Rationale: Managing patients with multimorbidity is complex. Frailty influences vulnerability, while prescribing patterns (polypharmacy, specific drug classes) may impact outcomes differently depending on underlying frailty. Understanding these interactions is crucial for optimising care and predicting health trajectories.
Aim: To investigate the predictive value of frailty phenotypes and longitudinal prescribing patterns on health outcomes in persons with multimorbidity.
Hypothesis: Prescribing patterns are causally associated with health outcomes within distinct frailty phenotypes among individuals with multimorbidity.
1. Cohort: UK Biobank participants with multimorbidity (2+ chronic conditions documented in linked health records – primary care, HES) between 2006-2025.
2. Data: Baseline data, linked primary care records (ATC codes), Hospital Episode Statistics (HES), and death registry data.
3. Phenotyping: Define frailty phenotypes using established markers (e.g., gait speed, grip strength, self-reported exhaustion, weight loss, physical activity levels, polypharmacy).
4. Exposure: Characterise longitudinal prescribing patterns (e.g., cumulative exposure to specific drug classes, polypharmacy count changes over time, specific drug combinations).
5. Outcomes: Ascertain key health outcomes including all-cause mortality, cause-specific mortality, hospitalisation rates, and incidence/progression of specific conditions.
6. Analysis: Time varying Cox regression with stratification by frailty phenotype will be used to investigate causal associations between prescribing patterns and outcomes, adjusting for baseline characteristics, comorbidities, and socioeconomic status.
This study will provide insights into how medication use interacts with frailty to determine outcomes in multimorbidity.
