There’s a lot to like about text messaging (Short Message Service: SMS) interventions for behaviour change: they can deliver cost-effective, brief, real-time and tailored messages at moments when individuals need them most. They reduce time demands on both the individual and health care practitioners, maintain the privacy of the individual and what’s more, given that the majority of the world’s population own a mobile phone, text messaging interventions can be delivered at a global scale.
Given these advantages, there’s been a great deal of research on the effectiveness of SMS interventions for health behaviours, finding mixed results. Last month, I blogged about a meta-analysis which found weak evidence for the efficacy of SMS interventions for smoking cessation. However, smoking cessation is a complex health behaviour and recent reviews have found that SMS interventions are more effective for more simple health behaviours such as medicine adherence and attending medical appointments.
Another recent meta-analysis by Orr and King (2015) published in Health Psychology Review was the first to examine the overall effectiveness of SMS interventions to enhance healthy behaviour, rather than focus on any one health behaviour (i.e. smoking cessation). Similar to the meta-analysis I blogged about last month, the authors also aimed to identify which SMS features have the biggest impact on intervention effectiveness.
The authors searched for randomised controlled trials (RCTs) which compared SMS interventions targeting health behaviour change to a non-SMS control, which did not attempt to change behaviour (i.e. RCTs which compared two active treatments were not included).
The authors examined the influence of six moderators on the effectiveness of SMS interventions:
- SMS dose (i.e. the frequency of the messages: multiple per day, weekly, once only etc);
- SMS message tailoring (i.e. standardised, tailored, personalised);
- SMS directionality between researcher and participant (i.e. one-way, two-way);
- Category of health behaviour targeted (i.e. unhealthy behaviour modification, chronic disease management, medication adherence, appointment attendance, disease or pregnancy preventive behaviours);
- Complexity of these behaviours (i.e. complex: chronic disease management, disease-related medication adherence, unhealthy behaviour modification; simple: appointment attendance, non-disease-related medication adherence, preventive behaviour);
- Participants’ mean age.
Thirty eight studies met the criteria for inclusion in this meta-analysis.
The meta-analysis found that there was an overall (pooled) positive effect of SMS interventions on healthy behaviour (g = 0.291, 95% CI = 0.219 to 0.363, p < 0.001). The heterogeneity between studies was low (I2 = 38.619, p = 0.009).
Planned sub-group analyses explored the impact of the six moderators on health behaviour change. There was little evidence that any of the six moderators impacted SMS intervention efficacy. However, when the authors regrouped the studies (i.e. unplanned analyses) for the tailoring, dose and complexity moderators, only SMS dose was found to impact SMS efficacy, with studies using multiple messages per day being more effective than those with reduced frequency.
The quality of the evidence in the 38 studies was judged to be relatively high, with all but three of the studies assessed as having high to moderate methodological quality. However, the authors judged that the risk of incomplete data was high in 40% of studies, despite efforts to contact the authors of the original studies for further information.
Further analyses of the data found that publication bias was not a threat to the validity of the estimated effect of SMS interventions for healthy behaviour.
The majority of included studies relied on self-report outcome measures, rather than actual behaviour, which is likely to have increased the observed effects of the studies. Future studies should use objective outcome measures.
The observation that more frequent text messages are more effective than less frequent messages was only found after regrouping the studies and running multiple unplanned comparisons between groups. This finding should therefore be treated with caution.
The authors only included studies which compared SMS intervention to no intervention at all. These findings therefore tell us nothing about how SMS interventions compare to other established interventions, such as verbal or other written reminders and messages.
Very few of the studies included in this meta-analysis were grounded in any health behaviour theory. The authors suggest that future research should examine the impact of established theoretical components on health behaviour change outcomes.
Using strict inclusion criteria for studies, this meta-analysis found that SMS interventions have a positive, albeit small, effect on healthy behaviour change. There was little evidence that moderators such as tailoring, directionality, health behaviour category or complexity or participant age influence efficacy. There was some evidence that higher SMS dose might increase efficacy.
These findings echo those in a recent meta-analysis of studies exploring the effectiveness of SMS interventions for smoking cessation where no moderator was found to be more effective than any other at increasing quit success and only a small positive effect of SMS intervention was observed.
Although this and previous meta-analyses have found only modest benefits of SMS interventions over control, given the low cost of delivery of SMS interventions and the potential to target large numbers of individuals, the public health benefits are still considerable and future research should continue to examine the efficacy of these interventions.
Orr JA, King RJ. (2015) Mobile phone SMS messages can enhance healthy behaviour: a meta-analysis of randomised controlled trials. Health psychology review (just-accepted), 1-36.
Maynard O. (2015) SMS texting to quit smoking: a meta-analysis of text messaging interventions for smoking cessation. The Mental Elf, 26 Aug 2015.