Allan, Victoria;
(2018)
Exploiting electronic health records for research on atrial fibrillation: risk factors, subtypes, and outcomes.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
eThesisVictoriaAllan.pdf Download (9MB) | Preview |
Abstract
BACKGROUND: Electronic health records (EHRs), collected on large populations in routine clinical care, may hold novel insights into the heart rhythm disorder atrial fibrillation (AF). AIM: To exploit EHRs to investigate, validate and extend evidence for AF risk factors, subtypes, and outcomes. METHODS: The CALIBER dataset (1997–2010) linking primary care, secondary care, and mortality records for a representative subset of the UK population was used (i) to model associations between cardiovascular disease (CVD) risk factors and incident AF, including AF with (AF+) and AF without (AF–) intercurrent CVD, (ii) to create EHR definitions for eight AF subtypes (structural, focal, polygenic, postoperative, valvular, monogenic, respiratory and AF in athletes) and (iii) to investigate stroke outcomes by CHA2DS2-VASc, sex, and warfarin use. RESULTS: Among 1,949,052 individuals, 50,097 developed incident AF: 12,652 (25.3%) with AF+ and 37,445 (74.7%) with AF–. Smoking (HR [95%CI] for AF+ vs. AF–: 1.66 [1.56,1.77] vs. 1.21 [1.16,1.25]), hypertension (2.19 [2.11,2.27] vs. 1.65 [1.62,1.69]), and diabetes (2.03 [1.94,2.12] vs. 1.45 [1.41,1.49]) showed consistent direct associations with AF+ and AF–, while heavy drinking (1.17 [0.81,1.67] vs. 1.99 [1.68,2.34]) and total cholesterol levels (0.99 [0.96,1.02] vs. 0.85 [0.84,0.87]) showed inconsistent associations with AF+ and AF–. EHR definitions for AF subtypes were created by combining 2813 diagnosis, medication, and procedure codes. There were 12,751 individuals with AF and valvular heart disease. Prosthetic replacements, mitral stenosis and aortic stenosis showed higher HR [95%CI] for stroke, thromboembolism and mortality (1.13 [1.02,1.24], 1.20 [1.05,1.36], and 1.27 [1.19,1.37] respectively). The net-clinical benefit (NCB [95%CI] per 100 person-years) of warfarin was shown from CHA2DS2-VASc≥2 in men (0.5 [0.1,0.9]) and CHA2DS2-VASc≥3 in women (1.5 [1.1,1.9]). CONCLUSION: AF is a heterogeneous condition associated with diverse disease mechanisms. EHRs can help refine understanding of risk factors, subtypes, and outcomes with relevance for clinical practice.
Type: | Thesis (Doctoral) |
---|---|
Qualification: | Ph.D |
Title: | Exploiting electronic health records for research on atrial fibrillation: risk factors, subtypes, and outcomes |
Event: | UCL (University Collge London) |
Open access status: | An open access version is available from UCL Discovery |
Language: | English |
Keywords: | atrial fibrillation, electronic health records, epidemiology |
UCL classification: | UCL |
URI: | https://discovery.ucl.ac.uk/id/eprint/10047654 |
Archive Staff Only
View Item |