To determine the extent to which higher costs in hospital-based skilled nursing facilities (HBSNF) can be explained by observable resident characteristics and unobservable selection effects, implying a design shortcoming of the skilled nursing facility prospective payment system (SNF PPS) implemented for Medicare-covered stays by the Balanced Budget Act of 1997 (BBA 1997).
Data on resident characteristics from the Minimum Data Set (MDS) are combined with staff time costs from the Centers for Medicare and Medicaid Services’ (CMS, formerly HCFA) 1995 and 1997 SNF Staff Time Measurement (STM) studies and nontherapy ancillary claim costs extracted from CMS SNF claim records. An endogenous switching model was estimated to measure the effect on costs of the relatively high acuity of HBSNF residents, net of differences purely attributable to the treatment setting.
It was found that virtually the entire HBSNF differential is attributable to setting effects with resident characteristics and selection effects playing a negligible role. In addition, it was found that marginal costs associated with particular services and conditions are often lower in hospital-based than in freestanding facilities.
HBSNFs incur high costs regardless of the characteristics of their residents. Their high fixed costs accompany relatively low marginal costs associated with admitting high-acuity residents. Consequently, a PPS casemix system that depends on resident characteristics and excludes consideration of facility characteristics (as mandated by BBA 1997) need not unfairly penalize HBSNFs, provided a powerful casemix system is used and HBSNFs specialize in the care of high-acuity residents.