Are Heart Failure Outcomes Measures Hurting Patients?

Analysis of Medicare's Hospital Readmissions Reduction Program

John M. Mandrola, MD; Larry A. Allen, MD, MHS


December 14, 2017

John M. Mandrola, MD: Hi, everyone. This is John Mandrola from | Medscape Cardiology. I'm here at the American Heart Association (AHA) 2017 Scientific Sessions and pleased to have Dr Larry Allen from the University of Colorado. He's a cardiologist, outcomes researcher, and second author on a provocative paper looking at the Hospital Readmissions Reduction Program (HRRP).[1] Larry, welcome.

Larry A. Allen, MD, MHS: Thanks for having me.

Dr Mandrola: Tell us about this paper.

GWTG Registry for Heart Failure

Dr Allen: There has been a lot of interest in the cardiology community around new outcomes measures for heart failure hospitalizations. These measures encourage hospitals to do a better job of transitioning patients out of the hospital after an admission for a decompensation and keeping them out of the hospital for the next 30 days.

We leveraged the Get With the Guidelines® registry, sponsored by the AHA, to look at relationships with the 30-day readmission rate as well as the 30-day mortality rate in hundreds of hospitals across the country.

Dr Mandrola: These are outcome measures, not process measures.

Dr Allen: That is right. When thinking about quality improvement and how care and quality can be graded, we can look at processes. Did you give a patient with a low ejection fraction a beta blocker? Or we can look at an outcomes measure. Did the patient live longer and feel better because of the things that you did? This is a shift from measuring exact things that were done to measuring how patients are doing.

Dr Mandrola: How did you go about studying this? Tell us about the Get With the Guidelines registry.

Dr Allen: Get With the Guidelines is a quality improvement program. Essentially, participating hospitals submit data on all of the patients admitted and discharged with the primary diagnosis of heart failure. One of the advantages of Get With the Guidelines is that it collects detailed clinical data, which are different from the coding data or the administrative data that many people have used to look at how hospitals are doing. The AHA collects these data and then feeds back to institutions benchmark-adjusted information about how they are doing.

Implications of HRRP

Dr Mandrola: Did you look at these data before and after implementation of this program?

Dr Allen: That is right. In 2010, Medicare put into place a value-based purchasing program called the HRRP. Essentially, they saw what rates of 30-day readmission were for key diagnoses, including heart failure, in all 4000 hospitals across the country. They began reporting those back publicly in 2010 and started penalizing hospitals in terms of their Medicare payments if they had rates of 30-day heart failure readmission that were higher than expected. A lot of attention was around this because there were significant dollars attached.

"One of the concerns [with HRRP] is whether there were unintended consequences of doing that."

Dr Mandrola: There is a big incentive now to change this outcome measure of readmissions.

Dr Allen: That is right. All of a sudden, you had hospitals around the country focusing on their patients with heart failure and watching what happened to them over the next 30 days with the goal of improving transitional care and keeping them out of the hospital.

One of the concerns is whether there were unintended consequences of doing that. If I tell you that I am going to penalize you if your patient gets readmitted to the hospital, do you deliver better care and start doing things to lower that rate?

On the back end, you could not admit a patient who might need to be admitted because they are going to count against you. Or, more concerning, you also might not admit patients who you think will get readmitted over time because they will be in your pool.

If I block patients from being admitted because I'm concerned about very high risk, or I do things to keep patients from getting readmitted after discharge, am I creating a scenario where rather than improving transitions and keeping people healthy and out of the hospital and living longer, I actually do the opposite? I am not giving people care in the hospital that they would need; thus, am I increasing the rate at which they are dying? The question we wanted to answer was whether this focus on the 30-day readmission rate improved or worsened mortality.

Dr Mandrola: You looked at a lot of hospitals and more than 10,000 patients. What were the findings?

HRRP Implementation and Heart Failure Outcomes

Dr Allen: Using detailed clinical data from Get With the Guidelines and adjusting for differences between hospitals and the patient mix (which is critical because not all hospitals are taking care of the same patients), we found that once the HRRP went into effect in 2012, the rate of readmission at 30 days went down by a little over 1%. We found that the mortality rate over the next 30 days went up by even a slightly larger degree.

Dr Mandrola: It is provocative that you could definitely show that readmissions were going down but that it was not associated with a reduction in mortality.

Dr Allen: It's very significant because, obviously, the intent of these value-based purchasing outcomes measures is to improve quality of life and survival for these patients. In focusing on the 30-day readmission, we may not necessarily be achieving both of those goals.

"All we are looking at here are data over time, and we cannot necessarily know what is causing what."

Dr Mandrola: In the discussion of the paper, you cite some literature that seems to align pretty well with this. Prior papers show similar observations.

Dr Allen: The Veterans Affairs healthcare system collected data and found the same kind of inverse association. As those 30-day readmissions were pushed down, it looked like there was an association of an increased rate of mortality.[2] There are a couple of datasets using detailed clinical data to adjust for patient differences where it does seem like this concerning inverse association exists.

Dr Mandrola: We have said the word "association" numerous times; that does not mean "causation."

Dr Allen: That is right. All we are looking at here are data over time, and we cannot necessarily know what is causing what.

Dr Mandrola: Dr Harlan Krumholz's group has a paper in JAMA[3] showing that readmissions are going down, and mortality may be slightly improved or at least not worse in specific hospitals. How does your paper contrast with that?

Dr Allen: We are not the first people to look at this. One of the key differences between Dr Krumholz and the Yale group's analyses versus the one that we have been talking about is that theirs uses Medicare claims data. They looked at billing information. The reason that we tried to leverage Get With the Guidelines is that it collects much more detailed clinical data out of the electronic health record (EHR)—essentially, people look back at the EHR and get detailed data that are not in claims data.

The methods are similar in terms of what is being looked at. With the Get With the Guidelines data, there may be a better ability to adjust and account for differences in patients and see the small increase in mortality that was not available when only looking at claims data.

What Are the Policy Implications?

Dr Mandrola: Anybody who practices knows that these metrics sometimes are not ideal. Can you speculate on what the policy implications are? Do you think this data signal will change policy?

Dr Allen: I do not want to be negative because quality improvement is working to create policies and payments that encourage ideal care and improved patient outcomes. It is an evolution. Many years ago, when we were just fee-for-service, patients got lots of care and maybe not as much quality associated with it.

There have been a number of changes over time to try to do a better job of encouraging quality with care. This move towards outcomes measures has created a lot of positive thinking and focus on what is important to patients and how to encourage hospitals to get to good outcomes that reflect the entire care that patients are getting.

I do not think that we want to go backwards and get rid of the concept of outcomes measures. We probably need to adjust how we are doing them and account for unintended consequences of the policies as currently written. My goal would be to use this kind of information to adjust what we are doing and to make policies considering all of the outcomes that are important to patients.

Dr Mandrola: Excellent. Larry, thanks for being with us.

Dr Allen: Thank you for having me.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.