3725 Developing a model to predict mobility decline in community dwelling older people

Abstract Introduction Being able to walk is a priority for older people and is key to maintaining independence. Declining mobility is an early predictor of loss of independence, reduced quality of life, increased health care use and death. The aim of this study was to develop and validate a prediction model to identify when an older person was at risk of self-reported mobility decline over a 2-year period. Method We used self-reported data from a prospective cohort study of 5409 people aged 65 years and over in England (The Oxford Pain, Activity and Lifestyle (OPAL) Cohort Study). Mobility status was assessed using the EQ-5D-5L mobility question. The outcome was any mobility decline at two years. Thirty-one candite variables were entered into the model including demographic factors, pain, walking, falls, comorbidities, general health and physical activity. LASSO logistic regression was used to select predictors. Models were internally validated using bootstrapping. Scores were assigned to identified predictors to calculate an individual’s risk of mobility decline. Results Over 18% of participants who could walk at baseline reported mobility decline at year two. The following variables were identified as predictors: Age Adequacy of income; Body Mass Index; Usual walking pace; Difficulties maintaining balance; Confidence to walk; Use of walking aid; Change in walking ability over 12 months; Lower limb pain; Current pain/discomfort severity; Number of health conditions; Physical tiredness; Self-reported general health; Current mobility level. Conclusions A prediction model for mobility decline were developed and internally validated. These questions could be used as an assessment tool within primary care or by older people themselves. External validation is required. We are working with stakeholders to understand how this model could be used to help older people maintain mobility.