Impact of frailty phenotype on chronic disease accumulation and healthy life years lost: A multi-state analysis of the UK Biobank

Abstract

Background Frailty is a systemic vulnerability syndrome linked to adverse health outcomes, yet its role in shaping the dynamic trajectory from health through incident disease to multimorbidity remains poorly understood. Methods Among 276,696 disease-free UK Biobank participants (mean age 55.3 years; 53.5% female; median follow-up of over 13 years), we evaluated baseline frailty phenotype (robust, pre-frail, frail) as a predictor of 43 chronic conditions using Cox models and restricted mean survival time (RMST) analysis. A four-state Markov multistate model mapped transitions across healthy, incident disease, multimorbidity, and death. Multimorbidity clustering was performed via multiple correspondence analysis and K-means. Population attributable fractions (PAFs) quantified the frailty-attributable disease burden. Results Frailty was significantly associated with 42 of 43 conditions (97.7%). The frail group lost 2.72 years (95% CI, 2.64–2.81) of multimorbidity-free survival over 12 years. In multistate models, frailty was strongly associated with higher hazards of transition from healthy to incident disease (HR 1.77; 95% CI, 1.72–1.83) and from incident disease to multimorbidity (HR 1.73; 95% CI, 1.68–1.79), but showed attenuated associations with post-disease mortality. Four stable multimorbidity phenotypes were identified. PAF analysis indicated that eliminating pre-frailty and frailty could theoretically prevent 9.85% of incident disease events and 9.40% of multimorbidity progression, with pre-frailty contributing the majority of the attributable burden (7.83%). Conclusions Baseline frailty was strongly associated with broad-spectrum chronic disease onset and accelerated multimorbidity accumulation. Systematic screening and early intervention targeting pre-frailty, particularly in middle-aged populations, holds substantial promise for delaying the multimorbidity trajectory.