By MATTHEW HOLT
Clarify Health has linked (but anonymized) data on about 300m Americans, including their claims, lab, (some) EMR data and their SDOH data. They then use it to help providers, plans and pharma figure out what is going on with their patients, and how their doctors et al are behaving. CEO Jean Drouin, a French-Canadian who incidentally at one point ran strategy for the NHS in London, explained to me what Clarify does, how it’s going to help improve health care, where these data products are going next–and why they needed to raise $116m in March to build it out. Jean thinks about creating a single source of truth, and I asked him a couple of tricky questions about whether his customers would want to know the answer. A fascinating discussion. (Full transcript below)
Matthew Holt:
Hi, Matthew Holt here with another THCB Spotlight. And I’m with Jean Drouin, who has a French Canadian name, but is an American who’s lived in London–a bit like me–who is the CEO of Clarify Health. So Jean, Clarify Health is one of the new startups. You guys raised over a $110 million a couple of weeks back, which I guess is a small round these days considering what everyone else is doing.
But essentially you are one of the new companies who are doing data analytics in a different way for the health care industry, by putting together a lot of different sources of data on a lot of people. So I hope I haven’t garbled that too much, but could you explain to start off with, what are the data sources that you’ve put together to form the base of all the products you’ve then built?
Jean Drouin:
Very happy to, and thank you for offering us the time today, it’s a pleasure to be with you. We have pulled together a data set now on about 300 million Americans that links their claims history, their lab data, their prescription data, some amount of EMR data, and then critically, social determinants of health, not at zip code level as most others typically do, but at individual level in the same thought process that a bank would in looking for credit scores for example, or that Amazon would use to predict what you might want to buy. And so think of us as being able to see credit card purchase history, whether someone has a driver’s license, a family member living within five miles that might have moved recently. So we’re able to stitch together both the clinical picture and the very important social picture to ultimately be able to deliver a far richer longitudinal patient journey.
Matthew Holt:
So before we dive into what you do with that stuff, I know one of the sources is CMS and you have a special relationship there-You talked about a lot of data there. I think you wrote a blog post about this saying that interoperability is a big problem in American healthcare. No shit. Before we even think about how you put the data together, where do you go to get all these different sources?
Jean Drouin:
Absolutely. So claims for example, as you know there’s a back office or a transmission set of pipes between the providers who submit the claims and the payers and the adjudication processes. The folks who manage those pipes are able to resell that data in a de-identified way for example. Same for the other categories of data, whether that’s lab, prescription, you can imagine there are folks who process that and are able to, again in a de-identified way, provide it.
Jean Drouin:
Now, up until about two or three years ago, that was a very complex, cumbersome process. There are companies that have emerged like Datavant and HealthVerity that now do what’s called tokenization of data. So think of it as a token being of that virtual patient identifier, and they have massively lowered the activation energy, if you will, or the barrier in terms of stitching together data. Such that in a very short timeframe for healthcare, really 18 to 24 months, they have made it far easier to bring together disparate datasets. My own view on this is, aside from maybe certain categories of data like genomics and specialty lab data, we are moving to a world where most other forms of healthcare data will become commodities.
Matthew Holt:
So, obviously there’ve been companies who’ve been playing in this game for a long time, IQVIA, the former IMS Health is obviously the most example. And then you’re talking about the change in technology where you can identify Matthew Holt without knowing the name Matthew Holt, but by figuring out some token that equivalents to it, but you can’t get back to the identity. Funnily enough, I was in a startup in 2000, which was trying to do this way the hell back when. At one point it had a provisional patent filed to do this — I know some other people built on it later — and then ran out of money before it could actually pay the lawyers ! Maybe we could come back years later but the patents would have expired by then. Anyway, by the by. So this idea of putting together data and selling information off it is not new.
Jean Drouin:
Correct.
Matthew Holt:
And what you’re saying essentially is that you yourselves are not necessarily in the data integration, well, you are doing data integration … You’re not in the data grasping business and the data cleaning business, you’re more talking about what you can do with that data set. But obviously there are other companies, and another one of your colleagues, competitors, Komodo Health, raised a big chunk of change as well last week- Slightly more than you did. I don’t know if you guys get jealous about that! But there are other players obviously in the game who are thinking about what they could build and what they could do with this.
Most of the activity so far — and you’ll now correct me — that I can see, is that most of this has been aimed at trying to identify the same sort of stuff that IMS was doing way back when, well, IQVIA was doing way back when, which is, on behalf of the big pharma and life sciences companies, trying to figure out where they are selling drugs, how they are selling drugs and how they can sell more of them. That’s probably an unfair characterization, but that’s the kind of characterization I’ve used.
You obviously have a bunch of life sciences companies who you’re working with. So number one, tell me a bit about what they use you for and how that works because you’re not getting identified data, right? So you’re not telling them that Matthew Holt is taking this drug, but you are telling them that Dr. Smith is prescribing this drug. Well, say that first, and then we’ll go on to other uses with the data with other players in the system.
Jean Drouin:
Okay. Terrific. Two quick thoughts before directly answering your question. So you’re absolutely correct that if we think of the analytics stack in healthcare, there’s the data layer, there’s if you will an intelligence layer or a transformation layer, and then there’s the workflow in which you want an actionable insight to be able to be consumed by someone who hopefully will make a better decision.
There are the classical data vendors like IQVIA, Change Healthcare, Optum, et cetera. And in fact, one of the things I would say we’re actively changing is that the typical model in healthcare for analytics is hey, new company, what data set do you have that I don’t have yet from those players? And then the mindset is oh, I’ll have someone go into a dark room for four to six weeks doing SQL queries to manually link and fork this stuff together and come out with what looks like very much an Excel spreadsheet sometimes dressed up to look a little bit better in Tableau. As opposed to what we have in other industries, which it’s all about time to ROI with on demand, self service insights off of what are now enterprise analytics platform that service multiple business use cases.
And so we think of ourselves — you’re absolutely right — as that intelligence layer that is the refinery or factory taking raw data and turning it into insights faster than anybody else. It’s actually a major differentiator with many of our peers, some of which you’ve mentioned, because if you really dig in, many of them still have an army of SQL queriers in the background, and where the world is going is, “how automated is your stack.”
Now, interestingly, Clarify started with providers in the value-based space. And I will give our colleagues at Komodo all the credit in the world, given what happened with the Trump administration around value-based payment models, it was a far better place to start in life sciences, in a world where as a CMS-qualified entity people were still in the mindset of “I’ll buy that dataset,” if you will.
So you’re right, in life sciences companies like us have initially often gone for the use case of helping the brand teams either identify the physicians that pertain to the drugs they want to prescribe, or the cohorts of patients that want to be matched. Where I’m most bullish though moving forward is the chance to move upstream into clinical trials, where for example, Novartis, a year ago, a billion spent on a heart failure trial, five patients short of statistical significance.
Matthew Holt:
How annoying! Could have gone out the street and found five people who want to use you, but …
Jean Drouin:
Exactly, right? So you would think that maybe if you could have found them the six patients, they might’ve been willing to keep you in business, you know? Now, the areas where we have found the most traction, particularly in this COVID year has been with payers. And for payers, we do three things at the moment. One is network design, the other is utilization management analytics, and the third is matching of populations to the right interventions.
The network one is particularly interesting because as you know, that’s one of the biggest levers a payer can pull in the US, is which PCPs and specialists do I put in my network in a particular geography, say Denver. The old process, two to three years, spreadsheets, and then people with gray knowledge a bit like brokers on the ground saying, “Oh, Holt’s a really good physician. You really need to have him in your network.”
Whereas now with big data, we’re able to bring money ball analytics from sports and meet clinicians emotionally where they want to be met which is, “Will you please, if you’re going to benchmark me, look at the difficulty of the patients that I faced.” And obviously, you know every clinician believes that their patients are more complex. Okay. But having done that normalization, we’re able to produce a far more precise scoring on the relative value-add or destroyed by clinicians against their case mix.
Matthew Holt:
You obviously also have a provider business that you mentioned at the start.
Jean Drouin:
That’s right.
Matthew Holt:
Tell me a little bit about what you’re doing for the provider systems.
Jean Drouin:
So it’s an old adage and you’ve seen many systems across the world right, that you get exactly the behavior you incentivize. So it won’t surprise you that in a fee for service system, there’s still a lot of interest around referral optimization And making sure the referrals come to the home team. Now, thankfully I’m hearing more and more discussion around smart referrals. So as the world moves to value, it’s not just, is it to the home team, is it the optimal referral? Long ways to go on that front? Right. I saw the smile.
Matthew Holt:
Well, yeah, and there are also people playing in that game so there’s specialist companies, like Kyruus and others who are trying to direct people specifically to the right place and yeah, in general, well I have a –as having worked in healthcare a long time — somewhat cynical view of what big health systems are actually really interested in?
Jean Drouin:
Hence my preamble on incentives. Right? And then the other is those that are going into value-based models. So say CMS now has this direct contracting model? Imagine that you automated all of the contract rules and you had an on-demand, self-service, digitized Dartmouth Atlas that allowed you to say, just name your health system, filter, name the program, DTC, MSSP, et cetera, and said okay, if you go in with all of your physicians, you’re going to lose a hundred million. But if you were to take these 10% out from that contract for the moment, you’d be in the positive. And if you want to help the ones you’ve got to improve, here’s the areas where they’re variated against the payment model and where they might improve. And that all of that can be rendered in a few minutes, as opposed to the weeks or months it would take to hire consultants to come up with a partial view. So there’s that use case as well.
Matthew Holt:
So that’s the provider thing and it sounds, I mean, a lot of these things I’m hearing from whether it’s analytics player like you, or some of these sort of AI automation of plain RCM or something, it seems to be replacing a lot of what consultants have done in the back office of these hospitals for many years. And it seems to be as you said earlier, we’re going from a bunch of people with spreadsheets to more automation in the stack.
So in terms of Clarify’s overall role, obviously it sounds like and I assume that the folks who chucked in the recent round all felt that things were growing on all fronts pretty well. As you mentioned, there’s some changes, and obviously the tailwinds of healthcare –of course we’re not yet out of the fee for service world –but the tailwinds of healthcare are at least heading in that direction and policies even from the Trump administration seem to be getting us slowly there. Give me a sense of the breakdown for your revenues and growth, as to where you’re seeing more activity across these different customer sales.
Jean Drouin:
So right now about 50% of our revenue is from payers, 30% from providers and 20% from life sciences. Now, part of that is because life sciences for us is a very young segment. If you ask me to project three years, I would think that you’re looking at a situation where it’s more 40% payer, 40% life sciences and 20% provider. And part of that is we’ve seen that historically adoption of new technology in providers tends to often take a little longer. And we see the same type of thing here.
Matthew Holt:
So, I’m going to ask you a couple of questions about the business and the market. And then because of your background working in the NHS and doing a lot of stuff, I’ll ask you a few off the wall questions. But the first one is, we’re in this crazy funding scene where no one understands the stock market, Gamestop is a million dollars a share, Bitcoin’s going through the roof. There’s money, obviously flooding into the venture community. And people looking to put it to work. And you’ve seen these huge funding rounds, right? Of companies like yours, not to mention a bunch of other companies doing, god knows what else, just in digital health alone. And this Tiger Global fund apparently is spending a hundred million dollars every two days on a digital health company. Not quite sure how much money they have left. But anyway.
So in that environment, obviously there’s a little bit of pressure too. Obviously you have to grow the business, but on the other hand, having a chunk of change helps. We mentioned a bunch of other companies already from Komodo all the way to IQVIA who are in this space with some data and analytics, and obviously you think you have an advantage over them. How many companies do you think can the industry sustain doing the kind of things that you’re doing? Take a look at the electronic medical record company market, where 20 years ago, there was a gazillion companies you couldn’t tell who was who, and now there’s basically Epic and Cerner and Meditech, and that’s about it. Where do you think we end up in the data analytics underpinning healthcare space?
Jean Drouin:
My sense is your analogy isn’t bad. I think over a 15 to 20 year period from now, we end up with two or three dominant players. One of the analogies I sometimes also use from another industry is Salesforce. They are 23, 24 years into their journey? It’s easy to forget that in the early 2000s, it wasn’t clear how that was going to play out either.
Matthew Holt:
In fact, you may remember Siebel stock was through the roof and all doing very, very well, and then they found that Salesforce was going to kill them and they had to go back to beg Larry Ellison to take them out of their misery!
Jean Drouin:
Exactly. So I also wonder whether you’ll end up with some that are more pharma dominant and some that are more payer provider dominant. And what I see particularly on the payer provider side, is we hear continuously that there is a desire for an objective single source of the truth platform sitting between providers and payers, that can baseline in a way the clinicians trust, here’s the performance today, here’s the performance a month from now. And based on the contracts that are loaded in say, okay, here’s what’s getting paid and bonused and penalties on an upside downside arrangement, without a need for two fighting revenue cycle armies, because that’s been automated out. That kind of platform to me, feels like it could be an independent occurrence vis-a-vis a clinical trial optimization platform in life sciences.
Matthew Holt:
That makes sense. You remind me of the conversation I had with Andy Slavitt 15 years ago where I said, “oh, you’re actually an arms dealer”, when he was back at Ingenix, pre name change to Optum, “you’re actually selling the same stuff to both sides”, right? The providers are up coding, the plans trying to down code and we’re trying to figure it out. Two more questions, which are sort of more esoteric. The data that eventually you’re using, got created about or by a patient at some point. There’s been conversations about people’s data and rights to that data, but also people’s rights to benefit from the data they provide. Obviously, this is going to go on.
Where do you stand on it? I mean, there’s people saying, well, Anne Wojcicki from 23andme said, it’s great, but even if I could make a billion dollar drug out of your data, the amount that we’d get back to you is so minimal it’s just not worth having the conversation. Whereas there are others trying to build blockchain places that people could protect and keep their own data and get paid every time someone puts it in some analytics. Where do you think we are on that and where do you think we’re going?
Jean Drouin:
This is quite visceral for me because I think of it as just you or me, or a loved one needing care. And I find it incredible given that there’s no technological reason preventing this, that you and I cannot have all of our healthcare data, and by the way our consumer data, available on my phone. But for us to then grant access in the way we would wish to, based on the need that we have. So if I have to walk into an emergency room and have my brother as my proxy have access to the whole thing, to say click, and give it to the doc. Or similarly, “hey, I have this condition. Yes. I would like to know about potential clinical trials that are relevant for me, but only for this little narrow use case”. So I’m a believer that ultimately we should and will move to a model where individuals are able to self-direct, if you will, the kinds of use cases they’re willing to share their data for, and what you get in return might not be monetary. It might be a benefit like learning about a clinical trial. It might be a benefit around receiving suggestions about clinicians that are a better match for you. Not just because of their outcomes, but because of their personality and their interests and their style. Okay, that’s not for today, but could we see something like that in 10 years? I would hope so.
Matthew Holt:
But if that were the case, does that put you guys out of business? You’re essentially relying on being able to collect all data on everybody?
Jean Drouin:
Yeah. I worry less about that because I still believe that there’s going to be a de-identified tier. See, HIPAA, you know, a lot of people criticize HIPAA. I actually feel that HIPAA is a fairly wise set up. It says, for the purpose of ensuring that there’s better quality, appropriate billing or innovation, essentially, it’s okay to use de-identified data. The minute you cross the line into direct marketing though, you know, it drops on you like a ton of bricks. So my sense is those that functionality will still be required across the system.
Jean Drouin:
I do think though it’s incumbent — and I wouldn’t mind if it were regulated to tell you the truth–for companies who have insights to share some of those insights publicly. And that was a feature of the QE program with Medicare. You do have to share some insights on a public basis every year. If the obligations there were even more robust? Totally fine. And you might even read into this that we’re thinking in the background of things we might be able to expose publicly on a free basis that would directly help consumers with the choices they make in terms of the care they seek, for example.
Matthew Holt:
Yeah. Now you can tell there potentially should be a lot of value to patients out of all the excellent work you’re doing, not just in that sense, you’ve hit on a bunch of them there. Let me run one more thing past you which is a little bit off topic from what you do in your day to day business, but with your experience of working in the UK and working at McKinsey, where you’ve looked at a lot of different stuff. There’s clearly going to be a bigger controversy coming over time, and I’m surprised it’s not a bigger controversy now, which is the concept of what is the taxpayer, you and me, getting out of Medicare Advantage.
Matthew Holt:
And the reason I say that is there’s just a MedPAC report just came out, which Dan O’Neill, a colleague, a buddy of mine did a great evisceration of on Twitter, which essentially said that the government has more or less given up trying to figure out whether there’s better quality or not better quality within Medicare Advantage, we’re pretty sure that everybody’s upcoding and this is being unfair not only to the plans who are not upcoding, but also unfair to the taxpayer.
And we know it’s a big, big source of many of the major health plans’ profitability. What’s going to happen, eventually at some point, once we get out of this pandemic recession we’re having, we’re going to start caring about how much money we’re spending on health care at some point, and if 35, 40, 45% of the Medicare recipients are costing more than the rest, and a lot of that is going in profits and big margin to big health plans … I can remember back to the late nineties where we said, that’s not so good, and then it changed again in the mid 2000s, right?
What does that mean to Clarify? You have a lot of that data, right? You understand a lot about patients, the rest of it, and what’s going on. If you were to come with that single source of truth, do you think that your customers on the health plan side would want to know the answer?
Jean Drouin:
It’s an interesting question, many facets to this. It’s also partly related to the recent rule on transparency that was introduced. There’s been obviously as you know a lot of focus on just the rate transparency? There’s quite a lot of other stuff in that rule, including having to provide by 2023 transparency on the total costs of 500 procedures and by 2024, all procedures.
Matthew Holt:
So far, I understand a little hospitals are blocking the Google searches of their price listing in the first bunch, not too surprising!
Jean Drouin:
Well, any transition from one regime to another is going to create as you know winners and losers, right? The other thing is pretty much any — and this is actually going to be an interesting question here, Matthew, which is, the payment models we have had in the past, whether they be fee for service value, whatnot have tended to be bell curved around an average. And you could reliably say, well, if I’m on the correct side of whatever game needs to be played, then I’ll end up in the right space. Right?
Jean Drouin:
What I will give credit for the Medicare Advantage incentives having done is caused a set of independent players, ChenMed, Oak Street, et cetera, to come out. And I do think that as part of what they’ve done, there have been helpful innovations in management of the care journeys beyond just the coding optimization. That said, it’s a question I think all of us in health care need to look ourselves in the mirror and ask periodically, which is how much real care model change has there been, and how much has it really delivered superior outcomes versus the cost trend.
And I almost feel like we’re maybe where we were as an economy, vis-a-vis productivity as a result of the internet in 2000, right? Which is, wow, this Internet’s amazing, where’s the productivity improvement. And I’m an optimist by nature so I’d like to believe that between CRISPR and other things on the biotech side, and some of the vastly increased trends, the democratization of transparency in a more real-time way in the hands of those making the decisions, that that will have a similarly positive effect on creating real care model change. It’s not going to happen overnight though.
And look, I applaud you for asking the question because those kinds of questions need to be asked more often. At the end of the day I believe in the adage of you do well by doing good, and transparency’s uncomfortable, I don’t think from a values point of view, we have a choice though. It’s where we need to go if we want … And this is not just in the US it’s equally true as you know, in the European, Australian, Singaporean systems. Yeah, we have to be willing to expose these secrets and uncomfortable truths.
Matthew Holt:
Yeah. I think, the way you said that that’s going to be winners and losers. I mean, I talked to Todd Parker at Devoted who I’m just not going to play that game, (RAF upcoding) and we’ll see if they grow it, but you have uncomfortable multiple masters if you’re a United or a Humana, right, you have Wall Street as well as the way to do good. And we still need to know a lot more about what’s working, what’s not working between the different players in that whole area, just in Medicare Advantage or in new models as well.
Jean Drouin:
Yeah, you’re absolutely right. One of the things is, the power of these new analytics is not just to show us where the problems are. It’s also to show us where the successes are. So for example, the research Arnie Milstein does at Stanford on finding positive outliers, and what is it about the positive outliers, right? The more we can also double down on that and create self-learning systems as opposed to what we’ve had, which is a more transactional, punitive type of, mercenary if you will, type of arrangement, that’ll set us off on a better trajectory. That’s going to take a generation, right? It’s not going to happen overnight.
Matthew Holt:
Yeah. And I was just talking with some of the folks in digital therapeutics all about how do you wrap these new types of tools that are like drugs, but are not drugs, into an ecosystem of care. And how do you then use that to have better patient outcomes. And you can see dribbles and drabs of that in the way that telehealth and remote patient monitoring is starting to come and the way that companies like Livongo and others are doing it, but, you’re moving, it’s not moving a battleship, it’s moving a fleet of aircraft carriers or something. I don’t know how you get it done quickly and easily.
But clearly the opportunity to use data sets like yours and to get them more public where possible, to start moving people and thoughts and ideas, and then transactions and decisions around is a big part of that. And so hopefully you and your cohort will be successful. All right.
So before I go, you got a big chunk of change that came in, hopefully you haven’t spent it all quite yet. What are you going to spend it on, what’s next for Clarify in terms of what you’re actually going to do, assuming you’re all not just going to run away and go to the beach. I understand that’s only relevant if you’re running a microbiome company!
Jean Drouin:
Yeah, yeah. No, so it’s interesting. We’re almost doing things flipped in reverse in order to prove to folks that it was possible to entertain a cross sector, provider-payer-life science enterprise analytics platform, we had to go the unconventional route and put into the market five or six business applications that we’ve discussed. A year ago people said, “Jean, you’re crazy. You should just do one or two.” And I said, “Sure. Then you won’t buy the premise.” So now that people say, wow, it’s actually happening, we need to turn around and be incredibly disciplined and focus around scaling what we’ve already got. I would say that’s about the next 18 months.
Now, what could happen in those 18 months though, is in the areas where we’ve chosen to play there are folks that might make good partners or potentially good acquisitions. And so that might be something that we would consider, not now, but maybe a year or two from now. I do think as we’ve discussed that there will be consolidation and at some point, one needs to start building that capability and muscle in one’s organization. You start more modest as you do bigger ones over time. And then you’ll see us based on customer feedback adding new business applications on the platform. But I’d say the biggest risk for us is actually lack of focus as opposed to really focusing on driving the impact from what we’ve already got.
Matthew Holt:
And in terms of numbers of customers, I don’t know if you will tell me about any revenue numbers, where are you now in those segments in terms of customers, and if you tell me revenue, ? And employees? And where do you expect to be in 18 months?
Jean Drouin:
Yeah. So about 130 employees right now, growing quite quickly as you can imagine. About 50 customers, roughly in the proportions we had discussed before, so half payer, 30% provider, 20% life sciences. And of course, revenues we’re not at liberty to share, but I would say this, they’re sufficient and growing on a trajectory where it is now reasonable to think about an IPO in the two or three year timeframe.
A lot of people ask us about SPACs and sure, they call. But just as we have benefited from having the right sets of investors and having always focused on what’s the very top valuation one might get with the specs. One has to be really careful with what ends up on the other side with the deSPACing and the sets of investors that end up in one’s base. Because ideally we would like to continue to have a terrific set of partners that have a long-term horizon.
So we’re laser focused on the things that ultimately matter, which is great team, great culture, and a set of delighted customers. We feel that if we do that and we stick to our values, that whether it takes three years, five years, 10 years we’ll end up in the right place.
Matthew Holt:
Sounds good. And I think your philosophy of saying, doing well by doing good is something that would happen, and there’s a lot of opportunity obviously for a company like Clarify to really shed some light on what’s a very murky, murky business, and look forward to seeing how that comes and interested to see how it actually gets used in the future. I’ve been talking with Jean Drouin. He is the CEO of Clarify Health. Jean, thanks for your time and congrats on the raise and good luck in the future.
Jean Drouin:
Thank you so much. Really appreciate it, Matthew.
Categories: Uncategorized