Abstract
OBJECTIVE: To explore the relationship between long-term physical activity (PA) participation and falls. METHODS: Participants in the Australian Longitudinal Study of Women's Health born 1946-1951 self-reported amounts of PA every 3 years since 1998 (mean age: 54 years, n=11 796). Latent class analysis described profiles of self-reported PA participation over 18 years. Associations between patterns of PA participation and self-reported falls measured in 2019 were examined using multinomial logistic regression adjusted for directed-acyclic graph-informed potential confounders, with the highly active group as the reference category. RESULTS: Women were grouped into five PA participation profiles. Compared with consistently highly active patterns (maintaining ≥300 min/week, 22%), consistently lower levels of PA<100 min/week (18%), consistently some PA<150 min/week (18%) and decreasing PA but maintaining≥150 min/week (n=3540, 30%) had higher odds of non-injurious falls (odds Ratio(lower level) (OR): 1.59, 95% CI 1.29 to 1.97; OR(some PA): 1.27, 95% CI 1.04 to 1.55; OR(decreasing activity):1.29, 95% CI 1.02 to 1.63) and injurious falls (OR(low level): 1.32, 95% CI 1.06 to 1.64; OR(some PA): 1.27, 95% CI 1.04 to 1.54; OR(decreasing activity): 1.47, 95% CI 1.18 to 1.83). No association was found between increasing PA (≥150 min/week, 11%) for non-injurious (OR 1.07, 95% CI 0.89 to 1.29) and injurious falls (OR 1.07, 95% CI 0.90 to 1.29). After adjusting for potential confounders, consistently lower levels of PA remained associated with increased non-injurious falls odds (OR(1998 survey): 1.40, 95% CI 1.11 to 1.77; OR(2016 survey): 1.35, 95% CI 1.07 to 1.71). CONCLUSION: The increased odds of falls among women with consistently lower levels of PA over 18 years supports ongoing participation of 150+ min/week of PA.