At HTN Now: April, were joined by Dr Gemma Donovan, head of behavioural insights at Generated Health, along with Dr Sabine Van Der Veer, senior lecturer in health informatics from the University of Manchester, for a discussion on the potential and pitfalls of digital health and behaviour change. In particular, Gemma focused on behaviour change for the Core20PLUS5 framework, the NHS England initiative aiming to drive down health inequalities by defining a target population and identifying five focus areas requiring accelerated improvement.
What is a behaviour?
To begin, Gemma defined what a ‘behaviour’ is, from a behavioural science perspective. “It’s anything that somebody does in response to internal or external events,” she said. “An internal event is where we think about something that we want to do and then we do it, whilst an external event is where we do something that is prompted by the external environment. In terms of health behaviours, we’re generally thinking about things that people are choosing to do or choosing not to do which may affect their health. Importantly, behaviours are observable – we can watch somebody perform a behaviour.”
Behavioural science is not about controlling what somebody thinks, she added, but rather to influence the action that they take.
Looking at behaviours that could be important for somebody’s health, Gemma shared a few examples. Self-monitoring involves a series of behaviours, she said; for example, using a device to track blood pressure and filling the results in. Taking medication is another behaviour; so is the action of requesting help.
It can help to think about behaviours through a model, Gemma continued, such as the Capability, Opportunity and Motivation model for Behaviour (COM-B), available to view at 07:35. “This is part of the Behaviour Change Wheel that was developed by Susan Michie at UCL. There are three different components; capability, either psychological or physical; motivation, reflective or automatic; and opportunity, social or physical.”
Gemma pointed out that on the model, both capability and opportunity feed into motivation. “If somebody lacks capability or they have enhanced capability, that will affect their motivation to engage in a particular behaviour,” she explained. “Opportunity will similarly affect this.”
Gemma also noted that all three components are bi-directionally linked to the performance of the behaviour itself. “If you can get somebody to initiate a behavioural change, that alters their capability, motivation and opportunity,” she said. “We can get a positive loop going sometimes with behavioural change – the more we encourage people to perform the behaviour, the more capable and more motivated they will feel, and the more they will feel that they have the opportunity to perform it.”
Behaviours and Core20PLUS5
Next, Gemma examined the components of the Core20PLUS5 framework with a view to what they might mean for behaviours.
Firstly, she provided a refresher on the framework, focusing firstly on Core20. “The core 20 are the most deprived 20 percent of the national population, as identified by the national Index of Multiple Deprivation. The domains of deprivation within this are income deprivation; employment deprivation; education, skills and training deprivation; health deprivation and disability; crime; barriers to housing and services; and living environment deprivation.”
“There’s a range of things, when you look deeper into Core20PLUS5, which could be potentially affecting people’s ability to engage in health behaviours,” Gemma said.
The PLUS part of the framework is recommended by NHS England to be defined at a local level, she continued, though they have identified some populations which they would expect to see in most areas, including ethnic minority communities; people with learning disabilities; people with autism; people with multiple long-term health conditions; groups with protected characteristics; coastal communities; and groups experiencing social exclusion, such as people experiencing homelessness, vulnerable migrants and sex workers.
The 5 part of the framework, meanwhile, focuses on five clinical areas of focus which require accelerated improvement: maternity, severe mental illness, chronic respiratory illness, early cancer diagnosis and hypertension.
“All of these areas can involve a range of behaviours; there are a lot of self-monitoring behaviours around chronic respiratory illness, and early cancer diagnosis involves a lot of detecting of symptoms. Blood pressure monitoring, for example, might be relevant for pregnancy-induced hypertension as well as hypertension case finding, and it’s also helpful in supporting clinicians to make decisions about treatment around hypertension management.”
Moving on to explore how this ties into the COM-B model, Gemma shared some example questions to detect behavioural influences from ‘The Behaviour Change Wheel: A Guide to Designing Interventions’ by S. Michie, L. Atkins and R. West. “These questions are designed to tease out the kind of things that might be getting in the way of people’s ability to change their behaviours,” Gemma explained. “They’re helpful to ask when thinking about digital health interventions, to help us understand people’s actions.”
The authors provide a range of prompts: “To change my behaviour, I would have to: know more about why it is important, know more about how to do it, overcome physical limitations, overcome mental obstacles, have more money, have it more easily accessible, have more people around me doing it, have more support from others, feel that I want to do it, feel that I need to do it, or believe that it would be a good thing to do.”
The questions, therefore, link into each of these prompts; are people not changing their behaviours because of physical limitations, mental obstacles, lack of money, lack of access?
“As you can see, there may well be things in the Core20PLUS5 populations where more barriers might exist,” Gemma commented.
Digital health equity
Sabine took over at this point to discuss digital health equity.
Firstly, Sabine clarified the difference between health equity – “the absence of unfair and avoidable or remediable differences in health among population groups” – and digital health equity: equity in design of digital health solutions, equitable access to digital healthcare, and equitable outcomes from digital healthcare and/or equitable experiences with it.
Sabine moved on to examine the digital determinants of health, sharing a framework illustrating the domains of influence over someone’s life in comparison to levels of influence. Domains of influence include the digital environment along with other influences such as biology or the sociocultural environment, whilst levels of influence range from individual to interpersonal to community to societal. The graph can be viewed at 22:12.
“It’s often about the individual level,” Sabine commented. “Do people have the skills, the knowledge and the confidence to use digital technology in a way that allows them to benefit? Not everyone has access to a smartphone, or maybe some people do have access but they don’t have an up-to-date version that allows them to use certain apps, for example, or they don’t have mobile data.
“At an interpersonal level, you see the influence of the clinician as the gatekeeper, giving people access to digital services. I see that factor quite a lot in my own research. That ties into implicit tech bias, which is another interpersonal influence; the clinician makes a judgement call on whether the person in front of them is willing and able to use digital services. There’s also interdependence, related to technology access as the individual level; some people might have access as a household to a device but share it among themselves, which can create barriers for using digital services.”
At community level, Sabine noted the importance of infrastructure; some communities, for example, may not have access to broadband or high-speed internet which means that they cannot have a virtual consultation with their clinician. “People living in an urban area might have a community centre around the corner that provides digital skills training,” Sabine said, “or they might have a technology walk-in clinic. People in more remote areas tend to have less access to that kind of infrastructure.” Community attitudes can also be significant, she added; if there’s distrust towards technology in the community, then they will be less likely to engage.
Finally, at societal level, Sabine said: “This is about how we organise digital technology as a society. Are there nationally-accepted design standards for developing accessible technology, to make sure that people with sight or hearing impairments or dexterity issues can still use them? Algorithmic bias is another factor; we know that people from certain communities are not represented in the datasets that we use to create algorithms, and we use those algorithms to help us make decisions or predict certain risks. If those people are not represented in the data in the first place, it’s very likely that the algorithm won’t work as well for them.”
So what is the link between these digital determinants of health and Core20PLUS5? Sabine displayed each of the domains the Core20PLUS5 framework and ran through the impact the domains can have on digital determinants.
Income and employment deprivation can affect technology access to devices and data, along with interdependence (sharing devices).
Education, training and skills deprivation can impact digital literacy, digital self-efficacy and attitudes to use.
Health deprivation and disability can affect the patient-tech-clinician relationship, healthcare infrastructure (including lack of integration of digital services) and design standards (with regards to accessibility in particular).
Barriers to housing and services can impact community infrastructure (such as internet) and community support (for example, accessing a local service for assistance).
Living environment deprivation and risk of crime can have an affect on people’s ability to access a private, safe space, and can lead to lack of representation in data sets through algorithmic bias.
“Gemma already explained that having certain deprivation characteristics means that you have additional barriers to behaviour change,” Sabine said. “This shows that if you are trying to implement digital behaviour change interventions, you have to take care not to create an additional barrier, because people from those groups have digital determinants of health that can make it even harder for them to engage with the technology effectively.”
Managing the impact of digital health equity
Next, Sabine shared a multi-stakeholder process to identify potential impacts of digital interventions and how to manage them called the Health Equity Impact Assessment.
“It’s not a checklist, although you could use it in that way,” Sabine explained. “The most benefit comes from going through the list together with stakeholders – the end users of your digital health interventions, but also the people developing it, the healthcare professionals who will be offering it to people, experts in the area. Put them together in a room and go through these steps.”
The process includes five steps:
- Scoping: consider social and digital determinants of health, start from initial testing results and published research and reports
- Potential impacts: explore positive and negative unintended impacts
- Mitigation: consider strategies to tackle negative and harness positive unintended impacts
- Monitoring: create a plan for the evaluation of the effect of mitigation strategies
- Dissemination: create a plan for sharing outputs and findings with stakeholders
Florence and behavioural change
Gemma moved on to discuss the role of Florence Intelligent Health Messaging, Generated Health’s platform, as an inclusive digital health tool for behavioural change.
“I don’t claim to have all the answers for how we tackle this,” Gemma remarked. “At Generated Health we’re also on a journey in terms of thinking about how best to do this – we’ve always been really interested in how we can be more inclusive. We’re purposefully an SMS-messaging solution because we feel that SMS is the most accessible form of digital health, and that comes with a couple of benefits from a behavioural change and inclusion point of view.” The SMS messaging delivery can also be adapted into other languages to try and improve inclusivity.
There is no cost for patients to use Florence, Gemma explained, as the cost of messages are recharged, which means there is no requirement for patients to have any credit on their phone. There is no need to access the internet and they can take part using any type of mobile phone handset. Standard adaptations on handsets which can be used to increase accessibility, such as increasing font size, do not affect how Florence is used.
“When we onboard our customers, we do a process of co-designing our content,” Gemma continued. “We have a completely adaptable solution. We use this a lot to try and improve our inclusivity. We’re used to working with particular populations in order to make sure that content delivered by Florence is suitable for that population. This means that we can adapt our content around the behavioural insights that we receive. When we work with our customers to understand their challenges, the likely behaviours that need to change, the best way to support patients to engage in that behavioural change, all of that can potentially be adapted to a very specific population if that is what is required.”
The project that Generated Health is currently working on with Sabine and the University of Manchester is taking that one step forwards, Gemma added, exploring how this co-design and personalisation can be developed further.
“We also make sure that our content follows clinical care processes,” she said, “and that it maps with what is happening in real life in the clinical setting. We hope that this means we’re less likely to exclude people on the way, in terms of who is being offered the technology, because it should be fitting around what is already there and hopefully enhancing existing care.”
An example of one of the ways in which Florence can support patients is the hypertension pathway; Gemma shared how the platform can assist patients in collecting blood pressure readings by sending them reminders, providing them with an easy way in which to record readings, and sending immediate feedback. This feedback could be instructions on what to do if a reading falls within a certain range, or it could be simple confirmation that all is well.
“Florence can respond to patients supplying word-based messages as well as figures,” Gemma explained. “For example, a patient can tell Florence about a symptom they are experiencing and Florence will record this and generate further action as needed. She can also pick up on trends to help identify whether patients are getting better, worse or staying the same, therefore helping clinicians to identify when an intervention is needed.”
Although SMS is very simple and easy to use for patients, the data is transformed into meaningful information on a data dashboard from the clinician perspective, summarised for ease of access.
Florence and the University of Manchester
Sabine shared some more information about the collaborative project taking place between Florence and the University of Manchester, which aims to explore how Florence can support hypertension self-management in people from South-Asian backgrounds.
“We’re going to do a focus group using the process I described earlier; so we’ll invite a group of people from the different stakeholder groups, involving lots of South Asian people with hypertension, and we’ll go through the process to work out the potential positive and negative impacts, and the strategies that we can think about for addressing them,” Sabine explained.
A second focus group will then hone in on the findings from the first group, with the aim of prioritising the actions that are most feasible whilst having the highest impact.
After that, the project will pick out the high-priority strategies selected through these groups and start co-designing them with stakeholders.
An expected outcome from this project might include altering the language of Florence messaging for this patient population, Sabine said. “But I think that we’ve learned so far is that understanding and reading English is often not the problem. It’s the way that you frame certain things which can mean different things to different populations. It’s especially important when you think about behaviour change. For example, if we’re trying to increase people’s physical activity, you might be working with a population who go to the mosque several times a day. Why not develop some culturally-adapted messaging prompts which take this into account?
“The other thing we’re hoping to get from this is to learn whether there’s anything to be done around how Florence is offered to people,” Sabine concluded. “We’re working with clinicians to think about their implicit biases, and how that can be tackled so it can be offered to everybody.”
Many thanks to Gemma and Sabine for sharing their time and experiences.
