On Jan. 23, leaders at the American Health Quality Association (Washington, D.C.) held a press briefing to highlight the publication of an article in the Journal of the American Medical Association (JAMA) in which leaders in AHQA documented an impressive decline in hospitalization and rehospitalization among Medicare patients in communities in which Quality Improvement Organizations (QIOs) have coordinated interventions that engaged whole communities in care improvement. The JAMA article, entitled “Association Between Quality Improvement for Care Transitions in Communities and Rehospitalizations Among Medicare Beneficiaries,” was authored by six healthcare researchers, including lead author Jane Brock, M.D., MSPH, chief medical officer at the Colorado Foundation for Medical Care, that state’s QIO.

As Dr. Brock and Joanne Lynn, M.D., M.A., M.S., a fellow study leader and article co-author, noted during the press briefing, 14 communities participated in the study; all 14 were communities in which QIOs had become involved in community-wide healthcare improvement efforts. Those communities participating in the study saw an average 5.7-percent reduction in rehospitalizations and a 5.74-percent decrease in initial hospitalizations during the same two-year period. The core innovation involved, Drs. Brock and Lynn, and Mary Ellen Dalton, R.N., Ph.D., noted, was the success of QIOs in working with coalitions of stakeholder groups within individual communities to create consensus-driven interventions to improve care delivery.

The 14 communities involved in the project were: Tuscaloosa, Ala.; Denver, Colo.; Miami, Fla.; Atlanta, Ga.; Evansville, Ind.; Baton Rouge, La.; Lansing, Mich.; Camden, N.J.; Albany, N.Y.; Pittsburgh, Pa.; Providence, R.I.; Harlingen, Tex; and Whatcom County, Wash. Within those communities, the researchers found that quality improvement interventions were associated with about 6,800 hospitalizations and 1,800 rehospitalizations averted per year.

Dr. Brock spoke shortly after that press briefing with HCI Editor-in-Chief Mark Hagland regarding her participation in the study, and her organization’s efforts more broadly. Below are excerpts from that interview.

Can you tell us about the origins of the program and of the study that led to the JAMA article?

CMS [the federal Centers for Medicare & Medicaid Services] creates its  national quality strategies, and issues a single core contract for each state and territory; and the QIOs are essentially companies or agencies that hold the QIO contracts. And when officials at CMS have an appropriate project, they release it for certain types of contractors to bid on. So they had released a contract for two special studies in 2006-2008, and we at the Colorado QIO had received both contracts. One was to recruit hospitals to target any elements of non-beneficial utilization, and to undertake a 15-month, four-state project, to reduce that utilization. The second study was given to Colorado to see if the care transitions intervention could be applied. The leading evidence-based intervention is called the Care Transitions Intervention, or CTI, and Eric Coleman [Eric A. Coleman, M.D., M.P.H., professor of medicine and head of the University of Colorado School of Medicine’s Division of Health care Policy and Research, and creator of the Care Transitions Intervention]

 is at the University of Colorado, right around the corner from us. And the purpose of the second study was to see if ordinary providers could be recruited into a care transitions project, and thereby improve readmissions. We got both contracts the same day and combined them. And New Jersey and California were two states involved. And about 18 hospitals decided to work on ICU [intensive care unit] utilization at the end of life. In Colorado and New Mexico, those hospitals decided to focus on readmissions. And CMS allowed us to put the hospitals in Colorado and New Mexico into both projects, meaning that we could use resources to feed them data.

Jane Brock, M.D., MSPH

So hospitals in Colorado and New Mexico were involved in both? And hospitals in New Jersey and California were only involved in the non-beneficial utilization project?

That’s exactly right. Especially in Colorado, we recruited the hospitals to participate in an effort to reduce readmissions, and fed utilization data back to them, and then also got them involved in the second project.

Were the Colorado hospitals the most comprehensive in their approach?

No, they were the most successful; they looked at utilization and readmissions. And the QIO provided them with the care transitions information and training. All that happened in 2006-2008; and Joanne spearheaded the effort to turn it into this 14-community effort. So the target was a community; and CMS didn’t mandate which interventions should be performed. So we in the QIO sat down and did root cause analysis: what causes readmissions in our communities? And then we decided who should be at the table in each of the 14 communities. And when we took a population health approach, it changed everything. Because in the past, the focus had been on developments within hospitals. But this project involved looking at population health across communities. So we entered the claims data and realized that we needed to work with particular hospitals and nursing homes, and then we brought the leaders of those organizations together, and asked them, well, what do you think causes readmissions?

And what did everyone learn?

At the beginning, the providers started out saying things like, ‘Yeah, we have a problem with our heart failure patients, so let’s look at discharge instructions to them.’ And that’s a reasonable approach; but you realize that the real roots of this are perhaps not heart failure as such, but the challenges inherent in home-based patient management. They go home from the hospital and eat a pizza; and why do they go home and eat a pizza, which is incredibly high in sodium, when you’ve just spent days educating them while in the hospital? It turns out to be very pedestrian reasons like, ‘Well, I was tired and all I had in my fridge was a pizza, because I had been in the hospital for seven days.’ So they might be old, confused, tired, without transportation, etc.

So we came to the realization that disease as such doesn’t matter so much; instead what matters is the level of community support. And that is exactly the premise of the coaching intervention: that people need to be prepared to take on the burden of self-management, and need resources to help them, and sometimes it might be Meals on Wheels or something like that. The constellation of support varies by community.  And increasingly, the area agencies on aging might begin to bridge the gap between community services and patients’ ability to manage their own situations.

The Pennsylvania group was amazing; they got involved in this work in just that way, and made fantastic drops in readmissions. I’m not saying that disease is not important, or that the activities of providers are not important, because they’re extremely important. But a readmission occurs when the burden of a patient creates a situation that can’t support their staying out of the hospital. If someone is homeless and demented, it’s going to be challenging. And look at heart failure; there’s no real, easy way to change a person’s diet overnight, especially at the point of discharge.

So we give people a lot of instructions; but when they get home, they often lack the capacity to function on their own. Do they live with family members? Do they have neighbors? What is their economic situation? A lot of the articles in the JAMA package talk about risk adjustment for various factors involved in this. But that gets very complicated. We’d have to know whether they live alone, whether they live in a safe enough neighborhood for community services to come into patients’ homes or not, factors like those. So it was easier for us just to throw out that risk adjustment and work on these interventions.

What types of IT and health information exchange (HIE) foundations will be required to replicate this kind of work elsewhere?

Several of these community-based organizations have Section 3026 awards. Section 3026 of the Affordable Care Act allows area agencies on aging to partner with hospitals to deliver interventions like coaches, and to build Medicare. And that means that the Hospital Insurance Trust Fund for the first time has funding to provide services in people’s homes after discharge. It’s one of the demonstration projects under the ACA. Through that section of the ACA, 75 sites have received awards, including Denver. So Medicare is now paying for community support under very specific conditions and contracts. And one area in which the community-based organizations are making faster progress, one thing that seems to make a difference, is if the community-based organization has access to the discharge planning record from the hospital, and if it becomes part of the medical record. Usually, the electronic medical record is often completely separate from the discharge planning record. And so physicians can’t directly see what the discharge planner is doing.

So when the discharge planning record is easily linked with the EMR and is viewable by the community support agency that is serving recently discharged patients, things get better faster. The other thing, and this is my opinion—I have felt that in some ways, the perfect has been the enemy of the good enough for some time in terms of linking physicians and hospitals. I’ve got to tell you that tremendous progress could be made if you could just notify the primary care physician about a patient’s discharge the moment it happens, even if you can’t do it in a robust, HIPAA-compliant way; because the physicians are technically taking responsibility.

There’s been this sentiment expressed by providers that, ‘Oh, we’ve got to get our e-mail systems set up a certain way.’ But honestly, if you can just get this connection to work, that helps. And everybody believes that the best transition occurs when the hospitalist has a conversation with the primary care physician at the point of discharge, but that never works out in terms of the primary care physician’s workflow. So if you can just post that notice to the primary care physicians, that would be great. I’ve long said, what we need essentially is a HIPAA-compliant bulletin board for this communication.

Perhaps through the federal DIRECT program?

Yes, perhaps; and maybe the primary care physician only needs to know one specific piece of new information, so maybe it’s just that Betsy Smith had one change in meds; that can make all the difference.

What could CIOs, CMIOs, and other healthcare IT leaders be doing right now to lay these needed foundations?

I would urge them to think about creating some version of this electronic bulletin board; it seems to me that there ought to be a way to facilitate that. And if there were a great way to securely post messages, then the information could be texted, and then the doctor could go and check, in their own time, on Betsy Smith [the patient]. And then if they see that, ohmygosh, Betsy Smith is a train wreck, his office could reach out to Betsy Smith and get her in for an appointment right away. But at least that would allow the person receiving the medical responsibility to be in charge of deciding the level of worry that is appropriate.