Epidemiological evaluation of oral anticoagulants prescribing and clinical outcomes in atrial fibrillation patients with and without cancer: analysis of primary care data in England

Abstract

Atrial fibrillation (AF) is a prevalent cardiac arrhythmia imposing a substantial global burden. Given its five-fold increase in stroke risk, prescribing oral anticoagulants (OAC) to intermediate to high-risk AF patients is crucial. However, OAC prescribing rates vary, influenced by factors beyond stroke risk, particularly evident in patients with cancer who face complex clinical conditions affecting stroke and bleeding risks. Yet, the efficacy of risk assessment tools in this population remains unexplored, complicating anticoagulation therapy initiation, often tailored to individual patients. Nevertheless, evidence regarding this matter remains limited. Using the Clinical Practice Research Datalink (CPRD), this thesis presents unique research on Nonvalvular Atrial Fibrillation (NVAF) epidemiology in England. It addresses five key questions: 1) NVAF incidence and OAC prescribing, 2) factors influencing OAC prescription, 3) MB incidence and OAC resumption in NVAF patients, 4) stroke and bleeding risk comparison in NVAF patients with and without cancer, and 5) CHA2-DS2-VASc and HAS-BLED score performance in predicting stroke and bleeding in NVAF patients with and without cancer history. Key findings reveal a temporal increase in NVAF incidence in England until 2015, subsequently plateauing. Disparities in OAC prescription correlate with comorbidities, ethnicity, and socioeconomic status, emphasizing the need for interventions to address inequities in NVAF patient care. Between 2009 and 2019 the incidence of MB in NVAF patients surged tenfold with many experiencing MB despite lacking OAC prescription at the time of bleeding. The decision to resume OAC post-MB appears contingent upon the initial anticoagulant used and does not significantly associate with recurrent MB risk. Examining NVAF patients with different cancer types revealed varying stroke and bleeding risks, with certain cancers exhibiting higher bleeding risks than stroke risks. Notably, certain cancer types, such as haematological and lung cancer, were less likely to receive OAC, highlighting disparities in care. Finally, it was found that CHA2-DS2-VASc score performed similarly in predicting ischemic stroke in NVAF patients, irrespective of cancer history. In contrast, the HAS-BLED score, while well- calibrated, lacked discrimination in predicting major bleeding events in the NVAF population overall and in specific cancer cohorts. Overall, this thesis contributes to the evidence around the pharmacoepidemiology of OACs in the NVAF population in England and highlights socioeconomic disparities in NVAF care. It addresses challenges in managing NVAF cohorts with major bleeding or specific cancers, where current risk assessment scores may inadequately predict clinical outcomes. Further research is necessary to explore health inequalities in OAC prescribing for AF patients in England and understand why certain cancers predispose individuals to bleeding or stroke events.