Identifying Fall Risk Factors: A 7-Year Home Study Using Innovative Marginal Predictions Strategy

Background Falls can occur unpredictably or follow patterns linked to modifiable risk factors and adverse outcomes. Identifying fall trajectories and their key predictors can help clinicians implement targeted prevention strategies. We hypothesize that distinct clinical fall trajectories exist, each with identifiable baseline predictors. Methods This seven-year prospective study followed 1,648 community-dwelling older adults (≥ 60 years). Participants were assessed at home. Data collection included cardiovascular risk factors, fall occurrences, socio-environmental characteristics and a comprehensive geriatric assessment summary score. Fall trajectories were identified using a Gaussian Mixture Model (GMM) and Multinomial Logistic Regression (MLR) determined the predictors of each trajectory. Marginal prediction allowed us to refine predictor analysis by identifying the category within each feature that contributes the most to the steadiest trajectory. Results Four distinct fall trajectories were identified during the 7 years follow up: Cluster Falls and No Falls (65.5%), Increasing Falls (6.7%), Chronic Recurring Falls (15.7%) and Low-Rate Chronic Falls (12.1%). At baseline the steadiest trajectory is Increasing Falls. Clustered Falls and No Falls trajectory is characterized by lack of leisure activities, functional impairment (Instrumental Activities of Daily Living [IADL] < 8) and pathological performance on the Single Leg Balance (SLB) test. The Chronic Recurring Falls trajectory was primarily composed of women with obesity. The Low-Rate Chronic Falls group also consisted mainly of obese women with IADL < 8 and pathological SLB. We also investigated predictors at the 18-month follow-up. Conclusions Falls in older individuals may occur at discrete intervals or follow recurrent patterns, including chronic recurrence, all of which are associated with increased risks. Women, obesity, impairment in activities of daily living, reduced physical performance and depressive symptoms should be prioritized for intervention.