Measuring the Hospital Length of Stay/Readmission Cost Trade-off under a Bundled Payment Mechanism

The Affordable Care Act creates incentives for hospitals to reduce readmissions under the Bundled Payments for Care Improvement Initiative.  In a bundled payments model, participating hospitals receive a single payment for an entire episode of treatment that includes an initial admission and any 30-day readmission.  This initiative creates a counter incentive for hospitals under the Medicare Prospective Payment System (PPS), which currently reimburses for each admission and readmission.  A key response of hospitals to PPS has been to shorten lengths of stay (LOS).  Yet if in response to PPS financial incentives, patients are discharged prematurely, many may need to return to the hospital for additional inpatient care within a short period of time.  Under bundled payments, LOS has interesting cost consequences because there is a trade-off between cost savings in the initial hospitalization realized though shorter LOS and reduced probability of a costly readmission associated with longer LOS.   

This paper examines this cost trade-off first with a theoretical model in which the hospital’s objective function is minimization of the combined cost of the initial hospitalization and readmission within 30 days.  Empirical application is through a two-part model that estimates the probability of nonzero cost of readmission separately from the level of costs conditional on nonzero costs, to obtain the expected cost of readmission for all Medicare patients treated for heart attack and discharged alive from acute care hospitals in New York State during 2008.  Two-stage least squares techniques control for potential endogeneity of LOS arising from unobservable variables that might affect both LOS and readmission probability.  The cost portion of the model uses generalized linear modeling in order to account for positive skewness in the distribution of patient costs.  A final step compares the expected cost of a readmission with the marginal cost of an additional day in the initial stay to examine the cost trade-off between an extra day of care and the expected cost of readmission.  The primary data source is the Agency for Healthcare Research and Quality’s State Inpatient Databases, supplemented with the American Hospital Association Annual Survey Database. 

Results showed no interior solution to the minimization problem.  However on average, the cost of an additional day of stay was offset by expected cost savings from an avoided readmission in the range of 15% to 65% of the cost of the additional day.  These results have important policy implications at this time as the CMS Innovation Center develops bundled payments models aimed at reducing Medicare costs under health care reform.  Hospitals might consider increasing the length of stay for certain heart attack patients as a strategy that simultaneously is cost-saving and quality-improving.