Four methods for characterizing disability in the formation of function related groups

l The Functional Independence Measure-Function Related Groups (FIM-FRGs) were developed to classify medical rehabilitation inpatients into homogeneous groups based on length of stay (LOS).Patients are lirst grouped into clinically relevant rehabilitation impairment categories, then by functional status, as expressed by the FIM. and in certain cases by patient age.The statistical approach used to form the final groupings was a recursive partitioning algorithm applied to the FIM scores and patient age within impairment category.This analysis compares four FIM-FRG classification schemes developed from four scale sets that combine FIM items differently: (1) use of the 18 FIM items as separate variables, (2) the combination of FIM items into six clinical subscales.(3) thle combination of the six clinical subscales into motor and cognitive subscales.and (4) the combination of all FIM items into a single scale.The FIM-FRG schemes explain similar amounts of variance in the logarithm of LOIS and contain approximately equal numbers of FRGs.The motor and cognitive subscale scheme is recommended for use in payment, however, this scheme and the other schemes have additional uses.Each FRt; scheme provides different insight into the clinical relationship between disability and LOS.C 199'4 by the American Congress of Rehabilitation Medicine and the American Academy oj'Physica1 Medicine and Rehabilitation Diagnosis Related Groups (DRGs).which were designed to characterize acute care hospital lengths of stay (LOS).do not adequately capture inpatient medical rehabilitation resource use.'~4 To address this problem.we developed a system based on the Functional Independence Measure (FIM)'~" to classify adult rehabilitation inpatients into homogeneous groups using LOS.'.'In our system.called the FIM-Function Related Groups (FIM-FRGs [presented previously"j), patients are classified by rehabilitation impairment category, functional performance on the admission FIM, and sometimes patient age.The variance in LOS explained by the FIM-FRGs compares favorably with that of Harada and colleagues' classification system.'It also compares favombly with the initial capacity of the DRGs to explain variance in L,OS in the acute care setting."This artic!le discusses four alternative FIM-FRG classitication schemes developed from four different methods of combining FIM items.The purpose of this work IS to determine whether any of the alternative classification schemes are preferable to the one reported in our original work and to identify relationships among impatrment.patterns of disability, and LOS.