Real Time Monitoring of Engagement with a Text Message Intervention to Reduce Binge Drinking Among Men Living in Socially Disadvantaged Areas of Scotland

Text messaging provides a method for delivering behaviour change interventions with the potential to reach large numbers of individuals at low cost. Recent systematic reviews conclude that text messages can promote smoking cessation [1‐3], increase adherence to antiretroviral therapy [4, 5] and address weight loss and physical activity [6]. A recent systematic review of text message interventions to reduce alcohol consumption reported that the evidence is promising but preliminary, and recommends further research [7].

Text messaging has become the most common method of electronic communication because it links simplicity with interactivity [8]. More than 90% of people under 65 years in the UK regularly use a mobile phone [9], and users frequently check their phones for messages [10]. Text messages are usually read soon after delivery [11] so individuals receiving an intervention by mobile phone are likely to open and read the messages. Little effort is required by recipients to receive the intervention, and text messages can be accessed at any time. Messages can be read quickly and re-read if desired. People who may not want to commit time for attending appointments or reading leaflets may be willing to receive concise text messages.

Despite a proliferation of text message intervention studies in recent years, there is insufficient evidence to describe the theoretical constructs and mechanisms that lead to modified intentions and subsequent behaviour change. Recent systematic reviews report that few studies provide details on the health behaviour theories that underpin the interventions [12‐14], and even among those that give the theoretical basis, few evaluate the effects of the components on behaviour [1, 6, 12, 14].

This study explored the participants’ real time responses to an empirically and theoretically based text message intervention. The aim was to assess the extent of engagement with the intervention. A key element of engagement with digital interventions is intended usage, which has been described as ‘the extent to which individuals should experience the content (of the intervention) to derive maximum benefit from the intervention, as defined or implied by its creators’ [15]. Yardley et al. [16] extend the approach to encompass ‘the process of achieving positive cognitive, emotional, behavioral, and psychologic change’. In this, they emphasise effective engagement which is sufficient to achieve the intended outcome.

This paper builds on the work of our feasibility study which showed that participants often responded to text messages [17, 18]. It uses the data from our recently completed randomised controlled trial [19] to address two research questions:

What was the frequency of responding to the text messages?

Do the responses enable an assessment of effective engagement with the intervention components?

This study used data from a four-centre, pragmatic, individually randomised controlled trial of a text message intervention to reduce binge drinking among young to middle-aged socially disadvantaged men [19, 20]. Participants were men aged 25–44 years who were recruited from areas of high deprivation in Scotland. Men were included in the study if they had two or more episodes of binge drinking (>8 UK units in a single session, corresponding to >64 g of alcohol) in the preceding 28 days. This paper focusses on the 411 participants in the intervention group only i.e. those who received the novel intervention. The results of the randomised controlled trial are described elsewhere [19].

The 411 men in the intervention group had a mean age of 35 years, just over half were living with a partner and almost one third were unemployed (Table 1). Over 60% of men had high school education only, and over three quarters lived in the most disadvantaged areas of Scotland, as measured by the Scottish Index of Multiple Deprivation (a tool that incorporates several different aspects of deprivation to provide a score based on postcode) [30]. The participants had at least one occasion of binge drinking per week, and mean alcohol consumption was more than 33 UK units per week.

This study has developed and tested a real time and real world method of assessing the impact of individual text messages within a digital behaviour change intervention. Almost all of the participants gave responses to the text messages which demonstrated that they were fully understood. The responses also showed that men frequently gained positive cognitive, emotional and behavioural benefit from the text messages, indicating effective engagement with the intervention. The frequency of disclosure of personal details, particularly to those text messages that did not prompt a response, confirms a high level of trust which is required for engagement.

Effective engagement was assessed using the approach developed for digital behaviour change interventions [15, 16]. We have applied this approach to components of the intervention, determining whether the responses indicate that the intended outcome of the component has been achieved. This study has shown that the important distinction between engagement with the intervention components and more general engagement with the study also applies to evaluation of the impact of intervention components. Some questions elicited responses that demonstrated positive cognitive and behavioural benefits, while others indicated no such benefit, but still motivated participants to reply.

An important finding from this study is the high levels of responses from the men. The study took advantage of the social rules for text messaging where the receipt of a message is likely to prompt a response [31]. Socially disadvantaged people are more likely to use text messaging and send and receive a higher number of text messages than people with higher education and income [32]. The intervention was tailored for men which could have encouraged feelings of camaraderie or cohesiveness which are found in single gender studies [33‐35].

Another factor which may have encouraged a high level of response is that the messages were constructed to be conversational, using unedited quotes from our feasibility study [18, 21]. These techniques were intended to impart the feeling of being part of a dialogue, using participants’ own language to describe experiences and situations familiar to them. The aim was to make the intervention appear authentic and credible. Dialogue moves away from delivering information in a passive way to learning through interactive engagement [36]. Taking part in a dialogue also allows individuals to learn about others’ values and interests [37]. Petralgia [38] reports that ‘contemporary learning theory suggests that learners’ perceptions of authenticity are critical because learning is embedded in our everyday experience of the world rather than in the world of formal information dissemination’. Authenticity ensures that the recipients understand not only cognitively but also emotionally how the information can relate to their lives.

Interactivity was encouraged by sending messages that prompted a response. Text message questions have been used in previous trials to promote interactivity [39, 40], but this study extended the use of this technique to reinforce key components of the behaviour change strategy and to monitor participants’ engagement in real time [18]. Responses to the questions on the key components of the behaviour change strategy indicated that participants responded to the intervention in the way that was intended.

A further notable finding is that the men frequently disclosed sensitive personal information. Self-disclosure can play a role in the development of relationships [41], and reciprocal disclosure has been shown to lead to more positive interpersonal outcomes, such as greater responsiveness, enjoyment and perceptions of being liked, than non-reciprocal disclosure [37, 42]. Internet surveys have shown that self-disclosure by the experimenter led to greater self-disclosure by participants [43]. The inclusion of anonymised quotes from participants in our feasibility study [21] may have encouraged the participants in this study to disclose personal details of their own.

The use of humour may have encouraged responses to texts and self-disclosure. It appeared to appeal to the participants, who responded with jokes and often self-deprecating humour. These matched the style of messaging in the intervention [43] because although they made jokes, many of the responses also had a serious side. Gray et al. suggest that laughter increases people’s willingness to disclose information about themselves [44].

At every step, the participants appeared to translate the messages to fit with their own lives. Instead of responding in general terms, the men gave specific, personal details. For example, discussion on the potential benefits of reducing drinking elicited responses on how their personal health could be improved or how their children could benefit. Similarly, for action planning, changes to be made were specific to the individuals’ drinking patterns.

This study has limitations. A key issue is the interpretation of non-response to text messages. Although 92% of the participants responded to the messages, some men replied on a few occasions only. This does not necessarily show a lack of engagement with the intervention. There may be many reasons why participants did not respond. Participants may not have had time to compose a response or may not have had credit on their phones. Some men may have reacted as intended but thought it was not necessary to respond. It is likely that the total number of responses underestimates the true level of engagement with the study. The nature of non-responders has had little attention in text message studies, but is well described in online discussion groups, where the term lurkers is use to describe non-responders [45]. Lurkers are a special group of web-site users who regularly log in to online communities but seldom post messages [46]. The literature suggests that 90% of participants do not post messages, 9% do so occasionally and only 1% do so regularly [47]. However, lurkers frequently read the online information and gain benefit from it [46, 48]. This issue needs to be addressed in text message interventions.

A further limitation is the brevity of individual text messages responses. Although some men gave extended explanations, most gave succinct responses. Further, the responses to these questions were often varied. For example, in response to the question on action planning, some men provided detailed action plans and others made statements such as ‘not buying any alcohol with my weekly shop!’ To cope with this brevity and diversity, a simple binary code (responded as: intended or not intended) was used.

This study has limited generalisability. Although these methods were acceptable to young to middle-aged socially disadvantaged men, the study provides no evidence that it would be attractive to other social groups. Nevertheless, the success of the study, in reaching and effectively engaging socially disadvantaged men in a behaviour change intervention, has implications for a range of public health interventions, not binge drinking alone.

This study has shown that the responses to text message interventions can identify whether the components of a behaviour change intervention achieve their intended outcomes. Responses to the questions on the key components of the HAPA model indicated that participants had understood the health messages that were being delivered and responded to them in the way that was intended. Participants appeared to translate the intervention to fit with their own lives and experiences. They were willing to disclose sensitive personal information indicating that the study had gained their trust. They responded with humour which showed they were at ease with the study processes. Text message interventions should include questions addressing key components of the behaviour change strategy to determine whether there is effective engagement with intervention components.