Several years ago it was indeed disturbing to find out that the national 30-day readmission rate for patients discharged with diseases such as heart failure, heart attack and pneumonia was almost 25%. Fortunately, since becoming a publicly reported measure and a financial determinant in the CMS Pay for Performance (P4P), there has been a great deal of attention given to this matter. As is typical with public and financial measures, there has also been a great deal of improvement. The national averages for unplanned readmissions for heart failure, heart attack and COPD are now 22%, 17% and 20% respectively, and they are continuing to decline. In the past year, CMS has expanded the list of readmissions that are considered in the P4P program to include hip/ knee replacement as well as coronary bypass procedures. As hospitals continue to ratchet down the rate, it begs the question as to what is the ideal target for patient readmissions. What is our end goal for population management? How should hospitals and health systems be measured?
Here are some points to consider:
1. The IDEAL readmission rate for any given hospital is not zeo. Unlike performance measures where the goal is to have zero deviations, or post-operative infections where the goal is to have zero events, the ultimate readmission rate that maximizes patient outcomes is NOT zero. Diseases such as heart failure and COPD are chronic in nature and they get worse over time. Patients are expected to have exacerbations as part of the course of the disease. It is not reasonable to assume that even with the best of care, these patients will not on occasion have setbacks that will require acute care within a hospital. In the present system, readmissions are now regarded as a “bad thing” or a “treatment failure”. This is not necessarily true given the nature of these diseases. I would argue that even if a hospital could reduce the readmission rate to zero they might be doing so at the risk of treating some patients unsafely.
2. Looking at all-cause readmissions may be efficient, but it is not scientific or even an appropriate results-driver. It just makes common sense that if a patient recovering from an admission for heart failure gets run over by a truck two weeks later, this readmission should not adversely effect the rating of the hospital. In the long run, these types of random readmissions should even out across hospitals, but in the short run, these occurences could make the difference between a good rating and a poor rating for an individual facility. We need to develop risk-adjusted, age-specific realistic targets and base these targets on the readmissions for related, preventable or avoidable conditions.
3. Successful readmission reduction strategies DO require a village. Probably the single most important component of successful programs is developing the appropriate communication between patient, family, primary care physician, and other care givers. No one component of the delivery system alone can substantially improve this measure. Patient identification, patient/family education and good communication throughout the system are essential.
4. What is magic about 30 days? Enough said there. A system where 31 days is a winner and 29 days is a loser can lead to gaming the program. There is no perfect window for optimum disease management, and this one is acceptable, but it should let us become myopic to the ultimate goal of long term community health.
5. The roles of prevention and end of life care have large impact in community readmission rates. Put simply, good preventive programs can decrease the initial admission, and sound end of life care measures can prevent the last admission. A successful community strategy must address both of these and not simply focus on a short term (30-90 day) window.
So as we continue to reduce unnecessary readmissions with the goal of continued improvement and better population management, we need to determine the appropriate age/risk-adjusted IDEAL readmission rate. Hospitals should have this number as the target. Those that perform significantly above or below this measure could appropriately be adversely rated under the P4P system.
For now, I do believe that few facilities have achieved this IDEAL rate. If I were to render a guess, the IDEAL rate of 30-day readmissions would be in the neighborhood of 10-15%. We do have a long way to go. But at some point, we will have to consider performance below this IDEAL rate as well as performance above this rate as a less than optimum patient safety environment.