Inter-rater reliability and predictive accuracy of the Shkuratova Assessment of Falls-risk in Rehabilitation (SAFER) tool

Abstract

Supplemental Digital Content is available in the text. Despite the recognised importance of falls prevention in rehabilitation settings, there is limited research focusing on falls risk assessment tools designed to guide both patient screening and therapy. This study evaluated the predictive accuracy and inter-rater reliability of the Shkuratova Assessment of Falls-risk in Rehabilitation settings (SAFER) tool. The study was conducted at a subacute rehabilitation facility in Australia. Patient assessments were performed on admission to subacute care by trained physiotherapists, and the incidence of falls was documented prospectively. Of the 147 patients, 45 had at least one fall and were compared to 102 who had no falls. The inter-rater reliability of the SAFER tool when used by trained physiotherapists was high with the level of agreement for individual items ranging from 74 to 99%. Thirty-two (76%) patients who experienced a fall during their admission and 46 (44%) who did not fall were identified as having a high falls-risk. Using a SAFER tool cut-off of 12/26: sensitivity is 69%, specificity is 66%, area under the curve 0.71 (95% confidence interval: 0.62–0.80). The high negative predictive values at a range of cut-offs provided strong evidence that patients identified as having a low falls-risk were unlikely to experience a fall. Performing a comprehensive assessment of specific deficits in gait, balance and mobility on admission provided a streamlined approach to identification of patients who would benefit from tailored falls prevention interventions.