Objectives: The present study evaluated alternative patient classification systems for skilled nursing facility and rehabilitation facility patients.
Methods: Medicare patients were selected from a random sample of 27 rehabilitation facilities and 65 skilled nursing facilities participating in a national longitudinal study of subacute care. Detailed casemix and resource use data was obtained on 513 patients with hip fracture and 483 stroke patients. The Functional Independence Measure-Function Related Groups (FIM-FRGs) classification system for rehabilitation facilities was replicated on length of stay and tested on resource use for rehabilitation facility patients as well as for skilled nursing facility patients. Modifications to the FIM-FRGs also were tested. The Resource Utilization Groups-Version III classification was tested on rehabilitation facility patients.
Results: The FIM-FRGs explained the same amount of variance in length of stay as in the original FIM-FRGs development sample (R2 hip fracture = 0.14, R2 stroke = 0.28), and similar variance in resource use. A modified version of the FIM-FRGs explained more variance in length of stay (R2 hip fracture = 0.19, R2 stroke = 0.39) and resource use (R2 hip fracture = 0.20, R2 stroke = 0.41). Neither model adequately predicted length of stay or resource use in skilled nursing facility patients. The Resource Utilization Groups-Version III rehabilitation groups accounted for little variance in rehabilitation facility patients' per-diem resource use (R2 = 0.11).
Conclusions: The FIM-FRGs are valid for resource use as well as length of stay for rehabilitation facility patients, but are not valid for skilled nursing facility patients. Similarly, the Resource Utilization Groups-Version III system does not apply to rehabilitation facility patients. Related work, however, suggests that development of a single episode-based patient classification system for skilled nursing facility and rehabilitation facility patients is possible and should be pursued.