Hospitals’ Role In Readmissions
- Yale University New Haven, Connecticut
- Harvard University Boston, Massachusetts
We are concerned that the work of David Chin and coauthors (Oct 2016), which relies on the intracluster correlation coefficient (ICC), may incorrectly suggest that hospitals cannot influence readmission outcomes beyond seven days after discharge.
The ICC describes the fraction of the variance in outcomes that is due to hospital factors, which is not an intuitive measure of quality. A more meaningful metric would focus on estimating between-hospital variation—specifically, the odds that an average patient is readmitted within thirty days of discharge from a hospital with moderately high readmission rates, compared to the case at a hospital with moderately low rates. An ICC of 0.01 corresponds to an odds ratio of 1.44, which is consistent with meaningful between-hospital differences in quality after the first postdischarge week.
The ICC has limitations as a marker of hospital quality. It does not capture the effects on readmission rates of care patterns that are similar across hospitals. As a result, it does not reflect opportunities that all hospitals possess to implement novel patient-centered strategies to improve readmission outcomes. The continuous increase in ICC after day 10 following hospitalization is not easily explained, as readmission reduction was not a major target for quality improvement during the years of Chin and coauthors’ study.
A patient-centered health care system should tailor postdischarge care to patients’ underlying vulnerability to adverse events, which remains high beyond the thirty-day period used for hospital performance assessment. 1 This profiling period was chosen to expand the role of hospitals to include stewardship of critical posthospital outcomes. Restricting accountability for readmissions to one week after discharge may largely undo recent national reductions in readmissions.
NOTE
- 1 Trajectories of risk after hospitalization for health failure, acute myocardial infarction, or pneumonia: retrospective cohort study . BMJ . 2015 ; 350 : h411 . Crossref, Medline, Google Scholar
