The paper by Bueno, et al. (JAMA, 2010) raises methodological issues worth flagging. Though I don’t have anything novel to say about their results, here’s a plain-word commentary on one finding (follow the link for more):
The most striking finding is that the period was associated with an increase in 30-day readmission rate. Although we cannot demonstrate that the shortened hospital stay caused these changes, it is certainly plausible that the effort to discharge patients quickly has led to transfers to nonacute institutional settings and occasionally sent patients out of the hospital before they were fully treated. Moreover, there is a paucity of studies that test criteria for readiness for discharge, adding to uncertainty about what constitutes appropriate hospital treatment for the condition. It is also possible that shorter lengths of stay in a system that supports the transition to outpatient status might not be associated with a higher readmission rate.
This was a study of fee-for-service (FFS) Medicare, heart failure (HF) patients aged 65 years or older in 1993-2006. The authors note some limitations
The inability to evaluate the quality of hospital care precludes the evaluation of the rate of premature discharge, as well as the demonstration that length of stay reduction is a causal factor in the increase in early postdischarge adverse outcomes. […]
[W]e were limited in our ability to determine if patients were in fee-for-service Medicare for the entire year before the index hospitalization, a period when we ascertained comorbidities. […]
[W]e were limited in our ability to determine if patients were in fee-for-service Medicare for the entire year before the index hospitalization, a period when we ascertained comorbidities. However, the short-term and adjusted analyses were similar and it is unlikely that this limitation substantively affected the results. In addition, our study may not reflect the changes over time in the managed care population, which tends to be a healthier population overall.[27]
It’s not entirely clear to me why the investigators could not determine if patients were in FFS Medicare for the entire year before the index admission. This is knowable from the data of the type they had. It is also true that measures of hospital quality exist, though perhaps they do not go back far enough into the past. In that case, one could compute them, though I’m sensitive to the fact that one can’t do everything on one study.
Nevertheless, these are very important methodological considerations and limitations. First, quality is a very common omitted factor in many studies. It’s omission can and often does bias results, though I’m not suggesting it does so here. Having said that, it’s also exceedingly hard to measure or even define. So, what I often look for is some attempt to do something about quality, perhaps a check to see if results are sensitive to some plausibly related measures of it. I also look for a discussion of quality as an unobserved confounder and what that might mean.
Second, Medicare Advantage (MA) enrollment grew over the period of study. Selection into MA is not random. It varies geographically. These must affect readmission rates, lengths of stay, and mortality. Unfortunately, there’s very little researchers can do since we do not have access to claims for MA patients (though we really should). This is a very serious methodological problem without a good solution.
What’s worse is that CMS does not factor the potential impact of MA into its readmission rates (PDF) used for penalties and bonuses. To do so fully, it would have to require MA plans to report hospital admissions by disease, as well as all relevant diagnostic information for risk adjustment. Perhaps some less precise adjustment could be done by controlling for MA penetration rates. Either way, this is a potentially large problem.
PS: As an aside, I’ve been wondering what the CMS Hospital Readmissions Reduction Program penalties/bonuses are a percentage of. This MedPAC presentation (PDF) says that they’re “applied to base operating payments [but not] to IME, DSH, or special rural payment add-ons.”
