A pilot investigation of the efficacy of falls risk assessment tools and prevention strategies in an elderly hip fracture population

Abstract

Background The importance of reducing the number of falls was highlighted by the UK National Service Framework for Older People [Department of Health, 2001]. In response falls risk assessment tools have been developed and implemented into clinical practice. However their predictive accuracy is generally unsubstantiated. Aims The study aimed to investigate the predictive accuracy of the Fall Risk Assessment Scale for the Elderly (FRASE) and St. Thomas's Risk Assessment Tool (STRATIFY) and determine if falls prevention strategies reduced the number of falls in hip fracture patients. Design A quasi-experimental design compared the incidence of falls between retrospective (n = 30) and prospective (n = 60) groups of hip fracture patients admitted to an acute NHS hospital. Receiver Operator Characteristic (ROC) calculated the predictive accuracy of both FRASE and STRATIFY in identifying the fall risk status of the sample. Also the inter-rater reliability of the two tools was tested. Results Inter-rater reliability for both FRASE and STRATIFY was high with correlation coefficients of 0.964 and 0.836, respectively, and p values of 0.001. Two patients fell in the prospective group and 3 in the retrospective group. There was no significant difference in the number of falls between the groups with the p value being 0.204. Both FRASE and STRATIFY were found to have poor predictive accuracy with ROC scores of 0.560 and 0.629, respectively, for the prospective group. The tools also demonstrated poor predictive accuracy with the retrospective group with a ROC score of 0.370 for FRASE and 0.463 for STRATIFY. Conclusions and relevance Although the study demonstrated good inter-rater reliability of both tools the level of predictive accuracy was poor. Both tools were demonstrated to over predict the risk of patients falling. At face value it would appear preferable for a tool to over predict risk of falling rather than under predicting, however this must be viewed in the context of finite resources. It is recommended that further research is needed to manipulate the threshold level for risk of falling with both tools and to ascertain if the inclusion of extrinsic factors would improve their predictive accuracy.