Use of internal data in insurance
This briefing note gives a high-level overview of how insurers can make better use of internal data to gain insight and drive competitive advantage.
Every country has its own, unique healthcare system, but within these disparate systems there are shared approaches that cross borders and languages, and we explore some of these similarities in this episode of Critical Point.
Announcer: This podcast is intended solely for educational purposes and presents information of a general nature. It is not intended to guide or determine any specific individual situation and persons should consult qualified professionals before taking specific action. The views expressed in this podcast are those of the speakers and not those of Milliman.
Jeremy Engdahl-Johnson: Hello, and welcome to Critical Point, a podcast brought to you by Milliman. My name's Jeremy Engdahl-Johnson, I'm with Milliman's Media Relations Team, and I'll be your host today. I'm excited to be joined today from our annual international internal conference. I'm joined today by Joanne Buckle out of our London office. Hi, Jo.
Joanne Buckle: Hi, Jeremy.
Jeremy Engdahl-Johnson: And also Kevin Manning, out of our Dublin Office.
Kevin Manning: Hi Jeremy.
Jeremy Engdahl-Johnson: So, Milliman works in a lot of different countries—we’ve got a unique global healthcare perspective. You both are taking advantage of that expertise and have put together a series of research papers looking at the commonalities in different healthcare systems around the world. I’d like to talk about those three papers today. The three topics are population health management, risk equalisation, and wellness programs. So having teed that up, Kevin, can you tell us a little about this research project and what you guys are trying to accomplish.
Kevin Manning: Thanks, Jeremy yeah, I think one of the challenges in healthcare and health insurance is that really—every country has its own health system, and each country's system is different and has its own peculiarities and idiosyncrasies. And often it has its own terminology, language, acronyms, and you name it. One of the challenges that brings is it can be very difficult for insights, thought leadership, and articles to be portable in other territories. One of the things that we wanted to do here was try and leverage some of that extensive expertise and experience we have in the US and elsewhere, but to try to do it in a way that's a little more stepping back from the detail of the specifics of the system and looking a little more at the issues that are common to different systems, the sorts of bigger-ticket problems that people are trying to tackle and address. And if you can focus on some of those issues, they are very common right across lots of different health systems, and you can therefore maybe get some insights that you can use in lots of markets or that could be helpful to people working in lots of different countries.
Jeremy Engdahl-Johnson: And what are some of those similarities? What kind of things are we seeing that are in common across different health systems?
Kevin Manning: Ultimately, people are really trying to tackle some of the same problems. They typically have limits on resources. They've got a budget. They've got some finite resource that they can use and they're trying to get the best outcomes, whether that's best outcomes in terms of patient experience, in terms of improving healthcare outcomes, or in terms of working conditions for the people delivering the health services. A lot of those very big picture kind of goals, which then translate into how incentives are built into systems to remunerate people, or how you go about investing in particular treatments or particular technologies. A lot of those issues are the sorts of things that are being tackled by health professionals and health systems in lots of different countries.
Our focus was to try to pick some of those types of issues and really focus on them.
Jeremy Engdahl-Johnson: How about you, Jo? What are the commonalities that you see? You have worked in some very different sorts of markets.
Joanne Buckle: Yeah, and I would say, typically, any new market that I go into—and I've worked in many different health markets in different parts of the world—I'm normally faced with somebody telling me that the issues that they have are unique to that market. My experience has been that that's absolutely not the case: that almost all of those big challenges are challenges across all the different health markets that we look at. They manifest themselves in different ways, but all health systems are struggling with how to get the best health outcomes and the best quality for their populations; how to deal with the new drugs and technologies that are coming on the market, and determining whether or not they are effective uses of funds in, as Kevin said, a resource-constrained environment. They're trying to work out how to get optimal productivity out of the health resources that they have in terms of people and technology. They're trying to work out how to align incentives across the system, so that health professionals are paid for outcomes rather than activity. So all of these things tend to be common in whatever health market you're working in. And, as well, some of those very patient-centered things around how to ensure that young people are starting their lives on a healthy basis and putting them on a healthy track for life and that they're looking after their own health. And things like end-of-life care, making sure that we're providing end-of-life care in a suitable, dignified manner and that it's a good cost-effective use of resources and allows people choices at difficult points in their lives. I think those issues are common across most systems and really don't depend on how any particular system is set up and paid for, and they don't really depend on whether the population is young, aging, or particularly skewed to one demographic or another, really poor or mature. Those issues are common across lots and lots of different systems.
Jeremy Engdahl-Johnson: All right, so let's dive into some of the topics you are exploring. One thing that actuaries are good at is looking at things on a population level. What kind of lessons have we learned worldwide about population management?
Joanne Buckle: I think the lesson that jumps out at me most often is that most systems at the moment are talking about population health, and they have a population health mantra. They have certain tactics that they're following, certain tactical things they're doing in the population. But the piece in the middle about turning that mantra into tactics via some kind of strategy seems to be missing. One of the things that we have tried to do is to set out some basic principles around implementing a population health management program. You need to be very clear about what your policy goals actually are. And certainly, in lots of systems I work in, that sounds really basic. But it is actually really difficult to do, because you have to go from saying, "We want to improve population health," as your mantra, to setting some really specific and measurable goals around what that actually means in practice. Thinking about some of the trade-offs, does that mean we want to improve population health in aggregate? Does it mean we want to reduce inequalities in outcomes of health? Does it mean some really specific things like reducing certain rates of chronic disease? Or certain rates of infection? Taking that sort of population health statement and turning it into a set of measurable goals is the bit that I see missing in a lot of the debate around population health management.
Kevin Manning: What we've tried to do in the paper is use some examples and case studies internationally that will help to underline those points that Jo is talking about. While some of this does sound quite basic, a lot of traditional actuarial control cycle ideas are very important—setting out goals, then designing and implementing a solution, and then the important phase of monitoring the experience, evaluating how that's going and feeding that experience back into your design so that you're constantly correcting, changing course, improving your system all the time, making sure that you're tracking the outcomes in a way that gives you actionable insights.
Jeremy Engdahl-Johnson: Any particular examples of programs that stood out?
Joanne Buckle: In the National Health Service (NHS) there is a fairly mature population health program in Manchester, which involves not just health but also education, adult social care services, housing, and a whole range of different aspects to health and what we call social determinants of health. That has been successful, I think, in setting out policy goals and population segmentation quite clearly. It indicates they want to focus on certain parts of the population, on certain outcomes. They have set themselves out a nice framework in a way that other parts of the NHS I think haven't managed to do as succinctly and now as successfully. It's too early to know whether or not it will be successful in achieving their aims. But at least they have, as Kevin said, started the control cycle. They have at least set out what they're trying to do. They've got something to measure against.
Jeremy Engdahl-Johnson: You mentioned the social determinants of health. It seems like the NHS would be well positioned to use that sort of information effectively since it's got kind of the entire population under its purview. I think some systems have struggled with how to deploy that sort of information and where it can be used, right?
Joanne Buckle: One of the great ironies of the NHS is it has the entire population, cradle to grave. It, however, does not have an integrated data store with even all of the health services that that population touches, let alone all of the other types of services that impact upon health. So some parts of the country have made quite a lot of progress in putting together that integrated data store, but it doesn't exist at a national level. We're still very reliant on local initiatives to bring together those data sources and draw inferences from them as much richer sources of data. But we are making progress. There has, over the last three to four years, been a recognition that that's the direction of travel we need to move in.
Sometimes as actuaries we get very frustrated because people ask us to do analytics when they haven't done the necessary investment in the data infrastructure to make those analytics possible. I think that still happens quite a lot in the NHS, and there are ambitions to do sophisticated population health analytics, but without necessarily the recognition that you need two or three years' worth of solid historical data to run those analytics on. And we're only in some cases starting upon that journey of putting that data into one place and linking it all together. Another thing is that bringing together data sources is very dependent on context. In some parts of the world there are obviously strict privacy laws against that, and in other places it's easier to bring the data together. I think what we've not done a good job of as actuaries and as other professionals is in articulating the power of having that data in one place and being able to draw inferences on it in terms of improving people's lives. And that's a conversation that needs to happen.
Kevin Manning: Just to maybe add to what Jo's talking about there. With population health management, the NHS is a great example, and some of the insights there, I think, are very useful. You can be led then to take a view that population health management is all about huge transformational, overarching systems, and that this sort of paper is only relevant to you if you're the Minister for Health, or somebody senior in the government department in a country somewhere. But the techniques, ideas, and considerations here apply really to any program that you're introducing to influence or manage the health of any population. Whether that's the population of the United Kingdom, or a small population for a health plan that you're responsible for, whatever that might be. population health management (PHM) types of initiatives on different scales. And some are successful and some not so successful.
We've seen an interesting one in Delaware where they introduced food prescriptions for people with issues around diabetes, and they thought, well, food security is a big challenge for people with diabetes. And rather than necessarily just prescribing medication, they used a system with a sort of mobile food bank that would come and give people access to staple goods and so on. It was a very nice idea, but it never really developed anywhere because they weren't gathering any data to understand how successful the initiative had been. There were only people who selected to enter the initiative and stayed in the initiative, and for those people there was already a selection bias. The clinicians weren't gathering anything other than survey data to try and understand how successful the initiative had been. I think there are lessons in this paper for anybody with a bright idea trying to influence the health of any particular population, big or small, to understand maybe how you might go about setting that up in a way that gives it the best chance of being successful.
Jeremy Engdahl-Johnson: It seems like not having the greatest data is an issue for many healthcare systems. Isn't that pretty universal across a lot of these situations?
Joanne Buckle: I think it is. A lot of data quality, rightly or wrongly, comes from how doctors and facilities are paid, and that obviously influences the granularity and the quality of the data that you get. Typically the data that's used for reimbursement is of a higher quality and accuracy than data that's not used for reimbursement. And I think there's a lesson there, particularly as we move to fewer fee-for-service (FFS) types of reimbursement strategies. Data quality needs to be protected, the granularity of data needs to be protected, and we need to come up with different contractual mechanisms and different ways to make sure that data quality is sufficient to give us the insights that we need going forward. And that's a big danger that I see in a lot of the movement away from fee-for-service. I think we all would agree that there are good reasons to move away from fee-for-service reimbursement, but that potential reduction in data quality is a big concern that we need to address.
Kevin Manning: Dealing with poor-quality data is a challenge in lots of countries in lots of different spheres, depending on what you're trying to do. It's one of the issues that's touched upon in another of the papers, the risk equalisation paper, which is again trying to take a practical approach to a challenge that lots of different countries and lots of different systems are facing, around how best to introduce, maintain, or update a risk equalisation scheme or risk equalisation system. One of the main challenges that lots of countries face is access to the right level of data, the most granular level of data. One of the things that paper tries to draw out is some lessons internationally for how you might deal with that situation if you don't have perfect data. We rarely have perfect data, so what sort of approaches can you take to deal with some of those types of issues?
Jeremy Engdahl-Johnson: Talk more about risk equalisation. How is it affecting populations around the world differently?
Kevin Manning: If we think about some of the bigger issues that health systems around the world are trying to tackle, one of the things that Jo mentioned earlier is around access to healthcare, equity. Risk equalisation is a way of protecting equity and access to healthcare for older and sicker people. It's also a really important way of understanding how you remunerate hospitals for the patients that they cover. There needs to be some fairness in that, otherwise it gives a sort of disincentive for hospitals to treat the sicker, more complicated people, because they're not going to be rewarded or remunerated appropriately for doing that. Risk adjustment plays a very important part there.
Joanne Buckle: It's a broad challenge across all health systems, in that risk equalisation or risk adjustment or predictive modeling, whatever you want to call it, really the essence of it is trying to work out what needs people have for health services. And what their state of health is, and therefore how that drives their needs for health services. The biggest confounding factor, or the biggest thing that makes those models difficult, is that often historical use of health services is driven more by the supply that's available in the marketplace and less by people's actual health status. So the biggest challenge of any risk equalisation system or any risk adjustment system is trying to disentangle those effects and work out how much of historical utilisation is driven by availability of health services, and how much is driven by true underlying health status. That challenge is common across countries, but has different impacts depending on the system. The NHS is a very supply-side-driven system. The budget basically exists to provide certain services and tends to be siloed into the suppliers of those services. Other health systems tend to be much more demand-driven, but still at their heart they have this issue that people use health services because they're there as much as because of their own health status.
Jeremy Engdahl-Johnson: Kevin, we’ve been talking a little about risk equalisation and I'm curious to hear what's been going on in your market. What's the Irish story of risk equalisation?
Kevin Manning: Risk equalisation has a slightly checkered history in Ireland. Risk equalisation tends to result in money changing hands somewhere, and often the people who are paying the money don't like paying it, and the people who are receiving the money think they should be receiving more, and that can lead to lots of challenges, including legal and political issues. The Irish approach to dealing with the legal challenges is an interesting one. We had a particularly unique sort of set of circumstances in Ireland that led to quite an extensive series of litigations through the Irish courts that ultimately ended up with the original risk equalisation scheme being thrown out by the Irish Supreme Court. It was back to the drawing board at that stage and the Irish government needed to introduce something. The need for risk equalisation hadn't gone away just because the scheme had been thrown out. So the Irish government went back to the drawing board and came up with a practical solution that involved money changing hands in a sort of a roundabout risk equalisation scheme that was initiated through the Irish taxation system. Part of the rationale for that, I think, was that the Irish taxation system was going to be more robust in terms of the legal challenges that could arise. It wasn't a perfect scheme by any stretch of the imagination, but it was a solution to a particular challenge that allowed risk equalisation payments to start to happen.
Joanne Buckle: Risk equalisation is a very difficult topic, and different countries have gone down different routes. Some have made very complex risk equalisation systems. Some have said, "Actually, given the current set of data that we have, let's do something pretty simplistic." At the moment, the Irish example is at the fairly simplistic end of the risk equalisation spectrum. But no doubt that will change over time, and every time there is a change in the way that the risk adjustment methodology works, there will be, as Kevin said, a winner and a loser. Trying to balance all of those different concerns at the same time is not to be trivialised. One of our jobs, really, as actuaries, is to advise on implications of changing the system, and to think about the incentives and the disincentives that we're creating in the system. And also to not create an environment where you are promoting behavior that's not necessarily desirable from a policy perspective. So an obvious answer to risk equalisation is to cherry-pick lives that get overcompensated for in the risk equalisation system, and try not to get lives that are undercompensated. That becomes an entire secondary market in guessing who is overcompensated and who is undercompensated. And that takes up a huge amount of time and energy, which you might argue is not particularly productive in the context of the health system as whole.
Kevin Manning: I think that's a hugely important point in terms of aligning incentives. The blunter the tool in terms of the risk equalisation system, clearly, the more obvious the opportunities for that type of cherry-picking behavior or for the wrong incentives to be built into a system, and for less than productive behaviors to be encouraged. It comes back to the challenge we talked about earlier around data. You know, you can build a very sharp and very agile risk equalisation system if you've got the perfect data to do that, but the less data you have, or the less granular your data, the harder it is to do that. And I guess that is one of the challenges of introducing a risk equalisation system. It may start at the simplistic end, but I think it's important then to build in the enablers to make it a more sophisticated system over time. And that might be a long play, but it may be about building the capacity to gather the data that you need to make more granular assessments of risk and to sharpen your system so that it isn't such a blunt instrument when it comes to trying to understand what the key drivers of risk are here.
Joanne Buckle: That agility point is actually really important, because obviously treatment patterns change over time. And if you take for example the historical cost of somebody who has type 2 diabetes versus somebody who doesn't have type 2 diabetes but is otherwise the same risk profile, there'll be a certain difference in cost, and that certain difference in cost will be driven by historical treatment patterns, maybe even at very specific areas of the country. Trying to roll that forward and say, "Does that historical disparity in cost hold true in two years' time for a country as a whole, or for a slightly different subpopulation, or when there's new drugs and treatments being rolled out the entire time?" You're forever chasing your tail, in effect, because you're always relying on the historical data and you're never taking a forward-looking approach to it.
Jeremy Engdahl-Johnson: So we're taking a global perspective, but given that I know a lot of American healthcare actuaries, I wanted to get both of your reactions to this question. What's the best thing and the worst thing that American healthcare has exported to the rest of the world?
Joanne Buckle: I think the best thing, that's probably the easiest one to answer, the best thing that it has exported is a whole range of different micro-experiments that we can learn from. I mean, one of the things that I often say to my non-US clients when they push back or express any skepticism about using lessons from the US is that the US is an amazing experimentation hotbed. So whatever you want to try anywhere else in the world, it's been done in the US at some point in the health system. You just have to find that use case and figure out the lessons from it. Because the US is not one system, because it is so fragmented, there are all sorts of little policy experiments going on all over the place and that's really helpful. If you can interpret that in a local context, that gives a huge number of really valuable lessons. So I think that that's great. I mean, it comes at a great cost to the US and US citizens, but I think, in terms of lessons for the rest of the world, that's really helpful. The worst thing that they've exported—I'm not sure I have a good answer to that.
Kevin Manning: I would struggle to sort of pinpoint the worst thing that the US has exported, but I guess it kind of links into the comment that Jo said about the best thing. I think the cost of healthcare in the US gives a kind of international jadedness when you start trying to talk about lessons learned from the US. You can find some really interesting stuff, and say, "Well, in the US they're doing this," and there's a very immediately dismissive, "Yes, but look at how much Americans spend on healthcare versus everybody else. Look at how the health outcomes that they get relative to the spend are so much worse than everywhere else." And because of the sort of the fragmentation in the US system and some of those inherent challenges that you guys have in the US, and maybe some of the inequities that are built into the system potentially, it becomes easy for somebody internationally to dismiss some of those really valuable insights and learnings and exports that Jo was talking about.
Jeremy Engdahl-Johnson: Has anybody worldwide figured out the overutilisation problem or is that just a chronic issue for all systems?
Joanne Buckle: I think it's a chronic issue for all systems. It depends on the incentive structure and it is different in different places around the world, but, yeah, it is a chronic problem. I don't think anybody's really got to the bottom of the old health needs versus health wants. And health wants is a perfectly valid thing in some societies. It depends, it's a real policy question, it's not really an actuarial question. It's a policy question as to how much in economic resources governments want to divert to fulfilling people's health wants rather than their strict health needs. But a lot of the countries that we work in, we have long discussion with them over—they want to set up, say, essential benefit packages, and what the word "essential" means is different in different countries, different cultures, and different contexts. So there is no actuarial answer to that. That's a policy question, but really our role in that is to highlight how much it costs to potentially offer different kinds of services and to fulfill those wants and to project out those potential costs over a long enough period that policy makers can make an informed decision over the coverages that they offer.
Kevin Manning: Yeah, I think that's a really important point. I don’t think it's appropriate for actuaries to come in and try to set health policy in any country. But if policy has been set, it needs to be set in a way that's informed by facts and evidence and information, and I think we can help with that modeling and provide that understanding. And I think also that if policy has been set in that way, helping to deliver that policy, helping to provide the tools and information to carry through on that, we have a very important role to play there.
Jeremy Engdahl-Johnson: OK, I know one of the other projects that you're working on is a global view of wellness programs, which are obviously pertinent to any kind of population health management. What are you seeing globally from wellness programs? And I always kind of brace myself whenever I'm asking actuaries about wellness programs, because they will cut to the quick in terms of pointing out what works and what doesn't. You know, the numbers tell the tale, right? But what are we seeing globally?
Joanne Buckle: Yeah, I think wellness programs are quite interesting, because they have evolved a lot over the last 10 years. The best ones that I see are really heavily evidence-based, which sounds very obvious, but actually most of the insurers and providers of wellness programs don't necessarily want to go back to first principles when they're setting up a wellness program and say, "Well, what does the evidence actually tell us?" Because that's a lot of work. You've got to go through the evidence, you've got to figure out, as Kevin said, your objectives, what population you're trying to target. You've got to then do your evidence review and work out what targeted interventions you need to put into your wellness program. You've got to work out whether or not it's cost-effective for you as a payer to put those interventions in. You've got to form third-party relationships often to provide the services and interventions that you need to offer a full program. You've got to follow good behavioral economics principles and behavioral psychology principles around engagement, which is still pretty undeveloped and immature in most wellness programs. So most wellness programs understand the principles around loss aversion, they understand the principles about people discounting future benefits ("I want to eat three donuts today, but I'm trying not to think about that effect on my health in 30 years' time"). They understand those basic principles, but we haven't yet seen the next generation of wellness programs where it's very targeted to you as an individual and your individual psychological profile.
So the example I always use is that if you look at what motivates me, as a person, versus what motivates my husband, to go for a run around the block, it's completely different motivations, and yet we would get offered the same wellness program. If you gave my husband five pounds, he would go and run five miles. But that's not going to get me off the sofa, because I'm just not interested. But if you sign me up for a race in 10 weeks' time, then I'm going to go out and run five miles. So it's trying to get to the bottom of those sorts of psychological differences and what motivates people and build those motivations into very personalised wellness programs. And we've not yet seen that next generation done well, I think.
Kevin Manning: I think that's a very interesting concept of personalised wellness programs, and that sort of next generation, which I don't think we have seen yet either. I'm wondering how technology fits in with that, Jo? Or is the ability of wearables and apps and apps on your phone, does that give you a capacity to target a little more accurately in some of those areas and develop something that's a little more personalised?
Joanne Buckle: I think so. I have seen some quite interesting work being done recently in terms of mental health and behavioral psychology apps and things which use both physiological and self-reported signs and symptoms to look at your mental state and your level of stress at different points in time, and then combine that with general demographic information to really make predictions about what's likely to motivate you at certain points in your life. Because, again, it's not going to be stable over time. What motivates you as a 20-year-old will be different from what motivates you as a 50-year-old. And what motivates you when you are particularly stressed, and there's a whole load of things happening in your life, will be different from what motivates you when you're going through a relatively stable period of your life. I've seen some interesting things looking at using wearable technology to try and really bring together all of those different aspects and move away from just the wearable that's tracking your heart rate and looking at a more sort of holistic view of you, as a person, and then engaging with you as a person on that level.
Kevin Manning: Of course, that type of holistic view brings in some of those concerns we talked about earlier around data and data privacy, and people naturally will have concerns around whether it's their employer or their insurer or whoever it might be having all of this information about them, this holistic information about their health and their mental health and the activities that they get up to and so on. Those concerns are always going to be challenges that wellness programs will have to try to address. The more tailored the programs become over time, the more data they're going to need to achieve that successfully and the bigger that some of those data challenges are going to be.
Jeremy Engdahl-Johnson: All right! Well, thank you both. This has been fascinating and I expect we'll be catching up in the future on this as we dive further into this project. Thank you, Jo. Thank you, Kevin. You've been listening to Critical Point, a podcast brought to you by Milliman. If you would like to subscribe to Critical Point, you'll find it on the Apple Store, Google Play, Stitcher, or wherever else you get your podcasts. We'll catch you next time.
Critical Point Episode 19: Lessons from healthcare systems around the world
Every country has its own, unique healthcare system, but within these disparate systems there are shared approaches that cross borders and languages, and we explore some of these similarities in this episode of Critical Point.