RWJF Year in Research: An Interview with Randall S. Brown

Voted No. 4 Most Influential Research Article of 2012, Brown directs a project that tests the ability of disease management and care coordination programs to control health care costs.

Randall S. Brown, Deborah Peikes, Greg Peterson, Jennifer Schore and Carol M. Razafindrakoto are the authors of "Six Features of Medicare Coordinated Care Demonstration Programs that Cut Hospital Admissions of High-Risk Patients," published in the June 2012 issue of Health Affairs. This article was funded through the Robert Wood Johnson Foundation's Changes in Health Care Financing and Organization program.

Almost all the talking I’ve done about this research has been by invitation... Organizations have called me up and said, ‘Wow, this looks like real interesting stuff. Would you come and talk to us about it?’ "

What was the inspiration for this studythe critical knowledge gap you were trying to fill or the challenge you were trying to address?

People have come to realize that if you want to control health care costs, you’ve got to put most of your effort into the high-risk cases. Since at least the mid–1990s, there’s been an ongoing effort at the Centers for Medicare and Medicare Services (CMS) to try to figure out what could be done to reduce costs for people with chronic illnesses—because that’s where we know the most costs are. As part of that, one of the things CMS sponsored was the Medicare Coordinated Care Demonstration project. The idea here was, let’s let people who think they could do this well propose their own intervention that they think works, for the group of chronic illnesses that they want to focus on. CMS then selected the proposed programs that they thought looked best and tested them out with a randomized trial. 

We did an evaluation that showed that only two of the 15 programs (in the demonstration project) seemed to reduce hospitalizations, and none of them reduced costs. The motivation for this particular study was to then examine whether maybe we’re missing some effects of high-risk cases. Maybe, while an intervention may have no effects for some of the healthier people, among the people who really can benefit most, it does—but we miss those effects when looking at the overall program effects because the two groups are lumped in together. Or on the other hand, maybe some people are too sick, and there’s no way to really reduce their need for hospitalization, but the program works for those who are not quite so ill. 

The other motivation from the very beginning was that we wanted to figure out why some interventions are effective and some aren’t, and come up with some lessons. So we can say, “Here’s some guidance on what we’ve learned across all these demonstrations about how you can make your program one of the successful ones.”

In a nutshell, what were the key findings and why are they significant?

On the targeting side, we found that four of the demonstrations had significant reductions in hospitalizations when we defined a high-risk subgroup of cases—people with one of three chronic conditions associated with high hospitalization rates: congestive heart failure, chronic obstructive pulmonary disease, and coronary artery disease, and whose condition was severe. We measured that by whether they had had a hospitalization in the year before the program began. We also included the “frequent fliers” in our high risk group—people who are in and out of the hospital a lot over the two years prior to the study.

For this group of people, the effects were significant reductions: about 15 hospitalizations per 100 people per year, over the full six years of the study period. So that’s real money we’re now talking in terms of reduced hospitalizations. In addition, we had visited all of these programs, talked to the people who run them and observed them in action. So our paper also describes some of the characteristics that at least three of the four successful programs had in common, and that few or none of the unsuccessful programs had. 

Finally, I think of the most important findings was that four different types of organizations were the successful ones. One was in an academic medical center in an inner-city area with a poor clientele, one was part of an integrated delivery system in a rural area, one was a quality improvement organization that hired itself out, and one was a home health hospice agency. The point is that there are a lot of different settings in which these kinds of care coordination programs can be successful. 

Who did you most want this research to reach, and what influence did you hope to have?

I think the main audience is policy-makers at the state and federal levels. Here’s a good example. During the 2000–2010 period, Medicaid programs were spending a ton of money on telephonic disease management—where people in these call banks off in some other state or city are calling people and trying to explain their diseases to them, but are not connected to the patients’ physicians at all. These companies were claiming to have huge effects, and Medicaid programs looking for savings were buying them and spending a lot of money. We did some studies that showed that these telephonic disease management programs don’t work at all, and so there’s a lot of money that needed to be spent on poor people that’s going down the tubes and not benefiting people or saving states money. I wanted to let states know some things that work.  That, for instance, telephonic disease management is not where you want to go, but if you fund programs like this, you might have a chance to really save some money.

The other point we were trying to make is that, if you don’t focus on this population of high-risk cases, you’re not going to generate much savings, and it’s not going to address the basic problem of bending that cost curve. For instance, some of the people who are pushing the medical-home idea say everybody should have a medical home, and we’re going to pay this amount. So maybe they vary the premium a bit between people who are very sick and people who aren’t so sick—but they don’t vary it enough.  So one of the things that we hope that our study would have effects on is how these programs are actually implemented. 

What are a couple examples of uptake and impact you are particularly proud of thus far?

Almost all the talking I’ve done about this research has been by invitation. I’m not sending in abstracts to do a conference presentation on it. It’s the Institute of Medicine asking me to talk to them about it, the Congressional Budget Office, the Medicare Payment Advisory Commission, Congressional staff, physicians groups, the National Health Policy Forum. All of these organizations have called me up and said, “Wow, this looks like real interesting stuff. Would you come and talk to us about it?” So we’re pleased that influential organizations think we have something important to say to policy-makers.

Are there other unexpected audiences who have taken interest, and/or new audiences you did not initially think about who you feel would benefit from this research?

There’s one example I can think of. A group that does educational webinars for Texas physicians called me up and asked me to do a webinar, which I did, and it was the most subscribed webinar they’ve ever done. More than 200 physicians attended, and there was very favorable feedback according to the people who run the program. So I wasn’t expecting individual physicians to be all that interested, but apparently they were.

Are there any lessons from this project that will inform your future research, or that you’d share with other researchers who want to maximize the impact and reach of their work?

I think the key thing is paying attention to the “so what?” questions. I think you see a number of evaluations where they kind of do a thumbs-up or thumbs-down assessment—did this program work or not, did the treatment group have better results than the control group. But that’s just the bare minimum. We need to think about what we might be missing, whether there are especially vulnerable groups that the program did work for, even though it didn’t work overall. But the important qualification there is that you can’t just go poking around after the fact, trying to find something significant, because that has no scientific validity—and it won’t (and shouldn’t) get published.

2012 Year in Research No. 4 Winner
No. 4 Year in Research 2012 Winner

No. 4 Year in Research 2012 Winner

Six Features of Medicare Coordinated Care Demonstration Programs that Cut Hospital Admissions of High-Risk Patients

Policy-makers seeking to slow the growth in Medicare spending have appropriately focused attention on beneficiaries with multiple chronic conditions.

Read the research

Read About the Grant

Read About the Grant

The Foundation's Changes in Health Care Financing and Organization program was designed to support research, demonstration, and evaluation projects examining major changes in health care financing.

Read more

Most Requested