Pilot randomised control trial

This chapter outlines the pilot RCT design, methods and data for screening, recruitment and follow-up. The methods of the pilot RCT are reported in the trial protocol paper.90

Pilot trial design

The pilot feasibility RCT comprised three arms: MET, SBNT and usual care. The pilot RCT was conducted in six sites across the north-east of England (Newcastle, Gateshead, Durham, Middlesbrough, Stockton and Redcar), to assess the acceptability of the adapted interventions (SBNT and MET) and the feasibility of taking one or both of the interventions to a full-scale multicentre RCT. Details of the flow of participants through the study are presented in Figure 7.

FIGURE 7 Randomised controlled trial flow diagram. ASAI, Adolescent Sexual Activity Index; ASSIST-Y, Alcohol, Smoking and Substance Involvement Screening Test–Youth; EQ-5D, EuroQol-5 Dimensions; EQ-5D-5L, EuroQol-5 Dimensions, five-level version; ESYT, Edinburgh study of youth transitions and crime; SDQ, Strengths and Difficulties Questionnaire; TLFB, timeline follow-back; WEMWBS, Warwick–Edinburgh Mental Wellbeing Scale. (Figure not reproduced here; see the original article.)

As previously highlighted in Chapter 1, Research objectives, the primary objective of the pilot RCT was to:

• conduct a three-arm pilot RCT [comparing MET, SBNT and a control (usual care)], to determine if rates of eligibility, recruitment and retention of children in care, and acceptability of the interventions, are sufficient to recommend a definitive multicentre RCT.

The secondary research objectives were to:

• establish response rates, variability of scores, data quality and acceptability of the proposed outcome measures for the future definitive trial (i.e. self-reported alcohol and drug use, health-related quality of life, mental health and well-being, sexual behaviour and placement stability 12 months post recruitment), to inform a sample size calculation for a definitive multicentre RCT
• assess acceptability of, engagement with and participation in the MET- and SBNT-based interventions by children in care, their carers and front-line drug and alcohol workers
• carry out a process evaluation to include fidelity of intervention delivery and qualitative assessment of the barriers to successful implementation, and to assess if key components from the MET and SBNT interventions can be combined to develop a new optimised intervention
• develop cost assessment tools, assess intervention delivery costs and carry out a value of information analysis to inform a definitive study
• apply prespecified ‘STOP/GO’ criteria and thereby determine if a definitive multicentre RCT would be feasible, and, if so, to develop a full trial protocol
• consider findings from the study as a whole, in order to develop a core intervention delivery package, potentially of a single optimised intervention, linked to a theory of change model to use in the definitive trial.

Randomised controlled trial methods

Participant screening and identification

As part of the SOLID trial, social workers were asked to screen all young people on their caseloads aged 12–20 years, with a standardised instrument. Training sessions were conducted with social workers from each of the recruitment sites. These sessions introduced the CRAFFT screening tool, which is a nine-item tool that has been used extensively with young people and is sensitive and specific to identify problem substance use.91 The researcher emphasised the importance of completing the CRAFFT screening tools with young people on their caseloads and explained how these data could be beneficial not only for recruitment into the study, but also for each local authority to develop a clearer understanding of the extent and nature of substance use within their local area. All young people aged 12–20 years, unless they were already in active treatment with drug and alcohol services, or were unable to access drug and alcohol services (e.g. due to currently residing out of the study area or an imminent move out of area), were eligible for screening.

Once training had been completed, each social worker carrying a caseload of children in care was provided with a screening log (see Appendix 1) and the appropriate number of CRAFFT forms, so that they could complete one form with each young person with whom they were currently working.

Prior to the social worker conducting the screening, a brief initial contact leaflet was provided to the child in care and discussed with them by their social worker, this provided details of the study. Following this discussion, the CRAFFT screening tool was completed. Page 1 of the form requested that the young person provide their contact details and asked if they consented to being contacted by a member of the research team to receive more information about the full study. Young people could elect not to provide their contact details and consent; in these cases, they could continue to complete the CRAFFT form anonymously.

Each screening tool had a pre-paid envelope attached for the completed forms to be returned to the research team. All of the screening data were entered and processed at Newcastle University, in accordance with the study ethics approval. Summary statistics on the data collected from the CRAFFT forms returned by the children in care within their locality were fed back to each local authority. These reports provided a breakdown of the number of children in care screened within that local authority area, along with the sex, age, placement type, self-reported drug and alcohol use, and self-reported risk taking behaviour of these young people.

If a child in care provided contact details, screened positive for substance use and agreed to be contacted, they were telephoned by a member of the research team, who introduced themselves, explained the purpose of the call, and checked that the young person could recall filling in the form and that they were happy to proceed. The researcher then arranged a date and time to meet the young person to take formal informed consent and complete the baseline questionnaire.

Trial inclusion and exclusion criteria

Inclusion criteria

• Children in care aged ≥ 12 years and ≤ 20 years.
• The young person must have screened positive for being at risk of substance misuse (i.e. indicate that they have used substances within the last 12 months).
• Provided informed consent to take part in the study [for children aged < 16 years consent from parent/guardian (local authority) and assent from young person was required; for young people aged ≥ 16 years consent from the young person alone was required].

Exclusion criteria

• Already in active treatment with drug and alcohol services.
• Unable to access drug and alcohol services (e.g. owing to currently residing out of the study area or an imminent move out of area).
• Unable to give informed consent (due to acute or severe mental health difficulties, mental capacity or language barriers).

The inclusion criterion regarding substance use was amended during the trial (June 2017), as outlined in Report Supplementary Material 2, after discussion with the TOC, the sponsor and ethics board. In summary, this saw the criterion change from having to score ≤ 2 on the CRAFFT to having used a substance within the last 12 months (CRAFFT = 1). This change was made because it was felt that children in care using any substance may be at risk of engaging in associated risky behaviours.

Consent to participate

Information on the study was shared with parents and carers and young people. Informed consent was taken directly from children in care aged ≥ 16 years. For participants aged < 16 years, assent was taken from the young person and consent was also obtained from an adult with PR. Those aged 12–15 years were seen by the researcher with an accompanying adult (such as a parent, carer, social worker or children’s home lead), who was asked to provide informed consent on behalf of the young person. If the accompanying adult did not have PR, the research team contacted the adult with PR (in most cases the allocated social worker) to obtain informed consent. If the parent was not contactable, or it was deemed by the designated social worker that it would pose a risk to the young person for their parent to be contacted, the social worker or local authority guardian with PR was contacted to sign the consent form. All of the RCT documents are shown in Report Supplementary Material 6.

Data collection methods

Data were collected on participants enrolled into the trial at two time points. Baseline information was collected on the first visit prior to randomisation and follow-up data collection were scheduled for 12 months post recruitment.

Baseline data collection

After informed written consent was obtained, the researcher collected baseline information from the child in care using a self-completed questionnaire, administered via a tablet computer to provide greater privacy for the respondent. The researcher was available to support young people in completing the questionnaire, if necessary, and to answer questions of clarification, if needed. The baseline questionnaire recorded demographics; placement type; drug and alcohol usage via the AUDIT92 and Alcohol, Smoking and Substance Involvement Screening Test – Youth (ASSIST-Y);92 mental health and well-being via the Strengths and Difficulties Questionnaire (SDQ)93,94 and Warwick–Edinburgh Mental Wellbeing Scale (WEMWBS);95 and health-related quality of life via the EuroQol-5 Dimensions, five-level version (EQ-5D-5L).96

Alcohol use Disorders Identification Test

The AUDIT is a 10-question screening tool developed by the World Health Organization (WHO),97 used to identify signs of hazardous and harmful drinking and identify mild dependence.98

When interpreting the AUDIT score, in adults a score of ≥ 8 is associated with harmful or hazardous drinking, and a score of ≥ 13 in women and ≥ 15 in men is likely to indicate alcohol dependency. An overall total score of ≥ 5 is deemed as AUDIT-C positive, and indicates increasing or higher-risk drinking.97 NICE advises professionals that, given the more harmful effects of alcohol consumption for young people aged 10–16 years, the referral and intervention threshold on the AUDIT should be lower than the standard adult threshold.99

Alcohol, Smoking and Substance Involvement Screening Test – Youth

The ASSIST-Y questionnaire is a ‘reduced’ version of the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST) instrument.100 The original ASSIST v3.1 questionnaire, developed by the WHO, is not validated for use by those aged < 18 years. ASSIST-Y, developed by Drug and Alcohol Services South Australia, under the guidance of the WHO, is suitable for use in this age group.

When interpreting the ASSIST-Y score, for young people aged 10–14 years, a score between 2 and 5 for tobacco, alcohol and inhalants indicates moderate risk, and a score of > 6 indicates high risk. Scores of > 2 for any other substance from the list indicate high risk.

For young people aged 15–17 years, for tobacco and cannabis, a score between 2 and 11 indicates moderate risk. For alcohol, a score between 5 and 17 indicates moderate risk. For amphetamines, inhalants and hallucinogens, a score between 2 and 8 indicates moderate risk. For cocaine, sedatives, opioids, NPSs and ‘other’ drugs, a score between 2 and 6 indicates moderate risk. Scores identifying moderate risk require brief intervention.

High-risk scores are a score of > 12 for tobacco and cannabis, a score of > 18 for alcohol, a score of > 7 for cocaine, sedatives, opioids, NPSs and ‘other’, and a score of > 9 for amphetamines, inhalants and hallucinogens. Scores identifying high risk require brief intervention and referral to specialist for assessment and treatment.

Strength and Difficulties Questionnaire

The SDQ is a standardised screening questionnaire to be used with young people aged 4–17 years. Although the SDQ is not validated for use with adults (aged ≥ 18 years), we used the same scoring systems for all participants to maintain consistency when reporting data. The SDQ comprises 25 questions that are arranged to assess four difficulty subscales (emotional, conduct, hyperactivity and peer problems) and to also measure pro-social behaviour. The four difficulty subscales are summed to give a ‘total difficulty score’ out of 40. The SDQ has been used extensively in mental health research with young people and used previously with children in care.41

The SDQ total difficulties score is generated by summing all scales, except the pro-social scale. Scores range from 0 to 40. Once the SDQ has been completed, a set of final scores is generated, along with a categorisation for that score [close to average (score of 0–14), slightly raised (score of 15–17), high (score of 18–19) or very high (score of 20–40)]. Each 1-point increase in the total difficulties score corresponds with an increase in the risk of developing a mental health disorder.

Warwick–Edinburgh Mental Wellbeing Scale

The WEMWBS is a 14-item scale covering subjective well-being and psychological functioning, with each item scored 1 (none of the time) to 5 (all of the time) on a Likert scale. Within the scale, all items are worded positively and address aspects of positive mental health. The tool has been used extensively with adults and has recently been validated for use in teenagers.95

The WEMWBS produces scores of 0–70, with higher scores denoting positive mental health. When interpreting the well-being scores (using the NHS self-assessment tool), 0–32 denotes a very low score, 32–40 is below average score, 40–59 represents an average score and 59–70 is an above average score.

EuroQol-5 Dimensions, five-level version

The EQ-5D-5L is a questionnaire covering five dimensions of quality of life (mobility, self-care, usual activities, pain/discomfort and anxiety/depression), each with five response options, ranging from no problem to extreme problems. Responses to the EQ-5D-5L can be used to calculate quality-adjusted life-years.96 The EQ-5D-5L is discussed further in Data collection methods.

Full details of each screening tool and the scoring thereof are provided in Appendix 3 (see Tables 29–32).

Randomisation and group allocation methods

After consent was obtained and baseline characteristics were collected, the researcher contacted the trial administrator and provided the participant’s unique identification number, area in which the child in care resided, placement type and age. These details were entered, by the trial administrator, into Newcastle Clinical Trials Unit online randomisation service to ensure concealment of allocation. Individual randomisation to the three trial arms was stratified by placement type (residential/non-residential), site (local authority) and age band (12–14 years/> 14 years). Once randomisation was completed, five allocation letters, indicating the group to which the young person had been randomised, were generated automatically and distributed through the post by the trial administrator, one letter to the:

1. young person outlining study enrolment
2. parent outlining study enrolment (when appropriate)
3. carer (foster carer, residential key worker, etc.) outlining study enrolment
4. general practitioner (GP) outlining study enrolment (if consent for this level of contact had been provided)
5. drug and alcohol service outlining group allocation (only for those allocated to MET or SBNT) and requesting an initial appointment.

The time from baseline survey completion to letters being sent by the administrator was, on average, 2–4 days. The research team did not discuss with participants which intervention they would be allocated to; rather, they explained that the next stage of the process would be that the young person would receive a letter informing them of the intervention to which they had been allocated. The research team also clarified that, if the young person was allocated to an intervention arm, a drug and alcohol worker would give them a call to introduce themselves and to organise a convenient time and place to meet for their first appointment. The child in care was notified that letters would also be sent to their social worker and the individual with PR, to inform them of the trial arm to which the young person had been allocated. Children in care allocated to the control group were eligible to receive usual care, which involved their social worker making a referral along the usual drug and alcohol service pathway as required.67

The allocation details were documented by the trial administrator on a password-protected drive to allow one member of the research team (HA) access so that they could liaise with drug and alcohol practitioners regarding trial participants. The trial administrator recorded the participant’s unique identification number, initials and date of birth, the arm the young person was allocated to and the date allocation took place. Blinding of group allocation was not possible for the children in care participants, or for those delivering the intervention; however, the trial statistician and health economist were blinded to group allocation until the statistical analysis plan had been agreed before the final analysis.

Delivery of interventions

Following randomisation, young people who had been allocated to an intervention arm were invited to attend face-to-face sessions of either SBNT or MET on a one-to-one basis. Six sessions of SBNT or MET were offered weekly to fortnightly over a maximum period of 12 weeks, with each session lasting approximately 50 minutes. The rationale for this number of sessions stems from learning derived from the United Kingdom Alcohol Treatment Trial (UKATT), pilot work using SBNT with young people referred to child and adolescent mental health services, and the formative research phase of this study (see Chapter 3). The treatment protocol stated that the drug and alcohol service should contact the child in care to offer treatment to the appropriate intervention group (MET or SBNT) to commence within a maximum of 6 weeks of randomisation in keeping with the service guidelines. For participants allocated to the control arm of usual care, responsibility was passed to the social worker to initiate a conversation with the young person and make a referral into drug and alcohol services, following their standard referral pathway, if the young person consented.

The drug and alcohol practitioners

In each of the participating local authority areas, the specialist young people’s drug and alcohol treatment was provided by voluntary sector organisations with charitable status. In each study area, a drug and alcohol service agreed to take part in the study by allocating a worker to deliver each of the three treatment arms. Separate experienced young people’s drug and alcohol practitioners were available to deliver the two active interventions (MET and SBNT) or usual care (control). The practitioners had varying professional backgrounds, including youth work, social work, counselling and unqualified workers; all were experienced in the delivery of drug and alcohol psychosocial interventions for young people. Drug and alcohol practitioners received training on the specific intervention they were to deliver throughout the trial as documented in Chapter 3, SBNT and MET training.

Practitioner intervention logs

When delivering sessions, each practitioner was provided with a practitioner intervention log; this provided a system of recording details for each study participant. Within each log, practitioners were requested to record their own name and the demographics of the participant, inclusive of the unique identifier for the young person, age, sex, placement type and geographic location. Details were also recorded regarding the interventions themselves, such as dates on which sessions took place; the type of casework that had been completed, inclusive of delivery of the intervention (service user contact and non-service user contact, which included any preparation work for sessions and/or completion of paperwork, such as referral to other services); the amount of time spent on each activity; the mileage and travel time attributed to each session; and the number of people present at the session. Finally, a summary of the interventions was requested, monitoring the number of sessions offered and attended, the reason for non-completion of sessions if known and any referrals into mainstream drug and alcohol services. Once a young person had completed a maximum of six sessions, depending on identified need, the practitioner returned the completed log to the research team.

The practitioner logs had two main purposes. First, they helped the research team to calculate the cost of delivering the interventions, as discussed in further detail in Chapter 7, Practitioner logs. Second, the logs contributed to the assessment of the acceptability and fidelity of delivering the interventions, in terms of the number of sessions delivered and number of people present within sessions.

The 12-month follow-up method

Participants were contacted again 12 months post recruitment to complete a follow-up questionnaire. A recruitment window of 8 weeks (i.e. 4 weeks either side of the 12-month anniversary) was put in place to maximise follow-up.

The method of follow-up contact the research team initially used was the preferred method of contact identified by the young person at baseline. However, if telephone contact numbers were no longer available, the research team contacted the child in care’s allocated social worker to request up-to-date contact details to enable a follow-up questionnaire to be completed.

Once contact had been made and the young person consented to take part in the follow-up, the research team arranged to visit the young person in their home or an alternative suitable location at a time and date convenient to them. If the researcher was not able to organise a convenient time to meet the young person for a face-to-face follow-up, a telephone follow-up was offered.

The questionnaire was administered by the researcher. At baseline, most data were self-completed on a tablet (unless completed over the telephone, in which case the researcher read out the questions and recorded the answers on the tablet on behalf of the young person); however, the researcher was also available to provide clarification on questions if necessary. The timeline follow-back (TLFB) substance use and self-reported occasions of ‘drunkenness’ in the last 30 days101 was completed at follow-up and this was researcher administered.

As at baseline data collection, children in care provided information on drug and alcohol usage via the AUDIT92 and ASSIST-Y,92 mental health and well-being via the SDQ93,94 and WEMWBS,95 and health-related quality of life via the EQ-5D-5L96 at follow-up. To minimise respondent fatigue, as recommended during peer review, questions relating to use of health and social services,96 placement stability and potentially sensitive questions on sexual behaviour [Adolescent Sexual Activity Index (ASAI)]14,102 and antisocial/criminal behaviour103 were asked only at follow-up. Details of the additional data collection tools are provided below.

Romantic and intimate behaviour/antisocial and criminal behaviour

The ASAI, as previously utilised in the Avon Longitudinal Study of Parents and Children, which was a UK longitudinal birth cohort study,102 was used in the SOLID trial to assess romantic and sexual activities. The ASAI is a standardised tool for measuring interpersonal heterosexual behaviours among youths.104 Questions relating to enjoyment of sexual experience from the ASAI were replaced by two questions from the European School Survey Project on Alcohol and Other Drugs (ESPAD) multicountry survey of alcohol and drug use, assessing regret in engagement in sexual contacts and engagement in unprotected sexual intercourse, which are particularly relevant in relation to sexual encounters preceded by substance misuse.14 The advantage of using the ASAI is that it is graded, with more intimate sexual contact not asked about if lesser contact, such as kissing and cuddling, has not yet been experienced. This instrument had been discussed with the study PPI group, who acknowledged the graded nature of administration and supported its use.

In addition, questions previously used in the Edinburgh Study of Youth Transitions and Crime105 were included to capture data on antisocial behaviour. This questionnaire contains 15 questions. For each of 15 questions, the tablet version used in the SOLID trial asks ‘How often in the last year have you done any of the following?’ The questionnaire seeks to evaluate co-occurrence of antisocial behaviour and alcohol use.

The romantic and intimate behaviour and antisocial and criminal behaviour questionnaires were completed only at the 12-month follow-up visit. Therefore, it is not possible to calculate and present the change from baseline for these questionnaires.

Four measures were derived from the questionnaire (see Appendix 3, Table 30, for details). Romantic and intimate behaviour was the standard questionnaire score calculated as per the skipping rules, but romantic and intimate behaviour (minor), romantic and intimate behaviour (advanced) and romantic and intimate behaviour (regret) were not standard scoring schemes from the questionnaire. The skip logic for the SOLID trial-specific outcomes derived from the questionnaire data is shown in Appendix 3 (see Table 31).

Alcohol timeline follow-back

The TLFB is a drinking assessment method that obtains estimates of daily drinking. A calendar was used to provide retrospective estimates of daily drinking over a 30-day period. The TLFB was used, as it provides a wide range of information about an individual’s drinking (e.g. pattern, variability and magnitude of drinking). The TLFB in the SOLID trial sought to identify the ‘number of occasions where five or more standard drink units are consumed on a single drinking day’.

Qualitative interview

At the point of completing the 12-month follow-up questionnaire, the young person was also asked if they would be willing to participate in a qualitative interview as part of the process evaluation (discussed in Chapter 6) for the SOLID trial. For those willing to be interviewed, the research team carried out the 12-month follow-up and process evaluation interview at the same time, to help reduce the burden placed on participants. If the young person was willing to take part in an interview, consent was obtained using the procedure outlined in Chapter 3, Consent.

Once the young person had completed the 12-month follow-up (and interview when appropriate), they were given a £10 ‘love2shop’ voucher to thank them for their time and for supporting the study. Following this, the researcher confirmed that their involvement in the study had now finished and that they would not be contacted by the research team again regarding this study.

Statistical analysis and sample size

As this study was a feasibility trial and had relatively small numbers of participants, the main outcomes are feasibility outcomes. Accordingly, we principally reported descriptive statistics in order to inform the design of a future definitive study. No formal comparisons are drawn, as the sample size was not powered to detect differences.67 All statistical analyses were carried out on an intention-to-treat basis, retaining participants in their randomised treatment groups.

The majority of the outcome data are presented below in simple descriptive tables giving percentages, means and standard deviations (SDs), medians and interquartile ranges or a five-number summary [minimum, maximum, median and upper and lower quartiles (as appropriate)], separately for each arm of the study. This information could be used to inform the design, choice of primary outcome, necessary sample size and approach to the analysis, of the anticipated definitive trial.

The pilot feasibility trial aimed to obtain data from a minimum of 35 respondents in each trial arm at 12 months’ follow-up to estimate the critical parameters to the necessary degree of precision for a continuous primary outcome [number of occasions drinking, ≥ 5 standard drink units in a single occasion and frequency of use of the most problematic classified substance in preceding 30 days, both derived from the timeline follow-back – 30 days (TLFB-30)]. Assuming a pessimistic 30% loss to follow-up, the sample size to be recruited was inflated to 50 young people in each of the three arms. Therefore, we aimed to recruit and randomise 150 children in care.

Statistical software

Statistical analysis was carried out by the trial statistician, a member of Institute of Health and Society biostatistics team, downloading snapshots of the data from the trial database (administered by Vera Solutions, London, UK) into a comma-separated values (CSV) file. Randomisation details were exported by Newcastle Clinical Trials Unit from the Macro database into a CSV file. These two sources of data were merged and imported into the statistical software package Stata^® 14 (StataCorp LP, College Station, TX, USA) for analysis.

Results

Screening data

Screening was conducted slightly differently across research sites in response to the capacity and standard procedures of each site team. The social workers across the Durham 16+ team (teams working with children in care aged ≥ 16 years and supporting young people who are transitioning out of the care system) and in the Middlesbrough sites screened children in care during their routine appointments. In Redcar, two PAs within the social work team led the screening. Across the Gateshead, Newcastle, Stockton and Durham ‘looked-after’ teams, a researcher spent time within the social work offices to support social workers in completing the screening. The research team kept in regular contact with a key point of contact within each service (in most cases this was the team leaders).

There were 1782 children in care registered with the local authorities, across the study sites, at the time of this study. A total of 332 (19%) were deemed ineligible by social work managers for the reasons shown in Table 6. Thus, the total number of children in care aged 12–20 years potentially eligible to complete the CRAFFT was 1450, 81% of all young people across the six study sites.

Social workers were tasked by their team leads with screening as many of the 1450 children in care across the study sites as possible. The time period over which sites completed the CRAFFT was variable, ranging from 19 weeks and 1 day in Stockton to 47 weeks and 3 days in Newcastle, as shown in Table 7.

The local authority sites experienced multiple competing demands while taking part in the SOLID trial, with other commitments influencing the progress made regarding screening. Two local authorities were preparing for and having an Office for Standards in Education, Children’s Services and Skills (Ofsted) inspection. Therefore, within that time frame, only essential work was completed and only a small number of participants were recruited. One site had commitments to another research project and three sites experienced significant periods of time without a senior manager in post.

Different techniques were introduced across the local authority sites in an attempt to support screening of children in care within the planned recruitment period. Bespoke support mechanisms were put in place by the research team following discussions with the TMG, social work team leaders, managers and on-the-ground social workers. Two teams introduced student social workers or PAs to support the recruitment process (Newcastle and Redcar). Members of the research team spent time embedded within four local authority sites to support the recruitment of participants (Gateshead, Newcastle, Stockton and Durham ‘looked-after’ teams). An alternative technique that proved to be efficient for one team (Durham 16+ team) was the use of ‘stop the clock’ sessions over a 3-month period (May–July 2017) for social workers to have a dedicated time allocated to contact every young person on their caseload.

Of the 1450 potentially eligible young people, 536 (37%) did not complete the screening, either due to them declining to complete the CRAFFT (n = 131, 9%), or the social workers not completing the screening tool with all the young people on their caseload (n = 405, 28%). Social workers did not consistently report the reason for non-completion, so it cannot be reported.

The flow of children from initial screening through to study completion is presented in Figure 8.

FIGURE 8 The SOLID trial Consolidated Standards of Reporting Trials (CONSORT) flow diagram. Reproduced with permission from Alderson et al.67 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure. (Figure not reproduced here; see the original article.)

Social workers within the six study sites collectively screened 860 (59%) of children in care, aged 12–20 years, for drug and alcohol use, using the CRAFFT tool.91,106,107 The numbers of children in care in each site, the number of potentially eligible young people (minus those identified in Table 6) and the number screened by each local authority site are shown in Table 8.

Of the 860 children in care screened for drug and alcohol use, 369 reported drug and/or alcohol use in the last 12 months as shown in Figure 9.

FIGURE 9 Self-reported drug and/or alcohol use. (Figure not reproduced here; see the original article.)

Out of the 860 young people screened, 2 (0.2%) 12-year-olds, 12 (1.4%) 13-year-olds, 18 (2.1%) 14-year-olds, 19 (2.2%) 15-year-olds, 57 (6.6%) 16-year-olds, 72 (8.4%) 17-year-olds, 69 (8.0%) 18-year-olds, 67 (7.8%) 19-year-olds and 52 (6.0%) 20-year-olds reported using one or more substances. Alongside reporting substance use, young people were asked to complete part b of the CRAFFT form to assess the number with ‘risky’ substance use. Just under one-third of young people screened (n = 278) reported one or more risk taking behaviour (Figure 10).

FIGURE 10 Self-reported risk-taking behaviour. (Figure not reproduced here; see the original article.)

Recruitment data

A total of 860 children in care were screened. Contact attempts were not made with 356 (41%) children, as they did not provide consent to be contacted, and 269 (31%) children provided consent but reported no substance use in the last 12 months.

The research team initially used the contact details provided by the young person. Contact details included personal mobile/landline numbers (n = 185, 88%), e-mail (n = 22, 10%) and address only (n = 3, 1.5%), and one person provided a number for the supported accommodation they were in (n = 1, 0.5%).

Of the 235 children in care contacted, 24 young people (3%) were subsequently excluded; 11 were already in drug and alcohol services; six had an imminent move out of area; three had been closed to social services; and four for other reasons [recall to prison/secure unit (n = 1), mental health problems (n = 1), a learning disability (n = 1) and turning 21 years old (n = 1)].67

This left 211 young people who were potentially eligible to be recruited into the trial.

Once contacted, 75 (35.5%) children in care declined to take part for a number of reasons (Table 9), including not perceiving themselves to have a problem, not being interested in taking part and, for young people aged ≥ 18 years, the fact that drinking was legal.

A total of 112 (7.7%) of the original 1450 children in care aged 12–20 years were recruited and randomised into the study. A minimum number of 35 children per study arm (n = 105 minimum) was required for us to proceed with the pilot RCT. Following extensive discussions with the National Institute for Health Research (NIHR), the study TOC and the study management team, a decision was made to continue with the pilot RCT, despite recruiting only seven young people above our minimum target. To our knowledge, a RCT delivering drug and alcohol treatment interventions to children in care has not been conducted previously; therefore, after consultation, the pilot RCT was continued to specifically assess engagement with the adapted interventions and the retention of participants at 12 months’ follow-up.

A visual representation of the recruitment progress by site is in Table 10.

The planned recruitment period was due to take place over a 3-month period from September to November 2016. The actual recruitment period was 1 year: 1 November 2016 to 31 October 2017.

Randomisation data

Children in care were distributed in a balanced manner across the trial arms and for each stratification variable (Table 11).

Baseline data

Demographic characteristics

Baseline demographics, clinical characteristics and trial stratification factors are summarised across treatment groups descriptively in Table 11.

Note that participant 36 (female, aged 19 years) was randomised, but her baseline data did not transfer electronically to the trial database due to a technical fault. This participant’s data could not be included in the baseline analysis, leaving 111 participants providing baseline data. Follow-up data at 12 months post randomisation was collected for participant 36. (Some demographics for participant 36 were therefore not available, so these are coded as missing.)

Sex was fairly evenly distributed across the MET and SBNT groups, although females comprised 67% of those randomised to usual care. Ethnic group was well balanced, although there were very few non-white participants in each arm, reflecting the ethnic mix of the local authorities.

Recruitment errors

Baseline data are presented in Baseline questionnaire summaries. Further analyses looking at variables by sex, age and residential type are presented in Appendix 4 (see Tables 33–50).

Baseline questionnaire summaries

Table 12 shows the distribution of scores at baseline for the AUDIT, AUDIT-C, ASSIST-Y, SDQ and WEMWBS; the data for the EQ-5D-5L are discussed further in Chapter 6. Further analyses looking at variables by sex, age and residential type are presented in Appendix 4 (see Tables 33–50).

Owing to the questionnaires being completed on a tablet computer, there was no option to skip any questions. Participants could choose to exit the baseline data collection without completing all the questionnaires; however, this did not happen. All 112 children carefully completed the baseline data collection questionnaires; however, as discussed in Demographic characteristics, a data transfer error meant that data were missing for one child.

Intervention attendance

In total, 15 (20%) out of the 76 children in care randomised to an active intervention arm attended at least one session of the intervention across MET and SBNT arms combined. Sessions took place at their chosen location, for all participants this was at their placement address (foster placement, residential home or independent living). Closer inspection of participation data shows an impact of the change of inclusion criteria. Twelve out of the 15 (80%) young people who attended any sessions were recruited under the original threshold of reporting two or less risky behaviours, compared with only three (20%) young people recruited under the new criterion of reporting any substance use in the last 12 months and reporting more than two risky behaviours. This would suggest that individuals reporting higher levels of risky substance use behaviour are more likely to engage with interventions and that future trials should use the original criteria set for this trial. The summary statistics for time from randomisation to first intervention session are reported by way of summary statistics in Table 13.

Of the 38 participants in the MET arm, five took up at least one session, three young people attended two sessions and one young person attended all six sessions. Similarly, of the 38 participants in the SBNT arm, four attended at least one session, one young person attended two sessions and one attended three sessions (Figure 11).67 Regarding the SBNT sessions and the principle of young people recognising support available to them, all six young people who attended sessions could identify at least one person who could provide them support and the two participants whom attended more than one session could identify how their network could support them. However, due to them discontinuing sessions, we are unable to report how their support networks performed in practice.

FIGURE 11 Attendance at MET and SBNT sessions. (Figure not reproduced here; see the original article.)

Seven young people provided reasons for early discontinuation of the intervention, as listed in Table 14 (note, pseudonyms are used within the descriptive analysis).

The 12-month follow-up data

In total, 60 children in care out of a potential 112 (54%) attended the 12-month follow-up visit, 19 (17%) declined follow-up, 11 (10%) did not attend follow-up meetings that were arranged, 18 (16%) were lost to follow-up due to the research team being unable to contact them and four (4%) could not be followed up for other specific reasons (follow-up deemed to be inappropriate by the allocated social worker for two and two could not be contacted as they were detained in a secure unit). When attempting to contact young people at follow-up, for 58 (52%) young people, their contact number had changed and for 54 (48%) young people their address had changed. Details of follow-up status are given in Table 15.

Of the 60 children in care followed up, 5 out of the 15 young people who participated in the intervention completed the questionnaire, the other 10 were lost to follow-up for a range of reasons [declined (n = 2), an appointment was made but the young person did not attend (n = 4), unable to contact them (n = 1), inappropriate circumstance (n = 1) and closed to social services (n = 2)].

In many instances, young people were lost to follow-up due to the multiple competing demands that they had going on in their lives, the frequent changing of mobile numbers and the placement instability, which meant that the contact details originally given were no longer current. Forty-two children in care (38%) were followed up within an 8-week window of 12 months, as indicated in Table 16.

The distribution of data for the proposed primary outcomes – episodes of heavy episodic drinking (≥ 5 units in 1 day) in the preceding 30- and 7-day periods as shown in Tables 17 and 18 – outcomes for a definitive trial were obtained from the TLFB-30 administered during the 12-month follow-up interview.

These questionnaires were assessed for completeness. For the questionnaire summaries, 60 young people had follow-up data, as shown in Table 19.

An (adult) AUDIT score of ≥ 8 indicates hazardous alcohol consumption and a score of ≥ 13 for females and ≥ 15 for males indicates alcohol dependency. An AUDIT-C score of ≥ 4 in males and ≥ 3 in females indicates hazardous alcohol consumption.

Data on romantic relationships and antisocial behaviour are presented in Tables 20 and 21, respectively.

Missing data

As the RCT demographic and questionnaire data were collected using electronic tablet devices during baseline and follow-up visits, the only missing data are baseline data for one participant whose data were collected but not transferred owing to a technical fault and 12 months’ follow-up data for the 52 participants who did not attend.

There were no partially completed questionnaires because the electronic tablets necessitated that all questions be answered before moving on. Only one participant did not complete the TLFB at the 12-month follow-up visit, but did complete all other questionnaires at that visit. As there were so few missing data when they were collected using electronic tablet devices, there was no gain in using shorter tools, such as AUDIT, ASSIST-Y, TLFB – 7 days and CRAFFT.

Non-completion of 12 months’ follow-up

Participants had the right to withdraw from the trial at any time without having to give a reason. No children in care formally or explicitly withdrew from the trial, but 52 did not attend the 12-month follow-up.

End of trial

The end of the trial was defined as the 31 October 2018, which was the last date a participant could be contacted to complete the 12-month follow-up questionnaire.

Safety data

There were no reported adverse events or serious adverse events occurring from the time of randomisation in the pilot RCT until follow-up was completed (12 months post enrolment).

Conclusions

The pilot trial identified that a definitive trial using the same design as in this pilot is not feasible. We aimed to recruit 150 participants into the trial which allowed for 30% loss (45 participants) at follow-up, so aimed to retain 105 participants 12 months post recruitment. One hundred and twelve participants were recruited (75% of the target) and data were collected from 60 participants (53.5% of those recruited into the study) at 12 months post recruitment, which was a 46.5% loss at follow-up.

Seventy-six participants were randomised into the SBNT and MET arms. Delivery of the SBNT and MET intervention sessions was very low, with only 15 out of 76 participants (20%) allocated to the interventions receiving any session. With such low rates of delivery, it is not considered feasible to deliver the interventions using routine care pathways.

The inclusion criteria were amended during recruitment into the study, from children in care scoring more than two risky behaviours, which identified 74 participants, to reporting any substance use within the preceding 12 months, which led to the identification of a further 38 participants.

There were very few problems identified with the proposed data collection measures. The use of tablet computers was acceptable to young people and the programming to force a response meant that rates of completion of the baseline and follow-up questionnaires were consistently high once the young people had consented to take part. There were no missing data at baseline data and only one TLFB was missing at 12 months post recruitment.