Readmissions Penalties Work. Here's Proof.

Hospitals with the highest incidences of readmissions (low performers) saw the highest reductions in readmissions when the financial penalties started kicking in.

A new study suggests that financial penalties provide an effective incentive to reduce avoidable readmissions, particularly at low-performing hospitals.

Researchers at Harvard and Beth Israel Deaconess Medical Center in Boston examined Medicare fee-for-service hospitalization data from more than 2,800 hospitals across the country between 2000 and 2013.

Based on 30-day readmission rates after initial hospitalization for acute myocardial infarction, congestive heart failure or pneumonia, researchers found that hospitals with the highest incidences of readmissions also saw the highest reductions in readmissions when the financial penalties started kicking in.

Study co-senior author Robert W. Yeh, MD, director of the Smith Center for Outcomes Research in Cardiology at BIDMC and associate professor of Medicine at Harvard Medical School, spoke with HealthLeaders Media about the study. The following is a lightly edited transcript.

HLM: What did your study find?

Yeh: Our initial goal was to examine whether or not the implementation of the HRRP [Medicare Hospital Readmissions Reduction Program] was followed by subsequent declines in readmission rates. We did confirm that.

It did look like there was an inflection point right about at the time of the passage of the ACA that looked like readmissions rates started to take a downturn.

Our second question was about whether some of these hospitals are penalized more than others. The way that the legislation probably was designed, you would hope to see that the hospitals that were penalized the most were the ones that had the worst readmissions rates and were the most highly incentivized to improve and hopefully the ones that improved the most.

That is what we observed. We observed that when we separated hospitals into four penalty groups, maximum, high, low and none, the hospitals that received the highest penalties did have the most rapid decline in their 30-day readmissions for those penalized conditions.

HLM: So, "skin in the game" works?

Yeh: I think so. It's a quite clear example that when hospitals are reimbursed, not just for how much they do but how well they do it, it makes an impact on their behavior.

That is what you would expect from an individual and this seems to incentivize organizations to act collectively to move in the same direction.

We are always careful about what is cause and effect. We know that readmissions penalties were followed by declines in readmissions, and they did so in a dose-dependent fashion; the more penalties you got the more you declined.

If I take off my epidemiologist hat, it does look like they work. The data is consistent with readmissions penalties working.

HLM: Were patients' socio-demographics factored into your findings?

Yeh: We did find that those hospitals that had the worst readmission rates and incurred the highest penalties were the hospitals that treated a higher percentage of minority of patients, a higher percentage of dual-eligible patients, a higher percentage of women, and the patient characteristics did look like they had more co-morbidities.

Those hospitals tended to be large, academic urban hospitals.

One of our concerns going in was that perhaps those hospitals that have high readmission rates were being penalized for things they can't control, but that is not support by the data.

If those hospitals had no control over their readmission rates, we would not have expected them to lower their readmission rates more dramatically than any other group. But it was quite clear that it was especially those hospitals that reacted the most favorably to the readmission penalties by dropping their readmission rates most significantly.

They dropped at about a 25% faster clip than hospitals that were not penalized.

HLM: Should socio-demographics be a factor in readmissions metrics?

Yeh: It's a broader debate than what we were able to look at in this study, but personally, I do think there are important socioeconomic factors that influence readmission rates that have nothing to do with the care those hospitals provide. They have to do with community resources and cultural influences on healthcare practices and outpatient medical care provided.

I sit on both sides. I see it as important to factor in. At the same time, these are exactly the types of hospitals that you want to see improve. In some ways by giving those hospitals a cushion for taking care of these patients, do you disincentivize them from improving? That is a two-edged sword and that is why there is such a fierce debate. The truth lies somewhere in the middle.

HLM: Are readmissions a valid quality metric?

Yeh: If readmissions correlated precisely 100% with mortality, then we would not need it as a metric, because we have mortality as a metric.

The fact that those are not correlated does not mean to me that readmissions are not a useful metric. It is true that there are many readmissions that are for reasons that probably have very little to do with hospital quality. It's a blunt instrument, no doubt.

As clinicians in hospitals, we can't say, 'we're doing everything right and there is no waste in the system.' There are preventable readmissions. We know that.

Readmission rates should not be zero, but the fraction of those readmissions that are preventable with better communication and care redesign should be prevented and that's what these readmissions measures are trying to target.

HLM: What can be tweaked to make readmissions a more effective measure?

Yeh: The main question is what are the actual interventions that are reducing readmissions? One of the challenges when we use administrative claims data is that they lack granularity; precise reasons for why patients get readmitted.

If you don't know the precise reasons for readmissions, it's hard to design rational approaches to prevent them. You need to go into your own health system and get detailed electronic medical record data or chart review and understand who are these patients who are coming back, not at the 1,000-foot level like we have done here, but on the individual patient level and understand what are the things that may have prevented that individual patient from getting readmitted.

It's not sexy, splashy research, but when we aggregate and understand those reasons we can design appropriate interventions.