My last post alluded to the JAMA theme issue on readmissions. I planned on writing a synopsis, but having read a related post from a friend—one I cannot improve upon—I will defer to his. However, a few housekeeping chores before the guest summary below.
First, I wish to make a few comments of my own on the release.
Overall, the findings and opinions did not surprise me. Condensing the conclusions in Cliff Note form, my take would be as follows:
–We readmit 20% of Medicare beneficiaries 30-days post discharge, and the majority return within 2-weeks with a different chief complaint.
–The interventions needed to reduce current levels elude us. We have a long list of options to choose from, but their application makes for a muddled enterprise.
–The readmission penalty program needs more than a little tweaking; the signal to noise ratio on what works and does not, the validity of current metrics, and how CMS computes the penalties need clarity. Woeful may be an overstatement, but we do not know enough given the attention readmissions receive. The public sees stories galore in lay publications every month and never flattering—usually in association with words such as “intolerable,” “awful,” “unacceptable,” and “how could this be.”
–How do we classify treat-and-release visits to emergency departments post discharge? Access problem in the community? Bad discharge planning? Unavoidable at some low percentage? Penalty worthy?
–Unlike CLABS infections or never events, all internally administered, the realization we (hospitals) cannot control everything post discharge, especially with limited finances, has not sunk in. Yes, we are the locus of control and have the current deep pockets, but the problem encompasses so much more than appreciated, especially by payers.
Second, an article in NEJM appeared a few weeks back relating to the disability patients undergo after leaving the hospital. Harlan Krumholz penned a perspective entitled, Post-Hospital Syndrome — An Acquired, Transient Condition of Generalized Risk, and the essay commanded attention in many venues. I do not know why.
For those who read Thomas Friedman, he often says, “If you can name an issue, you can own the issue.” Unfortunately, like metabolic syndrome, an amalgam of diagnoses that in sum create a new one, maybe, Krumholz conveys that hospitals decondition, underfeed, cognitively impair, and over stress patients. The result translates to unnecessary and sometimes permanent disability post-discharge. Consequently, we now have a name for a “new” problem. Or do we?
I posit, the challenge lies not in naming an impaired state most of us have always known about—but engaging staff to prevent what we have overlooked until now–and the term Hospitalization-Associated Disability works just fine (thank you). And I am certain checklists will coalesce to produce best practice models as the entity gets more attention.
Regardless, however renewed our intervention pushes might be, the awareness of hospital-induced frailty does not qualify as breaking news. We see infirmity at discharge every day, and have, for decades. The fixing part demands the elbow grease and requires our scrutiny. Our passivity to a known concern should no longer stand.
Third, I stole (actually, granted permission :)) the post below from a fellow blogger at The Incidental Economist. Austin has ramped up his posting of late on readmissions, and I encourage you to read his recent threads. His terrific insights coalesce in the final entry I co-opted underneath. Great stuff:
(Additionally, below his post, I checked in with Mark Williams for a brief Q&A. Mark’s seminal NEJM paper commenced the discussion on readmissions four years ago).
Though I am sure to blog more on hospital readmissions in the future, I’ve reached a pause, with no papers on the subject in my queue to read. Since it’s likely many readers weren’t following my posts on this in detail, below is a summary. You can follow the links back to posts in the series to date. All of them are tagged with “hospital readmissions.”
The proportion of hospitalizations leading to rapid readmissions is shocking. Whatever you think of Medicare’s Hospital Readmissions Reduction Program (HRRP), which penalizes or rewards hospitals based on their readmission rates, there is no reason to be satisfied with this. But what causes readmissions? In particular, what causes those that are potentially preventable? Many point to inadequate discharge planning, poor transitions of care, and insufficient outpatient follow up. Post-hospitalization is a vulnerable time, requiring more than the normal level of support. It is not implausible that many patients just don’t get what they need.
Is this the fault of hospitals? Evidence about the extent to which hospital readmissions are related to quality of care during the index admission is mixed. Readmissions may be related to some notions of hospital quality or safety but not others. They are also related to many other factors that are not amenable to modification by hospitals. Joynt and Jhaoffer the most compelling and complete case that variation in socioeconomic status and hospital resources play large roles in variation in readmission rates. Hospitals that lose resources due to high readmission rates may be the very ones that can least afford it. Quality may suffer for the most vulnerable populations, which is the opposite of the policy’s goal.
Whatever “potentially preventable” means (definitions vary), the proportion of readmissions that are such is likely fairly small, though varies by region. The ability to predict hospital readmissions is modest, and little work has been done on predicting potentially preventable hospitalizations. MedPAC’s estimate that 79% of readmissions are avoidable is almost surely way too high. Here’s an evidence-based rule of thumb: 20% of Medicare hospitalizations lead to readmissions, and 20% of those are avoidable So, of all Medicare hospitalizations, perhaps about 4% lead to avoidable readmissions. That’s not nothing, but I’ll bet you thought it was higher.
It is not evident how hospitals can reliably reduce readmissions. One recent study suggests that increasing nursing staff or improving their work environment can help, but I have some methodological concerns about it. Some interventions that reduce hospitalizations may not reduce the rate of readmissions, suggesting the HRRP is inadequate. More primary care may increase readmissions.
Researchers have observed a negative correlation between rates of mortality and readmissions. There’s a simple explanation for this: Dead people can’t be readmitted, a type of immortal time bias. Readmission rate estimates do vary depending on how mortality is handled in analysis.
I have concerns about how the HRRP computes readmission rates. Socioeconomic statusis not among the risk adjusters. Medicare Advantage enrollment is not considered. All in all, it’s very hard to be excited about the program. It’s unclear it is targeting the right thing and measuring it in the right way.
Still, it is possible readmission rates — perhaps more thoroughly risk adjusted — provide a more valuable signal for certain sub-populations. Perhaps, in concert with other measures, they reveal something of import about health systems in general or hospitals in particular. I don’t really know. I don’t want to say they’re useless. Right now, all I’m saying is that the way they are currently used by Medicare has been severely challenged in the literature. There’s a lot on the “con” side. Where’s the “pro”? I’m still seeking the counterpoint to Joynt and Jha. Who has offered the case for the HRRP? For all my looking, I haven’t found it.
Austin B. Frakt, PhD, is a health economist and an assistant professor at Boston University’s School of Medicine and School of Public Health. He blogs about health economics and policy at The Incidental Economist and tweets at @afrakt.
Mark’s comments below:
1. When you published your NEJM paper in 2009, did you grasp the magnitude of getting at the root causes of the readmission problem? In another words, did you expect the literature to become more, as opposed to less confusing four years later?
When we initially published, I saw the data, and felt hospitalists needed to know about it—we did not consider the aftermath and subsequent policy questions folks would raise. We also were not aware that the ACA would contain provisions related to readmissions, nor did we know CMS was thinking about the readmit issue strongly. MedPAC did have the problem on its radar screen though, and their 2007 report was probably the underpinnings of what became broader policy. We did not anticipate how impactful our paper would be and the changes it would bring about. We also did not expect over a thousand citations of it in subsequent publications.
2. Ten years from now, will we see a national benchmark for readmission rates, or, as the literature seems to show, as long as we have social disparities and underfunded hospitals and communities, marked variation will exist?
This approach was not as well informed as it could have been, but I am glad hospitals got the signal. Why? Hospitals required a wakeup call and the readmit, and admission in general mindset of fee-for-service needed to be broken. I recognize the dissonance, but the long-term benefits outweigh the short-term downsides of the penalty. Financials and motivation needed changing and we needed a first step. This was it. If CMS messages probably, and we succeed funding studies demonstrating what works, and the recidivism risks involved are elucidated, it will be worth it. But we must amend the policy as we learn.
3. Most folks don’t realize, the Hospital Readmissions Reduction Program–unlike other programs HHS oversees–cannot be amended unless Congress acts. The ACA dictates the terms bluntly and the message to hospital-based providers seems to be, CMS metes penalties despite our inability to impact outcomes as imagined. Cognitive dissonance aptly describes the circumstances, don’t you think?
We need more research. I would like to see at least $500 million (equivalent to the CCTP program) in the near term invested in projects to improve care delivery around transitions. Finding the right percentage is not out of reach, but we need more data and experience.
I think the absolute readmit rate average might drop 5%, and best performers could settle in at the 10% range. However, I don’t believe we can squeeze all the variability out of the system given the influence social and economic factors have on hospital performance. There are too many external factors, but the variability certainly could be diminished.