Introduction
Falling is the most common adverse event in U.S. hospitals, with approximately 1 million hospital falls annually, a third of which result in injury. Hospital falls are associated with increased hospital length of stay and higher rates of discharge to long-term care facilities.1 In-hospital falls cost U.S. hospitals $7 billion annually and lead to 11,000 excess deaths.2 Fall prevention has been the subject of intensive research and quality improvement efforts for years, yet it remains a major challenge for healthcare organizations, where substantial resources are devoted to prevention measures.
One key challenge is accurately assessing a patient’s risk for falling so appropriate and effective interventions can be put in place. Traditional falls risk tools require nursing assessment of specific factors. There are numerous tools available for use, including the Morse Fall Scale, the Johns Hopkins Fall Risk Assessment Tool (JHFRAT), the Conley Scale, and the Hendrich II Fall Risk Model. These traditional assessment tools are manual and require nursing time, multiple times a day, to account for changes in patient condition. Additionally, the most widely used tools have been shown to have low specificity and are poorly predictive of injurious falls in hospitals.4 Other studies have shown that, of patients who experience a fall, 43%-45% had either a low-risk score or no documented risk score.5,6
Clinical decision support (CDS) tools to ease documentation requirements and improve clinical practice have been developed for decades. CDS is generally defined as a process for enhancing health-related decisions and actions with pertinent, organized clinical knowledge and patient information to improve diagnostic or therapeutic processes of care.3 This includes assessment algorithms that leverage and update based on EHR data.7,8,9 There are perceived positive and negative impacts of CDS. In the positive, a CDS works as intended and improves clinical care and outcomes. In the negative, CDS may disrupt workflow and add noise in the form of additional data that does not inform care. In the case of falls risk prediction, a CDS tool may reduce documentation, but it is imperative that it also provide value in the form of superior accuracy to improve clinical care.
To better understand the current state of fall risk assessment we will review three fall risk assessment tools, including the performance of each and implications for clinical practice: The JHFRAT – as the benchmark for traditional tools, the Epic Risk of Fall (ROF) cognitive computing model, and the Kinometrix Fall Risk Assessment Solution (K-FRAS).
