Objective. To obtain information relevant to development of prospectiv
e payment for Medicare rehabilitation facilities (RFs) and skilled nur
sing facilities (SNFs): compares service utilization, length of stay (
LOS), case mix, and resource consumption for Medicare patients receivi
ng postacute institutional rehabilitation care. Data Sources/Study Set
ting. Longitudinal patient-level and related facility-level data on Me
dicare hip fracture (n = 513) and stroke (n = 483) patients admitted i
n 1991-1994 to a sample of 27 RFs and 65 SNFs in urban areas in 17 sta
tes. Study Design. For each condition, two-group RF-SNF comparisons we
re made. Regression analysis was used to adjust RF-SNF differences in
resource consumption per stay for patient condition (case mix) and oth
er factors, since random assignment was not possible. Data Collection/
Extraction Methods. Providers al:each facility were trained to collect
patient case-mix and service utilization information. Secondary data
also were obtained. Principal Findings. RF patients had shorter LOS, f
ewer total nursing hours (but more skilled nursing hours), and more an
cillary hours than SNF patients. After adjustment, ancillary resource
consumption per stay remained substantially higher for RF than SNF pat
ients, particularly for stroke. The adjusted nursing resource consumpt
ion differences were smaller than the ancillary differences and not st
atistically significant for hip fracture. Supplemental outcome finding
s suggested minimal differences for hip fracture patients but better o
utcomes for RF than SNF stroke patients. Conclusions. Much can be gain
ed from an integrated approach to developing prospective payment for R
Fs and SNFs. In that context, consideration of condition specific per-
stay payment methods applicable to both settings appears warranted.