Analyzing length of stay for inpatient rehabilitation: A simulation study on traumatic brain injury
Operations excellence in inpatient rehabilitation is critical to coordinate care transitions for people with serious illness and injury. Public insurers typically only pay inpatient rehabilitation services within a certain period of time regardless of the patient's recovery condition. Furthermore, the funded inpatient rehabilitation length of stay (RLOS) is determined without much quantitative analysis. Such an "ad-hoc one-fit-all" approach is often questionable, especially when inpatient rehabilitation service providers are under pressure of discharging patients early to control the rehabilitation cost, meanwhile maintaining their readmission risks. Keeping the right balance between these two outcomes is imperative in achieving a cost-effective system. In this research, we present a stochastic process to model the care transition. We report a simulation study to evaluate the process for a cohort of publically funded traumatic brain injury patients and explore the tradeoff between the rehabilitation cost and readmission risk. We systematically vary the funded RLOS within each patient cluster to identify promising discharge timing policies. Following the preliminary results obtained through the simulation experiments, we introduce a search heuristic within the framework of our model. The genetic algorithm based simulation optimization algorithm that arose out of this approach optimizes the RLOS within each cluster by identifying the ideal combination of RLOS over all clusters that would minimize both the rehabilitation cost and the readmission risk.^
Nelson A. Uhan, Purdue University, Nan Kong, Purdue University.
Health Sciences, Rehabilitation and Therapy|Health Sciences, Health Care Management|Operations Research