- Episode 5, Season 2
“The Data Knows” — Finding Answers To Drive Digital Health Innovation
DATAMASTERS Podcast SUBSCRIBE
Anthony sits down with Mike Alvarez, Director of Digital Services at Fuse by Cardinal Health, to talk about data innovation in the healthcare industry. They discuss leveraging data to help patients understand their best paths to care and better stick with treatments. Using data from points of care and external entities to help providers determine the best outcome and cost for value-based care. And, they tackle a popular question, “are dashboards dead?”
Mike Alvarez: It just feels like we spent the last 20 years extracting data, building warehouses, layering dashboards on top of them just to have the business turn around and say, “Hey, can you just give me ODBC access to that thing, because I want to do my own thing?”
Anthony Deighton: Welcome back to DataMasters. In this episode, we’re going to get some insider knowledge about how data is being used in healthcare to drive innovation, better patient experiences and outcomes. Our guest this week is Mike Alvarez. Mike is the head of the newly formed digital services team at the innovation organization Fuse by Cardinal Health.
As head of Fuse digital services, mike focuses on building a strong commercial foundation as well as sparking innovation through new products and business models. Fuse’s digital services group serves as a force multiplier to the enterprise with its data science center of excellence, commercial architecture, and new product research and development. The last several years of Mike’s career have been focused on leading change and driving strategy across Cardinal Health’s expansive medical and pharmaceutical segments with contributions to large modernization programs, major acquisitions, and reinventing Cardinal Health’s approach to data and analytics.
Aside from being a seasoned executive, Mike is also an avid podcast fan. And we’ll get some of his recommendations later in the episode. Mike, welcome to the podcast.
Mike Alvarez: Thank you, Anthony. First off, I want to thank you for the invitation and to have this conversation with DataMasters on this podcast. Since learning about DataMasters, I’ve listened to all of your episodes. I’m a big fan. You’ve had some amazing talent on this podcast. So congratulations on your success.
Anthony: Well, thanks. And coming from a podcast, it’s nice to hear we’ve made your short list. So that’s great. And let’s start from with you. Listeners may not realize how big Cardinal Health is. In a funny way, I’m sure everyone listening has had some interaction with Cardinal, but probably didn’t realize it. So maybe you could tell us a little bit about the company and your role within it.
Mike Alvarez: Yeah. So sometimes we call Cardinal Health the biggest company you’d never heard of. We’re right in the center of healthcare. Really I think back to DataMasters maybe it’s one of the better podcasts I had heard until recently. But let me share a little bit about what Cardinal Health does and then I can talk about kind of our size and then maybe talk a little bit about me.
But I like to think of Cardinal Health’s capabilities in kind of three buckets. So one, we’re a distributor of pharmaceuticals. So we don’t make drugs, but we ensure that 90% of our nation’s pharmacies and hospitals have this lifesaving medication they need. Secondly, we are a global manufacturer and a distributor rather of medical and laboratory products. So I think one thing that COVID made everybody understand was PPE, right? So the protective equipment. So we manufacture some of that. We also distribute it from other manufacturers. So our medical division is a critical aspect of our supply chain for everything from PPE, the stents, to continuous glucose monitors.
And finally, the third bucket is around digital health and digital healthcare products. So this is the area I’m really mostly excited about. My role, as you mentioned in the intro, some part of our Fuse organization really focused on product development for commercial technologies. And we like to say we’re on a mission to reimagine healthcare delivery.
Anthony: And I really love this idea that Cardinal’s a company that everybody knows but no one realizes that they know, because it really is critical to our daily lives, making sure that the drugs and healthcare products we really need are at our fingertips. And healthcare is a really interesting space. But if I step back a little bit from healthcare, something I’m sure our listeners can relate to is this idea that there are great insights trapped inside your data, and you sort of know the answers in there but you can’t get at it.
And that’s something that you’ve talked a lot about. I’d love you to sort of share maybe how do you think about finding these insights trapped in your data. And maybe there’s an example or two you can share of something that’s happened in the organization that reveals that.
Mike Alvarez: Yeah. It’s such a great question because you’re right. I mean, it just seems easy, right? So I had a question, I get the data, I add it together, I get my answer. But as we know, anybody in the data profession knows, it’s not that easy. I’ve always held this belief that the data knows. As in the data knows where I’m sitting right now because of my phone, right? The data knows how many trips I made to the grocery store this week because of my phone, my loyalty card, my credit card. And there could be other sources of data that are not structured like traffic cameras and that kind of stuff. So there’s plenty of data to answer the questions, but to the other part of your question, how do we get at that data, right?
So one of the challenges I think we all have in common is how do you identify the data, how do you tag it, and then most importantly, how do you index the data and align it and link it. And then when you get to healthcare space, it gets even more complicated because you start layering in a whole new set of constraints, like HIPAA. A lot of us deal with GDPR, but that kind of constrains our data. One of the really unique aspects of the data we deal with additionally is we have these contractual agreements with clinics and pharmacies, right? So we’re leveraging that data to build our insights, but it’s not our data, right? It’s not internal system data. So we have these contractual agreements or these BAAs that really kind of define how we can use that data.
So you have to take all that into account when you’re trying to build it or get to an answer. One of the mental pictures I have of this is kind of like a person, like we all see on TV, a thief cracking a safe, right? So they’re toiling at the safe and they break into the safe to obtain the treasure inside. I feel like that’s what our analysts go through in order to solve a problem because they’re… But they have to do this across several domains, right? So they have to crack a safe, get what’s inside, figure out what it is, go to the next safe. And it’s kind of this Daisy chaining of safe cracking. So it’s very time consuming.
On the how do we solve the problem, so I talked about kind a lot the context around the complexity of it with healthcare data. Within our company, we have two distinct approaches. Again, on the large enterprise side of the company, like I mentioned, the medical pharma side, we have aligned to a single data analytic solution called EDNA. So EDNA stands for enterprise data analytics. And I think of it as like the one stop shop for all clinical data related to our operations. Like many companies, sales inventory pricing, APAR, that’s like the one place to go for all of our business units to get that kind of data. And for us, that approach is working because we’ve grown through a lot of M and A. We have a lot of different business units. We’re getting everything in that one place into EDNA. So it’s kind of like the… It’s like the control room for this super tanker that allows us to see how the ship is performing and give us a little bit of forward visibility like radar into where we’re headed.
On the growing commercial side of our company, or Fuse side, the conditions are different, right? So we’re much closer to the points of care. And we’re gathering data from all these external entities, oncology clinics, pharmacies others. Again, it’s not our data, but we have rights to use the data. And the problems we’re solving are very different as well. We’re really trying to help patients understand the best path to access to care. And we’re trying to help is be more adherent on their medication treatments. And then we’re also trying to predict ways to get them back on that treatment if they’ve fallen off for some reason.
And also helping providers determine the best outcome with the cost of care, not just the best outcome for care but also the best cost, which the industry talks about that being value based care. So getting the answers to reveal themselves requires a different approach within this Fuse world. So one approach that’s unique about us is we’re building products and partnerships which allows to gain access to data and provide answers to these problems that I talked about a second ago. Additionally, to support this very complicated environment, my team, this digital services team, is building what we call a digital foundation to acquire and manage data. So we need data from electronic medical record or EMR systems in order to build our product for value based care. So we’re building this data acquisition product that will solve for the problems of acquiring and managing EMR data.
So the difference is we’re not standing up a shared services’ organization like you see in many data analytics solutions. We’re breaking the problem down into more autonomous chunks, forming product teams around those chunks to solve for the problems of that domain. And finally, we’re building matching technology within those products to allow us to create this longitudinal history of a patient so we can align signals along the patient’s journey, sort of like my analogy of going to the grocery store, but rather than trying to sell me a bag of potato chips we’re able to help predict who needs help in order to gain access to necessary care, or if a patient can receive the same care at a lower cost.
Anthony: So I’d love to pull on that thread for a second, because I think when we talk about healthcare and patient treatment there’s often this feeling that we want to pull out the stops. No matter what the cost, this is ultimately about your health. And in some cases, it literally is life and death conversations. But ultimately you are a health services company, and costs are big consideration. You want to run this in a smart way from a business perspective. That calculus is very different for Cardinal Health than it may be for other kind of commercial organizations. And it feels like in what you were just talking about, that data plays a really important role in thinking about the difference between what it costs to deliver that care and how to get the best possible outcome. How do you think about that and how do you sort of instruct those product teams to reconcile that challenge?
Mike Alvarez: Yeah. You’re right. And some people, I think, think of it as more of a zero sum game. I know for myself and in situations like that I like to think of it more of like a yes and, right? So while I would agree with you… So if you’re a patient, you want your provider to do all she can to treat your illness, or if you have a loved one in the hospital it’s like anything goes, do everything you can. But the cost side is really not about cutting corners to save money. It’s really about more creating the right incentives for the healthcare market to deliver the best outcome at a targeted cost, right?
So Medicare, Medicaid again, calls that value based care, right? So it’s balancing the best outcome at the right targeted price or at a targeted cost. And the other thing we can do, I think in this world as well, is creating incentives. Maybe we can talk about that later, but I think back to kind of balancing that care and cost, obviously the providers are highly trained physicians. They know how to treat their patients. But they don’t really have a lot of visibility into the cost side of the equation. They really don’t know what things are costing until maybe six months later. Within our fuse organization, we work with our specialty biopharma division to create a solution for oncologist under this value-based care umbrella. So again, oncologist are very highly skilled at treating cancer. They have this oncology care model with a lot of guidelines. So all that’s kind of scripted. But they don’t have tools or understanding of the cost side of treatment, and don’t really understand how they’re performing against these targets that come with this oncology care model.
So we’ve developed this data science driven application called Navistar. So Navistar will help oncologists track patient treatment, like I mentioned, but also will give them predictions on how much it’ll cost to treat that episode of care, giving clinic some important insights into care. And that’s just kind of the beginning. So back to that data notice theme. If we can find anomalies in the data to help clinics kind of benchmark and develop best practices and learn from each other, I think it’ll help drive outcomes and better incentives for care overall.
Anthony: Yeah. And I really like the way of thinking about it, which is when you give people the data, when you sort of share the information with them, they make better decisions. And I think the thing everybody can agree on is we want the best outcomes, and it’s not just about sort of throwing money at the problem but of actually about providing the best standard of care. It creates that best outcome. Now, we talked about this ahead of the podcast. So listeners won’t know this, but I’ll spill the beans. One way we’ve seen people give people visibility to the data is through a dashboard. And you said when we were talking before the podcast that dashboards are dead, which is a very controversial statement. And I love it. I happen to agree with you on it. I’d love you to… one answer to the conundrum of quality care is, well, just make a dashboard, throw the data up there, let people wallow in it. But it sounds like you wouldn’t agree with that. So maybe if you could share your perspective on dashboards.
Mike Alvarez: Yeah. Yeah. It’s a bit of intentionally provocative statement, you know, and I didn’t coin it. But when we started looking at some of our products and how we were going to expose those insights externally, we had to make some decisions around are we going to do the traditional model that’s been around for 20 years or are we going to do something different. And I really feel like there’s this shift underway where more and more people are getting used to data.
So to think about it another way, how many times have you heard maybe one of your customers are looking to re-platform or modernize and they say, “Hey, we have 3000 reports and I don’t know who uses them”? I heard that from one of our business units this week. It just feels like we spent the last 20 years extracting data, building warehouses, layering dashboards on top of them just to have the business turn around and say, “Hey, can you just give me ODBC access to that thing because I want to do my own thing?”
I really think we got to dig into the kind of the why behind that. Often the data work is being done by IT professionals, not business. And as IT professionals lack the business knowledge, right? They lack the knowledge of the data. They don’t know how to build the insights, not to mention they really understand what the business is trying to achieve with that insight. Right? Are they trying to drive sales? Are they trying to reduce inventory, understand where inventory is short or kind of stagnant? There’s thousands of questions that the business needs quick answers to. And I think our current approach is anything but quick, right? So what is happening is the business hires these analysts and these data professionals within their organization. They grab data from the underlying systems and they build their own insights. Right?
So I think this is how we end up with these thousands of unused reports and dashboards. And I know if any of your audience is part of a large company, they’re probably sitting there nodding their heads right now, because we’ve all have felt this there. I think there’s this shift underway right now. I think some of the factors to that shift that I’m thinking about… And again, I think this will play out over years, but I think the work workforce is becoming much more data savvy. I don’t have stats on this. I’d love to see them. But data is now in everybody’s vocabulary. A few years ago it wasn’t even part of our company’s mission. Now it is.
COVID made everybody aware and everybody kind of a consumer of insights, right? When we were talking about flattening the curve, they’re talking about an insight from data, right? So everybody’s kind of speaking this common language now. During the period of COVID I felt like we had kind of seven billion analysts overnight who were discussing all these data points related to COVID. I think the demand for data professionals is exploding. XLOne fact it said it was like 650% job growth since 2012. I think higher education is responding. So I’m seeing more and more of these data degrees coming out, these kind of blended data degrees coming out of universities.
I think all of this will culminate into like a more data savvy workforce that don’t want to stare at a dashboard. They want to be part of the curation process. So I think over time we may see analysts just being much more comfortable with engaging with the data. So within the Fuse world we’re trying to design with that aspect in mind. Don’t get me wrong. You always have your basic dashboard because you need some basic trending in KPIs, but I’m really interested more in building the data experiences across our product ecosystem where I can get out of the way. We can be build this data foundation. We can build these data products. And then the SMEs that understand the business problem they’re trying to solve can get access to that data safely and securely and then build the outcomes they’re trying to achieve.
Anthony: So that’s actually really interesting. So what you’re saying is that the problem with dashboards is that they’re dis empowering, that they put the onus on the analyst to come up with all the right metrics [inaudible 00:20:21] and display them. And they treat the user as just the dumb consumer of that. And what you’re trying to do is really put the data consumer at the center of that conversation, let them do the analysis and really be part of it and sort of be engaged in it.
It strikes me that you also mentioned something before, sort of like central versus decentral. And that’s a theme that we’ve had on DataMasters a lot, this question of whether we push the responsibility for data and analysis close to the decision maker or pull it and make it really centralized. And there’s often a question of cost here as well, right? Do I do this centrally at a low cost, do I do it in a distributed way, maybe at a higher cost, but with better decision making. I’m curious how you’ve thought about that, especially the context of your points around dashboards and your points around value based care.
Mike Alvarez: Yeah. It’s something I’ve obviously looked at over time. We’ve done it different ways as a company. I think there’s really kind of three different approaches. There’s this kind of this distributed approach or most like to think of it as organic, right? It just kind of evolves. And I think that’s probably the default of many companies, because if you don’t have a plan in place what’s going to happen is the teams are going to solve the problems within their teams. So they’re just going to hire the skill sets and grab the data they need, whether that be analytics or data science. So that’s one approach companies can take. I tend to think there’s a lot of redundant activity going on depending on the size of the company. There could be a lot of redundancy across a very large company like Cardinal Health.
There’s the centralized approach. So second approach is kind of centralized approach, which takes a great deal of alignment. Right? So I mentioned EDNA earlier. So that team formed is a central team. There’s a lot of energy and alignment going into getting everybody onto and aligned to that solution. And that’s a good approach too in certain scenarios, but it takes a lot of kind of top down alignment, right? Everybody’s got to agree like, “Yep, I’m going to go get my insights over here versus doing it on my own.”
I’d say a third approach is really just kind of federated or more like a hub and spoke. So this is how we’ve organized our data science organization within Fuse where we have sort of the five or so different business units I’m working across. I have a data science team within my organization. And I like to think of them as the data science kind of COE. COE sounds a bit bureaucratic honestly, but it’s really about how do we engage with teams and kind of meet them where they are on their data science and analytics journey and help them skill up. Right? We have a team that’s very mature. So I’m almost like staff out for them so that we need help we can put somebody over there. Talented data science can join their team and we can help augment their team capacity when they need it.
We have other teams that are maturing, right? So they’re kind of starting up the maturity curve. That’s kind of that Navistar example where we’re the data team and the data science team behind them. Over time if they want to take on that responsibility then we can kind of train them and help them get to that point. We also have built out our data science platform in the Google cloud platform. That’s a kind of a common solution that everybody can come and… They don’t have to worry about all the technology and networking and security that comes along with that. So we’re trying to remove in this federated approach, we’re trying to remove some of the barriers to them achieving their goals, achieving the outcomes. Again, getting these products to market, getting feedback on the products, and kind of iterating on them. That’s kind of how I like to think of it.
And then that hub and spoke, I guess the hub part, I try to bring the team back together every week and talk about lessons learned, right? So I’ve seen some great learnings where a data scientist will go out, make their part of a team. They solve the problem or they learn something new. They bring it back to that hub, and then we all kind of learn and grow from that.
Anthony: Excellent. I think that’s a really interesting… Again, a theme I’ve seen throughout DataMasters, this idea of executing in both the central and decentralized manner, but putting in place programs to make sure you’re sharing across those platforms so that you get the benefits of scale from a central approach but the benefits of insight which you get from working close to the data and the decision maker.
So I wanted to shift gears entirely. And I know we’re running short on time. I mentioned in the introduction that you’re a heavy podcast listener, and by its nature someone listening to a podcast must like podcasts. So I’m sure our listeners would be always looking for good, unexpected new recommendations. What great podcasts are on your listening list and would you recommend to listeners?
Mike Alvarez: Yeah. I think I’m always listening to something just in these little gaps of time I had because there’s so much to learn and so much to discover. So I’m always… Have my earbuds in and listening to audio books or podcasts. But like I said, I mean in this… In all sincerity, I’m not saying it just because I’m on the DataMaster’s podcast, but I highly recommend that to listeners. I’ve got great some great insights from some of the data leaders you’ve had on the podcast. I’ve been a long time listener of the Tim Ferriss podcast. I really like how he… He’ll get these I think he says world’s top performers on there and kind of he’ll dissect them and kind of decomposing what made them successful. And he does it across the very broad spectrum of professions, right? It’s not technology. It could be an artist, it could be an actor, it could be a CEO. But I kind of learn a lot from that.
Actually, I kind of do this cycle of learning from that and kind of what books they’re reading, then I’ll pick up a book and listen to books and I’ll go back to my podcasts. I like Freakonomics. It’s another good podcast. Just I like that kind of behavioral economics aspect of it. And then A16Z or the Andreessen Horowitz. Always great insights on kind leading edge of technology. If I can I’ll throw out a couple books too that I think are great listens. The Goal, which I think it’s a 30 year old book by Dr. Goldratt. It’s a really good book. It actually led to the creation of our data science process, which I know we’re out of time, so maybe we can talk about that another time.
On the personal side, I really like this book called Too Soon Old, Too Late Smart. It’s got like 30 different lessons in there around things we didn’t learn till later in our lives we wish we had learned earlier. I mentioned Loonshots by Safi Bahcall. Atomic Habits is another good one. And then for any company going through this digital transformation, I really like this, Designed for Digital by Jeanne Ross. It does a really great job of explaining the differences between digital and digitization.
Anthony: Love it. No, those are some great recommendations and a hearty plus one to The Goal. It should be required reading for almost anyone in almost any role that they take. It’s a great book. But those are some… Also a couple of those I hadn’t heard of before. So definitely be adding that to my short list. So, hey, Mike, really appreciate the insights you shared with us today. Fascinating conversation. And as you pointed out, there’s actually much more we could talk about then. So I think this… Fruit for a future DataMasters podcast episode. So really appreciate your time and thanks for joining us.
Mike Alvarez: Absolutely, Anthony. Thanks for having me on data masters. I look forward to future episodes.