Bleeding complications following acute myocardial infarction : time trends, risk assessment and associated prognosis

Abstract: Background: In patients with acute myocardial infarction (MI), bleeding complications are common and associated with worse prognosis. This thesis aimed to investigate the epidemiology, risk assessment and associated outcomes of bleeding complications in patients with acute MI. Methods and results: Study I: Patients with acute MI enrolled in the SWEDEHEART registry from 1995–2018 were included (n=371 431). The incidence of in-hospital and out-of-hospital bleeding at one-year was investigated parallel to treatment changes and ischemic outcomes. From 1995 to 2018, in-hospital bleeding increased from 0.5% to 1.3% and out-of-hospital bleeding increased from 2.5% to 4.8% along with increased use of invasive revascularisation and more efficient antithrombotic treatment. Meanwhile in-hospital and out-of-hospital ischemic outcomes decreased from 12.1% to 5.6% and 27.5% to 15.1%, respectively. Study II: Patients with acute MI enrolled in the SWEDEHEART registry from 2009– 2014 were included (n=97 597). A prediction model for in-hospital bleeding was created using logistic regression and the performance was compared to that of the CRUSADE and ACTION scores. Due to miscalibration, the CRUSADE and ACTION scores were recalibrated. The SWEDEHEART score, consisting of five baseline variables (haemoglobin, age, sex, creatinine, and C-reactive protein) plus one interaction term (haemoglobin and sex) had a C-index of 0.80 as compared with 0.72 and 0.73 for the recalibrated CRUSADE and ACTION scores, respectively. Study III: Patients with acute MI enrolled in the SWEDEHEART registry from 2007–2016 and discharged alive on any antithrombotic treatment were included (n=149 447). The incidence, associated outcomes and predictors of upper gastrointestinal bleeding (UGIB) was investigated. The incidence of UGIB within one year after discharge was 1.5% and experiencing UGIB was associated with increased risk of mortality and stroke, but not significantly associated with MI. Using both logistic regression and machine-learning models, new potential predictors of UGIB were found, such as smoking status and blood glucose. Study IV: Patients with acute MI enrolled in the SWEDEHEART registry and discharged alive on any antithrombotic treatment from 2012–2017 were included (n=86 736). The incidence and associated mortality risk of ischemic (MI or ischemic stroke) and bleeding events was investigated. Within one year after discharge, the incidence rate of ischemic and bleeding events was 5.7/100 person years and 4.8/100 person years, respectively. Both ischemic and bleeding events were associated with higher risk of mortality as compared with no event, with adjusted hazard ratios (HR)s of 4.16 (95% CI 3.91 to 4.43) and 3.43 (95% CI 3.17 to 3.71), respectively. In a direct comparison of ischemic vs bleeding event, the adjusted HR was 1.27 (95% CI 1.15 to 1.40.) Conclusion: In the past two decades, the incidence of both short- and long-term bleeding events has nearly doubled in patients with acute MI. The five-item SWEDEHEART score predicts inhospital bleeding in patients with acute MI more accurately than the recalibrated CRUSADE and ACTION scores. Among patients with a recent MI, upper gastrointestinal bleeding is common and associated with poorer prognosis. Beyond the known risk factors for bleeding, other predictors for upper gastrointestinal bleeding may be present. In patients discharged after an acute MI, ischemic events were more common and associated with higher risk of mortality than bleeding events.