Predicting Falls in Musculoskeletal Rehabilitation: A Retrospective Multisite Study

Abstract

Background. Fall represents one of the highest concerns in the healthcare system, especially in medical rehabilitation settings. However, there is a lack of instruments for the assessment of risk falls in the context of musculoskeletal rehabilitation. Methods. This retrospective multisite study aimed to assess the sensitivity and specificity of four fall risk assessment tools (the Functional Independence Measure, the Fall Risk Assessment, the Schmid Fall Risk Assessment Tool, and the ePA-AC) in predicting falls in patients admitted to musculoskeletal rehabilitation in Swiss inpatient facilities. Results. The data relative to 6970 individuals (61.5% females) were analyzed and 685 (9.83% of patients) fall events were registered. The area under the curve (AUC) relative to the Functional Independence Measure was 0.689, 0.66 for the Fall Risk Assessment, 0.641 for the Schmid Fall Risk Assessment Tool, and 0.675 for the ePA-AC. Among the four tools, the Functional Independence Measure had an acceptable discriminatory power in distinguishing between significant events (i.e., patients’ falls) and non-events (no falls). Conclusion. None of the assessed tools showed highly satisfying levels of statistical sensitivity or sensibility. However, the Functional Independence Measure could be used to assess the fall risk assessment in musculoskeletal rehabilitation settings, although with some caution, since this questionnaire was not designed for this diagnostic purpose. We strongly suggest urgently designing a tool for risk assessment that is specific to this population and the rehabilitative setting.