PURPOSE: Measurement of lower limb amputee mobility was investigated using the Rivermead Mobility Index (RMI). METHOD: Reliability and reproducibility were assessed in 62 patients. The timed waking test (TWT) was used to investigate concurrent validity. The RMI construct was examined in 200 established amputees. RESULTS: One hundred and forty-four males and 66 females, aged 13-90 were recruited. Intraclass correlation coefficients and kappa statistics showed good reproducibility. Spearman correlation coefficient between the RMI and TWT -0.58 (p<0.000). Psychometric properties of the RMI were tested using item response theory. Hierachical differences in RMI grades were identified in amputees compared with neurologically impaired patients for which the RMI was developed. The RMI construct was not unidimensional, with redundancy of items and local dependency. At the upper end of the scale there were insufficient items measuring high levels of mobility. Finally, differential item functioning showed items behaving differently for patient subgroups. CONCLUSIONS: Although initial impressions suggest the RMI is a useful measure of lower limb amputee mobility, further analysis shows it is not appropriate for all amputees, with a number of limitations of its psychometric properties. Its use is not recommended in this population.
PURPOSE: Measurement of lower limb amputee mobility was investigated using the Rivermead Mobility Index (RMI). METHOD: Reliability and reproducibility were assessed in 62 patients. The timed waking test (TWT) was used to investigate concurrent validity. The RMI construct was examined in 200 established amputees. RESULTS: One hundred and forty-four males and 66 females, aged 13-90 were recruited. Intraclass correlation coefficients and kappa statistics showed good reproducibility. Spearman correlation coefficient between the RMI and TWT -0.58 (p<0.000). Psychometric properties of the RMI were tested using item response theory. Hierachical differences in RMI grades were identified in amputees compared with neurologically impairedpatients for which the RMI was developed. The RMI construct was not unidimensional, with redundancy of items and local dependency. At the upper end of the scale there were insufficient items measuring high levels of mobility. Finally, differential item functioning showed items behaving differently for patient subgroups. CONCLUSIONS: Although initial impressions suggest the RMI is a useful measure of lower limb amputee mobility, further analysis shows it is not appropriate for all amputees, with a number of limitations of its psychometric properties. Its use is not recommended in this population.
Authors: Paul K Crane; Laura E Gibbons; Katja Ocepek-Welikson; Karon Cook; David Cella; Kaavya Narasimhalu; Ron D Hays; Jeanne A Teresi Journal: Qual Life Res Date: 2007-06-07 Impact factor: 4.147