Researcher Disputes Link Between Lower Readmissions, Higher Mortality

A controversial JAMA study faces a rebuttal from a researcher who believes readmissions policy has worked largely as intended, with no linkage between lower readmissions and higher mortality.  

A study that links lower hospital readmission rates higher mortality among heart failure patients is being challenged.

The study, which appeared this week in JAMA Cardiology, estimates that as many as 10,000 heart failure patients could die prematurely each year because of well-intended but misguided efforts that keep them out of the hospital to avoid the financial penalties attached to higher readmissions.

The findings flatly contradict a July study in JAMA that reviewed more than six million hospitalizations and showed no linkage between reduced 30-day readmissions and increased post-discharge mortality.

Kumar Dharmarajan, MD, chief science officer at Clover Health and lead author of the July JAMA study, spoke with HealthLeaders Media about the conflicting conclusions. The following is an edited transcript.

HLM: These two studies attempt to answer the same question, but come up with the opposite answers. How is that possible?

Dharmarajan: Their study – and I say this with respect – is technically limited for a few reasons.

What they did, essentially, is look at 30-day readmission rates and 30-day mortality rates over time. The problem with that is, just because two things are changing over time, it’s hard to know if they’re related. As a dumb example, in the summer people swim more and eat more ice cream and there are more drownings. Does that mean ice cream causes drowning?

Their analysis is interesting, but they don’t go to the next level, which is what we did. The premise in our paper was that, if it is true that readmissions are causing greater mortality, we would expect to see that when we looked at trends, not just nationally, which is what they did, but within each hospital.

We looked at 4,000 or more hospitals in the United States caring for patients with heart failure. For each hospital we plotted that trend in readmission and mortality. If readmission declines are harming patients, we would expect to see that the hospitals with the greatest decrease in admissions would have the greatest increase in mortality. We found that there was no relationship. To the extent that there was a relationship, it was the opposite. Hospitals that were dropping readmission rates had mortality rates that were improving.

The second limitation is that they didn’t use national data. They used a heart failure registry called Get with the Guidelines that includes data from a little more than 400 hospitals. Those 400 hospitals are not randomly selected. They wanted to participate in the registry. That would be OK if those hospitals were representative of the national picture, but they are not.

They actually see that readmission declines are mostly happening after 2013. Nationally, most readmission declines in the last decade happened between 2010 and 2012 and not much after 2013. They are clearly working with a different data set.

The second thing that proves they are looking at different hospitals is that they found that heart failure admissions were increasing over time. Nationally, they’ve been declining. They’re looking at a small non-representative sample. Even if they are absolutely right – and I still have my concerns – that those hospitals are demonstrating some sort of trend of harm, we clearly know that their data is not representative of what is happening nationally.  

The other thing is they are just looking at heart failure. We looked at heart failure, pneumonia and heart attacks, and the national data sets. We had 30X the number of hospitalizations in our study. What we found was consistent across all three conditions. They are more heart-failure centric as opposed to trying to understand what is happening across conditions.

HLM: It sounds like you’re dismissing this new study.

Dharmarajan: I don’t believe their findings, which simplistically plot a trend over time with readmissions and mortality. There is a real danger in looking at things like that. For every 1% decline in readmissions, they’re seeing a 1% increase in mortality! That doesn’t pass the sniff test!

Unfortunately, there is a predilection to want to believe this new finding because no one likes penalties in the first place.

HLM: Is it possible that there is something uniquely intrinsic in heart failure care that makes patients more vulnerable to readmissions reductions?

Dharmarajan: I would say those are the types of patients who benefit most from a focus on readmissions. What have hospitals done nationally in response to these penalties? They’ve improved transitions in care. They are more likely to focus on getting the medication right after discharge, and getting appointments set up when they leave the hospital and appropriate support services when they return home and coordinating with community-based organizations to help after discharge, making sure they’re seen soon after discharge in an outpatient setting. These are commonsensical things that were not a focus before these penalties came in. It hard for me to believe that things like that would cause harm.

HLM: Do you believe CMS’ readmissions policy has worked as designed?

Dharmarajan: Clearly readmission rates have declined. To the extent that the aim was to lower readmissions, they’ve had an effect there. The way that they’re being interpreted by the federal government and built into policy could be improved. Right now you have winners and losers. I would love to have a world where everyone won if the outcomes improve across the country and not so much zero sum, because people are staying healthier and out of the hospital that is going to save Medicare money and provide better care for beneficiaries. I am not sure you need to have a penalty to do that.

HLM: What about concerns that readmissions penalties harm safety nets that can’t cherry pick their patients?

Dharmarajan: It’s true that hospitals caring for vulnerable patients of low socioeconomic status are more likely to be penalized, but that is because the outcomes have been worse at those hospitals. To me the ultimate accountability is to the patient, not the hospital. There is a study that showed that readmission rate disparities between safety net and non-safety net hospitals have declined over time as these penalties have kicked in, which means that if you are patient at a safety net hospital, your improvement in readmission rates have been more marked that patients at non-safety net hospitals.

HLM: How are hospital administrators and clinicians supposed to respond to these diametrically opposed findings of these studies?

Dharmarajan: You need to do things that make sense for patients, like make sure their care transition is top notch. Make sure that they are getting great post-discharge care. Make sure their meds are correct and that they have adequate support and appointments in place. Let’s not send someone out of the hospital without having their meds checked or without setting up their discharge plan and expect a sick 85-year-old to set everything up for him or herself.

If the message from the new paper is that hospitals are keeping people out when they should be readmitted – and again I don’t believe it -- they need to stop doing that.