Hospital readmissions summary comments

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 Jha offer 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 status is 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.

@afrakt

Share on twitter
Share on facebook
Share on linkedin
Share on reddit
Share on email

Hidden information below

Subscribe

* indicates required
Email Format