Bleeding in cardiac patients prescribed antithrombotic drugs: Electronic health record phenotyping algorithms, incidence, trends and prognosis
BackgroundClinical guidelines and public health authorities lack recommendations on scalable approaches to defining and monitoring the occurrence and severity of bleeding in populations prescribed antithrombotic therapy. We aimed to develop electronic health record algorithms for different bleeding phenotypes, and to determine the incidence, time trends and prognosis of bleeding in patients with incident cardiac disorders indicated for antiplatelet and/or vitamin K antagonist (VKA) therapy. MethodsWe examined linked primary care, hospital admission and death registry electronic health records (CALIBER 1998-2010, England) of patients with newly diagnosed atrial fibrillation, acute myocardial infarction, unstable angina or stable angina to develop algorithms for bleeding events. Kaplan-Meier plots were used to estimate the incidence of bleeding events and we used Cox regression models to assess prognosis for all-cause mortality, atherothrombotic events and further bleeding following bleeding events. ResultsWe present electronic health record phenotyping algorithms for bleeding based on bleeding diagnosis in primary or hospital care, symptoms, transfusion, surgical procedures, and haemoglobin values. In validation of the phenotype we estimated a positive predictive value of 0.88 (95% Cl: 0.64, 0.99) for hospitalised bleeding. Amongst 128,815 patients, 27259 (21.2%) had at least one bleeding event, with 5 year risks of bleeding of 29.1%, 21.9%, 25.3% and 23.4% following diagnoses of atrial fibrillation, acute myocardial infarction, unstable angina and stable angina respectively. Rates of hospitalised bleeding per 1000 patients more than doubled from 1.02 (95% Cl: 0.83, 1.22) in January 1998 to 2.68 (95% Cl: 2.49, 2.88) in December 2009 coinciding with increased rates of antiplatelet and VKA prescribing. Patients with hospitalised bleeding and primary care bleeding, with or without markers of severity, were at increased risk of all-cause mortality and atherothrombotic events compared to those with no bleeding. For example the hazard ratio for all-cause mortality was 1.98 (95% Cl: 1.86, 2.11) for primary care bleeding with markers of severity, and 1.99 (95% Cl: 1.92, 2.05) for hospitalised bleeding without markers of severity, compared to patients with no bleeding. ConclusionsElectronic health record bleeding phenotyping algorithms offer a scalable approach to monitoring bleeding in the population. Incidence of bleeding has doubled in incidence since 1998, affects 1 in 4 cardiac patients, and is associated with poor prognosis. Efforts are required to tackle this iatrogenic epidemic. O_LSTWhat is already known?C_LSTO_LIClinical guidelines and public health authorities lack recommendations on how to define or monitor the occurrence and severity of bleeding in populations. C_LIO_LIThis is particularly important because clinical guidelines increasingly recommend the use of one, two or three antiplatelet and vitamin K antagonist drugs to lower the risk of subsequent atherothrombotic events in common heart diseases including atrial fibrillation, acute coronary syndromes and chronic stable angina. C_LIO_LIClinical guidelines lack consistent recommendations of how to reduce the main side effect of bleeding. C_LIO_LIFor acute myocardial infarction it has been shown that combining primary care electronic health records (which include information from hospital discharge summaries) and hospital admission data can generate valid EHR disease phenotypes and provide real-world estimates of disease occurrence. C_LI O_LSTWhat is not known?C_LSTO_LIIt is not known how to define bleeding occurrence and severity in large scale, unselected populations by combining available information on bleeding diagnosis in primary or hospital care, symptoms, transfusion, surgical procedures, and haemoglobin values. C_LIO_LIThe population-based incidence, time trends and long-term prognosis of bleeding have not been evaluated in people with common cardiac disorders. C_LIO_LIComparisons of the population burden of bleeding across common cardiac disorders, such as atrial fibrillation, acute coronary syndromes and stable angina, are lacking. C_LI O_LSTWhat this study adds?C_LSTO_LIPhenotype: We developed standardised replicable EHR phenotyping algorithms defining bleeding and severity measures based on available clinical information across structured primary and hospital care EHR sources. C_LIO_LIIncidence: At 5 years of follow-up, one in five patients with cardiac disease had a bleeding event and 6.5% had fatal or severe bleeding. C_LIO_LITrends: There was approximately a two-fold increase in incidence of primary care and hospitalised bleeding between 1998 and 2010. The rate of fatal bleeding remained stable. C_LIO_LIPrognosis: Patients with bleeding recorded in primary care or in hospital admissions are at increased bleeding between 1998 and 2010. The rate of fatal bleeding remained stable, risk of all-cause death and atherothrombotic events. C_LI