On Aug. 28, a team of researchers in the Waltham, Mass.-based Global Institute for Emerging Healthcare Practices, a division of the Falls Church, Va.-based CSC, published a new report, “Coordinated Care: Assembling the Patient Information.” The report, authored by Jane Metzger, Rebekah McLear, and Jordan Battani, examines what accountable care-type organizations, referred to in the report as “coordinated care delivery organizations,” or “CCDOs,” are facing with regarding to the data and information requirements needed to facilitate coordinated care or accountable care.

As lead author Metzger, principal researcher for the group, and her colleagues, write in their report, “Critical to success for any CCDO is assembling patient-specific information that provides a foundation for care and care coordination across the continuum in line with the new expectations. Thanks in large part to the HITECH [Health Information Technology for Economic and Clinical Health Act] incentive program, the pace of adoption of electronic health records (EHRs) has accelerated to the point that many hospitals and physician providers forming or participating in coordinated care delivery organizations are capturing important medical record information online and integrating EHR features such as clinical decision support and electronic patient tracking into patient management.

"This is a critical building block for the HIT needed for care that is coordinated across the continuum,” the authors note. “In many cases, however, electronic patient information captured locally in a hospital or physician practice (or medical group) is stored and managed separately in a local EHR,” and with patients regularly moving back and forth between and among hospitals, physician offices and clinics, nursing homes, and in-home care, the challenges embedded in accountable or coordinated care are thus many.

Jane Metzger

What’s more, the authors contend, even as many in healthcare are heralding the emergence of big data and data analytics as supports for coordinated or accountable care, the reality of leveraging data and information in still-fragmented landscapes remains profoundly challenging. As the new report was released, lead author Metzger spoke with HCI Editor-in-Chief Mark Hagland regarding the goals, revelations, and implications of her group’s important new document. Below are excerpts from that interview.

How did you start down this path looking at external sources of patient information?

In any conversation or publication about accountable care or coordinated care, everyone is talking about “data analytics” or “population health analytics.” We all know quite a bit about what EHRs are likely to contribute today and when providers achieve HITECH Stage 2, but we also know that won’t be enough because patients will often cross institutional and, therefore, also EHR [electronic health record] boundaries.  So we decided to look under the covers of the most likely other sources of information.

You’ve spoken of the “arm-waving” around data analytics. Obviously, analytics are important. But you believe that many healthcare IT leaders are underestimating the challenges involved. And obviously, it’s a very complex landscape out there, correct?

Well, I think it’s complex in many different ways. In any different coordinated care delivery organization, hopefully, all the clinical partners are participating in HITECH; but they aren’t all going to reach Stage 1 or Stage 2 at the same time; and even if they do, they won’t all have approached any given stage in the same way. So you’ve got a mix of conditions there. And then, even looking at a limited set of the most likely external sources that we looked at, you’re going to have a mix for different patients of what’s available and not available. Just to give one example, the eligibility verification information is really one of the only ways to know about patients who are showing up somewhere else, and they’re in trouble. And that isn’t even available for Medicare fee-for-service patients, because it’s not a requirement of that type of insurance coverage that there be any eligibility process.

The card the patient presents basically says they’re eligible. So the more deeply we got into this, the more we realized that, for any group of patients, there’s going to be a somewhat unique mix of external sources available to you. Also, getting access to some of those sources—for example, getting claims data from FFS Medicare—might be easier than with regard to a private insurer that’s not set up to give out that kind of information or doesn’t want to. So that’s a second type of mix.

And thirdly, if you think of patients in terms of their different health risks, you’re going to want information on them in different ways. In some cases, you’ll have such an urgent need that you’ll call people outside the organization; in other cases, you’re going to want to set things up so that someone calls you. And in the third instance, you may have an external source that is perfect for that need; let’s say for example, that you decide that it’s really critical for all your high-risk patients to know what their treatment compliance is. And one indicator will be whether they pick up their medications or not.

So for your high-risk patients, you may want to engage in obtaining medication histories regularly, because there could be an associated cost—so that you can compare those to the prescription information you have in your EHRs, so you know sooner than when they show up in the emergency room that they’re not picking up their meds for cardiac issues, or whatever. So this is really the complexity of it: not all sources are available for all types of patients, and there isn’t an equal level of urgency in terms of finding things out for all patients. So that really helped us think through what you might need to do here, because you might have multiple ways to do things.

Many of the sources you looked at are (or are related to) health plans or other payers.

Yes, the organizations that administer health insurance—public or private—have a wealth of information, much of it highly standardized and coded.  We looked at information stored from several related electronic communications between providers and health insurers or other payers around eligibility verification and referral authorization, claims, of course, and information about medications from Pharmacy Benefits Management (PBM) companies or pharmacy network providers such as Surescripts.  The advantage of all of these is that they are all more comprehensive than the information in the local EHR, except in the rare instance that a patient never accesses care outside of a health system with a fully implemented, comprehensive EHR.  That situation isn’t true for most patients.

Is all of this automatically available?  Is any of it in use today?

Iwouldn’t assume that anything is available without extensive negotiations and possibly even some financial agreements.  Plus there is additional work to put the information to work—integrate it with other information, get it to the right person at the right time, etc. But Coordinated Care Delivery Organizations [what we call ACOs] are already getting claims information updates monthly from Medicare and other payers.  Collaborative arrangements between private insurers are becoming more common, and some of the big national companies like Cigna and Aetna have national programs.  When I was looking recently at models for coordinated care, I found numerous examples where providers were already receiving notification when covered patients popped up in emergency rooms and as emergency admissions—the payer would provide work lists for case managers or care coordinators each morning.  And Surescripts reports that it provided medication history 586 million times in 2012.  So, yes, tapping these external sources is possible, but certainly not automatic.

You also looked at case management systems as a piece of the information puzzle.

Yes, that is likely to be the only electronic record of who’s on point for case management or care coordination and the current care coordination plan.  Care coordination—as opposed to clinical care—must also be “coordinated,” if you will to avoid overlapping, and, to the patient confusing, monitoring and support.  Plus, if data analytics are ever to help case managers cut down on telephone calls to see how patients are doing, the basic information about care coordination needs to be tied in.  Obvious examples would include lists of recently discharged patients who missed their scheduled follow-up appointment, or high-risk patients not complying with their regimen of prescribed medications.

You tied your look at exchanging information with external EHRs to HITECH meaningful use. Can you explain some of the thinking there?

HITECH requires specific capabilities for information exchange in certified EHR technology and makes HIE-related use a requirement of meaningful use by providers.  At least in the near term, the electronic structures and use cases defined for HITECH are likely to be most broadly available within any community of hospital and physician participants in a CCDO.  In Stage 1 requires exchange of very basic information—allergies, medications, problems, and lab results—a core information set for any patient.  Stage 2 advances to two very high-priority use cases for transitions in care:  for hospitals, the discharge summary, and for EPs, the referral request. HITECH also provides HIT capability for document exchange by requiring summary of care records to be displayable in human-readable form.

Is a lack of granularity of data exchange standards an obstacle in that regard?

I think it’s a continuing obstacle. I think we’ve made some progress because of HITECH. But I think there aren’t just problems with the coding structures; people still think there are gaps and lack of sufficient granularity in LOINC. I’m on a listserv with people who are currently having the same conversation about SNOMED, which, as applied to EHRs is still relatively new for most people. And there are all kinds of issues about usability, particularly since having the provider assign the code, at least for billing purposes, has been the norm under ICD-9, so there’s a lot of work to be done there; but also with regard to the packaging of standards. We looked at what HITECH says you need to be exchanging, because we thought that that was the near-term, most-likely-available, capability. And if you look at the details of that, even in Stage 2, only the most core information—problems, medications, allergies—really has to be data-coded to stated standards. So even this discharge summary has a lot of information in it that doesn’t have to be coded; it has to be extractable for human readability.

So yes, I think the work is proceeding, but you’ve referred to something I said in the paper, which is that this whole issue of data and the technology that’ll put data to work, and in this context, I mention true interoperability of EHRs, as well as large-scale databases just sitting there waiting to be tapped—these are going to remain works in progress for years to come. And the federal government, through the HITECH program, has driven a lot of the progress. So I can fully sympathize with asking the federal government to create additional standards, because I’m not sure how it can happen otherwise; voluntary efforts just don’t move as fast.

I actually decided that what’s recorded in the EHR becomes even more important when you look at these external sources, because really, except for claims, the external sources aren’t very comprehensive. And we all know the criticisms of claims: there can be an incredible time-lag, and the information has been recorded for reimbursement and doesn’t necessarily align with what the medical record might say. So at the core of this really has to be meaningful use of the EHR. So I don’t see that data analytics, fueled by a mix of sources, is a magic bullet; because unless the EHR is really well-designed and the truly important things are structured, and the problem list and medication list really are up to date and current, I don’t think it’s going to be possible to move the dial on coordinated care the way we have to.

You and your colleagues recommend developing an information plan, rather than a data plan.

Yes. Putting a robust data infrastructure in place for large populations of patients is realistically going to be the work of the next decade, with progress occurring in increments. It’s critical to work not just on that front, but also get information—in any useable form—to care providers and care coordinators on the front lines today. That’s why we recommend orienting planning around important things that can be accomplished with external information. An example is coordinating care for patients during a transition.  The quickest things to do might be sending text files of ED encounter notes and discharge summaries to medical homes, while planning to pilot exchange of electronic files as clinical partners implement Stage 2-certified versions of their EHRs.

What’s your advice for CIOs and CMIOs in organizations just starting down the path, ones that don’t even have a strategy yet?

Well, I would advise them to follow the principles we’ve talked about in the paper, while realizing that if they get certain basics pinned down in the EHRs in their clinical partners, they can start doing some pretty important things. I also would urge them to start planning for some kind of a platform that will assemble information from multiple sources, and start thinking about those external sources that are going to provide the biggest bang for the buck today. That could include medication histories from a particular PBM [pharmacy benefit management company] or from SureScripts, for a particular high-risk population. Another particularly high-impact action might be if they have collaborative contracts with commercial payers, to start getting lists of who popped up in the ED last night.

I think the real danger here is letting the perfect be the enemy of the good; and that’s why we called what we think people should be aiming for here, an information plan rather than a data plan. Of course, whenever data can be captured and used, it should be. But there are so many opportunities being lost today. I read somewhere about what trouble hospitals were having sending electronic discharge plans to medical homes, and the small percentage that were doing it routinely.

So if you have a high-risk population, it’s in the interest of the hospital to begin working with those populations, with regard to readmissions. And more and more patients are under risk-based insurance design, so it’s also in the interest of the medical homes or PCPs [primary care physicians]; and I keep reading that this still has a lot of challenges associated with it, so, working really hard on one very high-value information exchange like that will help you learn a lot in the process that will help you with a lot of other exchange processes. And I also think with regard to meaningful use, that organizations should pilot doing the Stage 2 summary of care records, as defined by HITECH, they’re very, very significant.

In the end, how hopeful are you that all this can be accomplished?

Well, I think that doing useful things with data is a little bit over the horizon, but not that far away. But I did come away with realizing how important the EHR is in all this work, because there’s no external data source that takes the places of having a meaningfully implemented EHR.