Yes, especially if you change the incentives, and you give them an “easy button.” That’s the second thing I learned—these groups are almost like exceptions that prove the rule, you know? Extraordinary leadership happened to be in place at the Palm Beach ACO and at the Rio Grande Valley ACO in McAllen, Texas, and these extraordinary leaders who managed to convince the doctors to take money from their own pockets to invest in the model; that’s so great. But studying these physician-led ACOs also convinced me that that model isn’t currently scalable. It was really difficult for them, and we need to make it easier. And to me, finding the capital, choosing and selecting and implementing the IT and analytics systems, hiring and managing people who can go in and do change management and practice coaching, understanding the regulations and risks, and the gotchas, and the Easter eggs in the practice, those things are difficult for your average doctor. That’s not hard for me. Taking care of primary care patients, that’s very hard for me. So why don’t I do what I’m good doing, which is helping them to become good at this? It just really made sense to me, in terms of what we’re doing now with Aledade.
Still, in the absence of extraordinary leadership, as demonstrated by the doctors in McAllen and Palm Beach, how do you get change healthcare system-wide?
You need systems. Systems are what ordinary people use to do extraordinary things. We aim to create systems—processes, data, protocols, a business architecture—that will help people do the right things.
And for them to spend the money on the overhead again and again—well, there is a cost for this. So the physician leader says, great, I got a check for $2 million, and my cost to create the infrastructure to get that check was $2 million. And we’re saying that that entire infrastructure should be spread out over a larger number of ACOs.
So you’re trying to create a template?
We’re trying to create a playbook; the word “template” sounds too rigid, too cookie-cutterish.
What kind of scalability is possible in all this?
It’s kind of like a pod or a village unit. There’s the theory that human beings in whatever organization, when the organization gets beyond 100 or 150 people, kind of breaks apart somewhat. So maybe the ideal size is to have these docs in units of 100 or so. So you can have medical directors and field staff and have a level of accountability that’s more immediate. If you’re at 100 or fewer, you have that sense of mutual community accountability to each other. And you want more? Let’s stamp out another pod, all of the different pods supported by the same data, technology, regulatory support etc., team. So we would like to try to keep each ACO at no more than 100 physicians; so you scale the units.
What were the hardest few things for the Rio Grande Valley folks, and how did they overcome them?
Honestly, if you talk to them, one of the hardest things for them was the IT piece of it—working with multiple electronic health records, because we don’t have a monoculture anywhere. So you get docs together, and you’ll have multiple EHRs—so getting the data out of the EHRs, for predictive modeling quality reporting. But also, getting intelligence back into the EHR workflow is really hard. As hard as it is to get intelligence out of the EHR, getting it back into the EHR is even harder. So that certainly was on the roadmap for meaningful use stage 3 and the Healthy Decisions standards and interoperability initiative. That was something that the [federal] Standards and Interoperability Committee [within ONC] had foreseen.
And here’s what I mean by that. You can get data out of your EHR and you could combine it with claims data, and admission discharge transfer notification, and then you could identify not just care gaps, but what actions to take for your high-risk patients. And you really want to push that into the system so they see the opportunity to shift the referral to a high-cost cardiologist of ophthalmologist, to a less-expensive, high-value one, or make sure they picked up their meds last time. It’s hard to take that intelligence and combine it into the electronic health record. It can be very hard, but it is the operating system for the practice, and you don’t want to treat it just as a dumb data slave harvesting intelligence, but how do you make the workflows smart with it? Whoever solves that problem will do very well.
What the vendors have available commercially for patient care organizations is not able to accomplish that, correct?
I have not seen it yet, no. There are vendors who I believe are working on their own population health analytics modules, with the idea that their own modules could write into and read from their own EHR. But I don’t think that’s what the market wants, to put it gently. What the ACO and population health market wants is the ability to use whatever analytic or population health software I want to use, and to be able to write those, using common methods, into a number of different EHRs.
What should healthcare IT leaders be thinking about this, and doing?