Sex- and age-specific temporal patterns of multimorbidity before and after incident atrial fibrillation: a longitudinal electronic health records analysis
R Kousovista, N M S De Groot, M KavousiAbstract
Background
Multimorbidity is highly prevalent in patients with atrial fibrillation (AF) and is associated with worse outcomes. Most patients present with multiple chronic conditions at AF diagnosis, yet how these pre-existing condition patterns influence post-diagnosis outcomes remains poorly understood.
Purpose
To identify sex- and age-specific pre-AF multimorbidity trajectory clusters and examine their association with post-AF disease accumulation and mortality.
Methods
We analysed 22 807 incident AF patients matched 1:1 to controls using linked general practice and hospital records. Pre-AF multimorbidity trajectories were clustered using network-based machine learning approach across twenty seven chronic conditions. We characterised prevalent conditions at AF diagnosis, post-AF incident conditions, multimorbidity (two or more chronic conditions in addition to AF) and all-cause mortality across clusters, stratified by sex and age (45-64 vs ≥65 years).
Results
At AF diagnosis, 72.6% had prevalent multimorbidity versus 43.7% of controls (p<0.001), with mean 5.65 prevalent conditions in AF patients versus 2.82 in controls. Network-based clustering identified distinct sex-age-specific phenotypes: males <65 showed respiratory-neuropsychiatric (COPD 31%, anxiety 30%) versus cardiometabolic-renal (diabetes 30%, heart failure 27%) patterns; males ≥65 displayed respiratory-thyroid-metabolic (thyroid 29%, COPD 26%) versus neuropsychiatric-cerebrovascular (anxiety 28%, stroke/TIA 20%); females <65 revealed severe premature organ failure (heart failure 33%, COPD 34%, CKD 28%) versus metabolic-neuropsychiatric (anxiety 21%, depression 21%, lipid disorders 22%); females ≥65 required three clusters. Post-AF, heart failure was the most common incident condition (13.8-16.5%). Neuropsychiatric/metabolic clusters with minimal prevalent heart failure developed the highest incident heart failure rates (16.0-16.5%). Post-AF multimorbidity accumulation ranged 33.0-45.5%. Mortality ranged 6.1-11.5%, highest in females ≥65 cardiovascular-renal (11.5%), lowest in females <65 metabolic-neuropsychiatric (6.1%).
Conclusions
Network-based clustering of multimorbidity trajectories from integrated primary-secondary care identifies clinically distinct phenotypes with differing post-diagnosis outcomes. This approach enables personalised prognosis and sex- and age-tailored management strategies for AF patients with multimorbidity.Fig1