A pilot feasibility cluster randomised controlled trial of screening and brief alcohol intervention to prevent hazardous drinking in young people aged 14-15 in a high school setting (SIPS JR-HIGH)

Article history

The research reported in this issue of the journal was funded by the PHR programme as project number 10/3002/07. The contractual start date was in October 2011. The final report began editorial review in September 2013 and was accepted for publication in June 2014. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PHR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Professor Eileen Kaner is a National Institute for Health Research (NIHR) Public Health Research funding board panel member. Professor Elaine McColl is a member of the NIHR Journal Editorial Group. Denise Howel is a NIHR Health Services and Delivery Programme board panel member.

Copyright statement

© Queen’s Printer and Controller of HMSO 2014. This work was produced by Newbury-Birch et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Research questions

The Medical Research Council (MRC) has presented a framework for the evaluation of complex interventions. 1 This work represents the development and piloting phases of the framework. Conducting a full-scale cRCT and economic evaluation of ASBI compared with ‘standard care’ in this population is likely to need many schools and to be resource intensive. As there are uncertainties regarding rates of eligibility, consent, participation in the intervention and retention for follow-up and regarding the feasibility and acceptability of the intervention for a range of stakeholders (teachers, learning mentors, young people and parents) this feasibility study was essential to inform the design and conduct of a larger scale definitive study.

The study sought to answer the following research questions: ‘Is it feasible to deliver ASBI in schools in England?’ and ‘What are the likely eligibility, consent, participation and retention rates of young people in a UK-relevant trial of ASBI compared with standard practice?’. Answers to these research questions will inform the development of a definitive multicentre cRCT to evaluate the effectiveness and cost-effectiveness of ASBI in reducing hazardous drinking in adolescents. Our hypothesis for the definitive cRCT will be that ASBI is more effective and cost-effective at reducing hazardous drinking in adolescents than a control condition of usual advice in high/comprehensive schools, as well as feedback on their drinking and an information leaflet.

Research objectives

1. To conduct a three-arm pilot feasibility cRCT (with randomisation at the level of school) to assess the feasibility of a future definitive cRCT of ASBI in a school setting.

2. To explore the feasibility and acceptability of ASBI and trial processes to staff, young people and parents.

3. To explore the fidelity of the interventions as delivered by school-based learning mentors.

4. To estimate the parameters for the design of a definitive cRCT of ASBI, including rates of eligibility, consent, participation and retention at 12 months.

5. To pilot the collection of cost and resource-use data to inform the cost-effectiveness/utility analysis in a definitive trial.

6. To develop the protocol for a definitive cRCT and economic evaluation of the impact of ASBI compared with standard advice to reduce alcohol consumption.

Chapters of the report

The report is structured as a series of eight chapters detailing the design, management and outcomes of the pilot feasibility study. The report begins by providing the background to the research and outlines key literature informing the design and conduct of the study (see Chapter 2). Following this, a chapter is dedicated to each core component of the study. Chapter 3 explores the design of intervention materials as well as the training and support provided to school staff in the delivery of the project. Chapter 4 reports the design, methods and results of the repeated cross-sectional survey. Chapter 5 provides the design, methods and results of the external pilot trial. Chapter 6 details the design, methods and results of the integrated qualitative process evaluation. Chapter 7 details the design, methods and results of the health-economic evaluation of the study. Finally, Chapter 8 provides a synthesis of the main findings from the pilot feasibility study, together with an assessment of whether the study met its aims and objectives, before detailing any recommendations for a future definitive cRCT.

Research ethics

The research study was granted ethical approval in November 2011 by Newcastle University, which acted as a sponsor for the research (reference 0508), and the trial is registered with the ISRCTN register as ISRCTN07073105. Approval was also granted by the local education authority in the study catchment area. Ethical approval was extended to accommodate a change in study protocol in October 2012, which related to measures completed at the 12-month follow-up of trial participants.

Changes to the original study protocol

The study protocol was published in 2012. 2

1. The published protocol indicates 6- and 12-month follow-ups for the trial group; however, it is not clear on the protocol that the full year group was followed up at 6 months and 12 months, as no identifiable data were taken at the year group level or the trial participant level at 6 months, therefore we have identifiable data for only the trial group at baseline and 12-month follow-up. The reason for this was that having a one-on-one interaction with the learning mentor could have acted as a ‘top-up’ to the intervention. We do not intend to include a 6-month follow-up in the proposed definitive study.

2. Objective 5 of the study – ‘to pilot the collection of cost and resource-use data to inform the cost-effectiveness/utility analysis in a definitive trial’ – was not included in the original study protocol.

3. The original protocol reported the control group as Personal, Social, Health and Economic Education (PSHE) only; however, the control condition was PSHE and also included the young person receiving feedback that he/she was drinking in a way that may be harmful and being provided with an advice leaflet. The reason that we added feedback and the leaflet (and therefore a change to the protocol) was that the research team and the University Ethics Committee believed that this was the minimally acceptable thing we could ethically do should a young person be identified as a risky drinker.

Research management

The Programme Management Group (PMG) was responsible for ensuring the appropriate, effective and timely implementation of the project. The PMG met once per month (more or less frequently dependent on the needs of the project) and comprised the Chief Investigator, Project Manager, co-applicants, named collaborators and researchers working on the project. Professor Eilish Gilvarry chaired this group. A Trial Steering Group (TSG) was also appointed to provide an independent assessment of the data analysis and to help determine if a future definitive trial is merited. This group met biannually and their remit was the progress of the study against projected rates of recruitment and retention, adherence to the protocol, participant safety and the consideration of new information of relevance to the research question. Professor Mark Bellis chaired this group. Written terms of reference were agreed and used by the PMG and TSG (see Appendix 1).

Research governance

The project complied with the requirements of the Data Protection Act 1998 and the Freedom of Information Act 2000, and other UK and European legislation relevant to the conduct of clinical research. The project was managed and conducted in accordance with the MRC’s guidelines on good clinical practice in clinical trials (www.mrc.ac.uk), which includes compliance with national and international regulations on the ethical involvement of patients in clinical research (including the Declaration of Helsinki, sixth revision 2008). All data were held in a secure environment with participants’ information identified by a unique participant identification number. Master registers containing the link between participant identifiable information and participant identification numbers were stored in a secure area that was separate from the majority of data. All staff working on the project were employed by academic organisations and subject to the terms and conditions of service and contracts of employment of the employing organisations. Where relevant, staff were trained in good clinical practice and all staff worked to written codes of confidentiality. The project used standardised research and clinical protocols, and adherence to the protocols was monitored by the PMG and TSC.

Patient and public involvement

Patient and public involvement (PPI) was sought at different time points and at multiple levels, and is reflected upon throughout this report.

Patient and public involvement representatives included local authority employees, parents, young people and members of staff at participating school sites. Their contribution to the development, management and delivery of this research included input into the design and conduct of the feasibility study (the local authority lead for education was a co-applicant for this research) and piloting of study documentation and intervention materials (parents and young people) to ensure readability and understanding (see Chapter 3). Participating schools were also heavily involved in the conduct of the feasibility study (trial and survey) and were regarded as key stakeholders (see Chapters 4 and 6). Finally, Chapter 8 includes modifications recommended for a definitive trial, which include input from PPI representatives.

Key points for Chapter 2

• Adolescents in England are among the heaviest drinkers in Europe, with consumption highest in the north-east.

• Young people are more vulnerable than adults to the adverse effects of alcohol owing to a range of physical and psychosocial factors that often interact.

• Literature shows that the ASBI for young people has been successful for selected individuals in certain settings.

• There is currently insufficient evidence to be confident about the use of ASBI to reduce risky drinking and alcohol-related harm in younger adolescents in a school setting.

• Despite well-documented parental influences over adolescent alcohol use, the evidence for interventions to reduce young people’s drinking that include family members is equivocal.

Prevalence

Adolescents in England are among the heaviest drinkers in Europe. 3 The percentage of young people who have ever had an alcoholic drink in England increases with age from 12% of those aged 11 years to 74% of those aged 15 years,4 and the prevalence of drinking in the last week rises from 1% of those aged 11 years to 25% of those aged 15 years. 4 Although the proportion of young people in England aged between 11 and 15 years who report that they have ever drunk alcohol decreased from 54% to 43% between 2007 and 2012, the mean amount of alcohol consumed by this age group has fluctuated between 10.4 units per week in 2011 and 14.6 units per week in 2008, with an increase to 12.5 units per week in 2012. There are, however, age-related differences in patterns of consumption. The amount consumed among those aged 14 years has increased from 13.2 units per week in 2007 to 16.15 units in 2012, whereas for 15-year-olds the mean amount has decreased slightly from 14.2 units per week in 2007 to 12.3 units in 2012. 4 This clearly shows that drinking increases throughout adolescence, but recent data show that this is not immutable with changes in trends between years and age.

In particular, the north-east has been shown to have the highest rates of alcohol misuse by young people in England, with 51% of 11- to 15-year-olds reporting having ever drunk alcohol. 4 This compares with 48% in the south-east, 46% in the north-west and 31% in London. 4 Further, the mean alcohol consumption in the previous week for young people in England in 2011 was highest in the north-east and north-west (15.7 units per week) compared with the south-east (11.0 units) and London (9.4 units). 4 Therefore, the north-east is a key place to study the issue of alcohol risk reduction in young people.

Consequences of drinking

The impact of alcohol on the development and behaviour of young people has been well researched in early,5 middle6 and late adolescence. 7 It is now well known that young people are much more vulnerable than adults to the adverse effects of alcohol due to a range of physical and psychosocial factors that often interact. 8 These adverse effects include physiological factors resulting from a typically lower body mass and less efficient metabolism of alcohol;5,6 neurological factors due to changes that occur in the developing adolescent brain after alcohol exposure;6,9–11 cognitive factors due to psychoactive effects of alcohol that impair judgement and increase the likelihood of accidents and trauma;12 and social factors that arise from a typically high-intensity drinking pattern that leads to intoxication and risk-taking behaviour. 13,14 The social factors are compounded by the fact that young people have less experience of dealing with the effects of alcohol than adults15 and they have fewer financial resources to help buffer the social and environmental risks that result from drinking alcohol. 7

Evidence suggests that hazardous drinking among young people occurs commonly in the context of other forms of ‘disinhibitory behaviour’, such as aggression and risk-taking. 16 Although these behaviours are well known to be linked,16 it is not clear if drinking leads to these behavioural problems or if they all arise due to a common linked trait. 17 A significant positive association between alcohol dose and aggression for both genders has been found. 18 As a result of the above risk factors, the list of negative consequences that result from drinking in young people is extensive and includes physical, psychological and social problems in both the shorter and the longer term. Immediate problems result from accidents and trauma, physical and sexual assault (including rape in young people), criminal behaviour (including driving while intoxicated and riding as a passenger with an intoxicated driver) and early onset of sexual intercourse and sexual risk-taking. 8,14,19 In relation to education, alcohol use can have a negative effect on school performance20 and those who have drunk are also more likely to have truanted from school. 4 Longer-term problems include the development or exacerbation of mental health problems,21 self-harm and/or suicidal behaviour. 22 Moreover, individuals who begin drinking in early life have a significantly increased risk of developing alcohol use disorders, including dependence, later in life. 23,24 Owing to this extensive array of damage, the prevention of excessive drinking in young people is a global public health priority. 25 In 2009, the Chief Medical Officer for England provided recommendations on alcohol consumption in young people,26 based on an evidence review of the risks and harms of alcohol to young people. 8 The recommendations state that children should abstain from alcohol before the age of 15 years and those aged 15–17 years are advised not to drink, but if they do drink it should be no more three to four units and two to three units per week in males and females, respectively, on no more than 1 day per week. 26

Young people’s views on their own health

It is important to note that young people often feel that they want to be empowered to be part of any decision-making in relation to their own health and feel that they have choices (C Sands, Newcastle University, 2013, unpublished data). For young people, confidentiality is a key issue, particularly within the school setting. However, to young people it is really important that they are familiar with the staff working with them, and therefore these issues should be taken into consideration when undertaking research with young people.

Primary and secondary prevention interventions for risky drinking

There is a large volume of evidence on primary prevention in the school setting,27–32 which is directed at all young people, whether they drink alcohol or not, with the aim of delaying the age at which drinking begins, and which uses general health education to prevent underage drinking. This body of work has shown mixed results and been reported to be methodologically weak,33 with only a relatively small number of programmes reporting positive outcomes. 30 One programme that has shown effectiveness is the School Health and Alcohol Harm Reduction Project (SHAHRP) project, a curriculum programme delivered across two consecutive school years in Ireland with 2349 pupils (mean age of 13.84 years at baseline and 16.48 years at final follow-up at 32 months). The programme had an explicit harm-reduction goal that explores knowledge, attitudes, alcohol consumption, and context of use and harms associated with a person’s own, or other people’s, use of alcohol. This showed significant improvements among young people in the intervention group in relation to alcohol knowledge and significant reductions in alcohol consumption. 34 Furthermore, research from the USA found that targeting young people and parents simultaneously but separately was effective in postponing the onset of heavy drinking among adolescents. 35,36 However, the results are equivocal, with some studies showing effectiveness and others not, and questions remain about the applicability to the UK setting. 27 As has been shown, there is limited evidence to support primary prevention programmes to reduce alcohol consumption in young people. Thus secondary prevention, i.e. targeting interventions at young people who are already drinking alcohol, is likely to be a more effective strategy, as the interventions will have more salience for the individuals receiving them.

Various screening measures have been used with young people to identify those who are at risk from their drinking including using measures of total alcohol consumption, levels of binge drinking and alcohol-related injury levels. 37 Research suggests standardised alcohol screening tools, such as the Alcohol Use Disorders Identification Test (AUDIT)38 are a highly sensitive and specific means of identifying current hazardous use of alcohol in adult populations, including college students. 39–41 Among adult drinkers, the AUDIT detects approximately 92% of genuinely excessive drinkers (sensitivity) and excludes approximately 94% of false cases (specificity),42,43 for which a cut-off score of ≥ 8 (out of a possible score of 40) is used to detect hazardous use of alcohol and alcohol-related problems. Broken down further, respondents can be categorised as ‘abstainers’ (0); ‘lower risk’ drinkers (1–7); at ‘increasing risk’ (8–15); at ‘higher risk’ (16–19); or ‘probably dependent’ (≥ 20).

There is some evidence from emergency department settings in the USA to suggest that the AUDIT is an appropriate means of detecting hazardous use of alcohol and alcohol-related problems among adolescents. 44–46 However, evidence remains equivocal whether it is either practical or appropriate for use with adolescents in other settings, including primary care and education. At 10 items, the length and wording of the full AUDIT may make it impractical for use with adolescents. 42,47 Evidence is especially equivocal as to the AUDIT tool’s ability to detect hazardous level drinking (the AUDIT positive score) among this age group or whether the concepts of hazardous or harmful drinking in adults are similarly meaningful in adolescents. Several studies have used AUDIT positive cut-off scores of ‘8’, designed for use with adults, to screen for alcohol use disorders among adolescents. 47,48 In comparison, other evidence supports using lower cut-off points, which generally fall between ‘2’ and ‘4’,43,49,50 when using the AUDIT in adolescent populations. For example, Chung et al. 44 recommend using a cut-off score of ‘4’ with young people aged 13–19 years (sensitivity 0.94; specificity 0.80) and Knight et al. 50 suggest that a score of ‘2’ is optimum for the identification of alcohol problems and disorders among those aged 14–18 years (sensitivity 0.88; specificity 0.81). Santis et al. 49 suggest different scores according to the level of alcohol consumption, with cut-off points of ‘3’ for hazardous, harmful and dependent alcohol use (sensitivity: 96%; specificity: 63.3%), ‘5’ (sensitivity: 75%; specificity: 64.5%) and ‘7’ (sensitivity 64%; specificity 75%), respectively.

Others suggest using a shortened version of the AUDIT tool, such as AUDIT-C (AUDIT-Consumption), which is scaled 0–12, and for which a score of ≥ 5 (among adults) is used to indicate increasing or higher risk drinking. 40,51 No specific score for young people has yet been recommended. It has also been shown that a single question screen based on drinking frequency can adequately identify youths with alcohol-related problems. 52,53 Bailey et al. 53 used the frequency of binge drinking (question three of the AUDIT tool – six or more drinks in one drinking session) to identify risky drinking in young people. 53 Thus, there is no clear consensus on which screening tool should be used, the validity of lower AUDIT or AUDIT-C cut-off points for use with adolescent populations or as to what this score should be, and whether the AUDIT or AUDIT-C or another measure should be the screening measure of choice. It could therefore be argued that in a school setting a shorter screening tool could be useful, and quick, to administer.

In terms of interventions for dealing with people who are drinking at harmful or hazardous levels, ASBI is a secondary preventative activity, aimed at individuals whose consumption level or pattern is likely to be harmful to their health or well-being. 54 They generally consist of screening (to identify relevant recipients) followed by structured advice or counselling of short duration, which is aimed at reducing alcohol consumption or decreasing the number or severity of problems associated with drinking. 55 They are based on social cognitive theory (from health psychology), which is drawn from the concept of social learning. 56 Here, behaviour is regarded to be the result of an interaction between individual, behavioural and environmental factors. It is assumed that each individual has cognitive (thinking) and affective (feeling) attributes that affect not only how they behave, but also how their behaviour is influenced and/or reinforced by aspects of the external world. Thus ASBIs generally focus on individuals’ beliefs and attitudes about a behaviour, their sense of personal confidence (self-efficacy) about changing it and how an individual’s behaviour sits in relation to other people’s actions (normative comparison).

A key feature of ASBI is that it is designed to be delivered by generalist practitioners (not addiction specialists) and targeted at individuals who are generally not experiencing severe problems (such as alcohol dependence) and who may not even be aware that they are experiencing alcohol-related risk or harm. Thus the goal is usually reduced alcohol consumption or a decrease in alcohol-related problems. 57 There is variation in the duration and frequency of ASBI58 but there are two broad types: simple structured advice – based on the FRAMES (Feedback, Responsibility, Advice, Menu, Empathy and Self-efficacy)59 – and behaviour change counselling – based on MI. This is a person-centred approach that aims to resolve conflicts regarding the pros and cons of behaviour change and thus enhance motivation. MI is characterised by empathy and an avoidance of direct confrontation. Elicited statements associated with positive behaviour change are encouraged so as to support self-efficacy and a commitment to take action. Since the time available for delivering BI may not allow for MI in its full form,58 its ethos and techniques have been59 distilled into a more directive format called Behaviour Change Counselling. 60

There is a large amount of high-quality evidence to support the effectiveness of ASBI with adults who have an alcohol use disorder. 58 Most of the evidence for ASBI demonstrates effectiveness for non-treatment-seeking adults in primary health care. 58,61–68 Furthermore, meta-analyses have consistently reported that students aged ≥ 18 years who received ASBI subsequently reduced their drinking behaviour compared with control group participants who typically received assessment only. 69,70 The key elements of the ASBI were personalised feedback on alcohol consumption, typically with a normative component70 and/or MI approaches. Such interventions typically achieved small to medium effect sizes71 across multiple measures of alcohol consumption, including quantity, frequency and intensity of drinking. The effects of BIs on drinking behaviour often peaked in the shorter term (generally 6 months) then diminished over time. 69 However, reductions in alcohol-related problems often took longer to emerge but were found in longer-term follow-up (12–18 months). Hence it is important to have BI outcomes measuring both consumption and alcohol-related problems and to follow-up participants at shorter- and longer-term time points.

Numerous systematic reviews have been published on ASBI in younger adolescents in recent years37,72–79 (details given in Appendix 2). Jackson et al. ’s review37 of ASBI for young people in health settings identified eight controlled trials53,80–85 for young people. The work was part of a larger review of ASBI in adults and young people. The trials were published between 1999 and 2009 and the majority (seven)80–86 were carried out in the USA. Study population sizes ranged from 34 to 655 young people and included young people aged between 12 and 24 years with two of the included studies being for those aged ≥ 18 years only. Five of the trials tested a brief MI, which lasted between 20–45 minutes,81,83–86 whereas one tested an audio programme;80 another involved a more intense programme of MI which included four sessions53 and one comprised an interactive laptop computer-based intervention. 82 The length of follow-up varied between 2 and 12 months. Four of the studies53,83–85 found statistically significant benefits as a result of the intervention. However, one80 of the studies found negative consequences following intervention, with an increase in heavy alcohol use among the intervention group. The authors offer two possible explanations for this. First, adolescents in the control group, unlike adolescents in the intervention groups, reported less bingeing after baseline, suggesting self-report bias in the direction anticipated if the control adolescents were trying to please the researchers. Second, the authors argue that the apparent increase in self-reported alcohol use in the intervention groups relative to the control groups was the result of an educational intervention influence leading adolescents to be more forthright. 80 Wachtel and Staniford77 also reviewed the literature in relation to alcohol misuse and binge drinking in adolescents in the clinical setting (hospital-based emergency departments, college health centres and adolescent healthcare clinics). 77 The review included 14 studies,53,80,82–85,87–94 12 of which were from the USA. 53,80,82–85,87–91,93 Nine of the included studies83,84,87–93 related to young people aged ≥ 18 years and included a heterogeneous range of interventions from very brief MI to four group sessions of 30–40 minutes, which meant that generalisability could not be ascertained. A review of the literature by Yuma-Guerrero et al. 78 around BI in emergency departments in the USA for young people identified seven randomised controlled trials (RCTs). 82–85,95–97 The primary studies included young people aged 12–20 years. Four82–84,98 of the included studies demonstrated a significant intervention effect; however, none reduced both alcohol consumption and alcohol-related consequences.

Mitchell et al. ’s systematic review75 identified 15 studies81,82,85,86,95–105 of alcohol and drug interventions delivered to adolescents in primary care (one study81), emergency departments (seven studies82,85,95–98,103), schools (five studies99,101,104–106) and other settings (one study with homeless young people86 and one in the community100) with young people aged 12–21 years. The authors identify the need for screening instruments to be brief to administer and quick and easy to score and interpret. 75 Because of the heterogeneous populations (ages 12–22 years), inclusion criteria (adolescents who use alcohol and drugs as well as those who reported being in a car with an intoxicated driver but who themselves had not used alcohol or drugs) and differences in outcome measures, the data did not allow for meta-analysis, although some individual studies did show reductions in alcohol consumption at follow-up. The review identified two studies101,102,106 (three articles) carried out in further education colleges in the UK, with older young people aged between 16 and 20 years, in which no differences were found between groups at 12-month follow-up.

Three systematic reviews included meta-analyses of ASBI for young people. 73,74,76 Tripodi et al. 76 carried out a meta-analytic review on interventions for alcohol abuse in a range of settings. Sixteen primary studies were included with young people aged 12–19 years. 81,105,107–120 The studies included various interventions including BI, cognitive–behavioural therapy (CBT) and multidimensional family therapy. The main outcome measures included abstinence and quantity of alcohol use measured between 1 and 12 months post intervention. Pooled effects of standardised mean differences indicate that interventions significantly reduce alcohol consumption [Hedges’ g = −0.61; 95% confidence interval (CI) −0.83 to −0.40]. Stratified analyses revealed larger effects for individual treatment (Hedges’ g = −0.75; 95% CI −1.05 to −0.40) compared with family-based treatments (Hedges’ g = −0.46; 95% CI −0.66 to −0.26). 76

Jensen et al. ’s review74 of the effectiveness of MI for substance-use interventions for adolescents included 21 primary studies, of which 12 had alcohol-related outcomes. 53,81,85,86,99–102,108,121–123 These studies were from a variety of different settings: educational,99,101,102,123 community53,86,100,108 and health. 81,85,121,122 No information was given on the nature of the interventions; however, the number of sessions ranged from one to four. The age range included in the studies was 12–23 years. Included studies that addressed alcohol and other drug use yielded a small, but significant, post-treatment effect size in reduction of substance use [mean d (standard mean difference) = 0.146 (95% CI 0.059 to 0.233), n = 16)].

Carney et al. ’s meta-analysis73 aimed to identify and evaluate early interventions that target adolescent substance use (alcohol and illicit drugs) as a primary outcome, and criminal or delinquent behaviours as a secondary outcome. They identified nine studies53,81,85,86,97,99,109,124,125 in emergency departments, juvenile correctional facilities, alternative high schools and a homeless drop-in centre – eight from the USA81,85,86,97,99,109,124,125 and one from Australia. 53 Study sizes ranged from 18 to 472. The age range was 15–17 years. Results showed that single sessions of BIs significantly reduced the frequency of alcohol use among young people (I² = 0%; z = 2.13; overall effect, p = 0.03). 73

Conrod et al. 126–128 have carried out a number of trials in London of group-based personality-targeted prevention for young people aged 13–14 years who are risky drinkers or drug users. The interventions consisted of two 90-minute group sessions that incorporated components of motivational enhancement therapy and CBT. The intervention was unique in that it targeted personality traits rather than problems. In fact, alcohol and drug use were a minor focus of the intervention. Young people have been followed up every 6 months for 2 years and long-term effects (at 2 years) have been found for problem drinking (measured using the Rutgers Alcohol Problems Index (RAPI) tool) (p = 0.02) and binge-drinking rates (p = 0.03). Finally, a study of US accident and emergency attendees97,129 who received ASBI showed reductions in aggression, as well as reductions in alcohol misuse following a brief alcohol intervention.

It has been shown that the family is a source of both risk and protective factors for adolescent alcohol use. 8 Parents in particular have been found to have a significant effect upon alcohol initiation and patterns of use. 130 Such parental factors include parental modelling,131–133 supervision and discipline,8 quality of parent–child relationship and communication among others. 134 It is therefore important to identify whether parents can play a role in helping to reduce their children’s drinking.

Parents

The majority of parents are aware that their children are drinking. 135 Parenting ‘style’ and ‘good’ family relationships have been demonstrated to have a positive effect on young people’s drinking behaviour regardless of family structure or whether parents consume alcohol. 8,134,136 Excessively authoritarian and permissive parenting styles are both associated with earlier onset of alcohol use or higher levels of drinking behaviour,137,138 and Foxcroft and Lowe139 identify a possible curvilinear relationship between control and adolescent drinking, where significantly stricter or lax parenting styles appear to increase the frequency of alcohol misuse.

Parents can also be a primary source of the supply of alcohol to young people. 140,141 This may be through the provision of money, by having alcohol available or by purchasing alcohol for young people directly. Easy availability of alcohol is associated with increased adolescent alcohol consumption142 and Elliott et al. 141 found that 65% of drinkers (aged 11–17 years) accessed alcohol via their parents. Further, it is implicitly assumed that if parents purchase alcohol for their children directly, the amount of alcohol consumed can be strictly monitored. In other words, that providing young people with alcohol will stop them from accessing it elsewhere, thus reducing the risk of alcohol-related harm. Again, the evidence for this is equivocal. On the one hand, Bellis et al. 143 found that (in contrast with other ways of obtaining alcohol) young people (aged 15–16 years) whose parents bought alcohol for them were less likely to drink in a public setting, ‘binge’ drink, drink heavily or drink frequently. On the other hand, receiving alcohol from a parent or getting it from home has been demonstrated to be the strongest predictor of increased alcohol use over time. 144 However, Gilligan et al. 145 found that negative outcomes from parental provision of alcohol are dependent on the context of supply. In other words, if parents supplied young people with alcohol, this did not increase the odds of risky drinking (although it also did not have the protective effect that motivated the behaviour). However, if alcohol was supplied for consumption without parental supervision then the odds of risky drinking were four times higher.

Given the well-documented parental influences over adolescent alcohol use, interventions that aim to involve parents who enhance parents’ awareness of the variables and strategies that can delay onset and reduce consumption levels in their child offer an opportunity for limiting the harms of adolescent drinking. 134

Alcohol Screening and Brief Interventions that include parents

Mixed effects have been found for ASBI for reducing young people’s drinking that include family members. 33,54,146,147

A RCT examining the effectiveness of 45–60 minutes of individual motivational interviewing (IMI) compared with IMI and a family check-up session found that both interventions resulted in significant reduction of drinking outcomes at 3-, 6- and 12-month follow-ups. The family check-up consisted of a 1-hour meeting, at which the parent(s) and the young person discussed family beliefs regarding alcohol and other drug use. Results show there was only one significant between-group difference on the number of high-volume drinking days at 3- and 6-month follow-up, with family check-up reporting lower alcohol prevalence compared with IMI. This effect had diminished at 12-month follow-up. 103

A RCT with three arms: (1) two 60-minute individual sessions of BI (young person and interventionist only); (2) two 60-minute individual sessions of BI (young person and interventionist only) and a BI session with the parent(s) [parent(s) and interventionist only]; and (3) control arm of assessment only found that both intervention groups showed significantly better drinking outcomes than the control arm for number of alcohol days and number of binge days with a small sample (n = 78). The intervention arm that included parental involvement reported significantly fewer alcohol days at 6-month follow-up than the intervention group without parental involvement but no difference in number of binge days. 105 This study was repeated with a large sample (n = 315) and, again, both intervention arms were found to be superior to the control condition. Significant between-group differences were reported by this trial in favour of the arm with parental involvement for drug outcomes but not alcohol. Indeed, the intervention arm without parental involvement reported significantly greater alcohol abstinence in the previous 90 days than the arm with parental involvement. 125

Mixed results have been found for intensive BIs for drug and alcohol using adolescents, with parental involvement (see Appendix 2). 112,115,117,148,149 However, significant variation exists between experimental conditions examined with regards to both the intensity and the frequency of the intervention (ranging from a single 1-hour family check-up to 64 hours of family and individual CBT), as well as the theoretical basis of the therapeutic approach. Moreover, the heterogeneity of the adolescent samples, which included dually diagnosed adolescents, risky drinkers, drug and alcohol users, runaways and gang-affiliated young people, made it difficult to compare the findings of the trials. Therefore, the evidence is equivocal.

Rationale for the present research

The literature shows that ASBI for young people has been successful for selected individuals, in certain settings. In particular, the current available evidence relates primarily to white, USA-based subjects, most often in educational settings and at the older end of the youth spectrum (see Appendix 2). However, there is currently insufficient evidence to be confident about the use of ASBI to reduce excessive drinking and/or alcohol-related harm in younger adolescents and in a school setting. Nevertheless, the current evidence base suggests that the most effective forms of BI are those containing personalised feedback about a young person’s drinking behaviour and MI approaches to help reduce levels of alcohol-related risk. Furthermore, there is some evidence to show that involving parents in ASBI may be beneficial. This present work builds on the evidence base by focusing on ASBI to reduce hazardous drinking in younger adolescents (aged 14–15 years). It is highly likely that if a BI was effective at reducing hazardous drinking, it might also result in a range of other positive behavioural outcomes, as has been found in the adult literature as well as work with older adolescents.

Key points for Chapter 3

• Learning mentors were identified to be best placed within a school setting to deliver an intervention about alcohol. They were trained in study procedures and intervention delivery.

• The study incorporated control, intervention 1 and intervention 2 conditions, all manualised and designed to be delivered on a one-to-one basis to young people who screened positive for risky drinking and left their name on the questionnaire.

• Young people who were in control group schools met with the learning mentor who explained the study to them, and provided feedback that they may be drinking at a risky level, along with an alcohol information leaflet to take away and read.

• In addition to feedback and an alcohol information leaflet, young people allocated to intervention 1 took part in a 30-minute, six-step, interactive intervention led by the learning mentor.

• In addition to receiving intervention 1, young people who received intervention 2 were invited to attend a subsequent session with parental/family involvement, designed to last approximately 30–60 minutes, led by the learning mentor.

• Learning mentors were asked to record time spent with participants using open-ended case diaries.

Introduction

The present study incorporated control, intervention 1 and intervention 2 conditions. All three interventions were manualised to ensure consistency of delivery across schools allocated to that arm of the trial and reproducibility by other deliverers (see Appendix 3). Owing to availability of resources, all tools and manuals were provided in the English language only. All young people recruited into the trial, regardless of arm, continued to receive ‘standard alcohol advice’, delivered as part of the school curriculum. The first section of this chapter is concerned with defining what this consisted of in the study catchment area. In addition, young people in schools allocated to the control arm received feedback that they may be drinking at a risky level and an alcohol information leaflet. Young people in schools allocated to the intervention 1 arm took part in a 30-minute one-to-one structured intervention session with a trained learning mentor. In addition to receiving intervention 1, young people in schools allocated to the intervention 2 arm were invited to attend a subsequent session, facilitated by trained school staff, with parental/family involvement, designed to last approximately 30–60 minutes. Young people from schools allocated to intervention 1 and intervention 2 received the same alcohol advice leaflet as those allocated to control. All young people recruited into the trial were followed up 12 months post intervention.

The rest of the chapter describes the design of intervention materials, as well as the training and support provided to learning mentors in the delivery of interventions. The rationale behind, and development of, each intervention condition (control, intervention 1, intervention 2) is detailed, and the outcomes of piloting of materials and consultation with key groups (parents and young people) are outlined with any resultant modifications to intervention materials reported.

Defining ‘standard alcohol advice’

In order to fully understand the control context, it was first important to determine the scope of ‘standard alcohol advice’ received by all young people aged 14–15 years at secondary school (Years 10 and 11). Provision of classroom-based drug and alcohol education continues to be recognised as an important aspect of the secondary school curriculum (for those aged 11–16 years) for England, Scotland and Wales, and is generally tackled within PSHE classes. PSHE is non-statutory yet the provision of high-quality PSHE forms a significant part of the Office for Standards in Education, Children’s Services and Skills (OFSTED) inspections and contributes to the statutory responsibility of schools to ‘promote children and young people’s personal and economic well-being; offer sex and relationships education; prepare pupils for adult life and provide a broad and balanced curriculum’,150 delivered as part of a wider ‘well-being’ remit through the National Healthy Schools Programme151 and the Social and Emotional Aspects of Learning (SEAL) strategy. 151

However, there are no prescriptive guidelines on what PSHE should actually entail, as long as it encompasses these wider statutory responsibilities. As a result, schools have developed their own versions of PSHE, and different ways to deliver it, rather than following standardised frameworks of study. 150 Our observations mirror this and the research team recorded different PSHE arrangements in each of the participating schools. For example, several schools timetabled weekly lessons dedicated to PSHE topics, sometimes described as ‘citizenship’ or ‘extended tutorial’. One participating school had no PSHE provision and instead timetabled a ‘health day’ once per academic year. Schools were also able to elect a key ‘well-being’ focus for the coming academic year. Thus, if they chose to elect alcohol rather than another area (such as self-harm or sexual health) then this could feasibly have an impact on educational provision. Thus, the control context was a highly variable condition, with ‘standard alcohol advice’ defined in this study as the regular provision of classroom-based alcohol education to Year 10 pupils as delivered at each particular school site.

School staff identified to deliver interventions

Learning mentors are specifically trained to provide a service complementary to that of teachers and other school staff, addressing the needs of children who require assistance in overcoming barriers to learning in order to achieve their full potential. All secondary schools have learning mentors working in them. They work with a range of pupils, but give priority to those who need the most help, especially those experiencing multiple disadvantages. Mentoring covers a wide range of issues, from punctuality, absence, bullying, challenging behaviour and abuse to working with able and gifted pupils who are experiencing difficulties. Learning mentors support, motivate and challenge pupils who are underachieving. They help pupils overcome barriers to learning caused by social, emotional and behavioural problems. Learning mentors need good listening skills and an understanding of health and social issues that affect children and young people’s development. The mentors mainly work with children who experience ‘barriers to learning’, including poor literacy/numeracy skills, underperformance against potential, poor attendance, disaffection, danger of exclusion, difficult family circumstances and low self-esteem. Thus, learning mentors were thought to be most well-placed within a school setting to deliver an intervention to young people about alcohol use.

Local areas vary in their essential qualifications for appointment for learning mentors. However, as a minimum, they need to have a good standard of general education, especially literacy and numeracy, as well as experience of working with young people. Within the present study, learning mentors were defined as the members of school staff trained in the delivery of the control condition/interventions to participating students. However, in practice, within each school, titles, roles and responsibilities varied, and this did not constitute a homogeneous professional group. Thus, for consistency, school staff responsible for the delivery of interventions are referred to only as learning mentors throughout the rest of this report.

Patient and public involvement: selecting an alcohol information leaflet (control)

All young people recruited into the trial were provided with an alcohol advice leaflet. It was important that this leaflet was age appropriate (for 14- to 15-year-olds) and suitable for use in a school setting, yet with a presentation style favoured by young people. Owing to time and resource constraints, it was not feasible for the study team to design a new alcohol information leaflet. Instead, we reviewed a large amount of national and regional resources (including materials from the Department of Health, NHS Choices, Home Office, Talk to FRANK, Change4Life, ‘Know Your Limits’ and resources from local youth drug and alcohol services) and liaised with experts in the field. Two appropriate packs or leaflets were sourced, both designed by the Comic Company (www.comiccompany.co.uk/: the ‘Cheers! Your Health’ alcohol leaflet, and ‘snapper’, a quiz question folding game). Both leaflets were discussed with colleagues at Newcastle University, who are experts in working with young people in the school setting and who supported their use. Following this, they were piloted during five focus groups held with young people from years 9–11 (aged 13–16 years) at participating schools.

Young people across all focus groups agreed that the ‘Cheers! Your Health’ alcohol leaflet and ‘snapper’ resources were suitable for young people aged 14–15 years, and these materials were selected as alcohol advice leaflets and provided to young people in all arms of the pilot trial (see Appendix 3). In particular, young people indicated that encouraging them to engage with anything in a non-pictorial way would be challenging. Young people wanted the information presented to them in a fun or humorous way, without too much text, and liked leaflets to include games or puzzles. In particular, positive comments about the ‘Cheers! Your Health’ leaflet included that it was ‘detailed’, ‘interesting’ and ‘interactive’.

Young people who were in the control group schools met with the learning mentor who explained the study to them. The young people were told that they may be drinking alcohol in a way which may be harmful to them. Once consented to the study the young people were given the alcohol leaflets mentioned above to take away and read.

Development of intervention 1

The intervention 1 session was a manualised intervention, which combined simple structured advice and behaviour change counselling techniques commonly used within the extended BI. The tool was a colourful, six-step intervention, intended to be an interactive discussion between the young person and the learning mentor (see Appendix 3). It sought to increase awareness of risks and enable the young person to consider their motivations for changing their alcohol use. The intervention was designed to last approximately 30 minutes and take place in the learning mentor’s office or alternative suitable space. It was expected that young people would be taken out of class to attend appointments with learning mentors. The rest of this section details each step of the intervention tool in turn. Intervention 1 consists of six sections.

Intervention 1

Development of intervention 2

Although BIs are mostly delivered on a one-to-one basis, intervention 2 sought to build upon the rationale for intervention 1 by involving parents. Young people from schools allocated to intervention 2 received an individual BI (intervention 1), followed by a group family intervention based on MI principles held approximately 1 month after. 155 By involving parents within a family intervention, the approach focused upon the dynamic between the individual, attitudes and the environment. 56 Indeed, the addition of a family intervention has elsewhere been found to improve drinking outcomes in adolescents at follow-up. 103

Intervention 2 was a manualised intervention based upon the principles of MI. It was intended to be a discussion that built upon intervention 1 described above, wherein the young person and the learning mentor explored the young person’s drinking and their motivation for change. Intervention 2 sought to build upon the young person’s motivation by encouraging the parents/family members to share their thoughts about the young person’s drinking. The young person and the parents/family member were then encouraged to consider an action plan for change. The intervention was designed to last approximately 30–60 minutes (see Appendix 3). At the end of the session parents were provided with a parenting information leaflet about young people and alcohol use. It was expected that this session would take place either during or after school hours, either within the school or in a community centre nearby, and would take place only if the young person consented to parental involvement and parents subsequently agreed to take part. Intervention 2 consisted of four sections.

Intervention 2

Patient and public involvement: piloting of interventions

Interventions 1 and 2 were piloted with one young person and their mother by the research interventionist who had experience in MI techniques (December 2011 and February 2012, respectively). The intervention 1 session lasted approximately 25 minutes, whereas the intervention 2 session lasted approximately 45 minutes. The young person suggested adding information about calories to the intervention 1 tool as a way of making information about alcohol use more memorable and pertinent to young people. In particular, they suggested that they would have found this to be an effective motivator to changing drinking behaviour. A focus group was also held, in February 2012, with a convenience sample of four female parents, to discuss the intervention 2 tool, as well as anticipated methods for contacting parents to take part in the intervention. In particular, this group highlighted that the initial approach of school staff would be very important when introducing the project to parents for the first time over the telephone.

Modifications to the intervention materials as a result of piloting and consultation

As a result of piloting and consultation, the following modifications were made to intervention tools and materials:

• Provision of information about calories on the intervention 1 tool. The number of calories in popular food items (depicted using pictures) was mapped against alcohol brands and quantities that were popular with young people.

• A slight change to the guidance that was provided to school staff in relation to contacting parents for the first time about taking part in intervention 2. Specifically, the importance of non-judgemental language was reinforced with the learning mentors.

Training and support

All learning mentors received training prior to commencing the study. The training was split into four sessions, with each session lasting a minimum of 1 hour and a maximum of 3 hours. PowerPoint 2010 slides (Microsoft Corporation, Redmond, WA, USA) were used to guide each training session. An intervention manual relevant to each arm of the trial was provided to supplement the training (see Appendix 3).

Training was conducted in a community venue or school, as outreach training has been found to be the most cost-effective implementation strategy for ASBI delivery in other settings. 156 The training was jointly delivered by an experienced interventions trainer and researcher, using training materials that were customised for the school setting. A total of 27 learning mentors across the seven schools were trained by the research team. The biggest individual school team of learning mentors comprised nine members of staff, whereas the smallest had two members of staff.

The first training session was delivered to learning mentors from all seven schools as a group to raise awareness of the risks associated with young people drinking and to introduce the study. The training included a PowerPoint presentation, group discussion and simulated young-person scenarios. Learning mentors were also trained to issue the participant information leaflet, gather informed consent from the young person, and deliver the alcohol information leaflet (control intervention). This concluded the training for learning mentors from schools allocated to the control intervention arm of the trial who, in order to prevent contamination, received no training on interventions 1 or 2.

Learning mentors at schools randomised to intervention 1 and 2 were then trained to deliver intervention 1. Again, learning mentors were trained together as a group. In addition to a PowerPoint presentation, training consisted of a demonstration of how to deliver the intervention and simulated young-person scenarios. Learning mentors randomised to intervention 2 were asked to return for a further half-day training session. For intervention 2, training sessions were delivered per school site, as a training date could not be identified which accommodated all learning mentors. This session trained learning mentors in how to gather informed consent from parents for the intervention and organise and facilitate intervention 2, as well as how to respond to difficult disclosures.

The final training session focused on delivery of the 12-month follow-up appointment. Training consisted of a PowerPoint presentation and a demonstration of how to deliver each of the measures included in the follow-up assessment. All learning mentors received the same training; however, training sessions were delivered per school site, as a training date could not be identified which accommodated learning mentors from all participating schools. A manual relevant to the 12-month follow-up appointment was provided to supplement the training (see Appendix 3).

Learning mentors were supported in the delivery of interventions and follow-up appointments by the research team, who organised weekly visits throughout the study period to answer questions or concerns, collect materials from completed interventions (such as consent forms and hard copies of intervention tools) and encourage learning mentors to complete outstanding interventions. The research team also provided telephone and e-mail support. Finally, learning mentors were provided with a case diary sheet, on which they were asked to record any interactions with the individual young people in the trial (see Appendix 3). A main resource-use component of the economic evaluation (for the larger definitive trial) will be the cost of learning mentor time required to prepare for and conduct interventions and follow up with the young people during the trial. Thus, time spent for the present study was calculated by observing the average minutes per case, as documented in self-completed case diaries. The appropriateness of the case diary tool for collecting these data was assessed according to rates of missing data (incomplete or wholly unused diaries) and of diaries missing relevant information (see Chapter 7).

Key points for Chapter 4

• Seven schools took part in a set of three cross-sectional surveys administered at baseline (TP1), 6 months (TP2) and 12 months (TP3). Surveys administered at TP1 facilitated screening for the pilot trial.

• Six per cent (n = 87) of parents indicated that they did not wish their child to take part in the study by completing and returning a tick-box opt-out form.

• Survey response rates among pupils whose parents allowed them to take part were 92% at baseline (TP1), 90% at 6 months (TP2) and 84% at 12 months (TP3).

• Levels of missing data were low for all variables.

• A comparison of the distributions of AUDIT and AUDIT-C scores between subgroups at TP1 demonstrated that gender, smoking and sexual behaviour were significantly associated with young people’s current drinking behaviour.

• Mean AUDIT scores were higher for young people who did not leave their names on the questionnaire than for those who did.

• Comparison of scores over three time points suggests there was little or no change in measures of alcohol use, alcohol-related problems and well-being within this age group over the course of a year, except for small but statistically significant shifts upwards in the distributions of AUDIT, AUDIT-C and A-SAQ scores between the first and second surveys.

This chapter reports the methods and results of a set of three cross-sectional school surveys, administered at baseline (TP1), at 6 months (TP2) and at 12 months (TP3).

Methods

Results

Recruitment and retention

The local education authority provided accurate pupil numbers for Year 10 for the seven schools participating in the study. There were 1388 young people across all seven schools that could feasibly complete the survey at TP1. On the days that the surveys were to be completed there were differing numbers of young people absent from school, making the final numbers of completed surveys 1280 at TP1, 1256 at TP2 and 1161 at TP3 (Figure 2).

FIGURE 2.

Completion of surveys at all three time points (TP1, TP2 and TP3).

Table 1 shows the number and percentage of young people who completed the questionnaire on the prearranged day and the number and percentage followed up in the days following.

TABLE 1 - Questionnaire completion by young people on day of survey (TP1, TP2 and TP3)

Young people missing on day of survey who completed the questionnaire at a later date.

Time point	Initial survey		Collected after initial surveya
	n	%	n	%
1	1178	92.0	102	8.0
2	1139	90.7	117	9.3
3	1028	88.5	133	11.5

Analysis of time point 1 data

Distribution of AUDIT and AUDIT-C scores

Figure 3 shows the distribution of the AUDIT and AUDIT-C scores for the whole sample. Scores had a positively skewed distribution, with 340 (28.4%) and 341 (28.2%) individuals scoring the minimum on AUDIT and AUDIT-C, respectively. Scores were recorded up to the maximum of the scales, with three (0.3%) young people scoring the maximum on AUDIT and 10 (0.8%) on AUDIT-C. The median AUDIT score was ‘2’ and the mode was ‘0’ (never drink); the median for AUDIT-C was ‘2’ with a mode of ‘0’. The figures also illustrate the differing proportions of young people who would be categorised as ‘positive’ using suggested cut-off values for adults (8+) and young people (2+).

FIGURE 3.

Distribution of the AUDIT score with (a) young person (score of ‘2+’) and adult (score of ‘8+’) cut-offs; and (b) AUDIT-C cut-offs (score of ‘5+’).

Twenty-six per cent (n = 307) of the sample had an AUDIT score of ≥ 8 (the cut-off used for determining alcohol use disorders in adults)162 and 58% (n = 691) scored ≥ 2 (the modified cut-off suggested for adolescents). 50 In addition, albeit using a breakdown designed for use with adults, 28% scored ‘0’ and could be categorised as ‘abstainers’; 46% ‘lower risk’ (1–7); 18% ‘increasing risk’ (8–15); 4% ‘higher risk’ (16–19); and 4% ‘possible dependence’ (20+) on the full AUDIT. Twenty per cent (n = 245) of the sample screened positive for hazardous or harmful drinking using a cut-off of ‘5’ on AUDIT-C.

Differences in AUDIT and AUDIT-C scores by gender

There was a difference in the distribution of AUDIT scores at TP1 by gender, with girls having a tendency to have higher scores (a median score of ‘3’ in girls as opposed to ‘2’ in boys). A similar shift in distributions was also seen for AUDIT-C scores, for which the median scores were ‘2’ in girls as opposed to ‘1’ in boys. This is illustrated in Figure 4. Mann–Whitney U-tests confirmed that these were statistically significant differences (p < 0.0001 and p = 0.0005, respectively).

FIGURE 4.

Distribution of AUDIT and AUDIT-C scores by gender. Box plots of (a) AUDIT score by gender; and (b) AUDIT-C score by gender.

Differences in AUDIT-C score by smoking status

There was a marked difference in the distribution of AUDIT scores by smoking status, with those who smoked having a tendency to have higher scores (a median score of nine in smokers compared with one in non-smokers). A similar shift in distributions was also seen for AUDIT-C scores, for which the median scores were ‘4’ in smokers as opposed to ‘1’ in non-smokers. This is illustrated in Figure 5. Mann–Whitney tests confirmed that these were statistically significant differences (p < 0.0001 for both tests).

FIGURE 5.

Distribution of AUDIT and AUDIT-C scores by smoking status. Box plots of (a) AUDIT score by smoking status; and (b) AUDIT-C score by smoking status.

Differences in AUDIT score by sexual behaviour

There was a marked difference in the distribution of AUDIT scores by sexual behaviour (measured here by use of condoms), with a tendency to have higher scores as the sexual behaviour was more risky. Young people were asked if they had ever engaged in sex without a condom after drinking alcohol, and the median score was ‘16’ in those who had not used a condom after alcohol; ‘5’ in those who had engaged in sex with a condom; and ‘1’ in those who had never had sex at all. Note that those who had never had sex will include some young people who had not ever drunk alcohol. A similar shift in distributions was also seen for AUDIT-C scores, for which the median scores were ‘6’ in those who had not used a condom, ‘3’ in those who had engaged in sex with a condom, and ‘1’ in those who never had sex. This is illustrated in Figure 6. Kruskal–Wallis tests confirmed that these were statistically significant differences (p = 0.0001 for both tests).

FIGURE 6.

AUDIT and AUDIT-C scores by condom use. Box plots of (a) AUDIT score by condom use; and (b) AUDIT-C score by condom use.

Distribution of Rutgers Alcohol Problems Inventory and Warwick–Edinburgh Mental Well-being Scale scores

The RAPI score was calculated on only those who had drunk alcohol and had a positively skewed distribution with a median of ‘2’ (n = 877). Six hundred and two (50%) individuals scored ‘0’ and three (0.3%) scored the maximum of ‘69’. The WEMWBS score was calculated for all young people who completed the measure and had a median of ‘48’ (n = 1123). This is comparative to other studies with young people aged 13–16 years (median ‘49’). 161 Twelve young people scored the minimum of ‘14’ (1.1%) and 21 (1.9%) the maximum of ‘70’. The distributions of these two variables are shown in Figure 7.

FIGURE 7.

Distribution of RAPI and WEMWBS scores. Histograms of the distribution of (a) the RAPI score excluding those who never drink alcohol; and (b) the WEMWBS.

Associations between measures at time point 1

The strength of association between AUDIT, AUDIT-C, RAPI and WEMWBS was assessed using Spearman’s correlation coefficients and is shown in Table 4 and Figure 8. Unsurprisingly, there was a strong correlation between AUDIT and AUDIT-C scores. The RAPI score showed a moderate association with both AUDIT and AUDIT-C score: this is illustrated in Figure 8. However, the WEMWBS score showed very weak correlations with all of the other measures.

FIGURE 8.

Associations between AUDIT and AUDIT-C and RAPI.

Results of comparison of surveys at different time points

The distributions of the numeric variables across the three time points were summarised in Table 1. There appeared to be a slight shift upwards in some variables, as seen in changes to medians or quartiles. However, the distributions of AUDIT, AUDIT-C, RAPI and WEMWBS scores were formally compared across the three time points using the Kruskal–Wallis test. There was no significant difference between RAPI scores and WEMWBS scores over time (p> 0.05). However, there were significant differences over time for both AUDIT and AUDIT-C scores, with both following the same pattern when data at pairs of time points were compared using Mann–Whitney U-tests. The median AUDIT scores were ‘2’, ‘4’ and ‘4’ at the three time points, and the consecutive median AUDIT-C scores were ‘2’, ‘3’ and ‘3’. The differences between TP1 and TP2 were statistically significant; however, there were no significant differences between TP2 and TP3. These results show a small shift upwards in the distributions of AUDIT and AUDIT-C over the 12-month period. UK guidelines recommend at least 1 hour per day of exercise for young people aged 5 to 25 years. 167 The median number of days in the previous week in the present study in which there was at least an hour of physical activity was four days at TP1 and TP2, and three days at TP3. The median number of days including physical exercise that were reported in a typical week was four days at all time points. The recommended daily intake of fruit and vegetables per day is five (www.nhs.uk/Livewell/5ADAY/Pages/5ADAYhome.aspx). The median number of daily portions of fruit consumed by young people in the present study was two at all time points, and the median daily portions of vegetables was also two at all time points.

The distributions of the categorical variables across the three time points were summarised in Table 3. For the alcohol-related categorical variables reported in Table 3a, the percentage of participants scoring ‘2’ or above (the cut-off suggested for adolescents) increased from 58% at TP1 to 66% (TP2) and 69% at TP3. The percentage of participants who scored ≥ 8 (the adult cut-off) was 26% at TP1, 29% at TP2 and 32% at TP3. For AUDIT-C, 21% of participants scored ≥ 5 (the adult cut-off) at TP1, 29% at TP2 and 32.6% at TP3.

The distribution of alcohol frequency categories measured by the A-SAQ appears to have shifted slightly upwards over time. Thirty-nine per cent of participants reported drinking at least four times but not every month (i.e. scored positive) with 47% at TP2 and TP3. As A-SAQ is an ordered categorical variable, the distribution over time was also compared using Kruskal–Wallis and Mann–Whitney tests. As with the pattern seen in AUDIT and AUDIT-C, the differences between TP1 and TP2 were statistically significant, as were the differences between TP1 and TP3; however, there were no significant differences between TP2 and TP3.

There was a slight decline in the number of young people who had never had sex over time and there was an increase over time of young people regretting sex after consuming alcohol and having sex without using a condom after consuming alcohol.

For the non-alcohol-related categorical variables reported in Table 3b, there was a slight increase in the number of smokers over the year. The majority of young people spent their free time going out with friends, although this reduced from 61% and 62% at TP1 and TP2, respectively, to 54% at TP3. After this, the next most popular answer was to spend time with friends in their own home or their friends’ homes. Spending time with brothers and sisters or with family were the least popular options.

Comparison of named and anonymous participants

The percentage of participants leaving their name on the TP1 questionnaire varied considerably between schools, from as few as 37% up to 74%. The percentage of participants who left their names and scored positive on the A-SAQ varied from 11% to 34% across schools (as a percentage of the total participants completing the survey). A further 21.3% of participants scored positive on the A-SAQ but did not leave their names, so were potentially eligible but not willing to participate. Combining results across the participating schools, the mean AUDIT score of those young people who left their names was lower than those who did not. Using a Mann–Whitney U-test on combined data across schools, there was a statistically significant difference between the distributions of AUDIT scores in those young people who did and did not (p = 0.0002), with a tendency towards higher scores in those who did not leave their name (means of 5.7 vs. 4.4). The results are summarised in Table 5.

Missing data within measures

Table 6 shows a breakdown of missing data for the items making up a questionnaire score. When an item was missing from a measure, an overall score was not computed for that measure. With the exception of the second item on the AUDIT scale, there seems little sign that items are problematic in terms of being missing more often. For the AUDIT score, the second item had a high number of missing values, because there was no tick box for young people who do not drink. This was accounted for when calculating the overall AUDIT score by automatically giving these young people an AUDIT score of ‘0’ if they had responded that they had not drunk alcohol in the last 6 months using the first question in the AUDIT scale (the lowest category). For the other AUDIT questions, missing data values ranged from 0.3% to 3.5%. There were slightly more missing data for RAPI questions, with the percentage of missing values ranging from 1.4% to 4.6%. WEMWBS has the most missing data of all of the measures, ranging from 2.8% to 9%. Overall, there were fewer missing data at TP3 and the most missing data at TP2. The response rate to the whole survey was lower at TP3, so the lower percentage missing on individual items at TP3 may reflect the fact that pupils present on the day of the survey were more likely to complete more items.

Summary

The survey response rates among pupils whose parents allowed them to take part were 92% at baseline (TP1), 90% at 6 months (TP2) and 84% at 12 months (TP3). Levels of missing data were low for all variables. The highest rate of missing data was seen for WEMWBS, which was the last set of questions in the survey pack. A comparison of the distributions of AUDIT and AUDIT-C scores between subgroups at TP1 demonstrated that gender, smoking and sexual behaviour were significantly associated with young people’s current drinking behaviour. The comparisons of scores over three time points suggests that there was little or no change in measures of alcohol use, alcohol-related problems and well-being within this age group over the course of a year, except for small but statistically significant shifts upwards in the distributions of AUDIT, AUDIT-C and A-SAQ between the first and second surveys. In every school, mean AUDIT scores were higher for young people who did not leave their names on the questionnaire than for those who did.

Key points for Chapter 5

• Seven schools were randomised to the three trial arms – control (n = two), intervention 1 (n = two), intervention 2 (n = three) – and retained at 12-month follow-up.

• Sixteen per cent of young people who completed the survey at TP1 met eligibility criteria for the trial; 80% of those eligible were recruited into the trial.

• Eighty per cent of those recruited into the trial completed the 12-month follow-up.

• Of the 75 young people recruited to intervention 2, only eight (10%) received both the individual and family-centred interventions: the remainder received only the individual-level intervention (intervention 1).

• There were very low levels of missing data at both baseline and 12-month follow-up.

• The TLFB was completed with all young people who attended at 12-month follow-up. There was some evidence that results on AUDIT, AUDIT-C and A-SAQ scales showed a slight shift to less alcohol consumption or risk behaviours at 12 months compared with baseline.

The external pilot trial was a parallel-group, three-arm cRCT with randomisation at the level of schools. A cluster randomised design was chosen to reduce the potential for bias due to contamination between young people allocated to different arms within the same school. The three arms were control, intervention 1 and intervention 2 (details of interventions are given in Chapter 3). The primary aim of the pilot trial was to assess feasibility and acceptability to plan a future definitive trial, including estimating rates of eligibility, consent, participation and retention at 12 months.

Process and measures: baseline

The questionnaire distributed at TP1 provided young people with the opportunity to volunteer their contact information or to complete the questionnaire anonymously. This TP1 questionnaire facilitated screening for the trial, and young people who screened positive for risky alcohol use using the A-SAQ and who provided their name at TP1 were invited to attend an appointment with a learning mentor to assess eligibility and provide consent. Young people were excluded from participation if they were already seeking help for an alcohol use disorder (AUD), receiving support from child and adolescent mental health services or had not been given consent by parents to take part.

Process and measures: 12-month follow-up

Twelve-month follow-up appointments with trial participants took place when young people had begun the next school year (Year 11). Collection of follow-up data began in January 2013 and was completed in April 2013. Trial participants who had moved schools during this time and were unable to be contacted for this appointment were lost to the trial. No trial participants had language or literacy problems that required additional support with reading the documentation. The session involved completion of three separate questionnaires: A-SAQ, AUDIT and 28-day TLFB, chosen to measure different aspects of drinking behaviour and anticipated to be primary or secondary outcome measures in a future definitive trial. All three measures were completed during a single one-to-one appointment with a learning mentor, which took place during school time. The order of presentation of A-SAQ and AUDIT were randomised and completed by the young person alone, with the TLFB being the last tool completed with the learning mentor. Wherever possible, the same learning mentor conducted both intervention and follow-up sessions. However, owing to staffing changes at participating schools, on some occasions a different learning mentor conducted the follow-up appointment. If this learning mentor was a newly recruited member of school staff then he/she was provided with a condensed training session, focusing on the intervention delivery phase of the study, in addition to training in how to deliver the 12-month follow-up appointment. The planned primary outcome for a future definitive trial is the 28-day TLFB questionnaire, completed by trial participants at 12-month follow-up. The TLFB has been validated for use in this population168–170 and involves a retrospective interview administered by the learning mentor to ascertain actual alcohol consumed over the 28-day period prior to the interview. Four alcohol consumption measures were derived from the 28-day TLFB: total alcohol units consumed in a 28-day period, percentage of days abstinent, mean number of drinks per drinking day, and number of days on which alcohol consumption was more than two units. The questionnaire invites participants to recall their daily alcohol consumption over the 28-day period and can examine total alcohol consumption as well as patterns of alcohol consumption (see Appendix 3). It is important that sufficient information is recorded to calculate accurately the units of alcohol consumed, including the type (and brand) of alcohol and the volume (or size of container) of alcohol consumed. To facilitate collection of data, learning mentors were provided with prepared copies of the 28-day TLFB questionnaire. Tools were marked with dates (such as Christmas, examination periods and local football games) in order to provide prompts and aid form completion. Other memorable dates specific to the young person were identified and used to aid recollection.

Design

Pilot trial sample size

As this was a pilot trial, a formal power calculation was not required. However, providing data to design a future definitive trial is an important function of a pilot study. A minimum number of 30 participants per intervention group at follow-up has been recommended to estimate key parameters for this purpose. 171 We used data from previous studies to estimate the proportions of young people who would be eligible, consent to enter the study and provide data at 12-month follow-up. 97,127 Our estimates suggest that the minimum number of 30 per arm providing follow-up data would be achieved if all pupils in Year 10 across seven schools were invited to take part (Figure 9). Note that we estimated that recruitment would be much lower for the intervention 2 arm, so two schools were randomised to each of the control and intervention 1 arms, but three schools were randomised to the intervention 2 arm.

FIGURE 9.

Estimates of eligibility, recruitment and retention used to plan the sample size of the pilot trial. The estimates are based on previous studies: a, 22% follow-up;127 b, 79% follow-up;126 c, 88% conservative estimate of take-up rate taken from Walton et al. 97 and Conrod et al. ;127 d, 65% follow-up rate. 172

Methods

Statistical analysis

The eligibility, recruitment and retention rates for the schools and young people have been summarised in a Consolidated Standards of Reporting Trials (CONSORT) diagram (Figure 10). The data collected for trial participants at TP1 and TP3 were summarised with descriptive statistics by trial arm, and combined across trial arms. This was to investigate suitability of scales and variables for a future definitive study, to establish baseline characteristics, and to summarise the outcomes. The percentage of missing and implausible values was reported for all variables, along with either a five-number summary (minimum, lower quartile, median, upper quartile, maximum) for numeric variables, or numbers and percentages in each category for categorical variables. In addition, at TP3, the variables derived from the 28-day TLFB and the AUDIT and AUDIT-C measures were summarised by their mean and standard deviation (SD): these were used to inform a sample size calculation for a definitive trial, and the comparison of mean and median values allowed consideration of the shape of each distribution. All analyses used the intention-to-treat populations.

FIGURE 10.

Trial CONSORT diagram. LM, learning mentor.

Results

Characteristics of trial participants at baseline (time point 1)

Results of outcome measures at 12-month follow-up (time point 3)

The four outcome measures derived from 28-day TLFB plus the results of A-SAQ, AUDIT and AUDIT-C collected at 12-month follow-up (TP3) are reported in Table 10. The five-number summaries show that there is a lot of variation within the groups. Across all the trial participants the range for the units of alcohol consumed in the 28-day period was 0–235 units, with a median of 10.3, a mean of 22.7, and a large SD of 36.3. There were occasional participants who reported consuming very high total amounts of alcohol; however, staff were trained in how to use the TLFB to maximise the validity of the answers. Typical levels are less well balanced between trial arms, with median levels of around eight in the control and intervention 2 arm and 14 in the intervention 1 arm.

Percentage days abstinent (from TLFB) had similar distributions across the trial arms (median per arm = 93%), as did days consuming more than two units (median per arm = 1 or 2). For drinks per drinking day, there was also some variability between the trial arms (medians 7.8, 7.6 and 5.8). Note that this variable cannot be calculated for those participants who do not consume any alcohol in the 28-day period.

There was a wide range observed in AUDIT scores at TP3 (0–28), but little variation between trial arms (medians 5–6). A similar pattern was seen in AUDIT-C scores, for which the trial arm median scores were ‘4–5’. For AUDIT and AUDIT-C we have data at TP1 and TP3. The median AUDIT score across all arms was lower at TP3 (‘5’) than at TP1 (‘8’), although there is no change in medians for AUDIT-C (‘4’) between TP1 and TP3. No formal comparisons were carried out, so any changes must be interpreted with caution. Some young people were lost to follow-up because of complex behavioural problems, repeated absence, moving school, or deciding that they no longer wished to participate in the trial. The median AUDIT score at TP1 of the 22 participants who dropped out by TP3 was ‘14.5’ and their median AUDIT-C score was ‘5.5’. Given some of the reasons for loss to follow-up, it is perhaps not surprising that typical AUDIT scores for young people retained at TP3 were lower than at baseline for all of those entering the trial. However, when only those participants who provided AUDIT scores at both TP1 and TP3 are included in the analysis, there does appear to be a slight tendency towards a reduction in AUDIT scores (although this was not seen for AUDIT-C scores). This is illustrated in Figure 11 where the distribution of individual changes in AUDIT scores is shown across trial arms. There was considerable variation in the change scores (AUDIT score at TP1 – AUDIT score at TP3) indicating both increases and decreases over the year. However, although the median change is zero for the control and intervention 2 trial arms, it can be seen that, in all arms, the positive changes (indicating lower AUDIT score at TP3) tend to be larger than the negative ones. However, any difference between trial arms must be interpreted with caution, as they are based on data from only two or three clusters.

FIGURE 11.

Distribution of individual change in AUDIT scores between TP1 and TP3 by trial arm.

We also had the distribution of the A-SAQ question available at TP1 and TP3. The summary statistics in Table 10 show that the distribution was shifted towards less frequent consumption at TP3 than at TP1. All had reported drinking four or more times in the last 6 months at TP1, but by TP3, 28% across all arms reported less frequent drinking than this. The percentage of young people reporting drinking every week had also approximately halved in all arms.

The reduction in alcohol use over 12 months in the AUDIT and A-SAQ measure was observed in all trial arms. This may have reflected a general change in drinking behaviour or socially desirable responses due to taking part in the trial, rather than a response to a particular intervention.

At TP3, levels of missing data were very low. For the TLFB measures (units of alcohol consumed in a 28-day period, percentage of days when abstinent, and days consuming more than two units), only the intervention 2 arm had any missing data. This amounted to 12% of missing data for units of alcohol consumed and days consuming more than two units, and 6% for percentage days abstinent. The higher numbers of missing values for drinks per drinking day is due to some of the young people not drinking at all during the 28-day period, and therefore not having a value for this. This applied to 10 young people in the control group (23%), eight in intervention 1 (16%) and 26 in intervention 2 (39%). For the other outcome measures at TP3, there were missing data again only for AUDIT and AUDIT-C in the intervention 2 arm – 6% and 1.5%, respectively. One young person failed to complete the A-SAQ in the control group.

Fidelity of the interventions

Fidelity of an intervention within research refers to the extent to which the intervention is true to the therapeutic principles on which it is based. 176 It requires the manualisation of the intervention wherein the philosophy, principles and procedures of the intervention are clearly described. This manual can then be used by the individuals delivering the intervention in order to deliver a standardised approach. 177 Moreover, a manualised intervention with verified fidelity enables the research to be replicated or the intervention to be implemented in practice.

Learning mentors were asked to record at least one session each; however, only six recordings of intervention delivery were made. In this study the BECCI was used to measure fidelity. BECCI is a tool developed specifically to measure the microskills of behaviour change counselling and MI. 178 The instrument focuses upon the practitioner’s consulting behaviour and attitude rather than the patient’s response. A qualified member of the team (RM) rated intervention 1 and intervention 2 audio-recordings. Rating was completed in line with the BECCI Manual for Coding Behaviour Change. 179 The mean BECCI score for the six recorded interventions was ‘2.5’, which suggested that the learning mentors were all found to be delivering behaviour change counselling to ‘some extent’ or to ‘a good deal’ as assessed with the BECCI. The median BECCI score was ‘2.55’, with the range 1.9–3.0 (individual scores were 1.9, 2.1, 2.3, 2.8, 2.9 and 3.0). Learning mentors typically performed well when discussing the risks associated with the young person’s alcohol use. Lower scores were assessed when measuring microskills relating to discussing and exploring behaviour change. Future training of mentors in intervention delivery should focus upon discussing behaviour change with young people.

The small number of interventions that were recorded is a weakness that would need addressing in a full trial. In the feasibility study, learning mentors, randomised to either group other than the control, were approached and asked to record a minimum of one recording. A more formal approach to fidelity measurement is required. In a definitive study, learning mentors will be asked to record a simulated intervention with an actor immediately following training but before commencing the trial. Further training can then be provided on areas of practice weakness. A specific date will then be agreed for a further recording of intervention delivery with a trial participant.

Summary

The required number of schools (seven) were recruited into the feasibility pilot trial and retained at 12-month follow-up. Ninety-two per cent of young people in Year 10 (aged 14–15 years) across the seven schools completed the survey used to screen for inclusion into the trial. Sixteen per cent of those completing the survey met eligibility criteria and 80% of those eligible were recruited into the trial. Eighty-eight per cent of those recruited into the trial completed the 12-month follow-up. However, of the 75 recruited into the intervention 2 arm, only eight (10%) received both individual and family-centred interventions: the remainder received only the individual-level intervention (intervention 1). The trial arms were not well balanced on all variables at baseline, but this is not surprising for a cluster randomised trial with very few clusters and heterogeneity between clusters. There were very low levels of missing data on each score or variable at both baseline and 12-month follow-up. Furthermore the interview held to complete the 28-day TLFB was successfully achieved in all who attended the 12-month follow-up meeting with a learning mentor. Finally, there was some evidence that results on AUDIT, AUDIT-C and A-SAQ scales showed a slight shift to less alcohol consumption or risk behaviours at 12 months compared with baseline.

Key points for Chapter 6

• Semistructured interviews were conducted with six lead liaisons, 13 learning mentors, 27 young people and seven parents (n = 53).

• Overall, the school was considered to be a feasible and acceptable environment to intervene with young people who are risky drinkers.

• Learning mentors were seen as being best placed to discuss alcohol with young people owing to their role within the school, their existing supportive relationships with young people and the trust that young people placed in them.

• The screening survey was found to be feasible, although in future work some consideration may need to be given to means of enhancing young people’s privacy in order to increase acceptability.

• Intervention 1 was found to be feasible and mostly acceptable. Some learning mentors expressed hesitation at informing young people for whom their drinking placed them at risk of harm and the calorie-focused content resulted in mixed views from both learning mentors and young people.

• Intervention 2 did not appear to be feasible. Learning mentors, parents and young people questioned the utility of an intervention that they believed was not engaging the ‘right’ people. Although parents who did engage in intervention 2 found the intervention to be acceptable, most young people and their parents who were offered did not express a desire to take part in this intervention or a benefit from doing so, and some young people who were interviewed told us that they did not want their parents involved.

For the integrated qualitative evaluation of the study, semistructured interviews were conducted with four key groups of participants: school lead liaisons; learning mentors; young people; and parents. This chapter begins with a description of the methods used in the conduct and analysis of these interviews, continues with a summary of the key findings, and concludes with a discussion of the overarching emergent themes from the qualitative phase of the study, alongside the limitations of the work.

Methods

Semistructured interviews were selected as the primary mode of qualitative data collection in order to inform a more in-depth understanding of the overarching research questions for the study. The aims for all sets of interviews were to explore the feasibility and acceptability of screening and BI approaches in the school setting, and to elicit participants’ views on the study measures and processes used in delivering the project. Key topics for interviews with young people and their parents included consent procedures; parental involvement in interventions; the comprehensibility and burden of study measures and follow-up procedures; and the appropriateness of school-led health promotion work across the school–home interface. All interviews were conducted between May and August 2012. Each participant was interviewed once and interviews were timed to take place as soon as possible after their involvement in study procedures had ended. Interviews with lead liaisons, learning mentors and young people were performed and analysed by KL. Interviews with parents were performed and analysed by SS.

Sampling strategy

As this was a qualitative study, the aim of sampling was to achieve data saturation and maximum variation of perspectives. School lead liaisons were defined as the seven key individuals who made or brokered the decision about participation in the study on behalf of their school. As there were only seven lead liaisons involved, all were approached for interview. Purposive sampling was undertaken within the remaining three participant groups to ensure maximum variation within the study population. For learning mentor interviews, defined as the members of school staff trained in the delivery of the control condition/interventions to participating students, sampling criteria were according to socioeconomic positioning of the school in which the learning mentor was used, and study condition.

For young people, sampling criteria were gender, socioeconomic status (SES) of school and the level of intervention received. SES and gender were considered important for this group because these factors are known to be related to drinking behaviour in this age group. 180 In addition, young people within the intervention 2 arm of the study were purposively sampled to include both those who agreed to family involvement and those who refused. This sampling frame resulted in 16 subgroups of young people to represent in interview.

For parent interviews, sampling criteria were SES of school and whether or not intervention 2 had successfully taken place. The latter criterion was included for two main reasons. First, there were a number of parents whom learning mentors had been unable to contact, using a range of different methods, to take part in a family intervention despite numerous attempts. Second, there were occasions when the parent agreed to take part in the intervention but the intervention did not take place, because either the parents or the young person changed their mind at a later date. It was felt that both of these groups of parents could give a useful insight into the complexities and dynamics of parental involvement in this form of intervention.

Recruitment and consent

A range of approaches was used in order to recruit interview participants into the study. Lead liaisons were approached directly by the researcher (KL) and learning mentors were, in turn, approached by their line manager (when not the school lead liaison) to ask if they would agree to be interviewed. Learning mentors acted as gatekeepers for interviews with young people, making the initial contact with the young person concerned, and setting up interview appointments on behalf of the researcher (KL). At each approach it was stressed that participation was entirely voluntary.

Learning mentors also helped facilitate access to parents for interview purposes. First, they contacted parents directly to confirm whether it was acceptable for a researcher (SS) to contact them about participation in an interview. If a parent had declined the family intervention (or learning mentors had struggled to contact them to arrange it), school staff attempted to contact them again (by phone, text message, e-mail and letter) to explain the purpose of interviews and to ask if their contact details could be given to the researcher. Parents were reassured that if they declined they would not be contacted about the study again.

Alternatively, if parents could not be contacted using these channels, learning mentors asked young people to invite parents for an interview and sent a message home with the young person, who was asked to provide contact information after checking with their parents that it was acceptable to pass this information on. It must be stressed that if young people withdrew their own consent for approaching parents then parents were not contacted to participate in an interview. Following an initial positive approach by the young person concerned, the researcher (SS) subsequently contacted parents, using a variety of methods (telephone, text message, e-mail and/or letter) to arrange interviews.

Interviews with lead liaisons, learning mentors and young people were all performed within their respective school setting. Interviews with parents took place at a time and place most convenient to the participant concerned. In practice, interviews were generally community based, comprising a mix of home interviews and interviews that took place in public locations, such as local coffee shops. One interview took place in the interviewee’s place of work.

Informed consent was taken at the beginning of interviews, after ensuring that the interviewee had read the Participant Information Sheet and been given an opportunity to discuss any questions or concerns with the interviewer. Interviews lasted between 20 and 90 minutes and were all digitally recorded, with the resultant data transcribed verbatim by professional transcribers. All interviewees were allocated a participant reference code to ensure anonymity and confidentiality, and an anonymisation log was maintained.

Interviews with participants

Table 11 summarises the number of interviews by interview group, according to school and study condition. Six of the seven lead liaisons were interviewed as part of the qualitative evaluation of the study. The remaining lead liaison was on maternity leave during this period and could not be interviewed during the study time frame. Thirteen participating learning mentors were interviewed. The majority of participating learning mentors were female and this dynamic is reflected in interview participants (male = 2, female = 11).

In total, 27 young people were interviewed as part of this research (male = 12, female = 15). Every attempt was made to ensure that the sampling frame was saturated (i.e. at least one respondent arising from each cell). However, it should be acknowledged that the potential pool of young people who had agreed to intervention 2 was extremely limited, thus, in reality, all participants were approached for interview. In particular, there were no high SES males who agreed to intervention 2 and so it was not possible to interview a young person from within this category.

Finally, semistructured interviews were conducted with seven parents, all of whom were mothers. Three (of seven) schools were randomised to the intervention 2 arm of the study, with family members from these three schools invited to take part in an interview. Although it was initially anticipated that a mix of mothers, fathers and other nominated family members such as grandparents would participate in the intervention 2 family intervention, in practice, with the exception of one family intervention with a father, only mothers took part. One father (who said no to the family intervention) agreed to take part in an interview but later changed his mind upon the researcher’s arrival. This parent appeared to be very uncomfortable with the prospect of taking part in an interview, stating to the researcher several times that his child did not drink and did not have a problem with alcohol. Six interviewees had taken part in a family intervention; one interviewee had not. This was because the young person did not want to give up their free time to take part and changed his/her mind – not because the parent said no. Although we set out to interview parents who did not participate in an intervention, owing to the small number of parents recruited to the parental component of intervention 2, interviews with parents proved the most challenging to arrange and there were clear barriers to participation. Nevertheless, despite interviewing only one non-participating parent, this account provided a rich and comprehensive insight into the complexities and dynamics of parental involvement. Further, the accounts of lead liaisons, learning mentors and young people were also instrumental in developing our understanding of parental involvement.

On completion of the 53 interviews it was deemed by the research team that data saturation had been reached: this was determined as the point at which no new themes were emerging from the interviews.

Data analysis

The interview data were analysed thematically,181 with the Framework approach, devised by Ritchie and Lewis,182,183 utilised to organise the analysis. The Framework approach, which is a structured organisation of themes, ensured that the analysis could be easily viewed and assessed by others in the research team. 184 Coding of transcripts was performed by the researchers who had conducted the interviews (KL for lead liaison, learning mentor and young person interviews; SS for parent interviews). A computer software program, such as NVivo, was not used during data analysis, as the research team felt that use of a program that ‘cuts’ the data into smaller chunks would inhibit us from looking at the data in its totality, risk information being taken out of its original context and potentially lead to ‘over coding’, through which a deeper level of interpretation is lost. 185 Instead, coding was performed by hand, using paper copies of transcripts. Later, resulting frameworks of codes were recorded in table format in a spreadsheet document. Each participant was listed as a column, and each code, and related subcode, listed as a row. When a participant discussed a code, the page and line number reference was placed in the relevant cell of the table. This enabled effective organisation, storage and retrieval of coded data. Each group of interviews was analysed separately from each other. Regular meetings were held with members of the research team with expertise in qualitative techniques to discuss and challenge emergent themes and exchange analytical thoughts. This is referred to as pragmatic double coding by Barbour. 186 The aim of these meetings was not to value one point of view over another, rather they aimed to ‘maximise the analytic potential of exceptions or potential alternative explanations’ (p. 1026). 186

Findings

Summary

It would seem that the school is both a feasible and an acceptable environment to intervene with young people who are risky drinkers. Learning mentors in particular are well placed to discuss alcohol with young people owing to their role within the school, their existing supportive relationships with young people and the trust that the young people place in them. Although it is acknowledged that the delivery of the interventions can be time-consuming, there was the sense that the activity remains feasible. The training provided to learning mentors was considered to fully prepare them for their role within the study. Importantly, acceptability of intervention delivery was high; intervening with young drinkers was often seen as important and necessary aspects of the learning mentors’ work.

Overall, the screening survey was found to be feasible, although in future work some consideration may need to be given to means of enhancing the young people’s privacy in order to increase acceptability. Teachers were often present, overseeing the class while the young people completed the screening survey. These teachers had not been trained in best-practice approaches to this research method, however, and had received only minimal information regarding the purpose of the survey. Delivering training to teachers regarding informed consent and the importance of enhancing and maintaining confidentiality is likely to improve the overall acceptability of the screening survey.

Intervention 1 was found to be feasible and mostly acceptable. Some learning mentors expressed hesitation at informing young people for whom their drinking placed them at risk of harm, choosing instead to advise the young people who had been selected at random. This is suggestive of an outstanding training need for the learning mentors. As such, future work should ensure that the training programme emphasises the importance of personalised feedback within the delivery of interventions. The calorie-focused content also resulted in mixed views from both young people and learning mentors. As this information is not central to the information, it is recommended that this is not included in an intervention within a definitive trial.

It would appear that intervention 2 is not feasible. Parents and young people did not express a desire to engage in this intervention or a benefit from doing so. Moreover, learning mentors, parents and young people questioned the utility of an intervention which they believed was not engaging the ‘right’ people. Although the parents who did engage in intervention 2 found the intervention to be acceptable, it should be noted that most young people and their parents who were offered did not participate in this intervention. Some young people interviewed told us that they did not want their parents involved. Although we did not interview any parents who chose not to participate in intervention 2, quantitative data presented elsewhere in this report reinforce the findings of the qualitative study that intervention 2 is not feasible, as well as suggesting that it is not acceptable to a large group of young people and parents. Furthermore, by not including an intervention that involves parents in future work, the time-consuming task of contacting parents, arranging appointments and rearranging appointments that are not attended would be alleviated, thus enabling learning mentors to use their time more efficiently.

Key points for Chapter 7

• The collection of data using the open-ended case diary tool highlighted a number of problems. A structured case diary tool would both be more precise and provide more reliable data while also reducing the data collection burden on the learning mentors in a definitive trial.

• Percentages of missing data for service use questions from the three survey time points do not seem to be problematic, suggesting that the tool is acceptable for use with young people in a definitive trial.

• However, some thought should be given to how we measure service use, especially in relation to certain categories (i.e. GP visits).

• It appears that the EQ-5D is an appropriate tool to use with young people. The majority of young people indicated that they had no problems on the first three dimensions of the EQ-5D-Y (mobility 93%; looking after self 99%; doing usual activities 94%).

• Higher levels of problems were found in the dimensions of pain or discomfort (19% having some level of problems) and being worried, sad or unhappy (24% having some level of problem). This indicates that there is some opportunity for the definitive trial to improve health, at least in terms of the final two dimensions.

This chapter presents findings from the health economics component of the study that aimed to rehearse the methods of data collection to inform the development of the economic evaluation in a definitive study. The definitive health-economic analyses will show how the costs of introducing and running the BI compare with the current practice; the reason for this is that a full economic evaluation should include current practice as a comparator, as it seeks to inform decisions about whether we should move from current practice to something else. 187 The analyses reported in this section will be used to produce the protocol for a definitive trial and attendant economic evaluation of the impact of brief alcohol intervention compared with standard practice (PSHE) in a school setting to reduce alcohol-related risk or harm.

This chapter focuses on examining what resource-use data we should collect and how these will be analysed. The focus is on the key elements of an economic evaluation, which are costs and consequences, which will be discussed below. The level of completeness of the data has been analysed and the suitability of tools is commented on accordingly. In each of the following sections the results of our analysis are presented with associated discussion and recommendations.

Sections of analysis

Costs Resources and costs required to provide the intervention.

Outcomes/consequences Health-economic outcomes of the intervention including NHS and public services resource-use and health-related quality of life (as measured by EQ-5D-Y).

The health-economic outcomes are based on the participant-completed questionnaires, specifically questions 14 (Resource Use) and 15 (EQ-5D-Y), administered as part of the non-randomised repeat cross-sectional survey. These data were collected at the three survey time points: TP1, TP2 and TP3. Questions 14 and 15 were not separately identifiable for the subgroup of survey participants at TP2 or TP3 when they were followed up within the trial. This pilot trial intended to test only the alcohol-related outcomes at 12 months for the trial participants. Therefore, the data we have available is for the entire survey cohort at these three time points, which is appropriate from a health-economic perspective, as our objectives were met and no economic evaluation was planned in this feasibility study.

Costs

Analysis of resources use and costs associated with both intervention (intervention 1 and intervention 2) and control groups relate to two specific areas: the resources required to provide the intervention; and the resources used subsequently after the intervention (or control). The details of such costs and resources are discussed below.

Learning mentor training time

Training for learning mentors will be provided by SIPS JR-HIGH research staff on site at all locations. The time to deliver the training, per location, will be documented by the SIPS JR-HIGH research staff in hours and minutes (i.e. 2 hours 15 minutes). A list of learning mentors in attendance will be recorded so that the training time (cost) per learning mentor can be incorporated into the cost of running the intervention. The learning mentors are not to record this training time in the intervention time diaries under the ‘Prep’ category, as that category is referring to intervention casework, not the training time. The SIPS JR-HIGH research staff delivering the training will record their time and pay grade so that the cost data can be incorporated into the analysis. In the definitive trial, the SIPS JR-HIGH research staff will keep a record of all time spent on training follow-up which is specifically related to the training of learning mentors. The methods to calculate these costs are discussed in the next section.

Resource-use and unit-cost information

To assess the full cost of the intervention to inform the definitive economic evaluation, both resource-use and unit-cost data will be collected and reported in tables similar to Tables 16 and 17. These tables illustrate an example for a single area of resource use and its associated unit cost. For the definitive trial, full tables will be populated with all applicable measures of intervention resource use and unit costs will be reported. For this feasibility study, the resource-use data and unit-cost sources are not reported as the definitive trial will go beyond the local authority school district.

In the definitive trial, once the tables have been populated with all relevant resource-use categories and corresponding unit costs, the two tables will be used together to calculate the total cost of running the intervention. All cost outputs will be reported in UK pounds sterling for the final financial year of the definitive study. For example resource-use data (i.e. 45 minutes of learning mentor time) is multiplied by the unit-cost data from Table 17 (i.e. cost per minute of a learning mentor) to calculate the monetary cost of that particular resource use. The data are an estimation of salary and on-costs (based on local authority data) to show the process that will be taken for the costing portion of a future full economic evaluation. For a definitive trial, these data will be sourced by school district centrally through their learning platforms. Cost per minute of staff time will be derived using the formula illustrated in Table 17, which shows that the average resource use for the learning mentor training portion of the intervention would cost £9.45 per learning mentor (45 minutes × £0.21).

Outcomes

Summary

In relation to collecting case diary data of the time spent by learning mentors on working on interventions, the open-ended format of the case diary proved to have many limitations; however, in order to identify the categories needed in a definitive trial this was important and has enabled us to identify the categories needed for the definitive trial tool (see Appendix 6). The revised tool should be piloted with a few learning mentors prior to beginning the definitive trial to ascertain whether it is ‘user-friendly’.

The majority of young people indicated that they had no problems on the first three dimensions of the EQ-5D-Y (mobility 93%, looking after self 99%, doing usual activities 94%). Higher levels of problems were found in the dimensions of pain or discomfort (19% having some level of problems) and being worried, sad or unhappy (24% having some level of problem). This indicates that there is some opportunity for the definitive trial to improve health, at least in terms of the final two dimensions. Results of this study show that the questions needed for health-economic analysis are acceptable for use with young people; however, some thought should be given to how we measure service use, especially in relation to certain categories (i.e. GP visits).

Objective 1

• The study succeeded in recruiting seven schools as planned. Part of this success was due to gaining the support and active involvement of the local authority in the study catchment area from the outset. The local authority provided the research team with written confirmation it was happy for the study to proceed in its geographical area, and schools were informed that the project was supported by the local authority.

• A range of factors influenced school participation in the study: the project presented direct benefits to participating schools in terms of boosting alcohol education provision through additional staff training and the provision of enhanced support for participating students in need.

• The screening and consent procedure produced sufficient young people to rehearse the trial procedures.

Objective 2

• Interviews were carried out with six school lead liaisons; 13 learning mentors; 27 young people and seven parents (total n = 53).

• School setting Qualitative interviews were specifically focused on feasibility and acceptability of the intervention and not on the wider engagement of parents in a school setting. Views from school staff were mixed regarding engagement of parents in the school setting, and appeared to reflect the focus of the school. Therefore, schools that were part of the ‘Extended Schools Agenda’ were more likely to describe school as a ‘hub’ of the local community (and felt that they regularly engaged with parents) than traditional academically focused schools. Although parents felt that school was the correct environment for an intervention aimed at young people’s alcohol use, they were unsure about their own involvement in school-based alcohol education, and suggested that they did not know whether their children would take them seriously if they were involved regularly in formal alcohol education, or whether other young people would always be open and honest in front of their parents. It would seem that the school is both a feasible and an acceptable environment to intervene with young people who are risky drinkers.

• Learning mentors Learning mentors in particular are well placed to discuss alcohol with young people due to their role within the school, their existing supportive relationships with young people and the trust that the young people place in them. Learning mentors were seen as appropriate members of staff to carry out the interventions by staff, parents and young people.

• Training The study showed that it was possible to train learning mentors in the research requirements (consent/intervention delivery); the length and content of training was seen as appropriate by learning mentors; learning mentors particularly liked the training manuals with which they were provided.

• Screening Overall, the screening survey was found to be feasible as has been found in the literature (see Chapter 2). Teachers were often present, overseeing the class while the young people completed the screening survey. These teachers had not been trained in best-practice approaches to this research method, however, and had received only minimal information regarding the purpose of the survey. Delivering training to teachers regarding informed consent and the importance of enhancing and maintaining confidentiality is likely to improve the overall acceptability of the screening survey. In the definitive study, consideration should be given to means of enhancing the young people’s privacy in order to increase acceptability. Study instructions for the young people should be made clearer on the front of the questionnaire at baseline. A standardised set of instructions should be provided for each class, perhaps as a video clip produced by the research team. We believe that these changes would improve the numbers of young people leaving their names on the questionnaires.

• Intervention 1 was found to be feasible and mostly acceptable. There was some hesitation among learning mentors around informing young people whose drinking placed them at risk. The calorie-focused content also resulted in mixed views from both young people and learning mentors. In the definitive study further emphasis will be placed upon the importance of personalised feedback within the delivery of interventions. All learning mentors randomised to the intervention arm will be audio-recorded while delivering the intervention within a simulated session with an actor (see Objective 3, Fidelity) and further training will be provided to learning mentors who continue to find this aspect of the intervention challenging. As learning mentors (and young people) expressed mixed views about the calorie-focused content of the intervention, this will be removed from the intervention in the definitive trial.

• Intervention 2 was not feasible to deliver within this study. Parents and young people did not express a desire to engage in this intervention or a benefit doing so, which has been shown in previous studies (see Chapter 2). Findings demonstrated that existing knowledge about young people’s drinking was the primary factor influencing parent participation in intervention 2. Thus, if parents did not know about their drinking, young people were far less inclined to consent to a family intervention. Although parents are a source of both risk and protective factors for adolescent alcohol use, as highlighted in our rapid review, evidence that interventions for alcohol involving parents are viable is equivocal.

• Moreover, learning mentors, parents and young people questioned the utility of an intervention, which they believed was not engaging the ‘right’ people. Although the parents who did engage in intervention 2 found the intervention to be acceptable, it should be noted that most invited young people and their parents did not participate in this intervention. Some young people interviewed told us that they did not want their parents involved. Furthermore, as shown in Chapter 2 and Appendix 2, the literature around parental involvement is equivocal, with no clear indication that involving parents in interventions to reduce their children’s drinking is effective. This suggests that the definitive trial should focus on working with young people rather than involving parents. 83

Objective 3

• Fidelity In this study the BECCI index was used to measure the fidelity of the delivery of interventions by the learning mentors. 178 This tool is used to measure the microskills of behaviour change counselling. As such, it focuses upon the practitioner. It is not able to measure the young people’s responses to the intervention or consider characteristics or compositions of the groups receiving the interventions. Six interventions were assessed. The mean score was ‘2.5’, with a range of 1.9–3.0, which suggested that the learning mentors were all found to be delivering the behaviour change counselling aspect of the intervention to ‘some extent’ or to ‘a good deal’, as assessed with the BECCI. The rate of recorded interventions was lower than was anticipated. We acknowledge the lack of detail regarding fidelity assessment and the low number of interventions assessed. Sessions that were assessed showed that learning mentors performed well when discussing the risks associated with young people drinking alcohol. Learning mentors performed less well when discussing motivation for behaviour change and strategies for behaviour change. The suggestion for a definitive trial is to include a minimum of one simulated intervention with an actor immediately after training for all learning mentors who are randomised to the intervention arm. A specific date to be agreed with each learning mentor for a further recording of intervention delivery with a trial participant.

Objective 4

• Six per cent (n = 87) of parents opted their child out of participating in the study. Discussions with young people and parents on the days of the survey indicate that many of these parents thought they were opting their children into the study, which implies that the letter was confusing. Ninety-two per cent (1280/1388) of Year 10 year groups completed the baseline survey, and of these students 18% met the eligibility criteria of reporting drinking at least four times in the last 6 months on the A-SAQ and left their name on the questionnaire, which showed willingness to be contacted later. This eligibility rate of 18% was slightly lower than we anticipated (presumed to be approximately 22%). At baseline, 40% screened positive on the study screening tool (A-SAQ), but only slightly over half of these young people left their name and so were contactable regarding participation in the pilot trial. Although young people who did not leave their names were drinking more, it is important to note that there was a considerable number who were drinking at risky levels who did leave their names. In the definitive study, instructions should be made simpler and clearer on the letter that goes to parents, with one tick box indicating opt-out with a clear instruction that the young person will be opted out only if the box is ticked and the letter signed. The A-SAQ should be used as the screen for coming into the definitive trial, as it is short and quick to answer, with the AUDIT being asked at both baseline and 12-month follow-up.

• Survey We found very low rates of missing data for virtually all variables. The highest rate of incomplete data (10%) was on the WEMWBS well-being questionnaire. This was the last set of questions in the survey pack, and it is possible that lack of time or fatigue led to more missing values. There was little evidence of implausible values being recorded, except for a few young people saying that they exercised on more than 7 days per week. There were a few very high values reported for alcohol use and problems but these could not be regarded as implausible. For the definitive study, consideration should be given to reducing the number of questions in the survey instrument.

• Survey At TP1, 50% of the sample were male and 94% were white. The prevalence of smoking rose from 20% at TP1 to 25% at TP2 and reduced to 23% at TP3. The median number of days that young people reported physical exercise was four at all three time points. The median number of daily portions of fruit and vegetables was two each per day at all three time points.

• Survey The proportion of young people who reported drinking alcohol fewer than four times in the last 6 months (A-SAQ) was 39% at TP1, 47% at TP2 and 47% at TP3. The proportion of young people who scored positive for an alcohol-use disorder using the AUDIT adult cut-off of 8+ rose from 26% at TP1 to 29% at TP2 to 32% at TP3. Using a cut-off of 2+, recommended for young people, this rate rose from 58% at TP1 to 66% at TP2 to 69% at TP3. The differences in all measures between TP1 and TP2 was significantly different but not between TP2 and TP3. Between the first two surveys, the median scores for AUDIT increased by two points, whereas AUDIT-C increased by one point, but there was no change in median scores between the second and third surveys. This highlights the differences in using different tools and cut-offs for identifying young people who are risky drinkers; however, all measurements show high levels of risky drinking at all three time points. The TLFB is a more robust measurement of alcohol consumption; however, it is more time-consuming to administer therefore for the definitive study the 28-day TLFB should be used as the primary outcome measure at 12-month follow-up.

• Survey The WEMWBS measures general psychological health, with a scoring range of 14–70, with a higher score indicating higher levels of mental well-being. At TP1 the median score for the WEMWBS was ‘48’, which is comparative to other studies with young people (median 49). 161 The RAPI was calculated only for those who reported drinking. At TP1 the median score was ‘2’. RAPI showed a moderate association with alcohol (using AUDIT 0.76 and AUDIT-C 0.65), whereas WEMWBS showed a very weak correlation (using AUDIT –0.13 and AUDIT-C –0.08).

• Trial The comparison between subgroups at baseline demonstrated that gender, smoking and sexual behaviour were significantly associated with young people’s current drinking behaviour, using the AUDIT and AUDIT-C.

• Trial Learning mentors recruited 80% of those young people who were eligible for the pilot trial. This recruitment rate matched that which we had anticipated (approximately 79%). Very few young people did not consent to the study (10%). However, 10% failed to meet with the learning mentor to discuss the trial for a number of reasons, including repeated absence, school exclusion and the existence of complex behavioural needs. This could be seen as a form of voluntary or involuntary withdrawal from the study and would need to be taken account of in a future study.

• Control Of the 60 young people eligible for the trial, three did not meet with the learning mentor (5%) and five did not give consent (8%). In total, 52 out of 60 were recruited to the trial (87%).

• Intervention 1 Of the 79 young people eligible for the trial, 15 did not meet with the learning mentor (19%) and 10 did not give consent (13%). In total, 54 out of 79 were recruited to the trial (68%). Therefore, both the control and condition 1 arms were found to be feasible.

• Intervention 2 Recruitment of young people to the intervention 2 arm was higher than expected. Of the 90 young people eligible for the trial, seven did not meet with the learning mentor (8%) and eight did not give consent to intervention 1 (9%). In total, 75 out of 90 were recruited to the trial and received intervention 1 (83%). However, having agreed to enter the trial, many of the young people in the intervention 2 arm did not receive the full intervention as planned. Of the 75 students recruited into this arm, 25 of these students agreed to their parents being contacted (33%); however, only eight (11% of the 75 and 32% of the 25) received both the individual intervention (intervention 1) and family intervention (intervention 2). There is more work needed to engage with parents in interventions in the school setting. Despite the input of lots of time and resources from the school and research staff, it was not, however, possible to engage parents in the third arm of the trial, reflecting experiences in other studies. 192

• 12-month follow-up Once enrolled in the trial, 88% of trial participants provided data at the 12-month follow-up meeting with the learning mentor (control, 83%; intervention 1, 91%; intervention 2, 89%). This was a higher rate than we had anticipated (65%) and it reflects well on the efforts of the trial team, learning mentors and school processes. The pilot trial has achieved the goal of demonstrating that outcome measures could successfully be collected on a high proportion of participants.

Objective 5

• There were very low levels of missing data in the baseline survey or the EQ-5D-Y (3.4–3.9%), with the tool being seen as appropriate. The majority of young people indicated that they had no problems on the first three dimensions (mobility 93%, looking after self 99%, doing usual activities 94%). Higher levels of problems were found in the dimensions of pain or discomfort (19% having some level of problems) and being worried, sad or unhappy (24% having some level of problem). This indicates that there is some opportunity for the definitive trial to improve health, at least in terms of the final two dimensions. For the definitive study the EQ-5D-Y and service use should be assessed at baseline and 12-month follow-up. Implausible values, in relation to service use, should be reassessed, especially in the case of visits to the GP, which showed a higher-than-average percentage of ‘implausible’ values, and different implausible levels could be given for different service use.

• In relation to service use, there was between 4.2% and 4.8% of answers missing at baseline. The majority of young people reported no use of services. The only possible exception was ‘GP visit’. Implausible data (values of seven or more) were found in 3% of all answers at baseline.

• The use of open-format diaries meant that differing levels of data were reported by learning mentors, especially in relation to preparation time. In the definitive study, case diaries should be made more concise and time categories should be provided. Time should be reported by ticking boxes of preselected times, informed by the summary statistics regarding the times reported in the feasibility study. Different forms will be needed for each arm of the trial. This will enable accurate data to inform an economic evaluation. These forms should be piloted with a small group of learning mentors to establish face validity.

Objective 6

For the definitive trial, we propose a four-region, two-arm, cRCT (randomisation at school level), with integrated economic and process evaluations. This would enable generalisable results and take into account geographical, ethnic and socioeconomic differences, as well as reflecting differences in organisation of education services. The literature shows that ASBI with young people is effective and the results of this present study show that it is feasible and acceptable to intervene with young people aged 14–15 years in the school setting. The intervention with parental involvement was found not to be feasible or acceptable. The hypothesis for the definitive trial would be that ASBI is more effective and cost-effective at reducing hazardous drinking in young people (aged 14–15 years) than a control condition of screening, feedback that the young person may be drinking at a risky level and an information leaflet, as well as usual advice in Year 10 of high/comprehensive schools in England. This research will have a broader impact on both the target community (young people) and wider society in reducing health and social harms and inequalities. Primary and secondary outcome measures will be the same measures used in the pilot feasibility trial.

• Screening tool A lifestyle survey, as used in the present study, which includes questions relating to risky drinking.

• Regions North East England, North West England, Kent and South London.

• Primary outcome measure Reduction in alcohol use using the 28-day TLFB questionnaire193 at 12-month follow up.

• Secondary outcomes Risky drinking using the A-SAQ2 and AUDIT;162 smoking behaviour; alcohol-related problems using the RAPI;163 emotional well-being using the WEMWBS;159 and quality of life and health utility will be measured using the EQ-5D-Y. 165 A modified S-SUQ will capture health and social resource costs for the integrated economic evaluation. 166 Learning mentor time will be assessed using a revised case diary sheet (see Appendix 6).

• Proposed design The multicentre, two-armed cRCT, incorporates a control and intervention condition. Schools will be paid £1000 for taking part in the research study for the time involved. Young people will not be given a £5.00 gift voucher, as in the pilot study, for two reasons: (1) it would be very costly and (2) this would not happen if the study was mainstreamed.

• Screening All pupils in Year 10 (aged 14–15 years) in each of the schools, whose parents have not opted them out of the study, will be asked to complete a voluntary questionnaire that will contain a number of tools including the primary and secondary outcome measure tools. All young people who screen positive and leave their name will be asked to consent to the trial by the learning mentor.

• Control condition Standard alcohol advice delivered in PSHE lessons delivered by class teachers, feedback to the young person that they are drinking in a way that may be harmful, and provision of an advice leaflet by the learning mentor.

• Intervention 1 In addition to PSHE, the young people who are eligible (risky drinkers) and consent to participate will be given feedback that they are drinking in a way that may be harmful and provided with an advice leaflet. They will then take part in a 30-minute personalised interactive worksheet-based session, developed during the pilot feasibility trial. This will be delivered by the learning mentor (at school) and consist of structured feedback about their drinking behaviour and advice about the health and social consequences of continued hazardous alcohol consumption. The intervention encompasses the elements of the FRAMES approach for eliciting behaviour change (Feedback, Responsibility, Advice, Menu, Empathy and Self-efficacy). 59

• 12-month follow-up All young people who come into the trial will be invited to meet with the learning mentor 12 months post intervention, during which they will be asked to complete the same battery of questionnaires used at baseline, as well as the 28-day TLFB.

• Training All learning mentors will receive school-based training in the study procedures and the intervention that is relevant to their school. Learning mentors will be brought together at one of the schools in each geographical area for this training. Such outreach training was found to be the most cost-effective implementation strategy for ASBI delivery in the pilot2 and in other settings. 156 Intervention training for learning mentors will be carried out by an experienced trainer. Learning mentors will be provided with support materials and will be assessed as competent by the trainer prior to embarking on the study. Changes to the training and manual will take into account learning from the pilot feasibility trial. Ongoing support and supervision will be provided by clinical staff working on the project.

• Fidelity We will carry out a minimum of one audio-recorded intervention delivered per learning mentor within the intervention arm of the trial.

• Setting High/comprehensive schools are governed by the local authorities in England. Screening will take place in the PSHE or registration class on a classroom basis. Interventions will take place in the learning mentor’s classroom or office space. This will be the anticipated setting for roll-out if the project is implemented.

• Patient and public involvement participation PPI has been imperative to the success of the pilot feasibility trial and this will be continued in the main trial with involvement from young people and parents; however, we acknowledge that more in-depth PPI work is needed in the definitive trial. We intend to set up a management group to steer the research in each of the schools that take part in the study, which will include teaching staff/learning mentors and young people. Views from these groups will feed into the PMG on a regular basis.

• Qualitative work Semistructured in-depth interviews will be carried out in each of the schools with staff and young people. The interviews will further explore factors that potentially hinder or enhance the use of ASBI approaches in the school setting and with the target age group, with the aim of exploring future roll-out of such work.

• Sample size As a two-arm trial, 100 responses would be needed per arm with individual randomisation and a significance level of 5%. We intend to use minimisation to balance out both school size and percentage free school meals. Using other trial parameters as above, this would equate to 220 young people per arm, and a total of 18 schools (nine per arm) when clustering is taken into account.

Study governance references

Contributions of authors

All authors read and agreed the final draft report.

Dorothy Newbury-Birch (Lecturer in Public Health Research, project management) was the Chief Investigator on the study and took overall responsibility for the study and the writing of the report.

Stephanie Scott (née O’Neil) (Research Associate and SIPS JR-HIGH Project Manager) project managed the study and co-drafted the report.

Amy O’Donnell (Research Assistant, qualitative research) co-analysed and co-drafted the qualitative chapter of the final report.

Simon Coulton (Professor of Health Services Research, health services research) contributed to the design of the study, conduct of the trial and the PMG.

Denise Howel (Senior Lecturer, statistics) contributed to the design of the study, conduct of the trial and PMG, supervised the statistical component of the research, carried out statistical analysis, and co-drafted the survey and trial chapters.

Elaine McColl (Professor of Health Services Research and Director of the Newcastle Clinical Trials Unit, methodology) contributed to the design of the study, conduct of the trial and PMG.

Elaine Stamp (Research Associate, statistics) carried out the statistical analysis of the study and co-drafted the survey and trial chapters.

Erin Graybill, (Research Assistant, health economics) conducted the health-economic analysis and drafted the health-economic chapter.

Eilish Gilvarry (Honorary Professor of Addictions Psychiatry, Consultant Psychiatrist in Addictions, BIs and addictions) contributed to the study design, co-developed the interventions, co-drafted the intervention chapter of the report and chaired the PMG.

Kirsty Laing (Research Assistant, qualitative research) carried out the interviews with school staff and young people and analysed the data.

Ruth McGovern, (Senior Research Interventionist, interventionist and qualitative researcher) co-developed the interventions, assessed intervention fidelity and co-drafted the intervention and qualitative chapters of the report.

Paolo Deluca (Senior Research Fellow, trial management) contributed to the design of the study, conduct of the trial and the PMG.

Colin Drummond (Professor of Addiction Psychiatry, addictions) contributed to the design of the study, the conduct of the trial and the PMG.

Christine Harle (Trial Manager, Newcastle Clinical Trials Unit, trial management) contributed to the design of the study, the conduct of the trial and the PMG.

Paul McArdle (Consultant Child and Adolescent Psychiatrist) contributed to the design of the study, conduct of the trial and PMG, co-developed the interventions and co-drafted the intervention chapter of the report.

Les Tate (Strategy and Commissioning Manager, education) contributed to the design of the study, conduct of the trial and PMG and co-ordinated school involvement in the study.

Eileen Kaner (Professor of Public Health Research and Institute Director, BIs, trial management) contributed to the design of the study, the conduct of the trial and the PMG.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the PHR programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the PHR programme or the Department of Health.

Publication

O’Neil S, Coulton S, Deluca P, Deverill M, Drummond C, Gilvarry E, et al. Brief intervention to prevent hazardous drinking in young people aged 14–15 in a high school setting (SIPS JR-HIGH): study protocol for a randomized controlled trial. Trials 2012;13:166. http://dx.doi.org/10.1186/1745-6215-13-166.

A feasibility trial of screening and brief alcohol interventions to prevent hazardous drinking in young people aged 14–15 in a high school setting (SIPS JR-HIGH)

A feasibility trial of screening and brief alcohol interventions to prevent hazardous drinking in young people aged 14–15 in a high school setting (SIPS JR-HIGH)

Parent opt-out letter

(PDF download)

Study information leaflet

(PDF download)

Screening questionnaire

Screening questionanire (PDF download)

Health and well-being leaflet

(PDF download)

Study manual

Information sheet for interviews

(PDF download)