Journal of Rehabilitation Medicine 51-8 | Page 31

Rasch analysis of UK FIM+FAM in patients with TBI Adapted from the original US version, the UK Fun- ctional Assessment Measure (UK FIM+FAM) was first published in 1999 (7). The 30-item scale comprises 16 motor items (including self-care and mobility) and 14 cognitive items (including communication, cognition and psychosocial adjustment). The UK Rehabilitation Outcomes Collaborative (UKROC) (8) provides the national clinical database for all specialist in-patient rehabilitation services in England. These services cater for a selected population of patients with severe complex neurological disabi- lity. The UK FIM+FAM is the principal measure of functional outcome within the dataset and, since 2013, its collection is a mandated requirement for reimburse- ment within these services. It is therefore pertinent to understand its psychometric properties in this context. In addition, transformation of the scale from ordinal to interval-level data (using techniques such as Rasch analysis (9–11)) confers theoretical advantages for clinical practice, potentially increasing sensitivity by stretching the existing ogive raw score points at both the upper and lower margins, as well as enabling the calculation of valid change scores. Traditional psychometric evaluations of the UK FIM+FAM in both a general neuro-rehabilitation cohort and in stroke patients (12, 13) have shown the UK FIM+FAM to have 3 distinct subscales: Motor, Communicative, and Psychosocial. Two studies from the 1990s have explored the linear Rasch transforma- tion of the original US FIM+FAM scale following TBI and showed partial conformation to the Rasch model (14, 15). However, approaches to Rasch analysis have changed significantly in the last 20 years (16), and as yet there are no published Rasch analyses of the UK FIM+FAM scale in patients with TBI. Our recent Rasch analysis of the UK FIM+FAM in patients with complex disability following stroke (17) provided evidence for uni-dimensionality. It showed that the best fit was achieved where responses for the 16-item ”Motor”, 5-item ”Communicative” and 9-item ”Psychosocial” subscales were summarized into 3 su- per-items, and were split for left and right hemisphere stroke location due to differential item functioning (DIF). This approach satisfied the expectations of the Rasch model without the need for re-scoring of item thresholds or exclusion of extreme scores. The aim of this study was to determine whether the UK FIM+FAM fits the Rasch model in patients with complex disability following TBI in a similar manner to patients with stroke, and if so, to derive a conver- sion table of Rasch-transformed scores for potential clinical use. 567 METHODS Setting and participants In the UK, rehabilitation services are broadly categorized on 3 levels (18). Following TBI, the majority of patients will have relatively simple rehabilitation needs and make a good recovery with the support of their local (Level 3) rehabilitation services. However, a smaller number have more complex disability requiring specialist rehabilitation in either local district (Level 2) or regional (Level 1) rehabilitation services. The UKROC database provides the national clinical database for all Level 1 and 2 rehabilitation services in England. The study sample therefore represents a selected group of adults with complex disability following severe brain injury. Principles of approach Where an instrument is already established in clinical use and the content and ordinal scoring system have some meaning for clinicians, a balance may need to be found between maintaining the integrity of the scale and finding the best fit solution for a transformed scale. In addition, patients with complex disability following TBI form a more diverse patient group than stroke patients, with deficits ranging from ambulant patients with highly challenging behaviours to severe physical disability. The diversity of this sample represents a challenge for the psycho- metric properties of the FIM+FAM. Hence, we expected that this analysis might not fit the Rasch model quite as readily as in more homogeneous conditions. UK FIM+FAM The 30 items of UK FIM+FAM are each scored on a 7-point ordinal scale as follows: 1 (Total assistance); 2 (Maximal as- sistance); 3 (Moderate assistance); 4 (Minimal assistance); 5 (Supervision/set-up); 6 (Independent with device); and 7 (Fully independent). A category of 6 or 7 implies no help from another person, whilst assessment for categories 1 to 4 is often based on frequency of intervention. Like the US version, the FIM components of the UK FIM+FAM are retained in order to maintain comparability for that component of the scale with units that use the FIM only. Amongst the FAM items, 3 are structurally different in the UK version (1999) compared with the original US version (1994): ”Concentration”’ replaces ”At- tention”; ”Safety awareness” replaces ”Safety judgement”; and ”’Use of leisure time” replaces ”’Employability”. The rationale for these differences is described in the 1999 source paper for the UK FIM+FAM. Data source The data source for this analysis was the UKROC database, which was initially set up by a National Institute for Health Research Programme Grant (19). It is now commissioned directly by NHS England to provide the national clinical and commissioning database for specialist inpatient rehabilitation in England. The dataset comprises socio-demographic and clinical data as well as information on rehabilitation needs, inputs and outcomes on admission and discharge from inpa- tient rehabilitation. Reporting of UK FIM+FAM data has been a commissioning requirement for Level 1 and 2 rehabilitation J Rehabil Med 51, 2019