OBJECTIVES: To examine variation in provision of assistive technology (AT) devices and the extent to which such variation may be explained by patient characteristics or Veterans Health Administration (VHA) administrative region. DESIGN: Retrospective population-based study. SETTING: VHA. PARTICIPANTS: Veterans poststroke in fiscal years 2001 and 2002 (N=12,046). INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURE: Provision of 8 categories of AT devices. RESULTS: There was considerable regional variation in provision of AT. For example, differences across administrative regions in the VHA ranged from 5.1 to 28.1 standard manual wheelchairs per 100 veterans poststroke. Using logistic regression, with only demographic variables as predictors of standard manual wheelchair provision, the c statistic was .62, and the pseudo R(2) was 2.5%. Adding disease severity increased the c statistic to .67 and the pseudo R(2) to 6.2%, and adding Veteran Integrated Network System further increased the c statistic to .72 and pseudo R(2) to 9.8%. CONCLUSIONS: Our research showed significant variation in the provision of AT devices to veterans poststroke, and it showed that patient characteristics accounted for only 6.2% of the variation. VHA administrative region and disability severity accounted for equivalent amounts of the variation. Our findings suggest the need for improvements in the process for providing AT and/or provider education concerning device provision. Published by Elsevier Inc.
OBJECTIVES: To examine variation in provision of assistive technology (AT) devices and the extent to which such variation may be explained by patient characteristics or Veterans Health Administration (VHA) administrative region. DESIGN: Retrospective population-based study. SETTING: VHA. PARTICIPANTS: Veterans poststroke in fiscal years 2001 and 2002 (N=12,046). INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURE: Provision of 8 categories of AT devices. RESULTS: There was considerable regional variation in provision of AT. For example, differences across administrative regions in the VHA ranged from 5.1 to 28.1 standard manual wheelchairs per 100 veterans poststroke. Using logistic regression, with only demographic variables as predictors of standard manual wheelchair provision, the c statistic was .62, and the pseudo R(2) was 2.5%. Adding disease severity increased the c statistic to .67 and the pseudo R(2) to 6.2%, and adding Veteran Integrated Network System further increased the c statistic to .72 and pseudo R(2) to 9.8%. CONCLUSIONS: Our research showed significant variation in the provision of AT devices to veterans poststroke, and it showed that patient characteristics accounted for only 6.2% of the variation. VHA administrative region and disability severity accounted for equivalent amounts of the variation. Our findings suggest the need for improvements in the process for providing AT and/or provider education concerning device provision. Published by Elsevier Inc.