Informing the Design of a Computerised Decision Support Tool for Falls Risk Assessment: Exploring Modifiable Risk Factors and Data Availability in Acute Hospitals

BACKGROUND: Falls are the most reported safety incident in acute hospitals, causing physical (e.g., fractures) and psychological harm (e.g., fear of falling). NICE guidelines recommend comprehensive falls risk assessment and tailored multifactorial interventions targeting modifiable risk factors, which can reduce inpatient falls by 25-30% and associated costs by 25%. However, current practice often categorises patients by falls risk level, applying blanket interventions to high-risk patients, missing opportunities to reduce inpatient falls through a personalised multifactorial approach. We propose (1) developing a computerised decision support (CDS) tool for comprehensive falls risk assessment and multidisciplinary delivery of tailored multifactorial interventions, including patient education, and (2) exploring whether an artificial intelligence (AI)-based CDS tool can improve on a rule-based tool by leveraging real-world data to recommend effective, individualised interventions, improving over time based on patient outcomes. First, we must establish a thorough understanding of modifiable falls risk factors and their interrelationships, to ensure the tool captures the most relevant factors while minimising documentation burden. We must also assess electronic health record (EHR) data to determine if: (i) data relevant to falls risk factors is routinely collected, reducing documentation burden; (ii) sufficient data about modifiable risk factors and delivered interventions to develop an AI-based CDS tool; (iii) delivery of tailored multifactorial interventions could serve as a secondary outcome measure in a future trial. AIM: To determine the feasibility of designing and evaluating a comprehensive falls risk assessment CDS tool through (1) developing a comprehensive understanding of modifiable falls risk factors; and (2) evaluating the availability and quality of EHR data related to these factors and delivery of falls prevention interventions. DEVELOPMENT WORK PLAN: WP1: Identify modifiable falls risk factors through a systematic review of hospital falls prevention guidelines. Findings will inform two stakeholder workshops where we will co-create with domain experts a directed acyclic graph to represent causal relationships between risk factors and falls. For newly proposed factors, we will undertake a literature review using the Evidence Synthesis for Constructing Directed Acyclic Graphs methodology. WP2: Assess availability and quality of EHR data on falls risk factors and delivery of falls prevention interventions through patient record review (manual and automated queries) and staff interviews across three hospitals with diverse patient populations. WP3: Produce a final report and work with collaborators to design a Programme Grant for Applied Research (PGfAR), informed by study findings. DELIVERY TIMELINES: The 19-month study concludes in April 2027 with a PGfAR application submitted to the next call. ANTICIPATED IMPACT: This study will enhance understanding of modifiable falls risk factors, their causal relationships, and EHR data availability and quality. Findings will underpin development of a tool supporting comprehensive falls risk assessment, multidisciplinary delivery of tailored interventions, and patient education, aiming to reduce inpatient falls, alleviate suffering, and lower healthcare costs. DISSEMINATION: We will work with lay researchers and Study Steering Committee to create a dissemination strategy, including open access publications, project website, and social media, e.g., Bluesky.