Modifying Alcohol Consumption to Reduce Obesity (MACRO): development and feasibility trial of a complex community-based intervention for men

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 12/139/12. The contractual start date was in May 2014. The draft report began editorial review in February 2016 and was accepted for publication in November 2016. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

John Norrie is a member of the National Institute for Health Research (NIHR) Health Technology Assessment (HTA) Commissioning Board and the NIHR HTA and the Efficacy and Mechanism Evaluation Editorial Board. Linda Irvine was the Trial Manager on the NIHR Public Health Research funded study 11/3050/30 [Texting to Reduce Alcohol Misuse (TRAM): a multi-centre randomised controlled trial of a text message intervention to reduce binge drinking among disadvantaged men] while the current study was being conducted.

Disclaimers

This report contains transcripts of interviews conducted in the course of the research and contains language that may offend some readers.

Copyright statement

© Queen’s Printer and Controller of HMSO 2017. This work was produced by Crombie 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.

Background

Heavy alcohol consumption by men who are obese increases their risk of death from liver disease 19-fold. 1 Having a raised body mass index (BMI) and consuming > 14 units of alcohol per week are both factors that independently increase the risk of liver disease. However, the combination of the two leads to a substantially greater risk. 1 The commissioning brief described this as the supra-additive effect of obesity and alcohol. The problem is compounded by the high prevalence of heavy drinking and of obesity. The English Health Survey 20122 found that 24% of men aged ≥16 years drank more than the then government-recommended 21 units per week. Data from 2011 revealed that 24% of men in England3 and 28% of men in Scotland4 were obese. Based on current trends, the UK government’s Foresight programme predicted an increase in the prevalence of obesity among men. 5 Furthermore, heavy alcohol consumption is associated with an increased risk of obesity. 6–8 Abdominal obesity is more common among men who drink beer. 9 Heavy drinking leads to overeating,10 thus increasing the potential for weight gain.

An intervention that could reduce both alcohol consumption and weight in obese men could make a significant contribution to improving public health. However, it is far from certain that obese men who drink heavily would engage in such an intervention. They may not regard their drinking as problematic and may not think it could contribute to weight gain. Many obese men want to lose weight, but are reluctant to try conventional weight loss approaches, because these are generally targeted at women. 11 Men are under-represented in weight loss interventions12 and attrition is high in weight loss studies. 13,14

Previous research has identified many features of an intervention that could make it attractive to potential male participants. Convenience and compatibility with a busy life are important,15–17 as is the use of community settings for intervention delivery. 18 The style and tone should be relaxed and friendly as opposed to strict and regimented. 16,17 The content should be personalised and accessible,15,18 but should be delivered in a matter-of-fact way. 18,19 The aim of the intervention should also be realistic and achievable. 17 The recently completed Football Fans in Training study20,21 provides an excellent example of an engaging community-based intervention that effectively changed behaviour. This previous research clarifies many of the requirements of an intervention for obese men. The present study aims to identify the specific content that would be required to change their drinking behaviour and to assess the feasibility of a trial to test the effectiveness of the intervention.

Study overview

This 21-month study was conducted in two phases. Phase 1 was to develop the recruitment strategy and the novel alcohol intervention. It was to comprise six focus groups with obese men aged 35–64 years, who would be recruited from a variety of community locations. In-depth interviews with six key stakeholders would also be conducted to help develop the strategy for the recruitment of participants.

Phase 2 was intended to recruit 60 men aged 35–64 years. Two complementary recruitment strategies were to be used. One would identify obese men from primary care records. The other would adopt a community outreach approach, time–space sampling (TSS), which was designed to identify hard-to-reach groups. 31 A key feature of the study is that trained laypeople would be involved in intervention delivery. The training programme used would be developed from the Department of Health manual,32 taking into account recent recommendations for training laypeople. 33 Each study co-ordinator would only deliver one type of intervention (either novel intervention or comparator) to prevent contamination across treatment arms.

The men in phase 2 would be randomised to a novel extended intervention group or a comparator group (a conventional alcohol brief intervention). The novel extended intervention would be informed by literature on changing health behaviours including recent systematic reviews34,35 and taxonomies of behaviour change techniques,36,37 including one specific to alcohol brief interventions. 38 The intervention would also draw on techniques developed for our successful NIHR feasibility study of a text message delivered intervention to reduce alcohol-related harm in disadvantaged men (project number 09/3001/09). 39 The comparator group was to receive a conventional brief alcohol intervention,40 which would be delivered in a face-to-face session by a trained layperson.

Modifications to the initial protocol

Three changes were made to the intervention. One part of the intervention was to be a series of text messages encouraging behaviour change. A small number of text messages were to include images, which would extend the ideas conveyed in words. This approach was used successfully in a NIHR Public Health Research (PHR)-funded feasibility study (project number 09/3001/09). 39 However, unanticipated technical difficulties were recently encountered in the NIHR PHR-funded RCT (project number 11/3050/30). 41 Exploration of the problem suggests that changes in the commercial technology used to transmit images (Multimedia Messaging Service images), and variations in the systems used were the cause of the difficulties. Text messages that do not contain images are sent without problems. The proposal to use Short Message Service (SMS)-only messages was approved as a contract variation by NIHR HTA before the intervention was developed.

The second change to the intervention was the omission of a volitional help sheet which participants were to have been given at the end of their face-to-face session. This help sheet would have identified ways in which participants could cope with situations that might weaken their resolve to reduce their alcohol consumption. 42 However, a preliminary assessment of the face-to-face session showed that it would be best used to focus on increasing motivation to drink less. The next phase of the intervention, the text messages, would cover the specific techniques of goal-setting, action planning and relapse recovery. As the volitional help sheet is intended to assist in these steps, similar approaches were incorporated into the text messages.

The final change concerned measuring weight. The protocol stated that men would be asked to monitor and report their weight during the intervention phase. On reflection, the intervention period of 2 months was thought insufficient to achieve substantial change in weight. Worthwhile, but modest, reductions in alcohol consumption would yield small reductions in weight over such a short period. Thus, measuring weight loss might reduce rather that increase motivation to maintain reduced consumption. Consequently, the men were not asked to monitor their weight.

Conclusion

This study seeks to address an important public health problem. The approach that was taken was to develop each component of the feasibility study and then test its suitability. The intention was to create an initial robust design, then identify opportunities to improve it during the conduct of the feasibility study. Based on an evaluation of the success of the study, this report will make a recommendation on whether or not a full RCT is warranted. This will be based on technical feasibility and the likelihood of whether or not a meaningful answer will be supplied.

Structure of the report

Feasibility studies involve the development of study methods accompanied by a careful evaluation of whether or not each method achieves its objectives. This study comprises a series of substudies. For clarity, this report presents each substudy as a separate chapter, together with its methods and findings and a discussion of the implications. Thus, Chapters 2–11 describe these substudies. The final summary and conclusions chapter (see Chapter 12) summarises the findings from the separate chapters and highlights issues that involve a comparison of the findings from the separate chapters. It assesses whether or not the study objectives were met and makes recommendations for improvements to the design of the full RCT.

Introduction

The remit of this study stipulated that weight loss was to be used as a motivator to drink less among men who are obese and who drink heavily. The rationale for this is clear: alcohol is high in calories and drinking increases snacking and the consumption of high-calorie food. 43 However, the use of one behavioural outcome, weight loss, as the motivation to change a separate behaviour, alcohol consumption, is unusual. This approach is not one that has been well covered in the literature, so focus groups were conducted to explore it. Because the logic of the intervention, although compelling, is far from straightforward, the issue was approached indirectly. Initially, participants’ attitudes and motivation to drink less and to lose weight were explored separately. Once ideas on each topic had been reviewed, issues around using reduced alcohol consumption as a means of weight loss were discussed. The aim was to clarify opportunities and strategies for an intervention to promote reduced drinking and to identify potential barriers to engagement with the intervention.

Methods

Six focus groups were conducted with middle-aged men who drank regularly and were overweight. Socioeconomic status was measured using the Scottish Index of Multiple Deprivation (SIMD),44 which is similar to the English Index of Multiple Deprivation. 45 Systematic reviews have shown that the provision of monetary incentives can significantly increase recruitment to research studies. 46,47 Consequently, men who agreed to take part were given a £10 gift voucher and were reimbursed for their travel expenses.

Results

The findings are organised by the three main sections of the topic guide: (1) reducing alcohol consumption, (2) being overweight and losing weight and (3) assessing intervention components.

Discussion

These findings have important implications for the design of the intervention to reduce alcohol consumption in obese men aged 35–64 years. They identify potential levers for behaviour change that could support the study aim of using the benefits of weight loss to promote reduced drinking (Box 1). In addition, barriers to reduced drinking were identified and an approach that was thought to have promise had the potential to be counterproductive.

The men identified many benefits of reducing drinking. Most prominent among these were the short-term consequences of heavy drinking, such as hangovers and loss of mental clarity. Although some longer-term health concerns were identified, they were much less commonly noted by the men. This finding not only identifies an important target for increasing motivation to drink less, but also provides one for maintenance of reduced drinking. Satisfaction with outcome is crucial for the maintenance of a new behaviour. 59,60 Benefits such as improved mental clarity are immediately appreciated, whereas many health benefits, such as avoiding diabetes or heart attack, will not be reaped for many years. Thus, the intervention could use the short-term benefits of reduced drinking to increase motivation, and then use the benefits obtained to promote maintenance.

The principal barrier to reduced drinking was the impact it would have on friends and social life. However, balanced against this was the finding that one benefit of reducing alcohol intake was a positive effect on relationships with family and friends. Thus, in the intervention, increasing the salience of the benefit could reduce or negate the impact of the barrier.

The calorie content of alcohol came as a surprise to many of the men. This makes it a high priority for inclusion in the intervention. The finding that drinking makes you eat more, and more unhealthily, is another important candidate for inclusion. The literature shows that moderate alcohol intake stimulates food consumption in the short term (within 1 hour). 55–57 Men also make unhealthier food choices when drinking. 43 Although this point did not feature on the self-completed cards, it emerged frequently in discussions and attracted no dissenting voices. Given the double benefit of less alcohol and fewer calories, the links between drinking alcohol and increased consumption of food should be emphasised in the intervention.

The way in which alcohol increases food consumption, particularly by snacking, was familiar to many men, although there was a perception that this effect was not widely known. This suggests that, although known, the effect has a low salience. This makes it an attractive target for intervention, as information about it may be readily accepted. One approach could be to ask men about their own experience, and then encourage them to reflect on what effect this has on their weight.

An unexpected finding was the hazard of describing calories in alcohol in terms of common food items. This was intended to be a simple method of conveying how many calories were in alcohol and it succeeded in that aim. However, the unintended consequence was that men could use the avoidance of high-calorie foodstuffs as an excuse to consume more alcohol; therefore, this topic should be avoided in the intervention.

Previous studies have shown that men care about body image61 and engage in negative body talk (making disparaging comments about their bodies). 62 Self-image and physical functioning are important motivators for weight loss. 18 These include the desire to improve appearance and physical attractiveness,63–65 but also to increase self-esteem and a sense of self-control. 63 A key finding from this study is that the men’s body image concerns were focused on the size and unsightliness of the stomach and the embarrassment that this causes. This presents an ideal lever to increase motivation to change, but it is one that needs to be handled sensitively. The implication that a participant is unattractive because of his shape could induce annoyance and rejection of the intervention. One way to overcome this would be to suggest that the men might recapture some of the vigour of their youth if they lost some weight.

The study also found that men were concerned about function, their ability to do everyday tasks, their engagement with their families and their ability to carry out work responsibilities. 66,67 Thus, another motivator for weight loss and/or reduced alcohol consumption found in this study is increasing effectiveness at work to become a greater asset to one’s workplace. 15 The men’s concerns about being able to function competently in society provide an opportunity to increase motivation to change, and have the benefit that, because they are conventional social obligations, they are likely to be readily accepted. Indeed, men may be more likely to welcome suggestions on how to increase their ability to meet these expectations.

In summary, the focus groups have identified opportunities for the development of an intervention to reduce drinking in a social group who have not been well studied: obese men who drink > 21 units of alcohol per week. These centre on improving physical, social and mental well-being by reducing alcohol consumption and losing weight. The proposed approach to weight loss, namely drinking less alcohol, offers benefits that are important to overweight men. The aim is to increase the salience of the benefits to increase motivation for change.

Introduction

One element of the recruitment strategy proposed to use community outreach to identify eligible men. It would focus on venues that obese men were likely to frequent. To identify potential venues and the best ways to gain access to them, interviews were conducted with key stakeholders. These were individuals in managerial positions who frequently encounter men in the target group, and who could thus act as gatekeepers for recruitment of participants. Such individuals were also likely to have insight into the willingness of potential participants to discuss taking part in the study. The stakeholder interviews investigated opportunities for meeting with potential participants and possible venues for delivering the intervention.

Methods

Results

The interviews explored three main themes: (1) views on the research topic, (2) recruitment of participants and (3) venues for intervention delivery. The results are presented in these themes. No details of the stakeholders’ job titles or organisations are included, in order to protect their anonymity.

Discussion and implications

These interviews have identified considerable enthusiasm among stakeholders for the study. Some of the stakeholders saw it as their role to provide access to health promotion for their members/customers/employees. Others did not see it as their role but still thought that weight and alcohol consumption were problems that needed to be addressed. Not only were they willing to provide access to potential participants, many could provide rooms for delivery of the face-to-face component of the intervention. This augurs well for recruitment.

The enthusiasm for the research is more than just altruism and a general willingness to help. It derives from a shared assessment of the scale of the public health problems posed by alcohol and by obesity. Furthermore, many of the organisations that were approached identified improving the health and well-being of their staff or their client group as a core value of their institution. This raises the possibility that, if the intervention proved effective, there could be opportunities for national roll-out through organisations that have similar core values. Naturally, there may be organisational obstacles that impede the progress of roll-out; however, the findings suggest that a possible approach could begin with discussions with senior staff in order to review shared values and the potential benefits arising from the intervention. This could lead to an evaluation of how the intervention could be implemented and an assessment of obstacles that would need to be overcome. A possible attraction for senior staff could be the low cost and time commitment of the intervention.

For the present study, the interviews have identified that although there may be many potential participants at some venues, accessing them may be difficult. Perceived lack of time may be the main issue. The stakeholders thought that the men would not be annoyed by being approached, although this falls some way short of being receptive to the invitation to participate in a research study. Another potential impediment is the preference of stakeholders to act as intermediaries in recruiting their staff. This could cause reluctance among staff to volunteer, as it could imply that they are overweight and drink too much.

In summary, these interviews have identified an unexpected enthusiasm for the research. This may extend to the provision of support for recruitment and of rooms for conducting face-to-face sessions. In addition, if managed sensitively, there may be opportunities for national roll-out of an effective intervention.

Background

The remit for this study included the design of an intervention that used the motivation of weight loss to promote reduced drinking. The study focused on men who were obese (BMI of > 30 kg/m²) and who drank > 21 units of alcohol per week. These men are at a 19-fold increased risk of mortality from liver disease because alcohol and obesity have a supra-additive effect on risk. 1 The rationale for the intervention is that alcohol is high in calories, and therefore reducing alcohol consumption could have a similar effect to eating less, thus reducing both risk factors at once. However, the logic of the health promotion message is complex. It takes the form of: if you take this action (drink less) then, because of this fact (alcohol is high in calories), you will get this benefit (consume fewer calories) and achieve that outcome (lose weight). The challenge is to deliver this logic in a form that is understood and that engages middle-aged men. This may be difficult, as the target group comprises men who are not seeking help, who do not think they have a drinking problem and who may well believe that they drink moderately.

The intervention was planned to involve a face-to-face session at which BMI (height and weight) would be measured and alcohol consumption would be assessed. This session would be delivered by trained laypeople (study co-ordinators). It was to be followed by a series of text messages developed using the techniques established in our previous NIHR funded studies (NIHR PHR 09/3001/09,39 NIHR PHR 11/3050/3041). Together, these components would deliver a complete strategy for behaviour change. Both components of the intervention drew heavily on the elements of effective brief alcohol interventions described in systematic reviews34,68 and a review of reviews. 35

The intervention was to be based on the causal model for behaviour change specified in the protocol, with modifications made in light of the findings from the focus groups. The initial causal model for behaviour change was (1) generate interest in the study, (2) increase awareness of consumption levels that are defined as harmful, (3) identify motivational beliefs, (4) increase awareness of susceptibility to alcohol-related harm for men who are already obese, (5) increase motivation to lose weight by reducing alcohol consumption, (6) alter alcohol expectancies, (7) gain commitment to change, (8) develop goals, action plans and coping plans, (9) increase refusal skills, (10) implement strategies to prevent relapse and (11) reduce total alcohol consumption, which would in turn lead to weight loss.

This chapter describes the rationale for the activities, discussions and behaviour change techniques that were incorporated into the face-to-face session and the subsequent series of text messages. Initially, several questions were identified:

• How should the causal model be modified by the focus group findings?

• How should the intervention be divided between the face-to-face session and the text messages?

• To what extent should the content of the text messages reinforce the face-to-face session?

Separate issues for the face-to-face session were how best to use the measurements of BMI and alcohol consumption in the intervention and what constraints would be imposed by the use of laypeople (study co-ordinators) to deliver the intervention. A further concern was how to design the intervention to overcome potential barriers to engagement and behaviour change.

Implications of the focus group analyses

The focus group analyses identified several opportunities to increase motivation to drink less. These included feedback of data on alcohol consumption, increasing the perceived relevance and importance of the benefits of reducing alcohol consumption, highlighting the benefits of weight loss, and feedback of data on calories consumed through alcohol. Feedback of data on alcohol consumption is a common component of alcohol brief interventions. 34,68 The focus groups confirmed that men in the target group (i.e. men who are overweight and regularly consume alcohol) are often surprised by how much alcohol they drink. As few men monitored their alcohol consumption, there could be value in encouraging this to provide motivation to drink less and to sustain reduced consumption. However, men also felt that the potentially detrimental effect on their social life would make them reluctant to drink less. This issue will need to be handled sensitively in the intervention. One approach would be to draw attention to the many social benefits of drinking less that the focus groups identified, such as improved relationships with family and friends.

The focus groups also confirmed that being overweight is recognised as a major problem, because of its adverse impact on self-image and on ability to carry out everyday activities. This is consistent with the findings of a recent NIHR HTA systematic review. 18 However, the focus groups also showed that the key issue was the so-called ‘beer belly’, as this caused embarrassment and made physical activities difficult. Engaging men with the benefits of losing this unsightly protuberance is a specific target for the intervention that could increase motivation to lose weight. Given that men volunteered no disadvantages of losing weight, suggestions of easily implemented techniques to lose this ‘beer belly’ should be well received.

A key finding from the focus groups was that men were aware that alcohol was associated with increased snacking. This effect of alcohol is well established in the literature,55–57 but it was unclear whether or not the men in this study would know about it. The finding that, when prompted, men agreed that they do eat more snacks when drinking could be used to increase motivation for reducing alcohol consumption. Thus, men would experience the dual benefits of reduced calorie consumption from both alcohol and food.

The focus groups confirmed that measuring alcohol consumption could provide useful data for behaviour change. The surprise expressed by men at their consumption could lead to a discussion of the benefits of changing behaviour. The measurements could easily be arranged in a sequence following the logic of the intervention. Initially, feedback on alcohol consumption could be used to develop a discussion of the benefits of drinking less, increasing motivation to reduce consumption. BMI could then be measured, leading to a discussion of the disadvantages of being overweight thereby increasing motivation to lose weight. Finally, the calories contained in alcohol could be calculated, personalising this by using the participant’s alcohol consumption. This could be supported by reviewing the ways in which alcohol stimulates appetite and leads to increased consumption of energy dense foods. Overall, the approach dissects the logic of the behaviour change strategy, illustrating each component with personal data and building simple steps into the complex argument. Crucially, motivation for action on each component should be gained before the next component is addressed.

The focus groups also provided approximate timings for making and discussing the measurements of alcohol consumption (10 minutes), calculating units in alcohol (5 minutes) and calories in alcohol (5 minutes).

The face-to-face session

The literature and the focus group findings identified ways in which measurements of alcohol consumption and BMI could be used to increase motivation to lose weight. As these measurements would be made in the face-to-face session, the most advantage would be gained by exploiting them as the data were obtained. Whether or not this session could also cover volitional components such as goal-setting and action planning would depend on the time available. The session was designed in a pragmatic way by exploring what could consistently be achieved for all participants.

The session had to be brief to be acceptable to participants who perceive that they have limited free time. 17 In the absence of direct information on what constituted brief, the arbitrary decision was taken to work initially with a 30-minute duration, although the exact length could subsequently be modified. Within this time, height and weight had to be measured, and the TLFB58 and baseline questionnaire had to be completed.

The other components of acceptability are the style and content of the session. Previous intervention studies have found that men appreciate a friendly, relaxed and non-directive approach20 and being given easily understood information. 18 The use of laypeople, who are likely to have an informal rather than professional style, could help achieve this.

A starting assumption was that participants will vary in their readiness to reduce their alcohol consumption. Thus, the intervention has to be appropriate for men who initially have no intention to change as well as those who are currently thinking about drinking less. The alternative would be to design a series of interventions tailored to individual motivation to change. This would create considerable logistical difficulties, as participants would have to be screened and stratified prior to intervention delivery and laypeople would have to be trained to deliver different versions of the intervention. The principle of having an intervention that could easily be rolled out nationally also militates against this approach.

Another constraint was that the face-to-face session component has to be delivered by laypeople. Although training would be given, there is a limit to the demands that can be placed on laypeople. The intention was to use some techniques from motivational interviewing (MI), but full training in MI would not be possible. The question then became how much training would they need to be able to deliver what could be covered in a reasonable time in the session. This was answered by reviewing how the alcohol and BMI measurements would be made and how they would be exploited.

The text messages

The process of writing the text messages

The development of the text messages followed the four stages shown in Figure 1: (1) establishing the provisional structure, (2) drafting the text messages, (3) review and revision and (4) piloting and final revision. It began with an exposition of the behaviour change strategy, the construction of a narrative to depict the process of behaviour change and the creation of characters to give life to the narrative. This established a provisional structure within which the text messages could be written. In practice, there was much iteration in the writing process, with modifications made to the narrative, the characters and the texts as the structure and content of individual text messages was developed.

FIGURE 1.

Stages in the writing of the text messages.

Delivery of the interventions

The intervention was designed to be delivered by a face-to-face session followed by a series of text messages. The process of recruiting and training laypeople to deliver the face-to-face session is described in Chapter 6.

Conclusions

A comprehensive and coherent behaviour change intervention was successfully developed. This required an explication of the theoretical processes of behaviour change, combined with a careful review of the opportunities and constraints of the study design and the modes of intervention delivery. The intervention followed the remit set by NIHR HTA, which provided a powerful underlying logic to behaviour change. The use of text messages enabled reinforcement of components of the face-to-face session.

The description of the intervention follows the Template for Intervention Description and Replication (TIDieR) checklist and guide items for the reporting of interventions. 91 Two items from the TIDieR checklist that are not presented here are covered in other chapters; see Chapter 6 for the training of the intervention provider and Chapters 9–11 for fidelity of delivery of the intervention.

The creation of the intervention was a lengthy process that required input from several members of the research team. The key to the design of the face-to-face session was the measurement of BMI (height and weight) and alcohol consumption. These will be the primary outcomes for the full RCT, so a function of the feasibility study was to ensure that participants were happy for the measurements to be taken and that both BMI and alcohol consumption could be accurately measured. They also provided personal data that could be fed back to participants, a key feature of many behaviour change interventions. 36,37 This provided the opportunity to develop discussions to increase motivation to change.

Text messages can offer several advantages. Modelling of behaviour change techniques by characters offers a potentially powerful method of reducing alcohol consumption. Direct quotations from focus groups provide concise, authentic phrases that convey lived experiences. It would be difficult to obtain this verisimilitude unless a highly skilled professional writer were employed. A longer narrative over 3–4 months might have been preferable. This would allow more development of the narrative and would provide additional time to deal more fully with relapse, recovery and maintenance of the new behaviour. The longer time frame would also give more opportunities for interactive texts. Nonetheless, despite this reservation the intervention delivers a detailed behaviour change strategy in an easily accessible format with content that has been carefully tailored for the target group.

Introduction

The remit for this study required that the control group be given a conventional alcohol brief intervention. The one selected was from the Screening and Intervention Programme for Sensible drinking (SIPS)92 funded by the Department of Health. This programme tested three levels of intervention: a participant information leaflet, a brief advice session and an extended intervention involving lifestyle counselling. These interventions were tested in three linked studies: one conducted in primary care,93 one conducted in emergency departments94 and one conducted in a criminal justice setting. 95

The present study used the mid-level brief advice session, as it provides an attractive comparator containing the components of a conventional alcohol brief intervention. The SIPS website92 provides an outline of the brief advice session together with the materials used to train staff to deliver the session. These comprise the Microsoft PowerPoint^® (Microsoft Corporation, Redmond, WA, USA) slides that describe the aims of the session, a brief advice tool and a video demonstrating how to use the brief advice tool. Permission to use the SIPS materials was kindly granted by Professor C Drummond.

The SIPS brief advice session

The SIPS brief advice session involved giving participants feedback on their drinking. This was followed by a 5-minute session of structured advice using a brief advice tool as a prompt. An initial version was developed for the World Health Organization collaborative study on alcohol screening and brief intervention. 96,97 The version used in the SIPS trials93–95 is available on their website. 98 The revised tool presented the components of a conventional brief alcohol intervention in five diagrams and tables. It gave details of the health and social consequences of hazardous drinking, compared the participants’ alcohol consumption to that of the general population and identified the benefits of reduced drinking. Participants were then encouraged to make a plan to reduce their consumption and set a target for achieving the plan.

At the end of the brief advice session participants were given a 16-page participant information leaflet. 99 This gives information on the units of alcohol in common drinks and presents recommended drinking levels. It describes the short- and long-term risks of exceeding the recommended limits, including the social and psychological consequences of alcohol misuse. It invites readers to complete a quiz to identify whether or not they are drinking too much and gives guidance on how to seek help.

Training for the SIPS brief intervention

In the SIPS trials those delivering the brief advice (professional staff in health care and criminal justice) received a 1-hour training session. This showed them how to deliver the brief advice session in 5 minutes following a standard protocol using the structured advice tool as a prompt. Those delivering the intervention then engaged in a role play session delivering the intervention to a coworker. This was followed by feedback from the trainer.

Training for the control package in this study

This feasibility study planned to use laypeople (study co-ordinators) to deliver the face-to-face session. Furthermore, the session involved more than delivering the intervention; the baseline questionnaires had to be administered, and height and weight had to be measured. Thus, the training was extended to three 2-hour sessions followed by an assessment session and two pilot sessions with men from the target group. The training sessions comprised short, interactive lectures followed by supervised activities and role play with supportive feedback. The first session covered the background to the study, the completion of the questionnaires and the measurement of height and weight. The second session introduced the SIPS brief advice tool, demonstrated its use and allowed candidates to practice delivering it to each other. The final session allowed candidates to run through all the components of the control package. A full description of the design of the training is can be found in Chapter 6, Recruitment and training of the study co-ordinators.

Conclusions

The well-established SIPS alcohol brief intervention formed an ideal active control for this study. The availability of the SIPS training materials greatly aided the development of the training programme. The training for this study was longer than that given in the SIPS because of the additional demand made by this study.

Introduction

Study co-ordinators, trained laypeople, were used to deliver the face-to-face component of the intervention. The decision was based on two considerations: (1) engagement with the intervention and (2) implementation should the intervention prove to be effective. The study was conducted in the community and recruited men who were likely to think that their alcohol consumption was normal and did not cause them problems. Using authority figures such as experienced researchers could make it difficult to build rapport with the participants. In contrast, laypeople may be perceived as less intimidating by participants, leading to more engagement with the intervention. The other benefit is the potential for roll-out of the intervention, if it proves effective. If laypeople can be trained to deliver the intervention with high fidelity, this would be a low cost method for its widespread delivery.

This chapter describes the processes by which the laypeople (study co-ordinators) were identified, selected and trained. An overview of these processes is shown in Figure 2. The details of the methods used, and the rationale behind them, will enable replication elsewhere. In addition to this report, detailed manuals of study methods were produced. These manuals were provided to the study co-ordinators to aid intervention delivery. The manuals were given out and used during the training sessions.

FIGURE 2.

Overview of the recruitment of the study co-ordinators.

Recruitment strategy

The recruitment was designed to identify individuals with the aptitude for delivering an intervention that addressed potentially sensitive subjects (alcohol and obesity) in an empathic and supportive manner. The recruitment strategy comprised three stages:

1. clarifying the key characteristics and skills that were required

2. identifying the locations at which recruitment efforts could be focused

3. developing an extended selection process to identify individuals with key attributes.

Key characteristics

The role of the study co-ordinator was to put participants at ease; to measure alcohol consumption, height and weight; and to deliver the intervention or control package with high fidelity. Several essential characteristics were identified for use in selection of suitable individuals. These characteristics included fluency in English, previous experience of working with people, and excellent communication, organisation and interpersonal skills such as friendliness and approachability.

Control study co-ordinators

The control group study co-ordinators attended three training sessions plus an assessment session. They were trained to deliver the established SIPS brief intervention. 93 The training for this was based on the training materials supplied by the SIPS team on their website. 98 The SIPS brief intervention is based on a leaflet that presents the components of a conventional brief intervention in five diagrams and tables. These were designed to require only a simple description by researchers and to be easily understood by participants. The leaflet includes (1) a table that indicates the risks from hazardous and harmful drinking, (2) a graph to show that the participant’s drinking places him at a higher risk of alcohol related harm than most people, (3) a list of the benefits of reduced drinking, (4) targets that the participants can aim for and (5) how to make a plan to reduce drinking.

As with the intervention study co-ordinators’ training, the first control session provided the study overview, and gave experience in completing the alcohol and demographic questionnaires and in measuring height and weight. The second session covered the sections of the SIPS leaflet. A presentation describing the contents of the leaflet was followed by a demonstration of how to deliver the brief intervention by two role-playing research staff. The study co-ordinators then practised delivering the intervention in role play. The final session gave practice in welcoming participants, running through the session activities and closing the meeting with participants. The control co-ordinators were given the study co-ordinators’ manual for the control group.

Preparation of the study co-ordinators’ manuals

Separate manuals were written to assist in the delivery of the intervention and control sessions. The manuals were prepared by creating an initial list of the tasks that were to be performed. These manuals were intended to be used during the sessions and had to be concise and user friendly. To achieve this, the manuals were laid out in short sections, one for each task, with numbered headings. Technical terms were avoided. Guidance on each task was written in the active voice, for example ‘ask for the participant’s permission to record the session’. Bullet lists were used when several items were needed for a task, and for complex tasks that involved a sequence of active voice statements. For those tasks for which participants would need some explanation, example scripts were given in text boxes. To enable the study co-ordinators to deal with unexpected events and to cope with questions from participants, a ‘What If?’ section provided guidance on how to respond.

A lone worker’s policy (see Appendix 3) was included in the manual. This explained why adherence to the policy was important and presented a flow diagram of the steps involved. Separate guidance, in numbered sections, gave the information necessary to carry out each step. All study materials, including questionnaires and participant information sheets, were also included. The manuals will be available from the corresponding author under terms of confidentiality.

Post-training evaluation

At the end of training all those attending (n = 9) completed a short questionnaire evaluating the components of the training (see Appendix 4). All attendees reported that the number of sessions and the pace of the sessions was just right. The following quotation is reflective of the responses:

Pace was good. Given enough time to take everything in.

Study co-ordinator 2

The content of the sessions was also thought to be about right, although one respondent thought that the addition of a short video would be helpful. The role play was particularly valued by all participants:

Always useful, allows errors to come out in a safe environment.

Study co-ordinator 1

The manual that was written to support the delivery of the control brief intervention was thought to contain all the necessary information and guidance and to be clearly presented. The examples provided in the manual of how to phrase introductions to sections and interpretation of data were also commended. One candidate suggested providing a checklist of activities for the face-to-face session, as it would provide a useful aide memoire. This was prepared and subsequently used in the delivery of sessions. In response to a question asking if the candidates had any additional comments the respondents expressed considerable support for the training: ‘it was a fabulous experience’ and ‘excellent training’.

Selection, distribution of study materials and piloting

Selection of the best candidates took place after an assessment session in which each study co-ordinator delivered the whole face-to-face session to a role-playing researcher. The choice of the most able co-ordinators was made by the research team, utilising notes from the training sessions as well as feedback from the assessment session. The six successful candidates attended a subsequent session at which mobile phones, audio-recorders, scales and stadiometers were distributed. Housekeeping matters (e.g. room booking, travel expenses, gift voucher distribution and completion of payment forms) were also explained.

Two pilot sessions were organised for each study co-ordinator. These involved men recruited from the community and were held at community venues, replicating study conditions. The sessions were audio-recorded and individual feedback was given. This was intended to promote confidence but also to ensure fidelity of delivery.

Results and discussion

In total, 45 individuals applied for the posts, after only 3 weeks of advertising. This was encouraging because the post offered only a few hours of work per week for a short time period (3 months). The advertising may have tapped into social networks, as some applicants reported that other people had told them about the job. In a future study, it could be helpful to encourage this process, possibly by including a statement on the poster such as ‘do you have a friend who might want this job?’.

Twenty-two applicants were selected for interview. Interviewing identified those who were ineligible because they had prior experience in behaviour change, because they had had strong views about alcohol or because they had limited experience in dealing with people. All of the candidates were content with the selection process that would take place at the end of training. Twelve suitable individuals were selected for training, although three of these subsequently withdrew for personal reasons prior to training.

The training programme was well received, and all those who attended the first session completed all training and evaluation. The methods for measuring height and weight were quickly acquired, but recording alcohol consumption accurately proved challenging. This occurred because of the large variety of drinks available, particularly those whose brand name is available in different volumes and strengths of alcohol. For example, cans of Stella Artois are available in 440 ml, 500 ml and pint (568 ml) sizes with strengths at 4.0% alcohol by volume (ABV) or 4.8% ABV. The study co-ordinators had to be prompted to probe participants for fuller details of drinks consumed. In further studies, additional training could be given to cope with the profusion of alcoholic drinks. A few of the potential control study co-ordinators struggled with interpreting the graph of the frequency distribution of alcohol consumption.

After several role-play sessions, with supportive feedback, the potential intervention study co-ordinators quickly learned MI-based skills. It is possible that the selection process had identified individuals for whom attentive listening and empathising were normal behaviours. Additionally, the amount of practice included in training would have improved their skills.

The delivery of training required the presence of at least three members of the research team at each session. Role playing was an essential part of the training, with members of the research team acting as amenable or recalcitrant participants. To meet the needs of study co-ordinators with full time jobs, these sessions were additional to the working day. In total, nine sessions were given (five intervention, four control), with at least three researchers attending each session. To ensure they were familiar with the materials used, the researchers were also involved in developing the training. Funding for the extra research staff should be included in any further studies.

The selection of six individuals to deliver the intervention and control sessions proved straightforward. Although based mainly on the evaluation session, in practice the researchers made notes during each training session. All of the decisions were unanimous, with remarkable consistency between performance during training and at the evaluation. However, it was thought useful to keep the evaluation session, as it provided a target for candidates to aim at and gave useful practice in delivering the session.

The pilot sessions conducted in community venues were essential. The study co-ordinators found these much less stressful than the assessment session and also benefited from the additional practice. Feedback based on the audio-recording of these sessions was intended to boost confidence as well as to tackle weaknesses. The most common problem was the one encountered in the training sessions: insufficient probing for detailed descriptions of drinks consumed.

In summary, able candidates for the study co-ordinator posts were successfully recruited and trained. Initial over-recruitment, to allow for withdrawals and the selection of the most able candidates, was essential. The extended selection process helped to identify those who could empathise with participants and who could master the components of the face-to-face session. The training was well received with 100% attendance at the sessions. Most of the necessary skills were mastered, although additional training should be introduced for measuring alcohol consumption.

Introduction

Recruitment to clinical trials is challenging, but recruitment to prevention trials is even more difficult. 101 This could be a particular problem in this study because alcohol consumption and obesity are sensitive subjects, and addressing both in the recruitment process could reduce participation rates. Thus, the study used two recruitment methods to ensure good coverage of obese men who drink heavily, one recruiting through primary care and the other by community outreach. In addition, information was gathered on the processes involved in recruitment to determine where challenges to recruitment lay.

Systematic reviews have shown that several techniques can increase recruitment in general population groups: repeated attempts at contact at different times on different days, financial incentives, personalised approaches, assurances of confidentiality, sponsorship by a prestigious body such as a university and minimising the length of questionnaires used. 46,47,102–107 These techniques were incorporated into the recruitment strategies.

Methods

Results

Recruitment began on 1 March 2015 and was completed on 2 June 2015. A total of 69 men were recruited, exceeding the target of 60.

Recruitment through primary care

Three GP practices kindly agreed to participate. SPCRN staff identified 5198 men aged 35–64 years, of whom 1022 (19.7%) were recorded with a BMI of > 30 kg/m². A random sample of 200 men was taken from each general practice. GPs screened these samples to exclude men who should not be contacted. This removed 167 of the 600 men as a result of concerns about health, family problems or other issues. A further 14 men opted out after receiving a letter of invitation from their GP. This left 419 men to be contacted by telephone. Figure 3 shows the outcome after several attempts at contacting these 419 men. Nearly half (44%) were excluded because they did not meet the threshold of drinking > 21 units per week. Invalid telephone numbers and failure to achieve contact accounted for almost one-third (32%) of the men approached. Only 2% were ineligible because their BMI was ≤ 30 kg/m². Thus, of the original 419 men, 283 (68%) were contactable. Of those contactable, 87 (31%) were eligible for the study. Of those eligible, 36 (41%) were screened, gave informed consent and were randomised.

FIGURE 3.

Flow diagram of GP recruitment.

Recruitment through time–space sampling

Some 370 locations were visited to identify potential participants. These included all premises on 12 industrial estates (329 businesses), four large employers, two supermarkets, five taxi ranks, two golf clubs and two football matches. In addition, 28 visits were made to high streets and shopping centres. In total, 470 men were approached, yielding 33 men who were recruited and randomised. Thirteen men were recruited from public places such as the city centre and high streets, 10 were recruited from small businesses on industrial estates, seven were recruited from small businesses throughout the city, two were recruited from taxi ranks and one was recruited from a golf club.

The process of recruitment, from being approached to entry into the study, is shown in Figure 4. More than half (53%) of the men approached reported that they did not drink enough alcohol to be eligible for the study (i.e. they drank ≤ 21 units per week). A small number (18%) were outside the target age range. Thus, almost one-quarter (24%) of the men approached were eligible on age and alcohol consumption. Of those eligible, 50 men (45%) were interested. Some men were lost because they could not be contacted by subsequent telephone calls (n = 7), or they did not meet the obesity criterion at telephone screening (n = 10). Thus, 33 men were recruited.

FIGURE 4.

Flow diagram of TSS recruitment.

Discussion

The recruitment target of 60 men was met, although achieving this proved challenging. The recruitment methods met their prespecified criteria, except for the community outreach criterion of number of men per venue visited. Several factors contributed to the challenge of meeting the recruitment target. More men than expected reported that they did not drink. Survey data for 2013 showed that approximately 10% of men in the same age range claimed to be non-drinkers. 108 The higher frequency of non-drinkers approached (24%) could have occurred because the target group, obese men, may be more likely to be abstainers. Alternatively, some men may have reported abstinence as a method of declining to participate in the study.

Recruitment from primary care registers was a straightforward process that required repeated attempts at contact by telephone. However, recruitment through community outreach proved difficult. A direct comparison of the two methods might suggest that they were similarly effective: primary care registers yielded 36 recruits from 419 approached, while community outreach yielded 33 from 470 approached. However, the workload involved in making telephone calls to the men on primary care registers was substantially lower than that involved in visiting and exploring venues for the community outreach. The researcher who did the recruitment estimated that at least twice as much time was required for the community outreach. (The collection of such data was not part of the study design.)

The challenge of recruitment by community outreach arose partly because two entry criteria (BMI and alcohol consumption) were applied simultaneously, so that many men were automatically excluded. This contrasts with the primary care selection whereby all those approached had been pre-selected to meet the BMI criterion. Recruitment through community outreach suffered because only a small number of overweight men were identified at the venues visited. Inspection of the community venues during preliminary fieldwork suggested that more than a sufficient number of men should have been available, but few were available when it came to recruiting them. This may be an example of the operation of Lasagna’s law109 in a community setting.

A surprising finding was that attempts at recruitment through large institutions (such as businesses or the local authority) proved unsuccessful. Interviews with key stakeholders had identified considerable enthusiasm for the project and a willingness for their staff to become involved. This enthusiasm was still evident during the fieldwork, but the assistance took the form of circulating leaflets (the participant information sheet) or asking men to contact us. In the event, no men from large institutions volunteered for the study. It is possible that many men were willing, but they may have had reservations about the study, or may not have wanted senior colleagues to think that they were heavy drinkers. Some may have intended to volunteer but never did so. In some cases, these barriers might have been overcome by direct contact with the researcher. The concerns could be addressed by explaining the non-demanding, confidential and supportive nature of the study. In future, stakeholder interviews should explore this issue and seek approval for direct contact with the men. This would require identification of appropriate times for contacting individuals in these institutions.

An observation by the researcher doing the fieldwork was that the financial incentive offered for participation (£40) often elicited little or no response from potential participants. This is in contrast with a previous study with disadvantaged men, during which the researchers noted a marked increase in interest when the incentive was mentioned. It is possible that, in the present study, the incentive was too small to encourage men who were, on average, more affluent than the disadvantaged group. There is some evidence that higher financial incentives may be more effective for recruitment110 suggesting that for a full RCT a larger incentive might aid recruitment.

None of the men who were approached in community outreach was reluctant to discuss his alcohol consumption. In contrast, four of the men approached by letter from the general practice registers asked why they had been nominated for the study by their GP. Some were concerned that this implied that they had a drinking problem. The men were reassured when it was explained that the researchers had no information about their drinking. This suggests that a change should be made to the approach letter, highlighting the point that selection is not based on the individual’s alcohol consumption.

In summary, although recruitment proved challenging, the target was exceeded within the time period allocated. This indicates that a RCT would be feasible. A multicentre study could recruit 30–40 GP practices from which some 400–500 men could be recruited. If community outreach were employed, it would be best to select centres with large populations to provide sufficient venues for recruitment.

Introduction

The men who were recruited and randomised were invited to attend a face-to-face session with a study co-ordinator. This session enabled baseline data to be collected (height, weight and the previous 4 weeks’ alcohol consumption) and the face-to-face intervention or control treatment to be delivered. This chapter assesses how well these activities were conducted and presents the data on the demographic characteristics and alcohol consumption of the participants. It also evaluates the suitability of screening tools that were used to identify eligible participants.

Methods

Results

The study co-ordinators made repeated efforts to contact the 69 men who had been randomised. The outcome of this is shown in Figure 5. Three men could not be contacted and four withdrew when contacted (three because of changed family circumstances and one due to ill-health). Four of these men were from the intervention group and three from the control group. The characteristics of the men who were not seen, based on information gained at the telephone screening contact, are shown in Table 6.

FIGURE 5.

Flow chart of conduct of the baseline interviews.

Appointments were arranged for the remaining 62 men. All these men were willing to take part, but other priorities sometimes made this difficult. Eight men cancelled their appointments and one did not attend. Of these men, seven were seen at their first re-appointment. The two remaining men required several appointments because of cancellation or failure to attend before they were finally seen.

The face-to-face sessions were held in a variety of venues: the participant’s home (n = 31), a community centre (n = 14), the participant’s workplace (n = 10), a library (n = 5), a church (n = 1) and a yacht club (n = 1). In total, excluding homes and workplaces, nine different types of venue were used. All sessions were conducted without difficulty and all men agreed to the session being audio-recorded. The lone worker’s policy was consistently adhered to, but in the event no problems were encountered.

Discussion

The assessment showed that the baseline interviews were successfully conducted and complete baseline data were obtained. Those recruited covered a wide range of age, education, alcohol consumption and BMI. The study found that screening, using self-reported BMI (height and weight) and alcohol consumption, identified eligible individuals. Many participants were found to be at very high risk of liver disease because of the combination of high BMI and alcohol consumption.

Almost all of the men (95%) were at a 19-fold increase in the risk of dying from liver disease1 (BMI of > 30 kg/m² and > 14 units of alcohol per week). Many men recruited drank very heavily and were also severely obese, placing them at a very much higher risk of liver disease. The recruitment methods of this study can identify men who are in urgent need of intervention.

The two recruitment strategies were intended to produce representative samples of the target population, albeit using quite different methodologies. TSS is often used to recruit hard to reach individuals. 31 If there were sub-sections of the target group who were reluctant to participate, this community-based face-to-face method of recruitment might be more successful in recruiting them. In the event, the two strategies yielded almost identical samples. As the evaluation of the recruitment (see Chapter 7) showed that recruitment through GP registers was simpler and quicker, this method is recommended for the RCT.

The data showed a marked imbalance in alcohol consumption between intervention and control groups. This was a small study and a difference of this size could be due to random variation resulting from the small number of men randomised. This is recognised to occur frequently. 22 The intervention and control groups were balanced on demographic characteristics and BMI: imbalance on one factor would not be unexpected. Current guidance on feasibility studies highlights the limitation of feasibility studies because of their small sample size114 and the likelihood of imbalance at baseline. 25

Screening using self-reported values for alcohol and height and weight proved successful. Previous studies report that self-reported height and weight show a modest tendency to under-estimate weight and to overestimate height. 115 In this study three men were below the inclusion criterion for alcohol and obesity, but all were close to the threshold values. This places them at high risk of developing liver disease, as their risk is only slightly lower than men just above the threshold. In a RCT these men could be excluded for violating the protocol, although this would seem unjustified as they all have the potential to benefit from the intervention.

Two important and related findings were that organising the face-to-face sessions proved more difficult than expected and that, despite repeated efforts, appointments could not be made for some of the men who had been recruited. Some men did not answer their phones, meaning that several calls had to be made and reminder text messages had to be sent. Appointments were sometimes cancelled at short notice and alternative times had to be arranged. This caused some difficulty for the study co-ordinators and should be addressed in the training session for any future study. To minimise the loss of men from the study the study co-ordinators could be encouraged to arrange appointments as soon as possible after the screening interview.

Many different venues were used to hold the face-to-face sessions, and most often these were provided at low or no cost. The initial fieldwork to identify these venues was crucial to the success of this part of the study. The fieldwork would be essential for scaling up to a RCT. A surprising finding was that many of the men requested appointments at their own homes. It is possible that this was simply more convenient for the participants; however, it raises the issue of safety of the study co-ordinators as lone workers. A lone workers policy was successfully implemented, but this relied on the research team to monitor safety. A more formal arrangement with a commercial firm would be needed in a RCT.

The use of study co-ordinators made more demands on the research team than was expected. Distributing baseline questionnaires and collecting them on completion, obtaining and evaluating the audio-recordings of study sessions, and the provision of mentoring support and feedback all took time. Collection of data on this workload was not a planned part of the study, so exact figures are not available. However, it is likely these activities would add substantially to the cost of delivering the intervention and thus should be included in any cost-effectiveness study that accompanied a RCT.

In summary, the selection and recruitment methods were effective in identifying men who met the inclusion criteria. A delay between recruitment and baseline interview may have led some men to withdraw from the study. This delay should be minimised in future studies. The two recruitment methods gave similar samples, and so the easier and quicker GP register approach would be preferred. The recruitment methods of this study identified men who are at very high risk of liver disease. There is an urgent need for an intervention to reduce this risk in these men.

Introduction

A major concern for behaviour change interventions is implementation fidelity, that is, the extent to which the intervention is delivered as intended. 116–118 Three elements of implementation fidelity were particularly important for this study: adherence, that is, whether or not all components of the intervention were covered; competence, that is, whether or not the components were delivered to a high standard; and treatment differentiation, that is, the extent to which the intervention and control packages differ in their immediate impact on participants. Other important issues were the acceptability of the study to participants and whether or not participants were actively engaged with the intervention components. Methods to assess these items were developed and tested as part of this feasibility study. This chapter describes the findings from these evaluations.

Methods

At the end of the face-to-face session, just before they were given their gift voucher and travel expenses, the men were asked to complete an evaluation questionnaire (see Appendix 8). The questionnaire explored the process of arranging the session, the acceptability of the venue and the men’s experience of the session. It also explored the communication skills of the study co-ordinator, particularly sensitivity to the participants’ interests and feelings. Finally, it explored the men’s responses to the information given. After completing the questionnaire, participants placed it in an envelope, which was returned unopened to the research team.

The face-to-face sessions were audio-recorded to enable assessment of the study co-ordinators adherence to the set tasks and competence in performing them. Detailed instructions on the conduct of the tasks were given during the co-ordinator training, and these provided criteria to assess adherence to each task. Separate task lists were used for intervention and control study co-ordinators (see Appendix 9). Performance at these tasks was assessed by a research psychologist (ED) using the audio-recordings. To facilitate feedback to the study co-ordinators, comments on desired improvements were listed, together with comments giving positive feedback on tasks done well.

The intervention co-ordinators, who had been trained in some MI-based skills, were assessed on their competence using relevant items from the Behaviour Change Counselling Index (BECCI). 119 This was performed by a research psychologist (ED) using the audio-recordings. The items covered were: showing empathy, being sensitive to the participants’ views, asking open questions and summarising at the end of the session. Twenty per cent of recordings were reviewed independently by two researchers to ensure a standardised approach to coding.

Results

The questionnaire showed extremely high levels of approval of components of the study: organisation of the session (100%), acceptability of the venue (100%) and ability of the study co-ordinator to hold the participant’s interest (100%). The written comments from participants show that the sessions were run in a relaxed, friendly atmosphere and that the information provided was perceived as useful by both intervention and control participants (Box 3).

More interesting were the responses to the question about what was useful in the session. Some (35%) mentioned only that the information given was useful, but a slightly larger group (39%) said it was directly relevant to them, and a further 15% said that it identified the need for a change in their behaviour. Furthermore, there was a marked difference between the intervention and control group participants in what they found useful in the session. Although similar numbers mentioned alcohol (12 intervention, 8 control), substantially more men in the intervention group than in the control group mentioned the calories in alcohol (13 vs. 1).

The comments in response to a question about what was useful in the study shed light on the impact of the control and intervention packages. The intervention group comments identified information on risks of alcohol, on weight loss and the calories in alcohol (Box 4). In contrast, the control group focused exclusively on alcohol; its harms, the need to monitor consumption, the advisability of cutting back and goal-setting (Box 5). A feature common to both groups is the surprise men express when they reflect on their alcohol consumption. The intervention group also expressed surprise about their weight. This suggests active engagement with the activities of the face-to-face sessions, with a clear differentiation between intervention and control.

The audio-recordings of the sessions showed that adherence to all tasks was very high for all study co-ordinators; only one item was missed on a single occasion. For the intervention group study co-ordinators, competence was mostly satisfactory or better for all activities; typical evaluation comments given by the research psychologist are shown in Table 14. These identify that there was scope for improvement on some items, particularly probing about alcohol consumption, transition between activities and provision of summaries.

The audio-recordings revealed that the control group study co-ordinators also performed very well, with few activities that could be improved (Table 15). A direct comparison between intervention and control is not possible, as they performed different sets of tasks. For both groups of study co-ordinators, these comments enabled verbal feedback (by telephone) to be provided. This consisted of one or two supportive comments followed by discussion on improvements that could be made.

Assessment of MI-based skills (intervention group only) showed some variation across the items, although the lowest score was adequate (Table 16). The maximum possible score was 4.0. Scoring was high on some items, such as displaying empathy (3.62) and sensitivity to the participants’ concerns (3.59). The more challenging items were encouraging discussion of current drinking (2.9), and providing summaries at the end of the sessions (3.17). This pattern was seen for all intervention group study co-ordinators.

Discussion

The evaluation revealed very high levels of acceptability of the study methods. The participants found the face-to-face session relaxed and non-judgemental; they enjoyed it and felt they benefited from it. This might not have been expected in a study that addressed two potentially sensitive topics: heavy drinking and obesity. Furthermore, the evaluation showed that the intervention and control packages were delivered as intended and that there was a major difference between the immediate effects of the two packages.

The very high levels of satisfaction with the face-to-face session were probably partly due to social desirability bias. This results from the desire of participants to avoid embarrassment in the presence of the interviewer. 120 However, the nature of the face-to-face sessions could have upset the men. The intervention group were told that they were very overweight and that their alcohol consumption could contribute to weight gain. The control participants were told that that their drinking would harm their health and that they should drink less. As these men were not help-seeking and did not think they had an alcohol problem, it might be expected that some would object to what they were being told. Thus, the almost complete absence of any critical comments suggests that none of the components of the session caused annoyance.

Another factor that might explain the high acceptability of the face-to-face session could be the friendly and supportive style of the study co-ordinators. Sensitivity and the ability to empathise were key features of selection and training. The pace at which the face-to-face session was conducted could also have been important. The intervention sessions lasted on average 45 minutes, and the control 33 minutes. This provided ample time for participants to ask questions and to discuss their feelings about the information that was provided.

Finally, the setting in which the session was conducted and the use of laypeople as study co-ordinators could have contributed to an atmosphere of relaxed informality in which potentially critical statements were judged to be inoffensive. Although these are only possible explanations, it would seem best to replicate this approach in a RCT as, at the least, the methods did not cause offence.

The use of laypeople (study co-ordinators) to deliver the face-to-face session proved effective. The wide range of venues and appointment times requested by study participants were coped with by the use of three co-ordinators for each treatment arm. The study co-ordinators engaged with the whole process and appeared motivated by a desire to help people. They found that the participants were willing to openly discuss their drinking. The intervention group were also willing to discuss calories and being overweight (this topic was not covered in control group sessions).

The participant questionnaire highlighted differences between the intervention and control groups in what participants perceived to be useful about the session. As would be hoped, the intervention group commented on weight loss and calories in alcohol. In contrast, the control group mentioned only alcohol, except for one man who mentioned weight loss. This finding should be viewed cautiously, as it reflected perceptions directly at the end of a session. Nonetheless, what can be said is that there would be grounds for concern if differences between the intervention and control groups had not been found.

The intervention and control packages were delivered with high fidelity. The audio-recordings of the sessions showed that adherence to the tasks and activities in the session was very high. This probably reflects the highly structured nature of the session, in which there were set activities (measurements), each of which was followed by a discussion of the measurements made. The assessment of competence identified that three activities could be improved: probing of alcohol consumption, transition between sessions and summarising at the end of the session. To improve performance during the study, sensitive feedback was given on these activities. This began with positive statements on what was done well and led to guidance on what could be improved. In future studies, more attention could be given to the problematic activities during training.

The monitoring of adherence and competence identified opportunities to improve this process. Providing a rating score for each task (poor, acceptable, good) would simplify the provision of feedback to study co-ordinators. In addition, the score could be used in quantitative analyses to see if it influenced behaviour change.

The remit for this study required that the comparator was an active control, a conventional alcohol brief intervention. The control package was based on the brief intervention used in three trials in the SIPS, funded by the Department of Health. 93–95 The training in this study was expanded from the 1-hour session used in the SIPS to five 2-hour sessions (three teaching, one assessment and one pilot). This was partly because training was needed in the completion of the baseline questionnaires and in the measurement of height and weight. In the SIPS, the intervention session was intended to last 5 minutes, whereas in this study the average length was 33 minutes. The extra time was taken in collecting the baseline data, but also in discussing the participants’ responses to components of the control package. These differences mean that the findings from the SIPS studies are not directly comparable with this study.

In summary, the evaluation provides support for the acceptability and technical feasibility for this component of a full RCT. In particular, adherence to and competence at intervention delivery were high. Furthermore, the men in the two treatment arms differed in their responses to their interventions. This approach to assessing intervention fidelity could usefully be adopted in future studies.

Introduction

The main purposes of the outcome assessment were to estimate the retention of participants, to assess the acceptability of the study methods and to measure the primary outcomes for a RCT. The protocol for the feasibility study identified the retention of study participants as a challenge. A recent systematic review has shown that men-only weight loss interventions have an average loss to follow-up of 22%. 121 A related concern is the acceptability of the study methods, as this could affect engagement with the intervention and retention in the study.

Methods

Participants were invited to attend a follow-up visit 5 months after the baseline session. They were sent a reminder letter and a text message 2 weeks before the intended date of interview. The letter also asked about the most suitable time to be contacted. The study co-ordinators were sent contact details of the men, together with the participants’ previous preferences for venue and time to be seen. Study co-ordinators were assigned participants whom they had not seen at baseline.

Results

Of the 62 men who attended the baseline face-to-face session, 61 (98%) were interviewed at follow-up. However, only 59 men (95%) attended a session at which their weight was measured and questionnaires were completed. The other two men were interviewed by telephone, during which they gave a self-reported weight. Because 4 of the 59 men were difficult to contact, their session was conducted by a member of the research team and not by a study co-ordinator. The Consolidated Standards of Reporting Trials (CONSORT) flow diagram for enrolment, randomisation, follow-up and analysis can be found in Appendix 12.

Alcohol consumption reduced substantially from baseline in both the intervention and the control group (Table 17). The reduction in consumption was greater in the control group than in the intervention group. The control group had a higher consumption at baseline and, despite showing the greater reduction, had a higher consumption at follow-up. More men in the control group reduced their number of drinking days, whereas more men in the intervention group reduced their number of binge drinking days (> 8 units in one session).

The average weight of participants did not change between baseline and follow-up (Table 18). Over this period some men lost weight, some remained unchanged and some gained weight. The patterns were similar for intervention and control groups.

The elements of the FAST112 shed interesting light on the effects of drinking on the participants’ lives (Table 19). The men seldom experienced instances where they failed to remember a previous night’s drinking, and they only rarely failed to fulfil obligations because of drinking. Only a few men had been told that their drinking was a problem or that they should reduce their alcohol consumption.

At follow-up, intentions and actions to reduce drinking were more frequent in the intervention group than the control group. More men in the intervention group had thought about reducing their drinking or made a plan to cut down (Table 20). Among those who tried to cut down, the main reason (n = 14) was long-term health benefits; money, short-term personal benefits and family concerns were seldom mentioned.

The intervention group were also more likely to think that they had successfully cut down and to have made a plan to deal with difficult drinking situations. It is surprising that in spite of these reported intentions and actions, few men in either group considered themselves at risk of the harmful effects of alcohol. Furthermore, when asked in general terms about possible harms of drinking, the participants were able to identify many different harms (Table 21). The only notable difference between the treatment arms was that more men in the intervention group identified overweight/obesity as a consequence of drinking.

The ability to refuse a drink in different scenarios was broadly similar in the intervention and control groups (Table 22). Possible exceptions were that more men in the intervention group strongly agreed that they could resist drinking while watching television and when friends were drinking. More notable is that half of participants did not think they could resist alcohol when drinking with friends.

In both the intervention and the control group, few men knew their BMI (Table 23). However, more men in the intervention group had counted the calories in the alcohol they consume. Almost all of the men in the study believed that their drinking contributed to their weight, with no difference between intervention and control.

Knowledge of the units of alcohol was assessed by asking the men to state the number of units in lager, wine and whisky. Because these drinks come in a range of strengths, the responses have been grouped, with one grouping covering the range of possible correct answers (i.e. 2.0–2.49 units for a pint of lager, 3.0–3.5 units for 250 ml of wine and 1.0–1.5 units for spirits). The results are broadly similar for the intervention and control groups (Figure 6). The men were mainly correct on the units in lager, but tended to either underestimate or overestimate the units in wine, and to overestimate the units in spirits.

FIGURE 6.

Participants’ estimates of the units in common types of drinks.

The question about the calories in common drinks was asked in the same way as the question about units in drinks. The correct answer for each type of drink fell in within a range: a pint of lager contains 150–249 kcal, a 250 ml glass of wine contains 150–249 kcal, and one measure of a spirit contains 50–99 kcal. The participants in both the intervention and the control group were more likely to be incorrect than correct, with a tendency to overestimate (Figure 7).

FIGURE 7.

Participants’ estimates of the calories in common types of drinks.

Discussion

The follow-up analysis has shown that the rate of retention was very high, that the outcome data were collected and that the study methods were acceptable to participants.

Rates of retention are often poor in obesity trials. 124,125 Although rates across individual studies vary widely,126,127 reviews have found average retention rates of 70–80%. 121,128 A recent study of a mobile phone intervention in overweight adults had 68% retention. 129 The retention rate of 98% in this study may reflect the efforts made by the study co-ordinators and the research team to contact men. The goodwill that the participants felt towards the study could also have aided retention.

A reduction in alcohol consumption was seen in both intervention and control groups. This study used an active control – a conventional brief intervention – so a fall in consumption would be expected. However, the size of the reduction was greater in the control group than in the intervention group. Interpretation of this finding is difficult because consumption at baseline was much higher in the control group than the intervention group. Imbalance at baseline is recognised to occur in small studies, and current guidance is that such studies should not be used to estimate effectiveness. 22,25,114

Self-reported alcohol consumption amounts to only 40–60% of alcohol sold. Recent survey data from four English-speaking countries account for only 31–57.5% of alcohol actually sold. The under-reporting for Scotland is 54%. 130 One explanation for this is social desirability bias, in which individuals report drinking less to make a good impression. 131 The bias would need to increase greatly from baseline to follow-up to account for the observed falls in consumption. This seems unlikely.

The findings on the fall in alcohol consumption could also be explained by regression to the mean. 132 Some men could have consumed an unusually large amount of alcohol just before the baseline session, resulting in a high mean consumption. However, by the time of the follow-up session the participants could have been drinking more normal amounts, producing the observed fall in consumption. These findings illustrate why an adequately powered RCT is required to provide a reliable estimate of the effectiveness of the intervention.

No loss in weight was seen in either the intervention or control groups. Although alcohol consumption was reduced, a longer interval would be needed before weight loss would be observed. It is also possible that an extended intervention period would be helpful. Weight loss interventions are commonly delivered over a period of several months. 18 Extending the text messages would enable additional time to be given to relapse recovery and maintenance of the new behaviour.

The acceptability of the study methods was high when measured by several indirect questions. Participants reported that they found the information given useful, they discussed the study with family and friends and they would recommend it to others. Some men suggested improvements to the study, but these mainly consisted of a wish for more information or for more contact with study co-ordinators. The only critical comments were on the content of a few of the text messages that the intervention group received.

An important finding was that most participants did not view their drinking as harmful, despite being knowledgeable about the harms of alcohol. Possibly this sense of invulnerability occurred because few suffered from acute effects such as forgetting what had happened during the previous night’s drinking or being unable to perform usual activities because of alcohol. Thus, the men may think that these harms will affect only other drinkers. The challenge for an intervention is to increase the perceived relevance and importance of these harms. In a full RCT, more attention should be given to the increased risk of liver disease that results from being obese and drinking heavily. In addition, efforts should be made to increase perceived susceptibility to these risks.

Recall of the units of alcohol in common drinks was poor in both the intervention and the control groups. One explanation for this could be that men do not often think about their alcohol consumption in terms of units. There is evidence that people measure their consumption in drinks rather than in units133 and find the arithmetical difficulty of the calculation a deterrent to the use of units. 134 In the light of this, it is not surprising that there is poor retention of the detailed information given on one or two occasions. For the RCT, it would be better to focus on reducing the quantity of alcohol consumed. This would allow men to visualise their consumption in ways that are familiar to them.

A few men felt that one theme of the text messages, five-a-side football, was not relevant to them. This was used as a device to encourage men to reflect on the physical capabilities of their youth, so that recapturing some of that lost fitness could be used as a motive for losing weight. An alternative theme that would be relevant to all men could be climbing stairs or walking a road with a steep incline. These are activities that the men in this study recalled performing with ease in their youth that they perceived to be important and relevant to them in the present and future. This theme could be incorporated into the text messages for a RCT.

The questionnaires that would be used in an economic evaluation produced mixed findings. The quality of life instrument, EQ-5D,123 detected problems that could be associated with obesity, mobility problems and pain and discomfort. It would be useful for a RCT. The Short Service Use Questionnaire recorded very few contacts with police/criminal justice, social services and the health-care system. In any planned economic evaluation, consideration will need to be given to the fact that the baseline contacts with the health system and criminal justice system will be very low. Any benefit of reduced alcohol consumption and weight is most likely to be observed in the future rather than during the study period.

In summary, the follow-up analysis has demonstrated that a high retention rate can be achieved with the target group and that the outcome measures for a RCT can be measured. It has also shown that the study methods are acceptable to the participants. Several findings are consistent with the intended mode of action of the intervention. More men in the intervention group than the control group reported making a plan to reduce their drinking, succeeding in reducing their drinking, thinking that alcohol led to overweight/obesity and counting the calories in the alcohol they consumed. The analysis has also identified several ways in which the intervention could be strengthened, particularly by increasing the perceived risk of the harms of alcohol and by extending the intervention period.

Introduction

The intervention group received a series of 95 SMS messages that were delivered over a period of 8 weeks. The control group were sent four messages during these 8 weeks as a way of keeping in touch with participants and maximising retention in the study. The results in this chapter are based on the data from the 31 men in the intervention group.

The text messages were designed to promote interaction and increase engagement with the intervention. Some messages were intended to prompt a response, but participants could reply to any of the text messages they received. The responses from participants were anonymised and collated by the Health Informatics Centre at the University of Dundee. The computer package that delivered the messages also stored all text message responses received from participants electronically (see Chapter 4).

Methods

Fidelity of the delivery of the intervention was assessed by monitoring delivery of the text messages to the mobile phones of the participants. Engagement with the intervention was assessed by recording the frequency of responses from the participants during the intervention phase. Content analysis of the responses given provided a method to ascertain the nature of engagement with components of the behaviour change intervention.

Results

Responses to the text messages

Text message responses were received from all but two of the participants (94%). A total of 456 messages were received from the remaining 29 men (Figure 8). The number of responses per participant ranged from 0 to 41 (mean 14.7, median 12). More than 60% of the men responded more than 10 times. However, there was marked variation between participants. Four men responded more than 35 times, while seven responded on fewer than five occasions.

FIGURE 8.

Number of responses to text messages per participant. ID, identification.

A total of 245 responses were received to the 14 text messages that prompted a response, giving a mean response of 17.5 per message (Figure 9). In addition, many of the participants responded to other text messages. Of the 81 messages which did not request a response, 40 received at least one reply. The mean response to these 81 messages was 2.6 responses per message.

FIGURE 9.

Number of replies to text messages that prompted a response.

Twenty-nine of the 31 participants responded to the text messages that asked questions (see Figure 9). More than 80% of the participants (25 men) answered the first and sixth questions. Overall, 10 of the 14 questions elicited responses from > 50% of participants. The first eight questions attracted ≥ 16 responses each, whereas four of the remaining questions attracted ≤ 13 responses. This may indicate that there was some attenuation during the second half of the intervention period. However, 20 men (65%) answered the penultimate question, indicating that interest was maintained throughout.

Discussion

This feasibility study has shown that the target group, men who are obese and drink more than the recommended guidelines, engaged with the tailored, interactive intervention that was delivered by text message. The methods used for monitoring delivery of the messages and evaluating the participants’ responses showed that the men received the text messages, they opened and read them, they carefully considered the content of the messages, they engaged with the components of the behaviour change strategy and they took the time to respond to questions asked.

Fidelity of delivery is essential for the success of an intervention. 135 This study has shown that a text message intervention can be delivered with very high fidelity. Almost all of the messages (98%) were successfully delivered to the phones of the participants. The high volume of responses received from the men, more than 450 over a 2-month period, confirmed that the messages had been read and understood.

Comprehension of the intervention and ensuring that participants are able to perform the new skills taught are also important measures of fidelity. 136 By incorporating questions that addressed steps on the causal model to behaviour change, the study provided a method to assess both of these requirements. All of the responses indicated that the participants had understood the questions. Many responses were what would be expected if there was a high degree of engagement with the intervention. Clearly, some of the men were willing to act on the messages in ways that could lead to behaviour change. For example, a precursor to action to change behaviour for men in this study is the recognition that obesity is affecting them personally, and the realisation that reduced drinking could help to address this. For many men who identified potential benefits of reducing drinking, the reason was related to obesity, for example getting ‘rid of the beer belly’, losing weight, and avoiding gout and diabetes. This gave a clear indication that the men were engaging with the intervention. Later in the intervention period, the responses demonstrated that the men were able to use the behavioural skills demonstrated in the text messages for example goal-setting and action planning, and the cognitive skills taught such as preparing for risky drinking situations. For example, some men had considered and implemented strategies to avoid excessive drinking, including taking their car on nights out, finding alternative activities to those that involved drinking, and simply buying less alcohol.

The study has shown that interest was maintained throughout the intervention period. Some attenuation in the number of men responding to the later questions was observed. However, this was partially due to the nature of the questions posed during the last two weeks of the intervention. Some of the later questions were relevant only to those men who had already made changes, that is the questions asked about coping planning, strategies to maintain reduced drinking and about the benefits of reduced consumption. The intervention period of 8 weeks was possibly too short to guide participants through the complete behaviour change process. For a RCT, a longer period of at least 3 months would give participants a greater opportunity to work through the steps to behaviour change.

A major advantage of this intervention is that participants respond in real time. They can read and re-read the messages at a time that suits them, they can take time to consider the content and consider their response or reaction to the messages and, if they so wish, send a response back. These responses give an indication of the impact of components of the intervention as it happens. These steps form a valuable part of the intervention, as well as serving as a tool for process evaluation.

The final benefit of monitoring the text message responses was the opportunity to detect any dissatisfaction among the participants or adverse events. By reviewing the text messages every day, the research team were able to act on any problems swiftly and effectively. Monitoring responses is important for ethical reasons in research studies. However, for a substantive service delivered over a wide area, an alternative approach might be required to reduce the workload of monitoring.

Conclusions

The text message intervention was delivered with high fidelity to the target group. Content analysis has shown that the text messages were understood, and that the participants engaged with the intervention. Without suggestion, men identified losing weight as the main benefit of changing their current drinking pattern. The responses also demonstrated that the men were able to use the behavioural and cognitive skills demonstrated in the text messages, for example goal-setting and action planning. More text messages, delivered over a longer period, may be required to allow the participants more time to progress through the behaviour change process.

Extent to which the feasibility study objectives were achieved

The protocol identified six key objectives for this feasibility study. A brief assessment of the extent to which these objectives were met is given here.

Patient and public involvement

This feasibility study has demonstrated the importance of involving members of the public in all stages of the design and conduct of research. Two user group representatives provided guidance on developing the recruitment strategy, the design of the intervention and the interpretation of the findings. The feasibility study involved focus groups that explored attitudes and beliefs about drinking and losing weight with men from the target group. This clarified opportunities to increase motivation to change behaviour and identified potential barriers to behaviour change. Their views shaped the content of the intervention. Laypeople also delivered the face-to-face component of the intervention. This helped the study participants, men who were confident that they did not have a problem with alcohol, to engage with an intervention designed to reduce their alcohol consumption. Key stakeholders, managers of local businesses and community groups, showed considerable enthusiasm and a willingness to be involved in research to improve public health. This has established a possible route for national roll-out of an effective intervention.

Discussing drinking and weight

Focus groups were held with men in the target group. These groups identified both levers and barriers to behaviour change. The men talked openly about their drinking and the benefits of losing weight. Thus, they would be receptive to an intervention that involved frank discussion of these topics. During the focus group discussions, the men described many benefits of losing weight, and identified no disadvantages to the prospect of weight loss. This suggests that the men might seriously consider guidance on how to reduce weight.

The main barrier to drinking less was the impact that the men perceived this would have on their social life. However, the men also thought that their relationships with close family and friends could improve if they drank less. The intervention could increase the salience of this benefit by capitalising on the surprise that men expressed when they calculated the amount of alcohol they consumed. The men could be encouraged to explore what benefits their close family would enjoy if they reduced their alcohol consumption.

The effect of alcohol on weight

The men in the focus groups required prompting to acknowledge the connection between alcohol consumption and being overweight; however, once prompted, they agreed that alcohol had an important effect. Prompting was also required to clarify the nature of the relationship, that drinking leads to increased snacking and additional late night meals. Again, the men were confident that this was a common phenomenon. In contrast, the men were unaware of, and expressed surprise at, the number of calories in alcohol. This suggests that men can readily appreciate the nature of the link between alcohol and weight. The intervention should reinforce this link and increase its salience to the participants.

The consequences of obesity

Men in the focus groups were concerned about the way they looked, to the extent that it caused embarrassment. Some men were also concerned about the ways in which excess weight limits carrying out usual activities such as climbing stairs and playing with their grandchildren. These factors could be used to increase motivation to lose weight.

The focus groups also clarified an acceptable style of delivery of components of the intervention. For example, information about the calorie content of alcohol could be conveyed as interesting and new. However, the benefits of drinking less were well known by the target group. Thus, the approach should be to increase the perceived personal relevance and importance of the benefits.

A counterproductive idea

Finally, the focus groups showed that an apparently attractive lever, illustrating the calories in alcohol by comparing them with the calories in common foodstuffs, could be counterproductive. Given the choice, the men were willing to forgo food in order to justify consuming alcohol containing the same number of calories as the forgone foodstuff.

Stakeholder interviews

Interviews with managers of venues in the local community identified a willingness to help with the recruitment of participants. The stakeholders requested direct involvement, by making contact with potential participants themselves. In the event, this appeared to act as a barrier, as no-one was recruited in this way. However, the enthusiasm for tackling obesity and alcohol consumption could offer a method for national roll-out of the intervention. Based on the findings of the present study, it would require prior negotiation to gain direct access to potential participants. This could be achieved if the initiative were part of a national campaign to tackle what the stakeholders perceive as major problems: heavy drinking and obesity.

Recruiting and training laypeople

The tailored recruitment strategy identified enthusiastic and empathic laypeople. Adequate training with continued mentoring equipped the laypeople to deliver the face-to-face component of the intervention. Each layperson delivered the intervention to an average of 10 people. The time required for training was of the same order as the time each layperson spent delivering the intervention. If the intervention were to be scaled up for a RCT or national roll-out, it could be more cost-effective to employ staff on half-time or full-time contracts. A comprehensive lone worker’s policy is necessary to ensure the safety of the laypeople doing the interviews.

Designing the intervention

The intervention was designed to follow the steps in the causal model to behaviour change. It was delivered in a face-to-face session followed by a series of text messages. The face-to-face session capitalised on the measurement of alcohol consumption and weight. Feedback of information on current behaviour is a common technique in alcohol brief interventions. Thus, the intervention gave men the opportunity to calculate for themselves the units of alcohol and the calories in the alcohol they consumed. They also plotted their height and weight on a BMI chart, revealing that they were in the upper extreme BMI category. Together, these measurements and calculations illustrated the intended logic of the intervention: reducing alcohol consumption could result in weight loss. This was reinforced by a discussion of the ways in which alcohol increased food intake (e.g. by increasing snacking) and the benefits that the men would enjoy if they lost weight.

The series of text messages was designed to strengthen the motivation engendered in the face-to-face session. More importantly, the text messages were intended to promote action to reduce alcohol consumption through goal-setting and action planning. Relapse recovery and maintenance of the new behaviour were also addressed. For a future trial, additional text messages should be written to reinforce maintenance of reduced drinking.

Recruitment of participants

In total, 69 men were recruited, exceeding the target of 60. Recruitment through GP registers identified a large pool of potential participants (men with a BMI of > 30 kg/m²), enabling recruitment to proceed smoothly. The high proportion of men who drank less than the entry criterion (> 21 units per week) was the main challenge in recruiting from the GP lists. Recruitment by community outreach was time-consuming because very few men were recruited from each venue visited. This resulted in a higher workload for community outreach than for recruitment through GP registers. Of the men approached in community outreach, a high proportion (49%) reported drinking < 21 units per week. The city centre was the most productive location for recruiting men by community outreach. Recruitment from large organisations was largely unsuccessful, and leaving Participant Information Sheets at potential venues for recruitment was ineffective.

Baseline findings

Baseline interviews were successfully conducted at a variety of community venues convenient to participants. Initial fieldwork is necessary to identify these locations. Many men requested visits at their homes. The participants covered a wide age range and the full spectrum of socioeconomic status.

The screening methods successfully identified men meeting the entry criteria on BMI and alcohol consumption. These methods could be used in a RCT. On average, the participants drank well above government guidelines and frequent binge drinking was common. The recruitment methods of this study identified men who have a 19-fold increased risk of dying from liver disease. There is an urgent need for an intervention to reduce this risk in these men.

Evaluation of the face-to-face sessions

The methods for assessing the fidelity of delivery of the face-to-face sessions were successful. They established that although the quality was high, there were areas for improvement, particularly in probing for details of alcohol consumption and summarising the content of the session. Individual feedback to study co-ordinators was well received. Mentoring the study co-ordinators and providing them with study materials was a time-consuming process.

The face-to-face session for the intervention group was restricted to increasing motivation to reduce alcohol consumption. This decision, taken when designing the intervention, proved to be sensible. The session was intended to last for about 30 minutes, but in the event the average duration was 45 minutes. Adding in volitional activities would have overburdened the session.

Fidelity of delivery of the text messages

Almost all of the text messages (98%) were delivered to the participants’ phones. As judged by the responses they made to these texts, the participants engaged enthusiastically with this phase of the intervention. The nature of their responses showed that the men had understood the text messages. They also showed that men responded as intended to key steps in the behaviour change sequence. Text messages provide a low-cost method of delivering an intervention over an extended period.

Responses to the text messages sent to the intervention group were monitored daily to determine if any action was required. In the event, four men (13%) were contacted because of concerns and all were resolved satisfactorily. It is important that all studies using text messages make provision for this monitoring.

Follow-up of participants

A very high follow-up rate was achieved (98%). Sessions were organised to be convenient for participants, in both location and timing. The sessions were run in a friendly, supportive manner, and participants’ questions and concerns were fully addressed. Several methods of contact (mobile phone, home telephone, postal address and e-mail address) were used to maximise follow-up. Text messages were sent to keep in touch during the follow-up period. Finally, the study co-ordinators and the research team made considerable efforts to arrange follow-up appointments. All these techniques could easily be replicated in a RCT; however, it is likely that the follow-up rate would be lower in a RCT because the follow-up period would be longer than that used in this feasibility study.

Outcome assessment

Two outcome measures that would be used in a RCT, alcohol consumption and weight, were successfully measured. The average weekly alcohol consumption remained high at follow-up (34.6 units) and most men continued to binge drink at least weekly. Both intervention and control groups identified a wide range of harms of alcohol. Despite this high consumption and knowledge of harms, most men (74%) believed they were at low risk of harm from alcohol. An explanation could be that their perceived lack of vulnerability was because they seldom suffered acute harms (e.g. hangovers) and made few visits to a GP or hospital. In a RCT, more attention should be given to increasing the perceived relevance of harms.

Weight did not change in either group, although alcohol consumption fell. However, given the short duration of follow-up, a large fall in weight was not expected. A longer period of follow-up would be required for a RCT.

In this study, the outcomes were assessed by the same set of study co-ordinators who did the baseline assessment. Although the study co-ordinators were blind to treatment group, it would be better for a RCT to have the outcomes assessed by an independent researcher.

Estimating effectiveness

The study did not attempt to estimate the effectiveness of the intervention. Current guidance on feasibility studies cautions against estimating effect sizes, as feasibility studies are under-powered and the estimates they provide are likely to be biased. In this study, there was an imbalance in alcohol consumption at baseline between the intervention and control groups, which confirms that an effect size should not be calculated. Instead, the study provided estimates of the variance of the primary outcome measures (alcohol consumption and weight), which could be used in the power calculation for a RCT.

Acceptability of the study

The study methods were found to be acceptable at baseline, with the face-to-face session being described as friendly, informative and worthwhile. The very high follow-up rate (98%) suggests that the men had participated enthusiastically and indicates that they were very willing to maintain their involvement. At follow-up, most men thought they had benefited from taking part, providing examples of awareness of their high alcohol consumption and their commitment to drinking less.

Economic evaluation

The quality of life instrument, the EQ-5D, detected problems that could be associated with obesity, specifically mobility problems, pain and discomfort. It would be useful for a RCT. However, the instrument used to detect usage of police/criminal justice, social services and the health-care system, the Short Service Use Questionnaire, recorded very few contacts. Nevertheless, it may be important to establish that this group of subjects is currently consuming few health-care resources. For that reason, the inclusion of this short resource questionnaire may be required. The design of the economic evaluation will need to consider that the benefits of reducing alcohol consumption and weight, in terms of improved quality of life and reduced resource consumption, are likely to occur some years after the delivery of the intervention. For this reason, any economic evaluation is likely to require an extensive modelling component based on surrogate endpoints included in the trial.

Conclusions

In summary, this feasibility study was conducted successfully. The focus groups made an important contribution to the design of the intervention. They provided a helpful way of testing intended components of behaviour change interventions, and thus we recommend that the use of focus groups should be standard practice in feasibility studies. The objectives for recruitment, retention and acceptability of the study methods were all met. An intervention that fully met the remit set by NIHR HTA was developed and delivered. The study showed that the men, who thought they were at low risk, were in fact at very high risk of liver disease. However, these men engaged enthusiastically with an intervention intended to change their behaviour. Furthermore, they responded as intended to the key steps in the behaviour change strategy. This suggests that an intervention targeted at these high men could be successful. There is an urgent need for an intervention to reduce the risk in these men. The feasibility study methods will enable a RCT to evaluate the effectiveness of the intervention to be carried out to a very high standard.

Modifications for a randomised controlled trial

A main purpose of feasibility studies is to identify improvements for RCTs. Although the methods of this study were shown to be robust, several opportunities for improvement were identified.

Contributions of authors

Professor Iain K Crombie (Professor of Public Health) was the chief investigator for the study. He designed the study and supervised its conduct. He led the design of the intervention and the training programme for the study co-ordinators, contributed to the training of study co-ordinators, analysed the data and wrote the first draft of the final report.

Dr Kathryn B Cunningham (Research Assistant) recruited focus group participants and key stakeholders, conducted the focus groups and stakeholder interviews, led the analysis of the focus group data, assisted in the design of the intervention, recruited and led the training of the study co-ordinators. She wrote the first draft of the Study Manuals.

Dr Linda Irvine (Senior Research Fellow/Trials Manager) contributed to the study design, intervention development, training and supervision of the study co-ordinators, analysis of the data and the writing of the final report. She also wrote the text messages.

Professor Brian Williams (Director of the Nursing, Midwifery and Allied Health Professions Research Unit) contributed to the study design and the text message intervention.

Professor Falko F Sniehotta (Professor of Behavioural Medicine and Health Psychology) contributed to the study design, and provided guidance on the recruitment of participants and the writing of the Study Manuals.

Professor John Norrie [Professor of Clinical Trials and Biostatistics, Director, Centre for Healthcare Randomised Trials (CHaRT)] contributed to the study design.

Dr Ambrose Melson (Research Associate) contributed to the study design, conducted the focus groups and analysed the transcripts, and contributed to the design of the intervention.

Dr Claire Jones (Senior Software Engineer) was responsible for the web-based randomisation system and created, ran and monitored the message delivery system, the system for recording receipt of SMS texts and the mechanism for storing text message responses from participants.

Professor Andrew Briggs (Health Economist) advised on approaches to economic evaluation and suggested questionnaires that could be used.

Dr Peter M Rice (Consultant Psychiatrist) contributed to the study design, advised on alcohol questionnaires and alcohol related problems.

Dr Marcus Achison (Research Assistant) recruited participants from GP practices and by community outreach. He co-ordinated the allocation of participants to study co-ordinators at baseline and follow-up, and assisted in the supervision of the study co-ordinators.

Andrew McKenzie (Research Assistant) recruited participants by community outreach. He assisted with the training and supervision of the study co-ordinators and analysed the trial data.

Elena Dimova (Research Assistant) assisted with the design and delivery of the study co-ordinator training, contributed to the Study Manuals, assessed the audio-recordings of the intervention sessions and provided feedback to the study co-ordinators.

Dr Peter W Slane (GP) advised on recruitment methods and contributed to the study design.

All authors have contributed to the final report.

Data sharing statement

For research purposes, data can be obtained from the corresponding author: Professor Iain K Crombie, Division of Population Health Sciences, School of Medicine, University of Dundee, UK, i.k.crombie@dundee.ac.uk.

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 HTA 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 HTA programme or the Department of Health.

Control group

• Introduction.

• Completion of questionnaire.

  • Demographics.

  • TLFB.

  • FAST.

• Provision of information on risks of drinking.

• Opportunity for participant to consider the risks.

• Comparison of participant’s drinking with general population.

• Provision of information about benefits of reducing drinking.

• Provision of information about strategies for reducing drinking.

• Provision of information about sensible drinking targets.

• Provision of ‘How much is too much?’ leaflet.

• Measurement of participant’s height and weight.

• Finish.

Intervention group

• Introduction.

• Completion of questionnaire.

  • Demographics.

  • TLFB.

  • FAST.

• Calculation of participant’s typical alcohol consumption.

• Discussion of pros and cons of the participant’s current alcohol intake.

• Discussion of pros of the participant reducing alcohol consumption.

• Measurement of height and weight.

• Charting of participant’s BMI.

• Calculation of the participant’s typical calorie consumption from alcohol.

• Discussion of how alcohol can contribute to weight.

• Finish.