Identifying and Accommodating Unduly Influential Facilities in Developing a Prospective Payment System for Inpatient Rehabilitation Care

• Activity Number: 40
• Type: Contributed
• Date/Time: Sunday, August 11, 2002 : 4:00 PM to 5:50 PM
• Sponsor: Section on Health Policy Statistics*
• Author(s): Susan Paddock*+
• Affiliation(s): RAND Corporation
• Keywords: Bayesian statistics ; influential observations ; outlier models ; regression

Abstract

In the Balanced Budget Act of 1997, Congress mandated that the Centers for Medicare and Medicaid Services (CMS) implement a Prospective Payment System (PPS) for inpatient rehabilitation care. This PPS was implemented January 1, 2002. One component of the PPS is a payment adjustment factor that compensates inpatient rehabilitation facilities for costs that are beyond their control, such as geographic variation in labor costs. Because these payment adjustment factors are based on data that were collected under the previous payment system, they will be refined once post-implementation data become available. In this talk, potential refinements of the regression model that was used to develop the payment adjustment factors will be explored, with a focus on how to identify and accommodate facilities that are unduly influential on payment factors. We will assess various methods for identifying unduly influential observations, such as Bayesian outlier models and case deletion methods.