We keep seeing all kinds of research on readmissions, some pushed to justify the readmission penalty programs, some pushed to suggest that those programs are seriously flawed and unfair. The latest is a study carried in the New England Journal of Medicine. (NEJM Article) The researchers looked at Medicare beneficiaries who had multiple admissions for the same diagnosis, but with the admissions being at different hospitals. They compared readmission rates among this group of patients with the pre-existing readmission status of the hospitals they were admitted to. The hospitals were divided into quartiles, but using all readmissions as the measure, not just the common, but limited, diagnoses used for the Medicare penalty program. Obviously, this could be misleading if it is used to judge the adequacy of the calculations under the penalty program, since there could be different factors at work in the more limited subset of diagnoses used by that program. In addition, there is something circular about using potentially flawed readmission rates to divide up the hospitals into the quartiles, and then judging their performance. You have in essence already decided that they were poor performing for the same reasons you are supposed to be checking.
Nonetheless, there were over 37,000 patients with admissions for the same diagnosis to different hospitals, with the hospitals being in a different readmissions quartile ranking. The spread in overall readmission rates across the quartiles was very low, lending further lack of credibility to the results. The lowest quartile was only 1.3% worse than the highest. Readmissions were higher, according to the study results, after the admission to a lower quartile hospital than after an admission to a high quartile hospital. (Low being good in this analysis.) But only the comparison between the lowest and highest quartile hospitals was statistically significant, and the difference was a whopping 2 percentage points. The authors say the results support the idea that hospital differences are being picked up by the readmission penalty program, but as you can tell, I am dubious. There may be something about the small subset of patients who used two or more different hospitals. Why did that occur instead of going back to the same hospital? And using all admissions instead of just the ones in the penalty program may introduce confounders. I still don’t understand at all why hospitals aren’t judged on the appropriateness of each specific readmission. That is the only logical way to identify avoidable readmissions, and then those could just not be paid for. Given that CMS already RAC audits the hell out of hospitals for every admission, wouldn’t seem too hard to just do readmissions.