Psychological intervention, antipsychotic medication or a combined treatment for adolescents with a first episode of psychosis: the MAPS feasibility three-arm RCT

Joint working between early intervention in psychosis services and child and adolescent mental health services

Child and adolescent mental health services are commissioned to provide care for young people up to the age of 18 years, and this includes young people with psychosis. 61 The NICE CG (CG155)55 recommends that primary care services make an urgent referral to either CAMHS (up to the age of 17 years) or an EIP service (aged ≥ 14 years) that has a consultant psychiatrist with training in child and adolescent mental health. The literature on the prevalence of psychosis cases in CAMHS is limited, so estimates are difficult to obtain. However, in 2011, audit data from Rethink (London, UK)62 indicated that just under half of the CAMHS teams who completed the survey were working with young people with psychosis who were aged < 18 years. CAMHS is structured in four tiers, with increasing complexity and specialism at each tier, as follows: non-mental health specialists working in the community with children (tier 1), CAMHS specialists in community primary care settings with children who experience mild mental health difficulties (tier 2), CAMHS specialists in multidisciplinary community child psychiatry outpatient settings (tier 3) and highly specialised CAMHS clinicians working in outpatient teams, day units or inpatient settings (tier 4).

UK policy for the treatment of psychosis stresses the importance of good communication between EIP services and other service providers, such as CAMHS, in ensuring appropriate assessment. 55 Examples of effective joint working include the provision of CAMHS psychiatry and prescribing for EIP service users aged ≤ 16 years; provision of EIP psychiatry for young people transitioning from CAMHS to adult mental health services (AMHS); input into CAMHS by EIP service team members, including training and attendance at CAMHS team meetings; and employment of FI therapists to work across CAMHS and EIP services. 62 Research conducted with UK primary care trusts (PCTs) and Strategic Health Authorities suggests that joint working is most effective when CAMHS and EIP service staff engage in joint training/education to align their philosophies, and that senior support from PCTs and Strategic Health Authorities is required to facilitate effective working. 62 Although there is evidence of effective working, a report published by Rethink62 found that 67% of EIP staff reported that they did not have training to work with 14- to 16-year-olds and 64% did not have training to work with 16- to 18-year-olds, and half of the EIP teams reported that CAMHS was not a source for identification of psychosis.

There is evidence to indicate that some young people may experience the transition from CAMHS to AMHS, such as EIP services, as an abrupt change that feels poorly planned. 63,64 Research has found that young people report largely negative experiences of the transition from CAMHS to AMHS, citing factors including feeling underprepared for the change in care provider, feeling abandoned by services and being unhappy with the quality of care received. 63–65 A recent Education Policy Institute report66 obtained data from 47 out of 60 CAMHS in England on their current waiting times for assessment and treatment. In 2017–18 the average median waiting time for treatment was 60 days. Waiting times varied substantially across the country, with the 10 CAMHS teams with the longest median waiting times to treatment reporting figures of 82–188 days in 2017–18. The authors concluded that there is still a postcode lottery in the access to time treatment from CAMHS. 66 Specifically, in relation to EIP services, several challenges have been reported in relation to joint CAMHS/EIP service working, including different philosophies,67 role confusion68 and different funding sources. 69 Although having separate EIP services and CAMHS is common across the UK, it has been argued that a unified service should be offered for young people up to the age of 25 years given the high risk of mental health issues starting between the ages of 15 and 25 years. 69 There are examples of such services developing in the UK, such as Forward-Thinking Birmingham (Birmingham, UK), which is an integrated service covering the age range 0–15 years including EIP.

Pharmacological treatment

Antipsychotic medication prescribing for young people with psychosis is increasing72 and, typically, AP medication is the main treatment available for adolescents with psychosis. 71 In the UK, NICE CG (CG155)55 recommends that for children and young people with FEP oral AP medication should be offered, and that the decision about the choice of AP medication should be made in collaboration with the young person and their parents/carers. 55

A number of systematic reviews and meta-analyses on the topic of AP treatment of psychosis in young people have been published. 70,71 In relation to current treatment practice in the UK, the NICE CG (CG155) was informed by the systematic review and meta-analysis published by Stafford et al. 71 This review identified 19 studies of AP medication, seven of which were placebo-controlled trials and 12 of which were head-to-head trials. A meta-analysis was conducted on the seven placebo-controlled trials, with a total sample size of 1094 participants and a mean age of 15.5 years (range 15.4–20.0 years). The drugs that were compared with placebo were quetiapine (Seroquel^®; AstraZeneca UK Ltd, Cambridge, UK), aripiprazole (Abilify^®; Actavis Ltd, Dublin, Ireland), risperidone (Risperdal^®; McGregor Cory Ltd, Bracknell, UK), paliperidone (Invega^®; Janssen: Pharmaceutical Companies of Johnson & Johnson, Beerse, Belgium; and Cilag AG, Schaffhausen, Switzerland), amisulpride (Solian^®; Delpharm Dijon, Quetigny, France), olanzapine (Zyprexa^®; Eli Lilly and Company, Indianapolis, IN, USA) and haloperidol (Haldol^®; Ortho McNeil, Raritan, NJ, USA). Results of the meta-analysis showed small benefits of AP medication over placebo for positive and negative symptoms, depression and psychosocial functioning and large effects for global symptoms. However, data quality across the studies was considered poor and the authors highlight that the placebo groups also improved substantially.

A more recent systematic review and network meta-analysis identified 28 randomised controlled trials (RCTs) of AP medication for children or adolescents with a diagnosis of schizophrenia, schizoaffective disorder or schizophreniform disorder published from 1976 to 2017. 70 The total number of participants across the 28 RCTs was 3003, with a median age of 14.41 years (range 7.7–18.3 years). In total, the studies included in the network meta-analysis provided data on 17 AP medications but the most frequently studied were haloperidol, risperidone and olanzapine. Results of the pairwise meta-analyses revealed that, compared with placebo, the following AP medications were significantly more effective in reducing overall symptoms: olanzapine, risperidone, lurasidone (Latuda^®; AndersonBrecon Ltd, Hereford, UK), aripiprazole, quetiapine, paliperidone and asenapine (Sycrest^®; Schering-Plough Labo N.V., Heist-Op-Den-Berg, Belgium). Results of the network meta-analysis indicated that clozapine was superior to all other AP medication for total, positive and negative symptoms. After clozapine, olanzapine and risperidone had the greatest effect sizes of –0.74 and –0.62, respectively. However, as olanzapine was associated with the highest weight gain, the authors recommended that it should be avoided in children and young people. Network meta-analysis revealed that haloperidol, trifluoperazine (Stelazine^®; Mylan NV, Canonsburg, PA, USA), loxapine (Loxapine succinate^®; Mylan) and ziprasidone (Geodon^®; Pfizer Inc., New York, NY, USA) were not more effective than placebo, and the authors recommended that these drugs should not be used in children and adolescents. Findings from this systematic review and network meta-analysis suggest benefits of AP medication over placebo (except for haloperidol, trifluperazine, loxapine and ziprasidone) and provide useful data to inform decision-making regarding the selection of AP medication and their relative superiority in comparison with each other. However, it should be noted that network meta-analysis relies on the use of indirect as well as direct evidence, and for clozapine the majority of the evidence was indirect. Krause et al. 70 also note limitations to the quality of data, with 50% of the RCTs providing inadequate information about the randomisation procedures and 75% not providing adequate information about allocation concealment. It should also be noted that inspection of the included studies indicates that 7 out of the 28 studies did not use either DSM or ICD-10 diagnostic criteria in application of clinical diagnoses; rather, the studies report that the diagnostic criterion was clinical diagnosis. This may draw into question the homogeneity of the participants included in the studies and the generalisability of the findings.

A review of the literature indicates that since publication of the systematic review and network meta-analysis by Krause et al. 70 two further trials of AP medication for young people with psychosis have been conducted. 73,74 The Tolerability and Efficacy of Antipsychotics (TEA) trial,74 conducted in Denmark, was a head-to-head, double-blind RCT comparing the safety and efficacy of quetiapine-extended release (ER) with aripiprazole in 12- to 17-year-olds with an ICD-10 schizophrenia spectrum diagnosis. The total sample size, across seven university clinics, was 133 (55 allocated to quetiapine-ER and 58 to aripiprazole). The primary outcome was a positive Positive and Negative Syndrome Scale (PANSS) score at 12 weeks, and the results of the trial indicated that there was no significant difference between medications in the primary outcome. However, in both groups positive PANSS scores significantly reduced from baseline to 12 weeks, by approximately 5 points, which is a modest reduction in PANSS score given that the minimum clinically important difference is thought to be about 15 points. 75 Quetiapine-ER and aripiprazole had differing adverse effect profiles. The Quetiapine-ER group had worse metabolic outcomes than aripirazole, with greater weight gain in the quetiapine-ER group. Furthermore, an analysis of the homeostatic model of insulin resistance favoured aripiprazole. However, there were significantly more cases of akathisia and sedation in the aripiprazole group. The number of adverse events (AEs) was high in both groups. In the quetiapine-ER group, the following percentages of AEs were reported: 79% experienced tremor, 92% had increased duration of sleep, 78% experienced orthostatic dizziness, 80% experienced depression, 69% experienced tension/inner unrest, 76% experienced failing memory and 87% experienced weight gain. In the aripiprazole group, 91% experienced tremor, 71% had increased duration of sleep, 81% experienced orthostatic dizziness, 77% experienced depression, 88% experienced tension/inner unrest, 77% experienced failing memory and 68% experienced weight gain. The results of the trial contribute to the relatively limited literature on head-to-head comparisons of AP medication for adolescents with FEP. However, concern has been expressed about the quality of this trial, including the limitations on the conclusions that can be drawn regarding the efficacy and safety given that one-third of participants were prescribed AP medication in addition to those being studied. 76 In the same year as the TEA trial, Correll et al. 73 published a double-blind, placebo-controlled, randomised withdrawal-design trial of oral aripiprazole maintenance treatment. In this study, all 201 participants were cross-titrated to oral aripiprazole and were stabilised on this medication for 17–21 weeks before randomisation in a 2 : 1 ratio to either oral aripiprazole or placebo. The results showed that treatment with oral aripiprazole was associated with a significantly longer time to exacerbation of symptoms of psychosis than treatment with placebo. Interestingly, the rate of serious treatment-emergent adverse effects (TEAEs) and discontinuation because of TEAEs was lower in the medication arm, although not significantly so. 73

Adverse effects of pharmacological treatment

Antipsychotic medications are associated with a wide range of adverse effects, which include metabolic effects (i.e. weight gain), cardiovascular effects, hyperprolactinaemia, antimuscarinic side effects (dry mouth, blurred vision and cognitive impairment), sexual dysfunction and movement disorders. 77 The adverse effects of AP medication are associated with increased stigma, physical morbidity and mortality, poor adherence to medication and reduced quality of life. 77 A systematic review78 of the effects of AP medication on brain volume in adult populations concluded that some of the structural brain changes found in people with a schizophrenia diagnosis may be the result of AP medication. When considering the risk of adverse effects of AP medication for young people, there are biological differences between adolescents and adults that are important to consider. 79 Adolescence represents a time of social and educational change. Medication side effects may affect important educational milestones. The stigma of taking an AP medication may be challenging for a young person’s relationships. Compared with adults, young people may be more susceptible to the adverse effects of AP medication,80 particularly in relation to weight gain. 81 A study by Correll et al. 82 showed that 12 weeks of AP treatment in children and adolescents who had less than 1 week’s prior AP exposure was associated with significant rates of obesity and new-onset categorical glucose and lipid abnormalities. For example, a weight gain of > 7% occurred in 56% of participants taking quetiapine and in 84% of participants taking olanzapine. 82 Krause et al. 70 make an explicit recommendation that this medication should not be used in young people. Moreover, there is an indication that weight gain is associated with high rates of medication discontinuation. 71 In 2016, the NICE CG (CG155)55 was updated to include specific recommendations in relation to olanzapine, advising clinicians who are choosing between olanzapine and other AP medications to discuss with the young person and their parents or carers the increased likelihood of greater weight gain with olanzapine, and that this is likely to happen soon after starting treatment. 55

Summary of pharmacological treatment

Although there is some evidence to suggest benefits of AP medication over placebo for young people with psychosis,70,71 the quality of the evidence is low. 71 Head-to-head comparisons are limited and inferences are often indirect, which limits the extent to which comparisons can be drawn between AP medication. Study durations are short, which means that the long-term adverse effects of AP medication in young people are unknown. Furthermore, the evidence base for AP medication is significantly limited compared with the larger body of adult-derived data, the risks and benefits may be less favourable for young people and the additional benefit of AP medication over placebo is small. 71

Psychological intervention

Psychological interventions for young people with psychosis are recommended in the UK NICE guideline (GC155);55 specifically, cognitive–behavioural therapy (CBT) and FI. The current guidelines suggest that CBT and FI should be offered in conjunction with AP medication. However, if a young person and their parents wish to try PI alone, the NICE CG (CG155) suggests that the care team should advise that PIs are more effective when delivered in conjunction with AP medication, but offer individual CBT and FI if the young person and family wish to pursue that option. In the next sections we will review both CBT and FI, outlining the approaches, the evidence base and potential adverse effects.

Efficacy and safety of psychological intervention for adolescents with psychosis

Although there is an established evidence base for CBT in adult psychosis populations,89–92 the availability of studies to determine its efficacy in children and young people is limited. 71,93 In adult populations, meta-analyses show CBT to have small to moderate effects when delivered in combination with AP medication. 89–92 In relation to FI, NICE conducted a review of the literature for FI for adults with psychosis and found that FI reduced the risk of relapse, the risk of hospital admission during treatment and the severity of symptoms both during and up to 24 months following the intervention. 92 To inform the NICE CG (CG155),55 Stafford et al. 71 conducted a systematic review of PIs for the treatment of psychosis and schizophrenia in children, adolescents and young adults. Searching the literature up to July 2013, Stafford et al. 71 identified eight trials of PI, of which seven included a treatment arm with CBT or FI (one trial was of movement therapy vs. dance therapy). However, no trials of PI (i.e. CBT or FI) carried out exclusively in an under-18 population were identified; rather, the studies included in the review included participants up to the age of 25 years (mean age 23.2 years, range 15–24 years). Meta-analysis of the data from these studies indicates no evidence of treatment effects on symptoms and low-quality evidence for the combination of CBT and FI on the number of days to relapse. In addition to the limited evidence base, the quality of studies in under-25-year-olds was questionable, with a risk of detection bias from inadequate concealment of allocation across all studies and four trials at high risk of selective reporting. 71

Since the publication of the review by Stafford et al. ,71 a further review has been produced by Anagnostopoulou et al. 93 that included studies of PIs published up to 2017. The aim of the review was to evaluate the efficacy of PIs for young people with EOP (before the age of 18 years) in relation to positive and negative symptoms, cognitive functioning and psychosocial functioning. The authors included papers in which the study was a RCT and the participants were aged 12–18 years and had received a schizophrenia spectrum diagnosis. The review identified one controlled trial of CBT versus FI94 and one RCT of a psychoeducational group intervention for adolescents with psychosis and their families,95 both in a strictly under-18-year-old population.

In their pilot controlled trial of CBT versus FI versus treatment as usual (TAU), Browning et al. 94 recruited 30 adolescents with psychosis who had been admitted to a psychiatric inpatient unit and allocated them to receive CBT and TAU (n = 10), FI and TAU (n = 10) or TAU only (n = 10). TAU included, as a minimum, medication (although the authors do not specify whether this was AP medication or other mental health medications), a nursing care plan and group-based activities on the unit. CBT was adapted for young people by delivering shorter sessions twice per week. Participants allocated to receive CBT could access 10 sessions. FI comprised five 1 hour-long sessions and was delivered over 4–10 weeks. The study conducted by Browning et al. 94 provides data to suggest that it may be feasible to recruit young people to a study comparing PIs. However, there was a high risk of selection bias owing to the non-randomised design and the numbers recruited were too small to make any inferences regarding efficacy, although there was an encouraging signal from the data that participants in the PIs reported greater satisfaction ratings than those receiving TAU alone.

A secondary analysis of the SoCRATES trial,96 which recruited people with psychosis who had a recent first or second inpatient admission, found that participants under the age of 21 years who had supportive counselling showed significantly greater improvements at 3 months on the PANSS positive, PANSS general and the PSYRATS (Psychotic Symptoms Rating Scales) delusions subscales than those who received CBT or TAU. 97 In addition, among those under the age of 21 years, the therapist-rated therapeutic alliance was significantly poorer in the CBT group than in the supportive counselling group. 97 The authors concluded that PIs for people with early psychosis may need to consider age-related factors. 97 Although these findings contribute to the literature, it should be noted that this is a secondary analysis of the SoCRATES study, which was not designed to evaluate these treatments specifically in young people with psychosis, and so the conclusions that can be drawn are limited.

In their RCT of a group-based psychoeducational intervention for young people with psychosis and their families, Calvo et al. 95 recruited 55 young people and their families and randomly allocated them to either a psychoeducational problem-solving group arm (n = 27) or a time-matched control arm (n = 28). At the end of the intervention, young people in the psychoeducational intervention arm had significantly fewer visits to accident and emergency and lower negative symptom scores than those in the control arm; however, these results were not sustained at 2-year follow-up. 95 The results of this study are a signal that, as with adult populations, FI with adolescent populations and their families may be effective in reducing relapse. However, to our knowledge, the study by Calvo et al. 95 represents the only trial of FI in an under-18-year-old population, and any inferences that can be drawn from this study with a small number of participants is limited.

Overall, there is a paucity of evidence from which conclusions can be drawn about the efficacy of CBT and FI for psychosis in children and young people. In producing the UK NICE CG (CG155),55 the guideline development group considered if there were grounds for recommending that treatment with PIs in young people should be any different from that in adults. Given the paucity of evidence for CBT and FI in young people, the guideline development group utilised the data from the much larger adult evidence base to make the recommendation to offer CBT and FI in conjunction with AP medication. CBT trials have been criticised for poor reporting of adverse effects98 and data on the safety of PIs for adolescents are lacking. There clearly remains an important need to address the question of efficacy and safety of CBT and FI for this population.

Psychological intervention in the absence of antipsychotic medication

It has been argued that evidence from meta-analyses of adult psychosis studies demonstrates that the superiority of AP medication over placebo has been overestimated and that the adverse effects of AP medication have been underestimated. 99 For this reason, there has been a growing interest in the efficacy and safety of PIs in the absence of AP medication. A systematic review and meta-analysis of psychosocial treatment with a time-limited postponement of AP medication versus initial AP treatment found that psychosocial treatment may be effective in the absence of AP medication, with psychosocial treatment having a small–medium effect size advantage over AP medication. 100 Although the findings from Bola et al. 100 suggest that it is feasible to conduct research into psychosocial interventions in the (short-term) absence of AP medication, the quality of the studies is limited, with only one RCT, and the psychosocial interventions evaluated varied across the studies and represented treatment programmes rather than specific PIs, such as CBT or FI.

In relation to CBT, Morrison et al. 101 demonstrated that it was feasible and safe to recruit adults who chose not to take AP medication to a RCT of CBT and TAU versus TAU alone. Data from this trial showed that CBT may be effective in reducing symptoms in adults who choose not to take AP medication, particularly in participants under 21 years old. However, the trial was not definitive and conclusions about the efficacy of CBT in this population are limited. The COMPARE (Cognitive behaviour therapy Or Medication for Psychosis – A Randomised Evaluation) trial,102 a three-arm RCT for people aged ≥ 16 years with FEP, allocated participants to AP monotherapy, CBT monotherapy or a combination of both treatments. The results showed that over the 12-month follow-up period the PANSS scores were significantly lower in the combined treatment group than in the CBT monotherapy group, but there was no significant difference between AP medication and CBT, or between combined treatment and AP medication. 102 Participants who received CBT monotherapy had fewer non-neurological side effects and less weight gain than those in the AP monotherapy and combined treatment groups, which may indicate that CBT monotherapy leads to fewer adverse effects. 102 However, although one-fifth of participants in the COMPARE trial (15/75) were aged 16–18 years, there are no head-to-head studies of PI versus AP medication in young people and, therefore, no data on which to draw conclusions about the clinical effectiveness and cost-effectiveness of PI versus AP medication versus a combined treatment in young people with psychosis. 102 The COMPARE trial also did not include FI as a component of the PI.

Trial design

MAPS was a Prospective Randomised Open Blinded Evaluation (PROBE) to explore the feasibility of running a definitive trial of three treatment options for FEP in 14- to 18-year-olds. Participants were allocated in a 1 : 1 : 1 ratio to receive AP medication monotherapy, PI monotherapy or a combination of both treatments. MAPS was conducted over 27 months across seven UK sites. The recruitment window ran from 1 April 2017 to 31 October 2018. All follow-up assessments were finalised by 31 May 2019. The current REC-approved trial protocol is available at www.journalslibrary.nihr.ac.uk/programmes/hta/153104/#/.

Settings

MAPS was conducted within NHS secondary care mental health services, namely EIP services and CAMHS. The seven UK sites were Manchester (central site); Birmingham; Lancashire; Oxfordshire and Buckinghamshire; Norfolk and Suffolk; Northumberland, Tyne and Wear; and Sussex. The study started with four sites: Manchester, Sussex and Oxford all commenced recruitment on 1 April 2017 and Lancashire commenced recruitment on 8 June 2017. The remaining three sites were added in 2018 to determine the feasibility of replicating effective recruitment approaches from the initial four sites. Norfolk and Suffolk commenced recruitment on 9 May 2018, Birmingham on 31 May 2018 and Northumberland on 20 June 2018.

MAPS sites differ considerably in their characteristics, including the levels of urbanisation and deprivation. For example, the 2015 English Indices of Deprivation104 rank Birmingham and Greater Manchester third and fifth, respectively, in having the highest proportions of neighbourhoods that are in the most deprived 10% of areas nationally, whereas Oxfordshire and Buckinghamshire have some of the lowest proportions (35th and 38th, respectively, out of 39 Local Enterprise Partnerships).

Participants

Sixty-one participants were recruited from EIP services and CAMHS in six sites: Oxfordshire and Buckinghamshire (n = 25 randomisations), Manchester (n = 21), Lancashire (n = 9), Sussex (n = 4), Birmingham (n = 1) and Northumberland, Tyne and Wear (n = 1). No participants were recruited in Norfolk and Suffolk.

Inclusion and exclusion criteria

We included young people who met all of the following criteria:

• aged 14–18 years (up until their 19th birthday)

• presented with FEP (defined as being within 1 year of presenting to services with psychosis)

• under the care of an EIP service or/and CAMHS

• scored ≥ 4 points on the PANSS105 for delusions and/or hallucinations at the baseline assessment to ensure current symptoms of psychosis

• met the ICD-10 criteria for schizophrenia, schizoaffective disorder or delusional disorder (as diagnosed by the treating consultant) or met the entry criteria for an EIP service

• able to provide written, informed consent and if under 16 years old have a parent/guardian willing to provide additional consent for the MAPS team to contact the young person (for ethics reasons).

To ensure that our sample was representative of young people with FEP as a primary problem, we excluded those who:

• had a diagnosis of ICD-10 organic psychosis

• had a moderate/severe learning disability

• had primary alcohol/substance dependence

• were non-English speaking (to ensure that participants were able to engage in assessments and therapy)

• scored ≥ 5 points on the conceptual disorganisation item of the PANSS (as above)

• presented with immediate risk to themselves or others at the time of referral

• had received AP medication or structured PI within the last 3 months (to ensure treatment naivety).

Exclusion did not include comorbid diagnoses, such as autism spectrum disorders, personality disorders or use of substances (e.g. cannabis or alcohol) (unless this was a primary alcohol/substance-dependence diagnosis, as outlined above).

Data collection

Potential participants were informed of MAPS by a member of their care team and, if interested, were asked for a verbal agreement for basic referral details to be provided to the research team and for a member of the research team to contact them. For people under 16 years of age, the MAPS protocol required the care team member to ask the young person’s parent/guardian to provide verbal consent to be contacted by the research team initially. If verbal consent was provided, the care team member shared with the RA basic referral information for the young person and contact details for the parent. The RA would then contact the young person (aged 16–18 years) or parent (for under-16-year-olds) to describe the study briefly and send them a participant information sheet. Each young person or parent was given at least 24 hours to consider the participant information sheet. If interested, RAs would arrange to meet the young person or parent at their preferred venue [including the individual’s home, school/college, mental health services, general practitioner (GP) surgeries or youth offending services]. If there were any concerns about the individual’s home environment or risk to others, the initial meeting would take place at an NHS site.

At the initial appointment, the RA discussed the participant information sheet with the young person or parent and asked them to reflect on the information, ask questions and raise concerns to ensure that the information was understood. Once the RA and individual were satisfied that all of the information had been provided and understood, they would complete the parent/guardian consent to approach form (parents of under-16-year-olds) or the participant consent form (16- to 18-year-olds). The RA read out each statement on the form to the individual, checked their understanding and asked individuals to sign their initials next to each statement if they agreed with it. Finally, the RA and young person/parent provided their signatures and the date of consent at the bottom of the form. For individuals under 16 years old, once the parent/guardian had provided consent for their child to be contacted about MAPS, the RA could meet the young person to complete the consent form, as described above. All participants, regardless of their age, provided written informed consent to enter the trial.

Following the consent appointment, the RA completed the baseline assessment. RAs were advised to minimise participant burden, that is ideally keep appointments to a maximum of 1.5 hours, avoid multiple appointments for any one time point and prioritise participant choice. It was recognised that for some participants the assessment battery may not be complete in full should the participant choose to decline an assessment or opt for a single appointment. Participants completed the PANSS first, before completing other self-report and physical health measures. Each participant was given a personalised crisis card with details of their care co-ordinator, GP and other crisis support numbers (e.g. Samaritans, Ewell, UK). All participants who attended baseline assessments received £10 for their time and contribution to the research.

We designed a follow-up period of variable length, such that participants recruited in the first 16 months were offered assessments at 3, 6 and 12 months post baseline, and those recruited thereafter were offered assessments up to the end of treatment (6 months). RAs contacted participants or their parents (with the participants’ consent) by telephone to arrange assessments. RAs confirmed and documented ongoing consent with participants prior to completing any measures. RAs began with the PANSS before moving on to the other measures. Participants were given another copy of the personalised crisis card and were compensated £10 at each follow-up appointment.

The RAs telephoned participants at 1.5, 4.5 and 9 months to maintain ongoing contact and engagement with the follow-up appointments, and ascertain any changes in contact details (no clinical outcome data were collected). RAs posted £5 gift cards to participants for their choice of shop after completion of these telephone calls.

Outcomes

Interventions

Training and supervision

Randomisation and blinding

Randomisation was independent and concealed, used randomly sized randomised permuted blocks of three and six, and stratified by (1) centre and (2) family contact (because participants without regular family contact would receive CBT only, not FI, if allocated to a PI arm). RAs carried out the randomisations via a web-based platform developed by the Centre for Healthcare Randomised Trials (CHaRT), a UK Clinical Research Collaboration registered clinical trials unit (CTU) (#7). CHaRT provided consultation on the development of the randomisation algorithms. Using the CHaRT platform ensured that investigators were unable to enrol participants to conditions or predict the allocation sequence.

Participants were randomised within 2 working days of their eligibility being confirmed at baseline. The trial manager, administrator, therapists, chief investigator and PIs were notified via e-mail of the participants’ treatment allocations, to allow monitoring of adherence to allocation and provision of supervision to therapists. Each site had a delegated staff member (either an administrator or a therapist) who sent allocation letters to participants and their care co-ordinators, psychiatrists and GPs.

The assessors (RAs) were blind to allocation outcomes until all participants’ outcome measures were completed. RAs and therapists received extensive in-house training on retaining the blind, including reading and signing a trial-specific standard operating procedure for managing and retaining blinding. The trial used numerous methods outlined in the standard operating procedure to maintain blinding, including separate offices, diaries, pigeon holes and databases for RAs and therapists; reminding participants, family members and clinicians about the blind; protocols for taking messages and secretarial support; and encrypting and password-protecting randomisation information. All accidental unblindings were reported to the trial manager and, where possible, another (blinded) RA would complete the assessment. The chief investigator, iDMC and TSC regularly monitored unblindings in each centre in case corrective action was required.

Statistical methods and analysis

Design

A nested, qualitative study aimed to explore the acceptability and feasibility of the trial and associated treatment interventions among key stakeholder groups, that is trial participants (young people with experience of FEP), their parents or carers and local clinicians (particularly those with prescribing responsibility). This qualitative approach allowed us to gain valuable views of this trial and recommendations to inform the design of a potential definitive trial. Individual semistructured interviews conducted with representative samples from each group evaluated views and experiences of recruitment to the trial, potential barriers to participation and solutions, assessment processes, randomisation to trial treatments and ongoing care and outcomes. Subjective experiences of receiving trial treatments were explored with participants and parents/carers, whereas clinicians were asked to share their perspectives on the delivery of treatments; all interview groups discussed the relative benefits and adverse effects of treatments.

Semistructured topic guides were devised for each of the three stakeholder groups of this interview study, and all interviews followed the same relevant pattern. Topic guides were considered flexible and open to amendment throughout the course of this study, as it was expected that novel areas of interest may emerge, generating additional interview prompts for subsequent interviews, or that original questions may prove to be inessential and, therefore, removed. For this reason, each topic guide was subject to several minor amendments between the first and the final interviews in each treatment arm.

All of the interviews were digitally audio-recorded. Interview recordings were returned to and stored within closed and secure NHS digital network locations. All interviews were transcribed verbatim at the central site (i.e. Manchester). All names, locations, highly idiosyncratic personal details or any other potentially identifying information were removed to ensure anonymisation of the interview data.

Young people

Parents and carers

Clinicians

Confidentiality and anonymity

All qualitative interviews were conducted individually and face to face to promote interpersonal engagement and ensure privacy and confidentiality. The exceptions were that several young people interviewees chose to attend with a friend, in two instances two parents attended the same interview and one young person attended their parents’ interview. All clinician interviewees attended individually; two clinician interviews were conducted by telephone.

Prior to commencement of each interview, researchers outlined the bounds of confidentiality and sought verbal confirmation of consent to proceed. All participants were assured that all feedback would be welcomed and valuable. This was highlighted particularly to encourage possible negative feedback about the trial or treatments, as it may be reasonably expected that consenting interview participants are more likely to represent those with generally positive or neutral views.

Data analysis

The interview study was conducted with leadership and ongoing involvement from individuals with personal or parental experience of psychosis spectrum difficulties. This involvement was expected to enhance attention to young person and parent/carer concerns during the conduct and analysis of all interviews. All interview data were thematically analysed. 129 The overarching aim of this analysis was to inductively code qualitative data at the manifest level (i.e. analysing only the immediate meaning of participants’ language) to produce an accessible body of coded data from which meaningful thematic representations of participants’ perspectives can be reported.

Consistent with established thematic analysis methodology,129 all transcribed interviews were initially read through in full to enhance familiarity with the data, and initial impressions of the emerging quality of the data set were recorded in memo form. All interviews then were coded systematically and iteratively within NVivo qualitative data analysis software (version 11) (QSR International, Warrington, UK). Researchers did not impose a rigid or literal ‘line-by-line’ coding approach at this stage, but sought to identify and code all sections of data that either directly answered a relevant interview question or offered additional relevant information. This led to many areas of data being coded under multiple topic areas, as many relevant aspects of this research area are inter-related (e.g. the relationship between diagnosis and treatment).

At regular scheduled points during the analysis processes, qualitative researchers met to discuss and refine coding and analysis strategies through reliability checking and establishing consensus. Periodic analysis meetings were also conducted with MAPS leads to further develop emerging thematic and conceptual outputs, and to ensure that valued insights emerging from interviews were transmitted to the trial team as a whole as quickly as possible. The substantive analysis meetings produced evolving thematic ‘maps’ or models representing the commonalities along with the diversity expressed in the views of each respective interview group. Over time, this mapping process elevated or reduced the prominence of key candidate themes until a fixed and final model of each group’s range of perspectives was agreed on.

Assessment and diagnosis: diagnostic uncertainty and clinical judgement

A central concern among all clinician interviewees was the importance of assessment and diagnosis of FEP when considering a referral to MAPS. They recognised that reliably identifying a FEP in adolescents is often very difficult. Along with uncertainty around distinguishing between sub threshold at-risk mental states and FEP, clinicians also highlighted the importance and complexity of assessing for neurodevelopmental difficulties, including autism spectrum features and personality, or the impact of environmental stressors among adolescents in particular:

Sometimes it can be more complicated in a child or an adolescent to make a very confident diagnosis, because you might have to watch and wait, you might have to see how symptoms develop over time, often things are not as clear-cut as they might be in somebody who is an adult, and you don’t always get textbook presentations in children, and also you have other developmental considerations to think about, such as learning, cognitive ability, there might be difficulties within the family or social settings, they might be looked-after young people living in a children’s home, you need to take all that into account so, sometimes it can be a complex assessment.

C07

Standardised psychiatric and psychological assessments for psychosis, such as the PANSS measure that was used in MAPS, were viewed as inadequate alone to resolve this diagnostic uncertainty:

I’ll see people who have experiences and are distressed by those experiences who might on a measure like PANSS might not quite make whatever the particular grade is at that PANSS in time to be part of the service but who I think very strongly that they would benefit from the service. At the same time, I might meet young people who would score on such instruments who I think are not best suited by the service and conceptualising their difficulties in that way is not the best to kind of formulate things forward.

C01

Instead, clinicians stressed the importance of using their own clinical judgement during assessment processes. Key additional dimensions central to meaningful assessment included risk, distress and severity of psychosis.

Treatment and management

There was strong consistency in clinicians’ experiences of treatment of adolescent psychosis. AP medication was commonly viewed as a necessary and helpful first-line treatment for FEP, and AP prescribing was perceived to be a key responsibility for clinicians:

If I was convinced that somebody had a psychotic illness I wouldn’t withhold antipsychotic medication to wait for therapy, you know I would see it as sort of my primary function as a psychiatrist to use the treatments that I can to try and help somebody get better, so if I thought that was appropriate I wouldn’t sort of wait to do that.

C07

Key considerations for treatment decisions included the young people’s levels of distress and risk, along with the objective severity of their symptomatology:

As the level of disturbance increases and the level of risk and the level of impairment increases, the priority is to try and reduce that risk as soon as possible, and antipsychotics can be very helpful with that.

C17

The clinical value of AP medication was nonetheless weighed against the potential serious adverse effects, particularly when prescribed to children and young people, who were viewed to be more prone to side effects. This was a further reason why such emphasis was placed on the care that is needed in conducting a reliable assessment:

You cannot just start medication based on PANSS [score]. You might realise your mistake after a year.

C05

Given that clinicians placed such importance on using their own clinical judgement to make individual treatment decisions, it was unsurprising that referring to a trial in which treatment was allocated by randomisation could be challenging:

I’ve referred a couple of people in and actually I found it quite difficult and they were kind of randomised to the opposite of what I might have chosen to do if I’d had a clinical decision, so that, that did require a degree of kind of trust.

C11

Time and timing

Clinicians also highlighted the importance of timing in assessment and treatment decisions, making it clear that, in contrast to the standardised phases of a clinical trial, in the ‘real world’ the timing of assessment and treatment decisions is far more varied. In some cases, the importance of timing in clinical practice relates to the need for urgent treatment to alleviate severe distress or risk:

It partly just really depends on the young person and where we sadly ended up last week was that her, her positive symptoms, her psychotic symptoms became far more apparent when we did the second home visit.

C03

Clinicians also discussed the value of prolonging assessment periods to ensure the reliability of a psychosis spectrum diagnosis, and subsequent treatment decisions:

There’s a degree of uncertainty when we start off about quite what might be happening for the young person so it’s not always clear-cut for me whether you know medication’s . . . something we’re going to be prescribing right at the start so you know there’s often a period of assessment before we’d be considering whether that’d be an option for someone.

C04

Duty of care and safety

A dominant idea within this theme was clinicians’ own responsibility for the prescription of AP medication. Most described caution about both diagnosis and treatment of psychosis in children or adolescents, especially in relation to AP treatment:

We need to be more careful about these younger people in (a) giving a diagnosis that is as significant as psychosis and (b) using antipsychotics.

C16

However, some clinicians expressed a concern around the potential harm of not prescribing AP medication. This often related to the perceived need for immediate pharmaceutical treatment to alleviate risk or distress, but was also discussed as an important factor in long-term outlook:

If I thought they had a psychosis I wouldn’t wait to treat them, because we know that duration of untreated psychosis is you know a risk for adverse outcome.

C07

Some also expressed some concern about the delivery of psychological treatment alone to young people with psychosis. Although this concern involved the absence of AP medication, echoing concerns detailed above, it was equally grounded in direct observation of suboptimal outcomes from CBT:

I do worry that people who go for a CBT alone approach do worse, I’ve seen them a few years after they’ve finished their CBT and they want more CBT, they’ve not taken any medication, they’re frankly psychotic . . . as a clinician that can be quite distressing to see, people who’ve really avoided medication or had very very tiny bits, have never got completely better, have had grumbling psychotic symptoms, but they’ve managed them better while they’ve had the high intensity support of a CBT therapist but when that stops and the person loses that support they continue to have the symptoms.

C12

Treatment beliefs

Treatment decisions were influenced by clinicians’ prior beliefs about treatment. Clinician interviewees held a range of views on the relative benefits and risks of AP medication and psychological treatments (CBT/FI). Along with the professional caution described in relation to treatment risks, interviewees described perceived benefits of both of the treatment types offered in MAPS. For some, AP treatment was seen as essential for the most unwell young people, in part because of its perceived capacity to deliver beneficial effects quickly:

We’ve had a few under-18s who, a couple who weren’t in the trial who were very floridly psychotic and were actually admitted to paediatric wards because they were so unwell, and there was no doubt there that the appropriate thing to do was to start antipsychotic medication first.

C11

Antipsychotic medication was also seen by some to be important or even essential in the longer term to prevent relapse in young people with psychosis or after previous treatments have not helped:

I can think of a few cases where you know we’ve had very young people . . . 14 to 18 [years] who’ve been admitted to the [adolescent inpatient] unit with kind of bona fide persistent paranoid psychosis with thought disorder, negative symptoms, horrendous kind of schizophreniform illnesses, and have responded very well to for example clozapine or other antipsychotics, after other treatments have failed.

C13

Both of the psychological treatments offered within MAPS (CBT and FI) were viewed as valuable. Participants believed that they could be very helpful for dealing with a range of difficulties, such as low mood:

. . . if they are experiencing hearing voices but actually the primary problem is actually they’re struggling more with their mood or engaging with education if we were to do a psychological piece of work that’s gonna have much more of an impact than medication possibly with that.

C09

However, several clinicians voiced caution around the potentially limited capacity of individual young people to engage with psychological treatments, a concern not evident in discussions around AP medication:

. . . particularly quite young people, often that they find it quite difficult to describe their experiences or to talk about what’s going on so I’m thinking that sometimes that decision between CBT and medication is about that ability to be able to talk and think and tolerate their distress, and we have had some people who you know are very distressed and you just can’t really get them to describe their experiences, you can’t really get them to talk to you very much.

C11

The psychological treatments offered in MAPS (CBT or FI) were sometimes described as a more appropriate treatment approach than medication for certain (non-psychotic) difficulties:

It’s thinking about what actually are the specific difficulties the young person may be having for example, if they are experiencing hearing voices but actually the primary problem is actually they’re struggling more with their mood or engaging with education if we were to do a psychological piece of work that’s gonna have much more of an impact than medication possibly with that.

C09

It was evident that psychological treatment was viewed as most useful when used in combination with AP medication, as there was potential for positive interaction between the two treatments. That is, as well as providing a benefit in its own right, AP medication could enable young people to undertake and make the most use of psychological therapy:

When people are at their most unwell, that’s really difficult to even engage with and usually they’re the people who are saying you know I don’t want CBT . . . and actually medication can sometimes move them forward to the point where actually they can then engage with the CBT, in a different way.

C08

Organisational context

Clinicians represented both EIP services and CAMHS (also known as Healthy Young Minds), in which the youngest young people referred to MAPS were seen. Clinicians identified that populations attending the different services presented differently and had different needs, and moreover that the expertise and capacity within the services and the treatments available differed:

Often I guess the younger ones . . . in terms of you know their developmental stage, some of the issues they will present with might be different, and also I think sometimes in that early adolescent period it’s obviously a time when many people might have, you know, emotional symptoms anyway in the population, so sometimes kind of teasing out what represents a sort of pathological process can be, you know, the task for anyone.

C04

Differences between service settings also included important perceived variations in the diagnostic assessment expertise located in either service type. CAMHS, for example, was seen by several interviewees to offer more holistic psychological/psychiatric assessments that were more appropriate for the youngest young people:

EI [early intervention] covers a wide age range and it’s purely staffed by adult psychiatrists, so it’s difficult for me when I’ve seen say a 15-year-old and felt they were psychotic and then for an adult EI clinician to come around and say ‘well I think you they’ve got a personality disorder’, when they don’t see, bluntly they’re not experts in adolescent psychopathology and what an adolescent truly, what a generically psychologically unwell adolescent looks like because all of them are a bit dysregulated, all of them are a bit chaotic.

C15

There was, however, a perceived strength in the psychosis-specific expertise offered by EIP teams that was not always available among CAMHS clinicians:

Because a lot of the first-episode psychosis now does not sit within CAMHS . . . the experience of that within CAMHS has actually reduced, so we don’t have staff members that are used to regularly asking kids about psychotic symptoms, their experience and comfort levels with that has gone down.

C10

Something that was similar across settings was the lack of availability of PI, and accessing this type of treatment was viewed as an important organisational driver to referring young people to MAPS:

Sometimes my kind of experience of psychological therapy is coloured by often that it’s something that’s harder to deliver in as timely a manner as might be ideal and that sometimes it means that it’s coming online often much later down the line.

01

Randomisation and control

Clinicians highlighted some concern around having less control over treatment, especially AP prescribing, if their young people were in a randomised trial:

What I’d worry about is if you came in and screened everyone who was 14 to 18 [years] with PANSS and got them all into the trial . . . there’s going to be some false positives where the psychiatrist might not want to prescribe an antipsychotic but perhaps feel pressure to because of the trial.

C06

It is important to note that clinicians recognised that MAPS did not prevent young people from accessing alternative treatment resources (e.g. psychotherapy) and did allow medication decisions to remain in prescribers’ hands, regardless of a young person’s trial allocation:

Yeah there’s certainly been people where I’ve seen them and I’ve decided not to on seeing them, even though they’re in the trial I’ve decided not to give them antipsychotics.

C16

Burden or resource

Some clinicians suggested that their involvement with MAPS involved a degree of additional workload or pressure, for example in decision-making or communication around treatment. One clinician recalled a specific incident of concern around communication with trial staff (subsequent communication of this concern ensured that further similar situations were unlikely):

Sometimes I’ve seen somebody I’m worried about and then I’ve been e-mailing and there’s not been anything and a week and a half has gone by and I’m thinking ‘if anything happens now’, I would have prescribed an antipsychotic when I saw this person, now nothing’s happened, that person has been in distress coming up to 2 weeks, they’re a bit risky.

C16

However, more frequently clinicians found that the trial offered important additional resources, especially clinical psychology (CBT/FI). Clinicians valued that the trial frequently offered access to psychological treatments far more quickly than was typical in services, and that trial-based therapists had the capacity to offer longer or more comprehensive courses of therapy:

So MAPS is great ’cos one young person who’s got shedloads of psychological therapy who we may have wished to treat with psychological therapy but without the MAPS trial we would not have been able to do that.

C01

Improving evidence with an uncertain population

Clinicians perceived that MAPS had the potential to improve the limited evidence base for treating psychosis in young people. However, some voiced uncertainty around the ultimate value of this trial’s results. It was recognised, for example, that the level of treatment resource in the trial may be unrealistic when compared with typical NHS service capacity and, therefore, study results may not be generalisable to everyday clinical settings:

The study might show that the CBT is effective, but we’re starting CBTp incredibly quickly. I think you would have to then be able to offer it very quickly if you were rolling it out as a treatment because also what you’re not testing in the study is waiting 6 months for your CBTp.

C15

The diagnostic uncertainty around young people with psychosis spectrum experiences could also affect the perceived validity of the trial’s results. Clinicians argued that if trial inclusion criteria do not adequately reflect clinician’s own diagnostic judgement then it is difficult to have confidence in generalising results from the sample included in the trial:

There will be many people in the MAPS study . . . that on reflection haven’t had first-episode psychosis . . . they might’ve met criteria but actually there’s you know there’s lots of reasons why people could meet criteria and it doesn’t mean they’re psychotic.

C16

Entry to MAPS

Randomisation

The relative acceptability of treatment allocations is an essential consideration when assessing the overall feasibility and acceptability of a clinical trial. Although all of the trial participants demonstrated acceptance of MAPS treatments by consenting to be randomised, participants’ recall of their entry to the trial indicated variations in degrees of treatment acceptability. The range of views describing this variation were characterised overall by a general belief that any treatment type could potentially be helpful and, therefore, acceptable:

Didn’t know what was like going on with me and stuff like I didn’t understand like why I heard stuff so I thought . . . if it’d help then I wouldn’t mind any of them.

YP12

Treatment preferences

Although some participants’ treatment preferences prior to randomisation were based purely on expectation (what may happen), others were informed by previous experience or observation (e.g. family members’ mental health treatment). Participants often reported that CBT was preferred over AP medication, and this was commonly influenced by the perceived benefits of ‘talking to someone’, along with concerns or uncertainty about side effects of the medication:

When I looked like through the different like [medication] side effects and symptoms . . . I started like freaking out a bit, whereas I think like talking to someone’s a lot easier.

YP02

By contrast, three participants indicated a preference for medication over therapy. For one participant this was a change of view after undertaking CBT within the trial, whereas the other two participants preferred medication from the outset. Although concern about AP medication was more common, in practice most participants accepted or believed that they would have accepted it:

Although I was quite hesitant I think I always said to myself that I’d go in to the appointment and find out all the side effects, you know the possible help it could provide . . . you know never say never.

YP10

Three participants voiced a preference for the combined treatment allocation, perceiving that more treatment would offer stronger benefits. One participant’s preference for combination treatment was informed by specific prior experience:

I think they work better together . . . I’m on antidepressants and stuff as well, and them alone will not work fully, but also therapy alone doesn’t work fully because you’re not in the right frame of mind to accept therapy.

YP03

Few participants expressed any initial views on the FI element of PI, although three participants were clear that they declined or that they would have declined FI, and two of those rejected FI as they did not want to worry family members:

I don’t really like worrying anyone so bringing like family or friends along would just you know worry them a bit more.

YP02

Antipsychotic medication

Of the nine interview participants allocated to receive AP medication during the trial, eight were prescribed AP medication and seven took them (three in the AP only arm, four in combined treatment). Both participants who did not take AP medication were in the combined arm.

Cognitive–behavioural therapy

Ten participants were allocated to PI (five to PI only, five to combined treatment) and all attended CBT sessions. One participant’s course of CBT ended early after three sessions, and they requested AP medication from their psychiatrist.

Family intervention

There was variability in participants’ engagement with the FI component of the PI, with just 6 out of 10 participants describing some level of FI experience during MAPS.

A chance to tell them what I’m going through

Although participants may have previously discussed their psychological difficulties with family members, such conversations were commonly limited. Therefore, family sessions served primarily to facilitate such discussions and to develop a shared perspective:

The family session was a chance for me to tell them what I’m going through and a chance for them to tell me what I’ve been doing wrong, and what I’ve been doing right.

YP06

Combined treatment

Five interview participants were allocated to receive AP medication and PI, although two did not take AP medication. All three participants who continued with both treatments described the combination of treatments as helpful. Notably, AP medication was perceived to facilitate engagement with PI, with each treatment helping in different ways, but also interacting beneficially:

In my opinion it’s [combined medication and PI] the best treatment for anything, because the drugs allow you to feel in a better state of mind to accept therapy, and then when therapy is opening up things you maybe wouldn’t be comfortable opening up with the drugs help make that more comfortable, so they all work together.

YP03

It was their choice

For some trial participants (14- to 15-year-olds), parental/carer assent was required for initial contact with trial staff and, therefore, entry to MAPS. Nonetheless, all family members voiced the view that the decision to enter MAPS had been the young person’s:

It was ultimately her decision . . . we just discussed which she wanted to do, what if it was one [treatment] or what if it was the other and she just said she knew she needed to have help.

F12

Family support and guidance

Some family members viewed themselves as having an active role that involved offering support and advice on starting the trial and throughout their young person’s involvement:

Yes it’s been discussed between us but it’s [young person]’s decision . . . if I can help with guidance as a parent, that’s what I’m here for.

F03

Although allowing the young person to make that initial decision, family members were mindful that there were constraints to this, which included the urgency of accessing treatment and the time limits that may have influenced entry to the trial:

I know that he didn’t feel pressurised in any way . . . I think what [psychiatrist] explained was, obviously it was an opportunity and you know he was free to take it up or not, but it was obviously time limited as well, the option to opt in was time limited but he chose to opt in so.

F11

In this example, a dad describes how he would have been willing to over-ride his son’s initial entry decision if he had seen that to be in his son’s best interest:

I think if he had have had fears about it and that we couldn’t have alleviated it I’ve gotta be honest we’d have probably pulled him from the trial purely because . . . we didn’t want him to worry about anything unnecessarily.

F09

Monitoring

An important demonstration of family support was evident from descriptions of what could be termed as ‘monitoring’, in which family members were active in assessing the impacts of trial treatments and intervening to address important concerns:

The [AP] dose that she was originally put on was too high so I tried that for 2 days and I just went back and said this isn’t my child, I need it changing, and to be fair the doctor’s worked with me and she reduced the dose.

F04

Any help was better than nothing

Family members’ own motivation to support the young person’s entry to MAPS was frequently characterised by a sense of urgency in seeking help; entry to MAPS represented a valued opportunity to access help and support that either was unavailable within local services or could be accessed more rapidly via the trial:

We thought that it would be a good idea, because I thought that she’d get more help than if she was just stayed through CAMHS.

F15

This was particularly important when family members perceived psychological therapy or ‘specialist help’ to be needed. Family members were often aware that these resources were often scarce, especially psychological support:

As soon as I heard that there was psychology involved or a possibility of psychology . . . it was like right well at least we’re gonna get a step in the right direction.

F04

Randomisation

The over-riding desire for help, any help, may explain family members’ general acceptance of randomised allocation to trial treatments:

We didn’t quite know whether it’d be the medication, if it would be the talking therapies or if it would be both . . . but we were just you know wanting something to start to just help him really.

F13

However, some family interviewees either had not been aware or made clear that randomisation occurred or had believed that trial treatments would be delivered based on individual needs:

I didn’t know it was random to be fair, yeah I thought she, she well not it’s not chosen on merit is it that’s the wrong word but you know chosen because . . .

Kind of needs-based.

On her needs yeah, yeah yeah.

F05

Two family members who did recall randomisation expressed some concern that this reduced the control around treatment:

I think the concern was obviously in terms of the, the control, you’re not sure which avenue he’s gonna be targeted down.

F11

However, more commonly participants seemed reassured that being in the trial did not prevent discontinuation or change of their young person’s treatment, or withdrawal from the trial:

I’m aware that you know it is a research study but actually if things changed, he could pull out and get the active treatment he needed at any point.

F11

Trial procedures

Along with discussions of decision-making at entry to MAPS, family members shared views relating to two additional important aspects of trial experience: study information material and assessments.

Antipsychotic medication

We interviewed family members of 10 young people allocated to receive AP medication, two allocated to AP alone and eight to combined treatment (one of whom was not prescribed AP medication). Given this complexity, analysis has sought to carefully isolate AP-specific features of interview feedback.

Cognitive–behavioural therapy

Fourteen out of 16 interviews were with the family member of a young person who had been allocated to receive CBT, either alone (n = 6) or in combination with AP medication (n = 8). Analysis has sought to distinguish CBT-specific aspects from more general treatment effects or changes.

Family intervention

Family intervention was offered to all young people receiving individual CBT, although young people did not always choose to involve family members in their therapy. During the qualitative interviews, 10 out of 14 eligible participants discussed FI. No family member discussed their initial views of FI acceptability during interviews; therefore, the following discussion relates to post-treatment views only.

Combined treatment

Trial outcomes

We collected demographic data and a range of potential primary and secondary outcomes to inform a definitive trial. The PANSS is a suitable choice of primary outcome measure for a definitive trial for a number of reasons. The completion rates of the PANSS were excellent at the end of treatment (6 months) and clearly met the prespecified progression criteria. The PANSS is a valid and reliable measure106 and it is the most commonly used primary outcome measure in psychosis trials. Therefore, use in a definitive trial, it would allow comparison of data with the wider evidence base and synthesis into systematic reviews and meta-analyses of treatments for psychosis. A review132 of 24 psychosis study outcome measures found that the PANSS was the most acceptable to service users. However, the PANSS is a clinician-rated measure and the same review of commonly used outcome measures for psychosis studies indicated that service users have a general preference for outcome measures in self-report format,132 and service users have expressed that global, recovery-oriented outcomes are as important to deliver as changes in psychiatric symptomatology. 133 Our service user-defined measure of recovery (QPR) had lower completion rates at the end of treatment; however, this may be a result of the approach to minimising participant burden, that is, when participants expressed difficulty or unease with assessment battery burden, all measures other than the PANSS were deprioritised.

At baseline and follow-up, the completion rates of secondary outcomes measures (other than the PANSS) were low, which indicated that the assessment battery should be significantly reduced for this population, with only the most essential outcomes retained. See Table 25 regarding the proposed outcomes for a definitive trial.

Trial feasibility

With regard to the progression criteria, we achieved a 50/50 mix of green and amber progression zones. Overall, these results indicate that it would be possible to conduct a definitive trial, but some modifications to the trial design are required. We provide information on these modifications in full in Table 24. Recruitment was in the amber progression zone, suggesting that modifications are required to recruit to target in a definitive trial. Two out of the four original sites recruited well, with the Oxford site over-recruiting and the Manchester site recruiting to 91% of the target. Site selection is crucial to ensuring recruitment to target for a definitive trial, and a number of considerations must be made in selecting the sites, as outlined in Table 24. In addition, further refinement of the inclusion criteria could have important implications for recruitment to target. Adherence to ≥ 6 consecutive weeks of AP medication prescription for participants in an arm that contained AP prescription was in the amber zone; however, we make specific recommendations for how this could be addressed in a full-scale trial. Both retention of participants at the primary end point (end of treatment) and adherence to six or more sessions of CBT for those participants in an arm that contained CBT were in the green zone. Retention of participants at the primary end point (6 months, end of treatment) was excellent. However, retention at the 12-month follow-up was lower (27–47%); we make specific recommendations for improving this in a full-scale trial, although a substantial proportion of these missing data were because of a planned variable follow-up rate in which participants recruited within the final 3 months of recruitment between August and October 2018 (12 participants, 19.7% of the sample) were never intended to receive a 12-month assessment.

Acceptability and feasibility of treatments

Evaluation of data on adherence to treatments indicates that both PI and AP medication are acceptable treatments, and it would be feasible to deliver both in a definitive trial. Although the data indicate greater adherence to a dose of PI (six sessions or more), adherence to AP medication (6 or more consecutive weeks) was lower. However, adherence to AP medication was within the amber zone for the progression criteria, suggesting that it would be feasible to deliver in a definitive trial with some amendments as outlined in Table 26. Adherence to AP medication within our trial compares well with the general literature on AP adherence, which finds high levels of non-adherence, including in RCTs.

We obtained data from medical records to characterise the prescribing of AP medication for the sample, self-report data from trial therapists to evaluate the quality of therapy delivered (both CBT and FI), including therapy milestones and change strategies used during therapy, and CTS-R ratings for fidelity to the CBT model. In addition, end of CBT and end of FI questionnaires completed by therapists provided data on treatment manual amendments for a definitive trial.

Qualitative interviews provided an in-depth understanding of participants’ experiences of the treatments and the acceptability of the treatments to both young people and family members. Several key lessons were learnt that inform treatment delivery in a definitive trial. We present the lessons learnt in Table 26.

Economies of scale

Each site had one full-time RA costed to work on the study. As outlined in Chapter 2, the RAs were tasked with liaison and recruitment from clinical teams and research assessments including the PANSS, psychometrics and physical health checks. Possible economies of scale include dropping the physical health checks from the RA tasks and reducing the number of assessment measures. This would reduce the length and/or number of research assessment visits per follow-up. However, for a number of reasons, we recommend that for a definitive trial a full-time RA be retained for each site. First, as outlined earlier, an assertive outreach approach was highly valued by participants and it is likely that this contributed to excellent follow-up completion rates at 6-month assessment. To offer an assertive outreach approach regarding the time, location and rearrangement of appointments plus text, telephone and letter reminders (with the associated paperwork to document participant contact) flexibility is required and this would be limited in a part-time role. Second, we recommend that more check-in telephone calls are scheduled to facilitate engagement and long-term follow-up; again this would require sufficient diary capacity. Third, with improved screening methods and accuracy of screening, as well a higher age limit of 25 years, we envisage an increased number of baseline visits. To ensure swift access to entry to the trial (and, therefore, treatment) a RA would require flexibility and capacity in their diary to act quickly in response to a referral from a psychiatrist.