Abstract 15650: Valvular Atrial Fibrillation: Electronic Health Record-Derived Phenotypes, Trends in Prevalence and Prognosis.

Introduction: Non-valvular atrial fibrillation (AF) has been studied extensively, but contemporary studies have neglected valvular AF. Electronic health records (EHR) enable high-resolution clinical research, but computable EHR-derived definitions of valvular AF are lacking. Hypothesis: Valvular AF is increasing in prevalence over time and is associated with worse prognosis, compared with non-valvular AF. Methods: We used the population-based CALIBER resource linking primary and secondary EHR on individuals with AF in England from 1998 until 2010. Using diagnosis, procedure and prescription terms from the Read, ICD-10, OPCS-4 and BNF controlled clinical terminologies, we developed and evaluated an EHR phenotyping algorithm to identify 18 valvular disease subtypes. Associations with a composite endpoint of incident stroke (ischaemic, haemorrhagic and unspecified), systemic embolism, and all-cause mortality were analysed. Results: Among 76019 individuals with AF with median follow-up of 2.2 years, the final algorithm combined 406 diagnosis terms identifying 8623 (11.3%) individuals with valvular disease at baseline and a further 4,128 (5.4%) with valvular disease after baseline. Between 1998 and 2010, increased prevalence of valvular disease was observed in individuals with AF aged 65-75, 75-85 and 85+, but not in those <65 years. There were 31934 endpoints (stroke, systemic embolism and all-cause mortality), 3764 (11.8%) in individuals with valvular disease at baseline. Compared with no valvular disease, individuals with prosthetic valves, mitral stenosis and aortic stenosis had higher adjusted hazard ratios[95% confidence intervals] for the combined endpoint (stroke, systemic embolism and mortality) of 1.13[1.02, 1.24], 1.20[1.05, 1.36], and 1.27[1.19, 1.37], while individuals with bioprosthetic valve replacements had a lower adjusted hazard ratio of 0.78[0.68, 0.88]. Conclusions: EHR-derived phenotypes of valvular AF subtypes are feasible. Among individuals with AF, prevalence of valvular disease increased with age. Between 1998 and 2010, prevalence of valvular disease has increased in individuals aged >65 years. Compared with non-valvular AF, AF associated with aortic stenosis, mitral stenosis and prosthetic valves was associated with increased risk of stroke and mortality. [ABSTRACT FROM AUTHOR]