Predictors of Anxiety in Middle-Aged and Older European Adults: A Machine Learning Comparative Study

Abstract

Anxiety in older adults is a prevalent yet under-recognized condition associated with significant societal and individual burdens. This study used a machine learning approach to compare the relative importance of 57 risk and protective factors for anxiety symptoms in a population-representative sample of middle-aged and older European adults (N = 65,684; ages 45–103 years; 55.7% women; 15 countries represented). The results revealed loneliness and self-rated poor health as primary risk factors (Nagelkerke R2 = 0.272), with additional predictive contributions from country of residence, functional limitations, financial distress, and family care burden. Notably, follow-up analysis showed that none of the 16 social network variables were associated with loneliness; rather, cohabitating with a partner/spouse was most strongly associated with reduced loneliness. Further research is needed to elucidate directional associations between loneliness and anxiety (both general and sub-types). These findings underscore the imperative of addressing loneliness for mitigating anxiety and related mental health conditions among aging populations.