John Mandrola, MD


November 13, 2017

Everyone wants quality healthcare. The problem is getting there, especially in the care of patients with heart failure (HF). If only every patient admitted with HF were 55 years old and had isolated left ventricular dysfunction . . . 

In patients admitted to the hospital with HF, current US policy holds that hospital readmission in the first 30 days after the first stay signals poor quality. This makes sense, because good care during and after hospital stay should result in lower rates of readmission.

So strong was this belief that policy makers financially penalized hospitals with high rates of readmissions. Said penalties induced lots of action from hospitals—some good (eg, efforts to improve after-hospital care), some bad (eg, aggressive care in the frail elderly and possible gaming strategies [see below]).

A paper presented at the American Heart Association (AHA) 2017 Scientific Sessions and simultaneously published in the Journal of the American Medical Association found these well-meaning policies to reduce readmissions are associated with higher death rates.[1] Lead author Dr Ankur Gupta (Brigham and Women's Hospital, Boston, MA) and his coauthors emphasized that association does not equate to causation.

Nonetheless, the findings were striking. The short take is here:

Using the Get With The Guidelines-Heart Failure registry data from 416 hospitals, the authors analyzed risk-adjusted outcomes (readmissions and mortality at 30 days and 1 year postdischarge) in more than 115K older-aged patients during three time periods: the policy implementation, after its implementation, and after the financial penalties went into effect.

They found that readmissions did indeed fall but the 30-day and 1-year mortality increased in the time period after implementation of the policy.  The degree of decrease in readmissions and increases in mortality easily reached statistical significance.


Given the millions of patients who are admitted with heart failure, the implications of a policy that could contribute to higher mortality is massive.

My take of this paper is that it is a careful analysis from respected researchers who looked at a large sample size. The results align with previous studies showing similar trends. And the registry they used contains detailed clinical data that allows for good risk adjustment of heart-failure cases. Their results also confirm my strong bias that heart-failure metrics have made heart-failure care even more heartless than it was previously.

But this paper has met strong criticism from Dr Harlan Krumholz (Yale University School of Medicine, New Haven, CT) and his coinvestigators. They published a paper in JAMA earlier this year that found reductions in 30-day readmissions were weakly but significantly correlated with lower 30-day mortality.[2] Essentially the opposite.

The Yale group's analysis used claims data (not a registry) and included almost 3 million hospitalizations for heart failure—a much larger sample. Led by primary author Dr Kumar Dharmarajan (Yale University), this analysis included a special twist: they did paired comparisons of specific hospitals and observed that hospitals with lower rates of readmission had lower rates of 30-day mortality—although the correlation was weak.

In short, we have two groups of distinguished outcomes researchers looking at the same question and coming to different conclusions. What should one make of this?

Both camps have plausible reasons for their findings. Gupta et al suggest gaming strategies (delaying admission beyond the 30 days, increasing observation stays, or shifting care to emergency departments) could worsen heart-failure mortality. That is surely possible. Dharmarajan et al would counter by saying it makes perfect sense that hospitals with lower readmissions have better care—because lower readmissions reflect quality. That too is reasonable.

I see another explanation. The vast majority of patients with heart failure admitted to the hospital rarely have decline of only one organ. The mean age of patients in these studies was 80 years. Heart-failure metrics nudge (perhaps force) doctors to treat these elders as if they were young patients with isolated nonischemic cardiomyopathy. For instance, soon after discharge, older patients come to a clinic and often get more tests and more meds. I believe more intense heart-failure care in the elderly could easily explain the higher mortality. Granted, this is just a theory. But . . .

In evidence-based medical practice, drugs and devices undergo great scrutiny before passing regulatory review. We don’t let a drug or device on the market just because we think they should work. They must be shown to be both effective and safe.

My sense is that healthcare policies don't face the same scrutiny.

The best-case scenario here is that we are uncertain that the policy of reducing readmission with financial penalties has any effect on mortality. The worst case is that a well-meaning intervention harmed a lot of people.

To me, the potential of harm warrants more rigorous study of policies before they are enacted and closer scrutiny of their performance after they are implemented.



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.