Ischaemic heart disease persists as the world's leading cause of mortality with serious consequences for unrecognised disease.1 Chest pain, the cardinal symptom of acute coronary syndromes (ACS), accounts for over six million adult emergency department (ED) visits in the USA and greater than 360 000 in England and Wales each year.2 ,3 If ACS is diagnosed early, there are effective interventions that improve outcomes. Therefore, increasing our diagnostic accuracy in ACS, specifically for acute myocardial infarctions (AMI), including both ST segment and non-ST segment elevation, is critical. A major part of these efforts is the use of biomarkers, primarily the serum troponin assay. To address the expanding clinical use of different troponin assays, the American College of Cardiology Foundation (ACCF) recently issued guidelines on interpreting troponin assays in the clinical setting.4 It notes several recent studies supporting high-sensitivity troponin (HS-Tn) assays as a mechanism to rapidly rule out AMI. These assays are quickly replacing troponin-I as the standard of care in routine clinical practice.4–8 However, these studies do not address one consequence of increasing sensitivity: the tradeoff of increased positive results in the absence of ACS and their associated patient health and financial costs.

Several studies have demonstrated the ability of HS-Tn assays to rapidly ‘rule out’ AMI in their respective populations when coupled with standard clinical assessment and ECG.4–8 Achieving a ‘rapid rule-out,’ that is, quickly obtaining accurate negative results, is helpful for the process of safely excluding persons without AMI. A perfect test would achieve a sensitivity of 100%, and no individual with negative test results would have active disease. However, the positive predictive value, which is a function of the incidence of a disease in the population being tested, is low in populations with low AMI incidence, in spite of high-sensitivity tests. Troponin assays are used for typical angina pectoris-related chest pain and are also frequently used in populations with atypical symptoms, including epigastric pain, back pain, arm pain, diaphoresis, transient loss of consciousness, nausea and vomiting.1 ,4 ,9–12 Elevated results in the absence of ACS frequently occur in low-incidence populations and are particularly worrisome, as they often lead to harms, including unnecessary testing (some with radiation exposure), hospital admissions and contribute to additional anxiety.4 Given the very large number of patients evaluated for ‘soft rule outs’, the inappropriate application of HS-Tn assays to broad populations, namely those with low AMI incidence, may do more harm than good. Such circumstances could result in a higher number of tests showing elevated troponin while capturing relatively fewer additional cases of AMI that may have been missed by lower sensitivity tests or observation alone.

To illustrate, we postulate an idealised HS-Tn assay which is 100% sensitive and 95% specific7—a conservative, but not unrealistic, estimate by currently available assay standards. Based on 2008 estimates of 6 396 00013 adults presenting to US EDs for chest pain, we analysed the relationship of AMI incidence and the positive predictive value (figure 1). Patients with a low risk of AMI (atypical history, symptoms and ECG findings) may have an estimated AMI incidence of 10%,4 and we would therefore expect there to be 1 215 240 positive tests with 639 600 myocardial infarctions in patients who presented to the ED complaining of chest pain (point B in figure 1). We would then expect 47% (575 640 individuals) of these results to have elevated troponin in the absence of underlying active disease—no better than a coin toss in accuracy but with many more downstream consequences. Although prior studies have demonstrated high sensitivity and effective rule out, there has been little attention paid to these ambiguously positive results and insufficient focus on applying these tests to the correct population at the optimal time and in the appropriate clinical context.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

The graph demonstrates changes that occur to positive predictive value as a function of acute myocardial infarction incidence in a population being tested. We demonstrate a range from low to high risk (defined by risk factors, history, symptoms and ECG changes). To the right, we include three 2×2 tables to demonstrate how changes in the incidence of myocardial infarction in patients presenting with chest pain dramatically affect treatment characteristics. Elevated troponin tests in the absence of underlying acute disease are in bold in each population.

Narrowing the use of HS-Tn assays to high-risk populations (considering age, sex, typical history, symptoms and ECG findings) will clarify the utility of positive results and avoid unnecessary interventions. The ACCF focuses on the clinical guidelines related to troponin usage by detailing an algorithm for better establishing pretest probabilities. Concurrently, we recommend targeted imperatives for future troponin research. First, studies should be designed to include information regarding elevated troponin without underlying acute disease, as well as the additional costs and harms for the affected population. By working to better characterise this clinical subgroup in addition to improving sensitivity, we can more carefully analyse the impact of assay results and better understand the possible burden of higher sensitivity assays on patients and the healthcare system. Second, research studies should clearly define the populations (using pretest probability characteristics as defined above) that would benefit from higher sensitivity assays as opposed indiscriminate use. For instance, research approaches have been varied, examining diverse groups of patients from those with high pretest probability to those with known stable disease that are not presenting acutely.13–15 It is not clear what can be done with troponin values that serve to prognosticate but that do not reflect underlying acute ischaemic disease. As such, the significance of positive results for ‘rapid rule out’ do not apply in ‘real-world’ clinical settings, as these patient populations differ from the studies. Studies of HS-Tn assays as tools for ‘rapid rule out’ need to be verified in the context of ‘real-world’ scenarios, for example, in patients with renal failure, congestive heart failure and atypical angina symptoms, all of whom present to the ED with chest pain. The ACCF has provided much needed clinical guidelines for both AMI and other causes of elevated troponin, but research with fewer inclusion limitations could give information on risks and benefits in more ‘real-world’ populations possibly to inform who may benefit the most from such testing. Finally, additional studies are needed to analyse how troponin assays are currently being used to evaluate their impact and understand how clinicians are implementing these assays in routine clinical practice.

Although, HS-Tn assays are powerful diagnostic tools, their application to populations with low incidence of AMI leads to an unacceptably high proportion of elevated results. Future studies must report on elevated troponin in the absence of underlying disease and the disposition of these patients, emphasise the need for the application of such tests to the correct populations, and examine the way in which high-sensitivity troponins are used in ‘real-world’ clinical practice.