Poster Paper: Luck of the Draw: Role of Chance in the Assignment of Medicare Readmission Penalties

Saturday, November 9, 2019
Plaza Building: Concourse Level, Plaza Exhibits (Sheraton Denver Downtown)

*Names in bold indicate Presenter

Andrew Wilcock1, Jose Escarce2, Peter J. Huckfeldt3, Teryl Nuckols4, Ioana Popescu2, Sushant Joshi5 and Neeraj Sood5, (1)Harvard University, (2)University of California, Los Angeles, (3)University of Minnesota, (4)Cedars-Sinai Medical Center, (5)University of Southern California

Beginning FY 2013, Centers for Medicare and Medicaid Services (CMS) has been penalizing hospital with high excess readmissions under the Hospital Readmissions Reduction Program (HRRP). Currently, the maximum penalty is 3 percent of the total Medicare payments. In the latest round of penalties for FY 2019, 2610 hospital are expected to be penalized with the total payments amounting to approximately $566 million. Even though the program has been now in place for 6 years, there are some unanswered questions. One question is whether excess readmissions for a hospital reflects the true quality of the hospital or whether excess readmissions are primarily driven by chance. The role of chance might come from two sources. First, readmissions for each beneficiary may be determined by factors beyond the control of the hospital such as weather or other random events. Second, readmissions depend on which patients are admitted to a hospital and the decision to go to a certain hospital might also be governed by random events. Even though risk-adjusters are used in the calculation of excess readmissions, these risk-adjusters may not fully capture the chance of readmissions or selection of patients in a hospital.

We evaluated the role of chance in hospital penalty status for FY 2015. To measure the role of chance in assigning hospital penalties, we calculated simulated penalty status using 1,000 bootstrap samples for each hospital that was exposed to HRRP. The bootstrap sample varies patients admitted to a hospital by sampling from the same underlying patient population. Therefore, it captures how variation in patient admitted due to chance also influences excess readmissions or penalties. We find that penalty status under the original sample of patients did not match penalty status under the bootstrap sample for a significant number of simulations. In particular, for hospital that were penalized with the original sample of patients there was a 23.3% chance that they were not penalized in bootstrap simulations. Similarly, for hospitals that were not penalized with the original sample of patients there was a 23.4% chance that they would be penalized under bootstrap simulation. The average difference in penalty was 0.52 percent of Medicare payment. We are currently refining the simulation methods to account for the role of chance in not only determining which patients are admitted to a hospital but also accounting for the role of chance in determining readmission conditional on a given set of hospital admissions.