6 minute read
Is Population Health Management past its sell by date?
Simon Swift, MD Methods Analytics, and Richard Oakley, Director of Data Science & AI at Methods Analytics, discuss the importance of including citizens in the data conversation
There are many fashions around healthcare at the moment. Many of them focus on smart services to provide personalised care, whether they are clinical devices informing long term condition management or smart devices to provide health information for individuals. And as healthcare moves out of formal settings and into homes, we need to look at how we identify individual health risks and how best to deliver this information to both citizens and health care professionals.
There are hundreds of apps and services out there collecting data, but it’s often only used for the purposes of the app itself, rather than for a collective good. If the data is collated and aggregated then the bigger picture emerges, one that can be used for the benefit of individuals and systems.
To a certain extent, I’m talking about what we call population health, because all roads seem to lead there. In my world of data analytics, we can aggregate
the data from individuals to a group level (a geography or condition for example) so we can do the maths to understand factors that are predictive of good and bad outcomes,; then our data driven services can predict individual risk. Then at the individual level, identifying the predictive factors and advising, enabling, ‘activating’ people to engage in activities that will decrease the impact of a negative predictive factor and increase the impact of a positive predictive factor, changing the prediction. Understanding what interventions make a difference, and to whom – ‘impactibility’ is key here.
Accessing the data
Through the pandemic, the general public have become more used to the idea of providing their data for the greater good. Whether it is logging a positive result on a COVID app or checking a box on a form to allow your information to be used, there is a sense that we all know how important it is.
But again, this is where the problem can lie. How does the healthcare sector encourage more people to provide their data? The insurance market is testing this at the moment, offering devices such as Apple watches to incentivise individuals to monitor their health and also hand over their results. Some people are suspicious
Richard Oakley Director of Data Science & AI Methods Analytics
of the motives, so the conversation we need to have is about the purpose of using their data. Is that purpose benign or malignant? If an insurance company is using that data to improve its service to you, to improve your health, that might be considered a benign purposes. If they use it to understand your risk profile, including your response to suggested beneficial interventions (exercise more, eat better, stop smoking….) so they can price your insurance accurately, you may think it is benign if your fee goes down. If their purpose is to understand who might be really expensive and deny them cover because they’re looking at the bottom line, then that purpose can rightly be seen as malignant. This kind of cherry-picking behaviour breaks the concept of risk pooling that underpins insurance.
At Methods Analytics, we look at it from the data ethics angle. We have a fairly simple approach when anybody asks us to come and do something with data. First ask why? Why are you trying to do this? What is it going to achieve? What impact is it going to deliver to your organisation or the users? And the next question: is this technically feasible? Is the data available and can it be used in this manner? Once you’ve understood the intent and feasibility, the next question is the ethical one – “should you?” Is this specific purpose an appropriate use of the data? If we can’t answer those three questions satisfactorily, the work is not going to go anywhere.
So what does a ‘healthy’ population look like?
Data from users can improve engagement and activation, and help people become proactive about their care.
We are currently working with an ICS to help them understand how to
link together data across disparate NHS and local authority entities so that they can build person-centric data models in order to understand the population at risk, to identify which interventions will mitigate that risk and to enable better targeted services to improve the health of individuals.
We’re also working with a local authority looking at vulnerable children, understanding the risk to individual children and what interventions can be put in place to mitigate that risk and improve their chances of a successful placement or adoption. Again, it’s a population health approach by any definition, but not within the normal sphere of ‘healthcare’, more ‘care’.
What about bringing these data together, adding in more data around small areas such as police data education data, benefits data in order to understand vulnerable adults, risk of violence, antisocial behaviour and so on. I would say this is all the same - mathematically it s. The purpose is to improve peoples’ lives and experience day to day just as much as a healthcare purpose is.
It’s about improving an individual’s health and broader circumstance by using the data to create a personal risk profile, and identifying which mitigating actions can be taken to reduce the risk of a negative outcome. It’s driven by an understanding of data at group level: populations and geographies, then understanding that within any group there are cohorts of individuals with similar patterns, similar shapes, and they are likely to respond to similar types of interventions.
I’ve always talked about population health management as a specific healthcare commissioning concept, but thinking about it more carefully, it is a general concept about using data to predict risk and the impact of interventions for groups and individuals.
The implications are clear for anyone looking after a group of people, whether it is an insurer, central or local government, a health organisation, a buyer or provider of health care, even the police, all of whom are in the business of gauging risk. The basic principle is using data to predict a specific risk and understand the impact of interventions on that specific risk, then to engage people, to activate them, and adopt those interventions. It’s by using data in this manner we can create not just population health but broad improvement in life circumstance and social value.
Engaging the public
So everything boils down to a common concept: understanding factors about an individual and how they will respond to an intervention. This concept is not a special one that only works in the context of health. It works for understanding early years education, anti-social behaviour, anything where we can use data to predict.
In health, the changes are ones that are largely down to individual. Outside of health the changes may well not be, but interventions at a group or geography level such as creation of green space or restriction of the density of fast food shops.
Discussing these ideas and bringing people into the conversation about the use of data to improve our lives, individually and collectively is necessary if we are going to achieve meaningful change. The scope for benefit is enormous but we won’t achieve this if people are not bought into the way their data is a vital part of the approach.