would continue as normal.

Site-04_Wellbeing Service and Site-07_Charity

Similar to the Site-14_School Nursing and Site-04_Wellbeing Service pathways, an agreement was set up for Site-04_Wellbeing Service and Site-07_Charity. Site-07_Charity is a charity that offers brief, ‘light touch’ early interventions to young people experiencing mental health difficulties. In terms of the severity of mental health difficulties they offer treatment for, they sit below Site-04_Wellbeing Service, and young people can self-refer. This service is largely made up of counsellors who, due to their training and qualifications, were deemed ineligible to provide IPC-A as part of this study. However, Site-07_Charity offered to provide additional TAU capacity to Site-04_Wellbeing Service, thereby minimising contamination and offering a clear alternative TAU. Site-04_Wellbeing Service would go through their usual process of checking if appropriate for ICALM and then checking if they were interested. The referral form was then sent to the research team to provide an information sheet and consent form, as well as confirming eligibility. If eligible, the young person was then randomised. If the young person was randomised to the TAU arm of the study, Site-07_Charity and Site-04_Wellbeing Service discussed who would deliver the TAU. The staff member of the respective team then completed the participant consent form, and the intervention and the follow-up continued as normal.

Site-04_Wellbeing Service and Site-07_Charity have a contract, which means that any referrals from Wellbeing will be seen within 4 weeks.

Site-10_Central referral point

In a meeting held with team leaders in August 2021, managers highlighted challenges in getting appropriate referrals from the central referral point as a key barrier to recruitment. Managers fed back that due to pressures at the central referral point, the Early Help teams were being allocated more complex cases, instead of the mild cases they are commissioned to take on.

As a result of this feedback, in November 2021, the ICALM study team initiated discussions with representatives from Site-10_Central referral point, Site-02_Early Help and NEET Team and Site-04_Wellbeing Service to set up a fast-tracked referral pathway for suitable ‘green’-rated (low-risk) cases from Site-10_Central referral point to ICALM. The process also received support from a representative from the commissioning group as it was in line with government agenda. Site-10_Central referral point was not a full site for the study; this site supported referrals to the study only.

Delivery site characteristics

An overview of the characteristics of the 13 sites that agreed to take part in the feasibility study is provided in Table 1. Site-10 has not been included as it was a referring site only. Detailed site characteristics can be found in Table 1.

Staff participants

In May 2021, the format of IPC-A training was significantly altered by being undertaken online due to COVID-19 restrictions and included several sessions devoted entirely to explaining the trial. This included detailed explanations of equipoise and randomisation. These training materials had been adjusted and routinely delivered and distributed to new teams or new staff when the study was explained. This was changed to help understand randomisation and to support practitioners within therapy delivery teams, as it was observed that the clinical practitioners within these teams were not familiar with research processes such as randomisation, which was seen as a barrier to recruitment.

In total, the study team trained 19 staff members as IPC-A practitioners who all gave consent to take part in the feasibility study. However, subsequently, eight withdrew from the trial. Four consenting IPC-A practitioners were allocated clients. Two IPC-A practitioners delivered the intervention. For those practitioners who did not deliver IPC-A, reasons given were (n = 1) client was offered IPC-A but did not engage with the service and did not attend any IPC-A appointments offered and (n = 1) client was not appropriate for IPC-A due to complexity and risk and was referred for a more appropriate intervention. Staff providing TAU were not consented until they were allocated a client. In total, three staff members providing TAU gave consent to take part in the feasibility trial. Two staff members delivered TAU. One staff member did not deliver TAU as their client withdrew from the study.

Participant flow

In total, 32 referrals were received, with 16 eligible participants being recruited and randomised (Figure 2). Two of these participants were recruited and randomised before the suspension of the study due to COVID-19. These participants were informed about the decision to discontinue follow-up data collection due to the current COVID-19 crisis causing research to be halted. Therapy (either IPC-A or TAU) continued, but the research practitioner did not contact participants for any more assessments. Of the remaining 14 participants, only 7 received an intervention, with 3 out of 6 participants receiving IPC-A and 4 out of 6 receiving TAU. Among participants who did not complete IPC-A, one did not engage with the service nor attended appointments. Two participants were found inappropriate for ICALM and IPC-A due to complexity and/or risk at rescreening by the IPC-A clinician, highlighting a progression of symptoms between baseline assessment and start of IPC-A. This led to the clients being discharged and signposted to another service. Concerning the two participants who did not complete TAU, not all data were able to be collected as to whether they received any intervention due to staff members not completing the Attendance Questionnaires.

FIGURE 2 Consolidated Standards of Reporting Trials diagram for feasibility trial (includes two before suspension of study). (Figure not reproduced here; see the original article.)

Recruitment

Due to remote working, the protocol was changed to allow the young person (or their parent/carer if under 16) to be asked if they are willing to photograph and electronically send a picture of the completed consent form to the research practitioner so that baseline assessments are not delayed by the added complexity caused by postal delivery services. It was also changed to allow interviews/focus groups for the qualitative analysis and treatment and assessment to take place face to face, by telephone/teleconference or videocall/video conference, depending on wishes of participants and safety issues arising from infection risk. Participants were posted copies of the consent form and participant information sheets, with the researcher going through the consent form remotely with the participant. The participant would then take and send a picture of the signed consent form to the research team so further parts of the study could continue. The participant, meanwhile, would post the form back so there was a physical copy for study records. The protocol was changed to accommodate recording online, as this was historically in-person with a dictaphone.

In March 2021, further changes were made in consenting procedures to account for the ongoing changes due to COVID-19. This included having multiple options for carrying out consenting processes, including face to face; via photo and post; and through incorporating the form into the Research Electronic Data Capture system (REDCap) and capturing consent via audio-recording.

Looking at Figure 3 for those who were recruited after suspension (n = 14), the overall rate of recruitment was slower than anticipated: on average, 1.7 per month (18 months) versus a target recruitment rate of five participants per month (12 months). Recruitment, however, started earlier in September 2020 and ended later than predicted in February 2022, compared to predictions starting in January 2020 and ending in December 2021. The alterations to the referral pathways did not increase the number of referrals made to the study.

FIGURE 3 Accrual curve for feasibility trial. (Figure not reproduced here; see the original article.)

Following previous unsuccessful attempts at recruitment, the fast-tracked referral pathway for suitable green-rated cases from Site-10_Central referral point was agreed in November 2021. Prior to this agreement, there were concerns raised from the services around the following areas

Caseloads of therapists

As Early Help managers were aware of the pressures faced by Site-10_Central referral point due to staff shortages and the backlog of cases, they expressed a concern of ‘flooding the services’ by highlighting an incident where approximately 600 cases had been referred from Site-10_Central referral point to another team within a short period of time.

This is an emotive subject for the services because Site-10_Central referral point can see this as a way of offloading cases and may lead to a lot of disquiet. You don’t want to send the message that Site-10_Central referral point is offloading cases. I am concerned that we are not opening the flood gates.

Concerns about the suitability of the cases for their services

The managers felt a number of referrals received from Site-10_Central referral point were not appropriate for their service (even after being advised what the criteria for the service are) and that although they may come through as low mood, there may be other confounding factors which are the cause of this presentation. The main challenge with the referral of cases from Site-10_Central referral point was that there was not a feedback loop which allowed for cases to be returned to the hub if considered inappropriate. The managers of the service therefore felt that there needed to be a feedback loop or support mechanism in place in case the identified cases were beyond the service’s remit.

Site-10_Central referral point not triaging cases before referring

Team managers felt that Site-10_Central referral point did not triage the cases fully, leading to teams not having the full information on the case to decide whether they would be apprioriate for their service. Site-10_Central referral point needed to identify cases that have low-level issues and exclude those that have complexities, for example, violence, safeguarding issues, eating disorders, adoption or learning disabilities.

Usual support

Some teams reported not having TAU for low mood. They would need to ensure provision is made for young people randomised to the TAU arm. This also impacted on practitioners’ willingness to randomise a case they thought suitable for IPC.

Following these concerns from the teams, the plan was to identify a small number of eligible young people (maximum 5) for each of the participating teams based on the capacity of participating therapists and how many cases the practitioners needed to qualify as an IPC-A therapist. The agreed plan was as follows:

Step 1

Site-10_Central referral point clinician to identify potential referrals based on ICALM inclusion and exclusion criteria. Site-10_Central referral point clinician would need to go through the general practitioner (GP) areas to ensure that the cases are being referred to the correct location and right teams. Site-10_Central referral point to refer cases to one of three teams, depending on location: Site-01_Early Help and NEET Team/ Site-02_Early Help and NEET Team/ Site-04_Wellbeing Service to speak to potential participants to see if they would like to find out more about the study:

• If no – no further action with person for ICALM.
• If yes – referral is made to the teams.

Step 2

The cases get reviewed by a team manager for appropriateness before assigning them to the team. If appropriate, the team manager will notify Site-10_Central referral point clinician that the case has been assigned to a manager so that it can be closed at the Site-10_Central referral point and noted that it has been assigned to the ICALM study.

Step 3

Team to refer suitable participants to ICALM team for baseline assessment. If they are eligible for ICALM team consent (young person and family), complete baseline measure and randomise participants (this can take up to 3 weeks).

Step 4

ICALM Trial manager to inform the team of the randomisation outcome. When referred to the team, the young people will be in contact with family support practitioners (FSPs) as per usual.

The perceived benefit of the process included reducing the long waiting list for ‘green’ cases while ensuring that services are not inundated with inappropriate referrals.

The Site-10_Central referral point clinician screened 1300 cases, of which only 31 cases were identified as potentially having ‘low mood’.

I filtered out the queues and removed all cases for potential or a diagnosis of ASD or ADHD, l removed learning disabilities, young people suffering from trauma or who are being violent and having anger issues. I also screened the list for the correct GP practices based on the areas of Bury and Ipswich, leaving only 31 cases.

In total I believe there are 19 possible ones for one of the teams and 12 for another – the ones in red (not referring to RAG rating) I am pretty sure on, however this is purely due to a lack of referral info.

For the 1300 referrals that were screened, only 4 young people were then referred to the ICALM team. None of these referrals consented to take part in the study due to not being able to be contacted prior to the recruitment deadline, and one person declined to participate.

Retention

After suspension of the study, of the 14 eligible participants who were randomised to the study, 12 (85.7%) reached the 23-week follow-up, with 2 (14.3%) participants withdrawing from the study at the allocation stage due to testing positive for COVID-19.

Baseline characteristics

Baseline demographic and clinical characteristics of participants recruited to the feasibility RCT are presented in Tables 2 and 3.

Suitability of outcome measures

Rates of completion

Outcome measures were well completed. Of those successfully followed up at 23 weeks, almost all participants (83%) completed all outcome measures; one RCADS measure was missed at week 23. The Trial Manager/Research Assistant was able to check self-reported measures for missing items and follow up with participants to complete the measures. The researchers then facilitated the assessments at week 10 and week 23, which resulted in no measures being missed.

Descriptive statistics for outcome measures collected

The objective of the study was to assess the feasibility of this study to inform a future RCT. This study was not powered to detect any significant changes in outcomes. Mean changes from baseline by allocated arm are presented in Table 4.

Safety and adverse events

There was one serious adverse event (SAE) recorded during the feasibility trial. The SAE was in the IPC-A arm and was an instance of a 3-day hospital admission due to urosepsis. The urosepsis was considered to have been caused by a urinary tract infection. Details of the SAE were reported to the Sponsor’s Participant Sub-Committee for independent review. It was deemed not to be related to the study procedures.

There were five adverse events (AEs) recorded during the feasibility trial, four in the TAU arm and one in the IPC-A arm. The AEs in the TAU arm were a positive test result for coronavirus (COVID-19), bereavement due to the death of an uncle and an ongoing investigation for a gluten allergy. The AE in the IPC-A arm was one instance of perceived difficulty breathing due to anxiety a few hours after baseline measures were taken. It was unclear whether this was related to the baseline assessment.

Health economic assessment

Follow-up was obtained on 12 participants at the 10- and 23-week periods. For a CHU-9D score to be generated, it is required that a valid response be given for all nine of the questions. Only in this case can a summary score be generated. It can be seen from the results outlined in Table 5 that scores could be derived in all but one case. For one participant at baseline, there was one answer to one question missing. In all other instances, complete data were obtained. Although sample sizes are small, the data indicate that health, as determined by the CHU-9D, was increasing over time. The use of t-tests suggests that the baseline and 23-week scores were statistically significantly different. There was no statistically significant difference between baseline and 10-week scores or 10-week and 23-week scores.

The results of the CHU-9D by individual response to each question are shown in Table 6 and Figure 4 for the baseline responses. The number of responses is given by level for each of the 9 CHU-9D questions. Each question has five possible responses, labelled 1–5 in Table 6 and Figure 4. In all cases, level 1 for each question represents the best response, and level 5 is the worst. Response categories differ by question. For example: level 1 for ‘Worried’ is ‘I don’t feel worried today’, whereas level 1 for ‘Activities’ is ‘I can join in with any activities today’.

FIGURE 4 Baseline responses by CHU-9D dimension. (Figure not reproduced here; see the original article.)

Again, numbers are small, so there are limitations in what can be taken from these data. However, there are some observations that can usefully be made. In most cases, there were responses across the entire range of the possible responses, with only ‘sad’, ‘pain’ and ‘routine’ dimensions recording no responses in the worst state. This indicates that baseline responses in ICALM cover the full range of potential responses in the CHU-9D. This would also indicate that the CHU-9D has the potential to show improvements from baseline in this group of individuals. By contrast, if almost all responses had been at level 1, there would have been little opportunity to show improvements. This may also indicate that the CHU-9D is sensitive to changes caused by depression in adolescents. A second point that could be noted with these data is that the question relating to feeling tired appears to show the most negative responses.

Table 7 shows the completion rates for the CSRI and its various questions at the three time points. At least some sections of the CSRI were completed for 16 participants at baseline, 11 at the 10-week follow-up and 12 at the 23-week follow-up. Responses by category of resource use (individual questions) are shown in Table 7, that is, this shows the number of times this section was completed at each follow-up. Where sections do not appear to be completed, it is likely because a question was left blank. This was particularly common for ‘inpatient stays’ and ‘other drugs’. There were more sections/questions left blank at follow-up than at baseline. There was one CSRI form at the 10-week follow-up that was returned but left blank. Generally, the CSRI appeared to have been completed for participants, particularly at baseline.

Resource use at baseline and follow-up, as indicated by the CSRI, is given in Table 8. Here, categories are only given if there was reported use of that type of resource. The ‘education, other’ category includes items such as personal tutors, attendance officers and success centre support workers. The modified CSRI used asked specifically about a number of named mental health drugs (fluoxetine, sertraline, melatonin/circadin, promethazine, citalopram, propranolol). No participant reported any of these drugs at any time period. The ‘other meds’ category in Table 7 gives a count of the number of different drugs reported. Generally, these are not mental health related. There was only one reported inpatient stay in any time period, corresponding to a 3-day stay. For other uses of secondary care, there was some limited use reported, generally for reasons not related to mental health.

The majority of healthcare uses we asked about were not services that were used by any of the respondents. Contacts were reported for: school nurse, mental health nurse, GP, paediatricians, physiotherapy and other healthcare contacts. There was also reported use of therapist services. Generally, this was reported by a small number of individuals reporting multiple contacts. There was no reported use of social care services, though a few individuals used after-school clubs. There were a number of reported contacts with services by participants’ families – this was mostly in the baseline period.

In general, the CSRI appeared to have been completed by most respondents. There were some categories where the response had been left blank and it was not clear if this was ‘missing’ or zero. However, in the majority of cases, a value of zero was used if there was no reported service use. Generally, there were only one or two missing cases in the data for the majority of data types.

There appears to be a clear pattern of reducing service use over the follow-up period. The exception to this appears to be the use of therapy services.

The number in Table 8 generally corresponds to number of visits or contacts or number of stays. The exception is use of ‘other drugs’ where numbers represent prescriptions.

As part of the economic component, we estimated the costs of providing the IPC-A intervention. Given the very small numbers who had the intervention, it was difficult to estimate what the costs may be if this was applied in a larger sample as it would be in any future trial or if the scheme was rolled out into practice. The costs calculated here can be divided into three categories: costs of providing training; costs of supervision; and costs of providing sessions. In all cases, costs are in UK 2020–1 Great British pounds.

The costs of training are given in Table 9. Training was provided for 2 whole days for 19 attenders. It was assumed that two members of staff provided the training; these were grades 8a and a consultant psychiatrist. It was assumed that these receiving training would be equivalent to grade 4 [Agenda for Change (AfC)] NHS staff. In all cases, costs were taken from a published source.41 The costs of those giving training were a mean of costs for a grade 8a psychologist and for a consultant psychiatrist (£75 and £123 per hour, respectively). The largest cost was the time of the 19 individuals being trained. No cost for accommodation was assumed, but the costs per hour would allow a component for overheads. Based on the 19 individuals trained, a cost per person trained of £636 was estimated. This only includes staff time (with overheads) but does include costs of both trainers and those being trained. There are a number of uncertainties here, so it is very difficult to estimate the training component of the cost of IPC-A. Firstly, it should be noted that of those trained, only two therapists provided intervention sessions. In any future study or routine care, it is likely that a much higher proportion of those trained would go on to practice what they trained, but that this may never be 100%, so the cost per person who goes on to offer IPC-A may be higher than the cost per person trained. However, in this study, there was high turnover of therapists, so it may always be the case in practice that significantly more people need to be trained than will ever provide the intervention. Secondly, in a future trial or in clinical practice, those trained may well provide care for a number of individuals. Also, the training may continue to be relevant for a number of years, so this may also increase the number of individuals to whom training is provided. The 2 staff who provided the intervention gave care to 3 participants or 1.5 participants per therapist. In Table 9, the average cost per person trained (£636) is divided by 1.5 to give an average cost of training per person receiving ICP-A of £42. However, in actual practice, training would enable a person to provide sessions for a larger number of individuals and would presumably be valid for a number of years before retraining was required. So, in a large trial or in clinical practice, the actual costs of training as a component of IPC-A are likely to be considerably lower than the values presented here.

The second component of the therapy cost considered here was that of providing supervision. It was assumed that this was only given to those providing the intervention. To simplify data collection and to reduce burden on participating therapists, no detailed logs of supervision were kept so the costs are based on assumption and expert opinion. Supervision was provided on a weekly basis by two practitioners (a grade 8a and a grade 7) over the course of 6 weeks. This gave an estimate of six sessions of supervision per therapist providing IPC-A. It was assumed that these supervisors provided equal amounts of supervision. Therapists receiving supervision were assumed to be equivalent to AfC grade 4. It was assumed that this would be provided while the therapist was providing the intervention sessions (over a 6-week period) and that no therapist was providing sessions to more than one study participant at any one time. We assumed 18 sessions of supervision were given equally by the two supervisors for the three individuals receiving the intervention. The costs were £75 and 65 for the grade 8a and grade 7, respectively. Assuming 30 minutes gave a cost per supervisory session of £37.5 and £32.5 and a total cost of supervision for the 18 sessions (three sets of 6 weeks) of £630. The cost of the therapist's time for receiving supervision was based on an assumption of 30 minutes and a cost per hour of £35, giving a total cost of receiving supervision of £315 (£35*0.5*18). This gave a total estimated cost of supervision for the therapists providing the three courses of IPC-A of £945 or £315 per person receiving IPC-A.

The third component of costs was the cost of therapists providing the IPC-A sessions. It was assumed that this would be provided by the equivalent of an AfC grade 4. Sessions were assumed to last 55 minutes, and there would be a ratio of contact to non-contact time of 1 to 0.3.42 For the four participants who received TAU, there was an average of 9.25 sessions per person, giving a total cost of £386 per person. For the three individuals who received the intervention, there was an average of 6.7 sessions per person, giving a total cost per person of £278. The costs of the various components for the IPC-A intervention are given in Table 10, with a total cost of £1017 per person.

However, it is important to bear in mind that this estimate is subject to large amounts of uncertainty due to the small numbers. In particular, the costs of training per person receiving IPC-A would be much lower if training costs were spread over a much larger number of individuals receiving therapy. This would be the case in a larger trial or in clinical practice.