Introduction: Frailty is a complex geriatric syndrome with a multifaceted etiology. We aimed to identify the best combinations of risk factors that predict the development of frailty using recursive partitioning models.
Methods: We analyzed reports from 1,724 community-dwelling men and women aged 65-74 years participating in the International Mobility in Aging Study (IMIAS). Frailty was measured using frailty phenotype scale that included five physical components: unintentional weight loss, weakness, slow gait, exhaustion, and low physical activity. Frailty was defined as presenting three of the above five conditions, having one or two conditions indicated prefrailty and showing none as robust. Socio-demographic, physical, lifestyle, psycho-social, and life-course factors were included in the analysis as potential predictors.
Results: 21% of pre-frail and robust participants showed a worse stage of frailty in 2014 compared to 2012. In addition to functioning variables, fear of falling (FOF), income, and research site (Canada vs. Latin America vs. Albania) were significant predictors of the development of frailty. Additional significant predictors after exclusion of functioning factors included education, self-rated health, and BMI.
Conclusions: In addition to obvious risk factors for frailty (such as functioning), socio-economic factors and FOFs are also important predictors. Clinical assessment of frailty should include measurement of these factors to identify high-risk individuals.
Keywords: Frailty; Logistic regression tree; Prediction; Recursive partitioning.
© 2022 S. Karger AG, Basel.