To the Editor: In the article by Davey1 no evidence other than deviation from a protocol published months before the study is produced to document the implied inappropriateness of single troponin assays.

Emergency physicians are experienced in assessing undifferentiated chest pain. Acute coronary syndromes are but one cause of presentation to emergency departments (EDs) of patients with chest pain, and indeed are but one cause of elevated serum troponin levels.

Many reasons may justify the “appropriate” ordering of single troponin assays. Some patients present to EDs many hours after their episode of chest pain. A single troponin test may be a very useful and sensitive test for a patient whose chest pain occurred yesterday. How many patients in the study group had their single troponin test done more than 12 hours after their episode of pain? How many patients discharged themselves against medical advice as they were unwilling to wait 6–8 hours for a second blood test to triage their risk for an acute coronary syndrome? How many patients died or were transferred to another hospital? How many patients had their single troponin test ordered in the investigation of a primarily non-cardiac illness, such as sepsis or pulmonary embolism?

I have no doubt that many troponin assays ordered in the study population were inappropriate. But, by failing to conduct an explicit medical record review of those patients whose tests were deemed inappropriate, the author has failed to answer his stated aim of determining if the troponin assay is used appropriately when chest pain is encountered. We are left with no knowledge of whether this problem is small or large.

Finally, does it matter? Are two consecutive negative troponin assays required to triage patients with chest pain? Recently, the Journal published a clinical outcome study that examined the implementation of a chest pain assessment protocol at a metropolitan university teaching hospital in Bankstown, Sydney.2 Patients presenting to the ED with “possibly cardiac” non-traumatic chest pain who were deemed to be low risk did not receive a second, late troponin assay, and yet this approach appeared to be safe.

Those of us who have an interest in the rational use of diagnostic testing for patients with acute coronary syndromes eagerly await the publication of further evidence on this important matter.

In reply: In acute myocardial infarction (AMI) diagnosis, the sensitivity and specificity of troponin rise with time after symptom onset. The sensitivity of troponin-I testing was shown to increase from 35% at 0–4 hours to 97% at 12–24 hours after an infarct.1 Similarly, a meta-analysis found that “multiple testing of individual biomarkers over time substantially improves sensitivity, while retaining high specificity” for AMI diagnosis.2 This position is taken by the National Academy of Clinical Biochemistry3 and European and American cardiologists.4

Furthermore, the diagnostic clock starts from a patient’s emergency department presentation if there is any unreliability suspected in the patient’s assessment of pain onset. Pain onset may have been stuttering, indeterminate, simply forgotten, some combination of these, or even absent. Bailey describes several situations such as these, thought to justify, or to explain, singlicate troponin testing, and suggests that, as I did not audit records, I was not able to quantify the true extent of inappropriate ordering. I acknowledged this shortcoming, but believe it does not detract from the endpoint found.

The Bankstown low-risk patients5 are only a confounder here. In our protocol they probably would not have been thought to have cardiac pain, and all the remaining Bankstown patients had serial biomarker testing.

In my audit,6 93% of singlicate troponin test orders did not diagnose an AMI, and if AMI were still considered, then the tests contravened the protocols.3,4 This is why they were called “inappropriate” — no arbitrary whim.

I assumed, moreover, that no clinician ordered a troponin test unless seeking a cause for chest pain. Our protocol begins with chest pain, recognises uncertainty, and leads through to treating an AMI or reconsidering the diagnosis.

Obliquely invoking Ockham’s principle is also dangerous here. Illnesses such as sepsis may “provoke” an AMI, but this must then be investigated independently, and according to its own rules of engagement, which do not alter solely because of the primary (co-)morbidity.

In short, we did know what is appropriate troponin use, and surveyed it. Like Bailey, we look forward to seeing further evidence.