Accountable care organization (ACO) work continues to move forward at the Springfield, Massachusetts-based Baystate Health, a six-hospital health system serving large swaths of western Massachusetts. Anchored by flagship facility Baystate Medical Center in Springfield, the health system encompasses six hospitals, 80 medical groups, and its own health plan, Health New England, which has 200,000 members. Baystate Health serves 900,000 patients across four counties.
Meanwhile, Baystate Health joined the Medicare Shared Savings Program for ACOs in 2012, with its Pioneer Valley Accountable Care organization, or PVAC, which encompasses 90,000 lives. Joel Vengco, vice president and CIO at Baystate Health, spoke recently with HCI Editor-in-Chief Mark Hagland regarding his organization’s experiences to date with accountable care-driven care delivery. In this first part of a two-part interview, Vengco shares his insights on the early phases of ACO development.
Tell us a bit about your MSSP ACO.
It’s called Pioneer Valley Accountable Care, or PVAC. We started in 2012 with the other MSSP ACOs. Our Medicare ACO serves roughly 90,000 covered lives. We’re going to be pushing that above 120,000 this year. We we’ve seen some fairly decent savings.
Have you received shared savings back from Medicare?
The first year, we saved roughly $2.5-3 million. The next year, we increased that a bit, but were still around 1.5 percent savings for Medicare, and you have to hit 2-percent savings to get savings back. We haven’t quite gotten to that threshold yet.
What have been your biggest learnings so far in your participation in the Medicare Shared Savings Program?
Obviously, the first big hurdle is just really creating the necessary environment for collaboration across the continuum of care and the region. Part of what’s challenging about this environment is that you don’t own all the physicians, and yet everyone is responsible for that patient population, and is sharing both the expense, and the benefit if you do it right. So a lot of it has to do with sharing that infrastructure—everything from a governance structure, to the policies and procedures around information-sharing, and secure communications, to analytics that enable you to actually track the progress of your ACO. So it’s really back to the basics of collaboration, which is in many ways a fundamental principle, but is also sophisticated and complex with regard to an operation like that.
A lot of organizations are still having problems with attribution. Are you still?
Oh, sure. We do it [successful attribution], but it’s very manual. And our systems are still not quite at the level that they need to be at in order to manage attribution at a satisfactory level. Our data requires a lot of cleansing, because the systems are still in many ways dis-integrated. So it’s still hard to tell which payer or primary care physician a patient is associated with. We’re talking about dual-eligibles in that case.
And there are all sorts of risk elements that make tracking fairly complex, for those who are in this space.
What have your learnings been around the leveraging of analytics for population health in the ACO context?
So to start at the beginning, the biggest challenge at the outset is the liberation of data from various EHR [electronic health record] systems, various registration/scheduling (“reg-schedge”) systems, and all these require analytics to attribute a patient to a PCP and stratify that patient. So we’ve just recently gotten over the hump around the basic extraction issues. We’ve now gotten more data than we have over the past two years; we’re now working with a vendor partner that is developing an EDW for us. We’ve got a variety of analytics that we can start to run now.
But the first challenge, and the first learning around it, is that you’ve got to pull that data out of the EHR, and then normalize and cleanse it and aggregate that data if you’ve got multiple systems across your organization, and standardize it, to get a comprehensive view of the patient. So if you extract data from your EHR and your reg/schedge, and those are different, and associating patients to their providers is a very basic, fundamental thing to do, but it’s actually not as easy as one would think. A patient is not always associated with the correct PCP. So it becomes an ongoing thing. But the lesson here is that you have to take your time in curating and normalizing and standardizing the data, so that your analytics will be as precise as you can make them. Stratification becomes possible when predictive modeling becomes possible, but only when you’re able to extract the data and normalize it. That’s number one.
Number two is, creating reports and dashboards is great, but if you can’t make the data actionable by clinicians, it’s for naught. If you’re not able to give them something they can really use to manage their population, it’s a waste of time. So the challenge is, how do you make these analytics packages and reports, upfront and center, for the clinicians to make actionable?
And how do you do that? What turns the key for physicians and other end-users of this data and information?
What turns the key is not hyper-alerting or over-alerting, but rather, creating an environment for clinicians and their affiliates to sit down and to reviews with their practice or medical group or the entire physician organization. In our case, the Baystate medical practices meet quarterly and review their metrics. They’ll review their metrics also in smaller groups, but in the larger group, they’ll review their 30-day readmissions and see who’s having high levels of non-reimbursable utilization. And those reviews influence the clinicians and the practices to do better. And they’re also seeing benchmarks of how they practice and others like them; and those are always good conversations to have, to share learnings.
If you leave them to their devices and give them a report card, but there’s no carrot or stick, it can be ignored. So having those collaborative sessions is very important.
Do you have a CMIO?
We do, we have a chief medical information officer, Neil Kudler, M.D., who works with those medical practices. And he’s just added another duty, of being the chief operating officer of [Baycare Health Partners, the physician-hospital organization supporting the ACO]. And you’d think that that’s a lot of work and might not make a lot of sense, but in fact, it does make sense to have a chief medical information officer who understands how these systems work, and also understands the operations of an accountable care operation.
What works in a physician dashboard, and what’s not necessary?
We have a care management platform that we use at both the health plan level and the provider, or medical group, level. You can drill in and out. You can see the individual patients and all their attributes, and also go up to the highest level, in terms of panels of patients categorized by disease states, and categorized. What we’ve seen as really effective is, number one, high-level views of the individual physician’s patient population—and the impact of that population in terms of their current disease progression. How are they doing on certain measures related to the ACO 33, or core, measures ? And ultimately, in terms of interventions, what percentage do the patients themselves adhere to? So really, the high-level metrics that help them manage their populations.
That ultimately drills down to tasks. If your patients aren’t adhering to specific treatments or procedures, complying by undergoing certain tests. And we’ve got nurse case managers, known as care managers, affiliated with the health plan, and also affiliated with the actual practices. It’s an interesting model, to be sure. There’s still a little bit of disconnect in terms of who really owns the care managers—the health plan or the medical groups? That’s still being worked out. Not unlike Geisinger or Intermountain, other organizations fortunate enough to have a health plan, there’s still some alignment that needs to e worked out. We do see some redundancy in terms of care managers affiliated with both, involved. So there’s still some coordination that needs to be worked out.
But the consistent component is that we’ve got one care management platform shared between those two groups. So that’s one learning, at least, that we’ve agreed to use one care management platform, so you get the same metrics, the same care platform that’s hard. So you can see that data, and it’s consistent across those two organizations.
In part two of this interview, Joel Vengco will discuss issues around aligning his organization’s health plan and ACO, and the implications of the learnings taking place at Baystate Health, for CIOs of other organizations.