Behaviour change. It may be one of the most common buzzwords in digital health right now, but how do we actually change behaviour through digital interventions? This is a question we’re often asked by clients looking to improve medication adherence rates. Luckily, I had the opportunity to brush up on my skills at UCL’s Centre for Behaviour Change Conference earlier this year, and I’ll let you into a few secrets.
The future of patient monitoring
It has long been recognised that medication adherence rates are low, with the World Health Organization (WHO) quoting as little as 50% adherence to long-term therapies in developed countries. Couple this with a move towards evidence-based treatment methods, a growing focus on preventative medicine, increasing rates of access to technology and more technology capabilities, and suddenly behaviour change via digital interventions becomes both more important and more plausible than ever before.
If we take the case of asthma medication, there are already some examples of successful behaviour change interventions. Propeller Health report an increase in adherence rates from 21% to 47% with their sensor and app, while Adherium report a 59% increase in controller medication adherence with theirs. These examples are unfortunately fewer and less impactful than one might expect, given the potential benefit to users and profits to companies, and this is primarily due to human behaviour being incredibly complex and difficult to change. So, what guidance is there for the development of successful behaviour change interventions?
What does the research say?
Firstly, and fundamentally, we must have a clear understanding of the existing user behaviour, unsurprisingly making user-centred design one of the biggest themes of the conference. As a Human Factors Engineer this was music to my ears, but it’s evident that successful implementation requires user-centred design at the deepest level, beyond usability, safety and efficacy. We must comprehend the subconscious reasoning which underlies the existing behaviour, as well as the barriers to the goal behaviour for all of the intended users of the product.
However, this may be something that not even these users understand, let alone can articulate concisely. Would you be able to explain, concisely and honestly, why you don’t eat more vegetables? Johnson & Johnson’s Behavioural Science Team provide a fascinating example of this, identifying seven phenotypes in patient adherence which enable them to develop effective interventions accordingly, to suit each one.
Secondly, digital interventions must be both tailored to the individual and adaptive over time. Leeds University’s research on behaviour change in cancer prevention demonstrates how some people were open to behaviour change interventions and nudges at different times of the day, or in different situations. Clearly it’s fundamental to understand the role of variables, such as time of day, on the effectiveness of an intervention to each individual and to use this to make interventions that are flexible and adaptive.
Digital interventions must also address the needs of all stakeholders. Many different people play a role in the success of the intervention, which Public Health England’s Behavioural Insights Team discuss in depth. They talk candidly about what they, the NHS and NICE require from digital health interventions. For example, they expect to see a clear description of the intended outcome, evidence for the effectiveness of the intervention, consideration of data security, as well as a clear business case. NHSX (the NHS Digital Roadmap and the development of the NHS app) is indicative of the NHS’s support for digital health interventions, particularly when it comes to preventative healthcare. Additionally, numerous presenters emphasised the need to prevent unnecessary clinical burden on healthcare professionals, avoid them taking on the role of tech support, and stressing the importance of their buy-in for sustained success.
So, how can we translate this research into practice?
Thankfully, progress has been made to identify the underlying causes of poor medication adherence, particularly for inhaled medications. For example, the WHO’s Adherence Report identifies causes for nonadherence to inhaled corticosteroid therapy as:
- Erratic nonadherence - includes forgetfulness due to a busy and changing schedule/not prioritising asthma management;
- Unwitting nonadherence - due to not fully understanding the treatment regimen; and
- Intelligent nonadherence - due to the fear of side effects or the cessation of symptoms.
We’re not starting from scratch in this field and this learning should be built upon when developing interventions.
Additionally, the field of behavioural science provides several theories and frameworks. Possibly the most widely recognised example is the COM-B Model which provides a taxonomy for characterising behaviour change interventions. There are also theories and ideas from seminal authors such as Thaler and Sunstein’s Nudge and, further afield, from the Samaritans’ Small Talk Saves Lives campaign which combines the behaviour change techniques of role modelling, enablement and persuasion into a multimedia campaign which is saving lives.
Models from the field of behaviour science should be implemented flexibly to complement user-centred design methods. For example, the COM-B Model has been adapted into a larger and more comprehensive model by The INHERIT Project to change behaviour with regards to the environment and sustainability. It is also being used in conjunction with customer journey maps by companies such as MadPow in the areas of disease prevention and condition management. With regards to medication adherence, we could call upon the many well-established user-centred design techniques already used within the medical field, such as contextual inquiry, personas and user journeys, in combination with these models to really understand our users’ behaviour.
Excitingly, recent years have seen both improved access to technology and increasing technology capabilities. This provides an opportunity to utilise technology not only to provide digital interventions but to develop them too. It’s now much easier to conduct longitudinal studies of interactions with apps and health behaviours in order to understand users over a longer period. For example, we can remotely monitor interactions with an app screen, daily routines with GPS and movement trackers, and other health activities such as sleep and exercise, to gain a fuller picture of the end user. Combined with an iterative design process, this could lead to more effective development and refinement of concepts to ensure they meet user needs in a tailored and adaptive way.
UCL’s Centre for Behaviour Change Conference gave me inspiration and new insights into developing digital interventions for behaviour change. It’s a large but not an insurmountable challenge, and it has the potential to benefit us all. To develop successful interventions, we must gain a deep insight and understanding into user behaviour, utilise existing theories, models, frameworks and technology, and consider all stakeholder needs. Although the challenge is great, what’s clear is that behaviour change is one buzzword which isn’t going away any time soon.