Larisa Dina, Department of Psychology, Kings College London
Giving people agency within a health behaviour change intervention may be both desirable and effective. Smartphone apps make this approach possible for a large part of the population. If this works, however, it is also vital to understand how and why an app is effective. Writing in an early issue of Behavioural Public Policy, Ralph Hertwig, calls for ongoing evaluation of decision-making “boosts” (and other types of behavioural interventions). This blog describes the process evaluation of a recent trial run by UCL researchers evaluating how engagement with an alcohol reduction app, Drink Less, and participants’ behavioural characteristics underpin its effectiveness.
Setting the Scene
Let’s imagine you’re at a doctor’s office, and your doctor recommends a mobile app to help you reduce your alcohol consumption. For both you and your doctor to trust that this app will work and do what it’s supposed to do, it needs to have been proven as effective through research. This is typically done through a randomised controlled trial (or RCT).
In an RCT, participants are randomly assigned to two groups. One group gets the intervention of interest (in this case, the mobile app), while the other group does not, and participants are allocated to either group at random. By comparing the main outcome (alcohol consumed) between these two groups at the end, researchers can determine whether the app really helps reduce alcohol consumption, compared with the alternative.
Now, let’s assume this app is effective, meaning that people who use it drink less than those who don’t. This is great news! Delivering a health intervention via a mobile app can help reach many people who might not otherwise access treatment—especially since over 87% of adults in the UK own a smartphone. But understanding how and why an intervention works is just as important as it being effective. Having this understanding can inform recommendations of how to use the current intervention and can also help design better interventions in the future.
Importantly, for an app to be effective, people need to actually use it, or in other words, engage with it. This is where process evaluations come in. In our recent paper published in npj Digital Medicine, we wanted to see if the positive effects of the Drink Less app on alcohol consumption were related to how people used the app.
The Study
First, we conducted an RCT comparing usual digital care (the NHS alcohol advice webpage) with an alcohol reduction app called Drink Less. We found that the recommendation to use the Drink Less app led to a 2-unit bigger reduction in weekly alcohol consumption after six months. For reference, that is equivalent to a medium glass of wine or a pint of beer less each week.
Next, we wanted to understand how the Drink Less app worked, or its mechanisms of action. This was done through a process evaluation. The mechanisms of action we investigated in this process evaluation were:
- Engagement with the app. We asked participants in both groups if they used the recommended digital tool (yes/no). For those randomised to the Drink Less app, the app also tracked metrics like the number of sessions, minutes spent, screens viewed, and days used.
- Influence on behaviour. Participants’ behaviour was measured with questionnaires at the start and end of the study. For example, we asked all participants how strongly they had felt an urge to drink in the 24 hours prior to responding, how motivated they felt to drink less, about their self-regulatory behaviours (how difficult they found it to control their drinking) and about their self-monitoring behaviours (how often they kept track of the alcohol they consumed).
What We Found
We used a series of statistical techniques to identify the ‘middle steps’ that lead to the 2-unit weekly reduction in alcohol consumption. The following two were found to partially explain how the Drink Less app worked to reduce alcohol consumption:
- self-reported usage of the intervention (with 78% of participants self-reporting using the Drink Less app)
- tracking alcohol consumption
Take-Home Messages
Our main findings are that self-reported usage of the app and tracking alcohol consumption partially explained the Drink Less app’s beneficial effect at six months. Our results are consistent with a previous study in a large sample of Drink Less users (>19,000) which found that 85% of pages accessed in the app happened in the Self-monitoring & Feedback module of the app. This indicates that this is how users primarily engage with the app. They also align with recent summaries of the literature which point out that tracking behaviour is linked to higher user engagement and behaviour change.
In summary, following the recommendation to use the intervention and self-monitoring behaviour (tracking one’s own behaviour) appear to be important mechanisms of action of the Drink Less app.
The Drink Less app is currently available on the Apple App Store, but not yet for Android devices.
