Optimising the use of cardiac biomarkers in the diagnosis and risk stratification of patients in the Emergency Department

Patients with suspected acute coronary syndrome and acute heart failure account for a very large number of emergency presentations to hospital, however the diagnosis can be challenging because many other life-threatening conditions often produce very similar clinical signs and symptoms. Clinical guidelines therefore recommend the use of cardiac troponin and natriuretic peptides to aid in the diagnosis and risk stratification of these patients in the Emergency Department. Despite these recommendations, there remains significant uncertainty in the diagnostic performance of these biomarkers. My aim was therefore to evaluate the diagnostic performance of current guideline-recommended diagnostic thresholds of cardiac troponin and natriuretic peptides in large patient cohorts and subsequently develop novel approaches to optimise the use these biomarkers. In 918 consecutive patients attending the emergency department without suspected acute coronary syndrome, I evaluated the prevalence, determinants and clinical outcomes of those with elevated high-sensitivity cardiac troponin concentrations. One in eight patients without suspected acute coronary syndrome had cardiac troponin concentrations above the guideline-recommended 99th centile. The majority of these patients have non-ischaemic myocardial injury. Troponin concentration was strongly associated with age, co-morbidities, adverse physiology at presentation and subsequent poorer outcomes. Using data from 48,282 (47% women) patients enrolled in a stepped-wedge cluster-randomised controlled trial across ten hospitals, I evaluated the impact of implementing a high-sensitivity cardiac troponin I assay with sex-specific diagnostic thresholds for myocardial infarction in women and men with suspected acute coronary syndrome. Use of a hs-cTnI assay with sex-specific thresholds identified five-times as many additional women with myocardial injury than men, such that the proportion of women and men with myocardial injury is now equivalent. Despite this increase, women remain less likely than men to receive treatment for myocardial infarction and the rates of subsequent myocardial infarction or cardiovascular death were not substantially reduced in either women or men following implementation of high-sensitivity troponin testing into clinical practice. To improve the diagnostic performance of high-sensitivity cardiac troponin, I developed a decision-support tool that combines high-sensitivity cardiac troponin concentration as a continuous variable and other objective clinical variables using statistical models to calculate an individualised probability of type 1 myocardial infarction. I found that this decision-support tool was able to rule-in and rule-out type 1 myocardial infarction more accurately than high-sensitivity cardiac troponin thresholds alone. I performed an international, collaborative individual patient-level meta-analysis to evaluate the diagnostic performance of N-Terminal pro-B-type natriuretic peptide (NT-proBNP) for the diagnosis of acute heart failure. Using data from 10,369 patients with suspected acute heart failure across 14 cohorts from 13 countries, I found that the performance of guideline-recommended NT-proBNP thresholds for acute heart failure varied significantly across important patient subgroups. I subsequently used this data to develop and externally validate a decision-support tool, which combines NT-pro-BNP as a continuous measure with clinical variables using statistical models to determine the probability of acute heart failure for individual patients. This decision-support tool accurately ruled-in and ruled-out acute heart failure and performed consistently across all subgroups. My findings suggest that current guideline-recommended thresholds of cardiac troponin and NT-proBNP have suboptimal diagnostic performance for acute myocardial infarction and acute heart failure. Decision-support tools that incorporate these biomarkers as a continuous variable with other important associated patient factors using statistical models have significantly improved diagnostic performance. Prospective studies are now required to evaluate the impact of implementing these decision-support tools on healthcare resource utilisation and patient outcomes