Hospital Readmissions are Not the Enemy

The Centers for Medicare & Medicaid Services has all but declared war on readmissions. But one researcher suggests that the relationship between readmission rates and quality is flawed.

Researchers and physicians at The Johns Hopkins Hospital are challenging the notion that readmissions are an accurate measure of quality.

In a study this month in Journal of Hospital Medicine, hospitalist Daniel J. Brotman, MD, and his colleagues examined nearly 4,500 acute-care hospitals' hospital-wide readmission rates and compared them with those hospitals' mortality rates in six areas used by the Centers for Medicare & Medicaid Services: heart attack, pneumonia, heart failure, stroke, chronic obstructive pulmonary disease, and coronary artery bypass.

The researchers found that hospitals with the highest rates of readmission were more likely to show better mortality scores in patients treated for heart failure, COPD, and stroke.

Adjusted odds ratios indicated that patients treated at hospitals that had more readmitted patients had a fractionally better chance at survival than patients who were cared for at hospitals with lower readmission rates.

Brotman spoke with HealthLeaders Media about the findings. The following is an edited transcript.

HLM: What prompted you to raise questions about readmissions as a quality metric?

Brotman: We've been involved at Johns Hopkins Hospital with a number of readmission initiatives and we have tried to reduce readmission rates using a lot of the interventions published in the medical literature. We were struck by the difficulty of reducing readmissions in outpatient populations.

When we did deep dives into causes of readmissions for individual patients, sometimes we saw situations in which providing more comprehensive, detailed, or sophisticated care was leading to readmissions. The defects that lead to readmissions are usually not related to the care provided during the hospitalization.

We were particularly alarmed to see that CMS Star Ratings rated readmissions similarly to mortality. That raised a question: Are readmissions really a quality metric? Certainly readmissions are a measure of how much care a patient is getting in the inpatient setting to some extent, but are they a measure of quality or do they measure something else?

HLM: You talk about "unintended consequences" with the readmissions metric. Please elaborate.

Brotman: One of the ways to prevent a readmission is to keep someone out of the hospital at all costs. That is not necessarily good for patient care. With increasing financial pressures to reduce readmissions, there are going to be unintended consequences.

We thought it would be worthwhile to ask if readmission rates at the hospital level and mortality rates at the hospital track. If hospitals that attempt to have better mortality rates also tend to have lower readmission rates, that would lend credence to the notion that readmission measures are one measure of quality of care that track with other measures of quality care, such as mortality.

In fact, we saw the opposite.

It wasn't a relationship that was so strong to say that people should flee from hospitals with low readmissions rates because their patients are more likely to die. Correlation does not always mean causation.

But it raises the question as to whether these all-cause readmission rates are a valid measure of quality. The purpose of the study was to shine a light on that and make sure we are not assuming that hospital readmission rates are truly a measure of quality. They are a measure of a lot of factors and quality can be one of them, but it is not a dominant factor.

HLM: What caveats would you attach to make readmissions a better metric?

Brotman: It's hard to make it a proper metric because for most patients who get readmitted, and this has been validated by others, the readmission isn't preventable by things that could have been done differently by the hospital or the discharging provider.

Most readmissions are either a function of the patient's illness, or a function of the patient's quality of outpatient care, or a function of the patient's engagement and follow through. Most are not due to defects in care.

The benefit of looking at readmissions is in making sure that you're addressing the particular care defects that might lead to readmissions. Doing root cause analyses on readmissions that you know are preventable can be a useful enterprise to identify system defects. But, you have to look at actual care defects.

CMS is hoping to use administrative data without having to do a deep dive on individual patient charts to calculate this metric. This is an experiment that is failing. It is not a proven quality and I want that discussion to be out there.

HLM: What are the practical effects of this misapplied readmissions metric on patient care? 

Brotman: In the aggregate, it probably is helping patient care to focus somewhat on readmissions, in that it helps institutions pay attention to potential care defects and to use the episode of care as part of a continuum rather than once the patient is out of the hospital you can be done with it.

There shouldn't be an incentive to deliver sloppy care, and when the patient gets readmitted you get paid again.  

Readmissions, like length of stay, should be a utilization measure, not a quality measure. There are some readmissions that are good. There are some readmissions that are bad. If you shorten length of stay with a bad diagnosis and send the patient out prematurely or kick someone out of the hospital who doesn't have a safe disposition plan, that is not short length of stay for a good reason.

By the same token, if you are restricting patients from accessing acute care hospitals because you are trying to keep your readmissions rates down, that is not good for patients.

I don't fundamentally have a problem with CMS trying to get hospitals to lower their readmissions or general admissions rates. We shouldn't admit patients who don't really need to be in the hospital.

But we also shouldn't be incentivizing hospitals to do their best to turn away patients who do need hospitalization regardless of whether they've recently been hospitalized.

HLM: Your study linked lower mortality rates with higher readmissions. How big of a factor was that difference?

Brotman: We are talking at the hospital level analysis. You wouldn't tell a patient "Hey, if you get readmitted you are less likely to die." That would be a misinterpretation of our data. What we are saying is that there is a small but significant association between good mortality rates as defined by the measures used by CMS and high readmissions as defined by CMS at the hospital level.

It does demonstrate that these measures do not correlate in tandem the way you might expect to see and there is a reason for that, which is that readmissions are not a quality measure.

HLM: How should readmissions be used in the Five Star Ratings?

Brotman: If they were constructed to look at avoidable readmissions where process of care and readmissions were coupled, that would be a valid quality measure. But if we are not tracking process of care, they have no role in quality measure assessment. They should not be in the Five-Star Rating.

However, they can potentially be a target for reimbursement, as with length of stay. If you have a high readmission rate, you are using more services and as a society we need to be circumspect about the resources we are using.

It shouldn't be a double whammy though, where not only are you having some financial impact for having high readmissions rates, but also a reputational impact as it is a measure of quality. It is a measure of utilization and not a measure of quality.