Many readers commented on the possibility that too-early discharge from the hospital is a major cause of bounce-back admissions. I went to the original article to see whether it included any data on this matter. Table 3 shows that people whose index hospitalization was short (less than half the DRG expected length-of-stay [LOS]) actually had a lower rate of bounce-back hospitalization, and those whose index hospitalization was long (more than twice the DRG expected LOS) actually had higher rehospitalization rates. At first glance, these figures seem diametrically opposed to the idea that too-early discharges are a major cause of readmission. But on further thought, I’ll bet these figures aren’t very helpful, for the following reason: An extremely short LOS -- less than half the DRG expectation -- is probably a marker for a less-sick patient. And a very long LOS -- more than twice the DRG expectation -- is probably a marker for sicker patients. What we really need to know is individual patient-level data on whether patients are clinically stable on the day they’re discharged. The Medicare database that was the source of the New England Journal study doesn’t provide this information.
I just saw another study that sheds some light on this topic (see Palacio C, et al. A comparative study of unscheduled hospital readmissions in a resident-staffed teaching service and a hospitalist-based service. Southern Medical Journal 2009;102:145-149). In this study from a large urban teaching hospital, the probability of readmission decreased significantly as LOS increased. This observation doesn’t prove cause-and- effect, but it supports readers who believe that too-early discharge is one cause of bounce-back admissions.
Allan Brett, MD, Editor-in-Chief, Journal Watch General Medicine