Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Troponin tests for cardiac damage

Troponins in acute coronary syndromes
Troponins and cost
Algorithms and outcomes

One of the problems with new diagnostic tests is getting one's brain around what they mean and how and when to use them. There will be masses of papers that are completely useless, where test results are compared in really sick patients with those with no disease. We know that such studies are hopelessly biased ( Bandolier 70 ). What we need are more real world examples of how and when tests can be used to advantage.

For some years new tests of damage to cardiac muscle have been available, measuring proteins called troponins. There are two varieties, T and I, and different forms of tests are used with different levels of sensitivity. There's little doubt that they look useful. Hard evidence has been difficult to come by. Three new publications help a lot.

Troponins in acute coronary syndromes

The evaluation of patients with chest pain suggestive of heart attack is a common and costly problem. We know that patients with chest pain and elevated ST-segments of an electrocardiogram are at risk and should be treated. A meta-analysis from California [1] tells us that patients with non-ST elevations but with a positive troponin result are also at increased risk.


The review sought clinical trials and cohort studies evaluating patients with suspected myocardial infarction. Excluded were studies only on patients with infarction, case-control studies, articles not reporting mortality and those that included patients with ST-elevations. Outcomes chosen pre hoc were mortality, and the combined end point of death or infarction for comparison between patients who had at least one positive troponin test and those patients whose troponin test was negative. Where more than one time point was used, information closest to 30 days was chosen.


After exclusions, seven clinical trials and 19 cohort studies were used, with 5,400 patients examined with troponin T and 6,600 with troponin I. The mean age of patients was 63 years, and two thirds were men. About 25-50% of patients had previous myocardial infarction. Patient cohorts included acute coronary syndrome and unstable angina.

Results for troponin T for death are shown in Figure 1 and for death or myocardial infarction in Figure 2. Outcomes for patients with a positive test occurred more frequently than in patients with a negative test. Similar results were obtained with troponin I.

Figure 1: Death after positive or negative troponin test

Figure 2: Death or infarction after positive or negative troponin test

Rates of death were about 1.5% with a negative troponin test and 6% with positive tests. Rates of death or myocardial infarction were about 5% with a negative troponin test, and about 16% with a positive troponin test. The relative risk was between 3 and 4 (Table 1).

Table 1: Outcomes in patients suspected of myocardial infarction without ST-elevation on electrocardiogram

Number/Total (%)
Test and outcome Positive test Negative test Relative risk
Troponin T
Death 99/1635
4.1 (2.9 to 5.7)
Death or MI 143/872
3.0 (2.3 to 3.9)
Troponin I
Death 108/1981
3.3 (2.5 to 4.4)
Death or MI 13/51
4.9 (2.4 to 10)

Troponins and cost

Barriers to introducing new tests are not restricted to lack of evidence of efficacy. New tests, like new drugs, are more expensive than the ones they replace, and budgetary constraints frequently hamper introducing new tests even when there is abundant evidence. Good evidence that new tests can reduce overall costs is usually lacking, but for troponins a randomised study from Connecticut concludes that their use can reduce overall costs [2].


Patients presenting to the emergency department consenting to the study were randomised (method not stated) to receive a standard clinical evaluation with electrocardiogram and CK-MB determinations, or the same package but with serial troponin T evaluations at presentation, and three and 12 hours after presentation. Exclusions were ST-elevation at presentation, defibrillation or resuscitation.

Clinical data were collected, with information about subsequent care, length of hospital stay, and total charges. Patients were followed up to 30 days. A pre-specified subgroup analysis was by presence or absence of acute coronary syndrome (confirmed unstable angina or acute myocardial infarction), and the final diagnosis was made by a cardiologist using WHO criteria and unaware of troponin values.


The groups were well matched at baseline. The mean age was 65 years and half were women. Serial troponin T measurements were used in 409 patients, but not in 447.

Length of stay was significantly shorter for patients who had troponin measurements with acute or non-acute coronary syndrome (Table 2). Total hospital charges were lower for patients who had troponin measurements, significantly so for those with acute coronary syndrome (Table 2). On average, total costs for patients with troponin measurements were about $900 less, representing a potential annual saving to the hospital in Bridgeport of about $4 million.

Table 2: Use of troponin tests on resource use in hospital for patients with non-acute and acute coronary syndromes

Diagnosis Outcome With troponin measurement Without troponin measurement
Nonacute coronary syndrome Stay (days) 1.2 1.6
(N=654) Charges (US$) 4,487 6,187
Acute coronary syndrome Stay (days) 3.7 4.6
(N=202) Charges (US$) 15,004 19,202
Red type indicates statistically significant reduction with troponin

For a diagnosis of myocardial infarction a positive test in serial troponin measurements had a positive likelihood ratio of 17 and a negative likelihood ratio of 0.04. The overall rate of myocardial infarction was 6%, so that the post-test probability of myocardial infarction with a positive troponin test alone was about 45%, while the post-test probability with a negative test was about 0.2%.

Algorithms and outcomes

So we know that elevated troponins are associated with bad things happening in people without ST-elevations, and we know that using troponins to evaluate people attending emergency departments with suspected heart attacks can save money. The cynics among us will then say that we need evidence about how the new tests can best be incorporated into work patterns in busy emergency departments. For instance, can the laboratory produce results within the time needed? If the results are not available to clinicians making decisions about whether to admit a patient, then why bother? A study from San Diego [3] claims that they can form part of a 90-minute triage using point of care testing.


This study tested a cardiac marker algorithm in 1,285 consecutive patients presenting to an emergency department with symptoms of cardiac ischaemia over nine months to April 1999. The pathway cannot be described in detail here, but involved electrocardiograms, clinical examination and judgement, and the use of troponin I, CK-MB and myoglobin tests done using a point of care testing machine at presentation and 30, 60 and 90 minutes later. All MI diagnoses were confirmed by a cardiologist using WHO criteria. The destination of patients (CCU, direct observation unit or DOU, ward or home) was compared with a six month period to March 1997.


The mean age was 64 years and 98% were men. Of the 1,285 presenting patients, 66 (5%) had a final diagnosis of myocardial infarction. All of these were diagnosed within 90 minutes by use of the cardiac marker algorithm and/or electrocardiograph changes.

Of the 508 patients discharged to home, 90% were discharged within 90 minutes and all within six hours. Thirty-day follow up of these patients showed that one (0.2%) was readmitted with myocardial infarction, and 7 (2%) were readmitted with unstable angina.

Of the 771 patients admitted to hospital, 32% were admitted to CCU, 35% to a direct observation unit, and 34% to a ward. Comparison of patient destination with a previous period showed a 40% reduction in the use of CCU beds (Figure 3).

Figure 3: Destination of patients after 90-minute triage


Bandolier 's ideal is to have the triple whammy of evidence of effectiveness, evidence about costs, and evidence about how to make a change. These three papers together come a long way to fulfilling those requirements for the use of troponin tests. They demonstrate that patients without ST-elevations but with positive troponin results have increased risk of death or myocardial infarction, that use of troponins has the capacity to reduce hospital resource use, and that people have figured out how to aggressively triage patients with suspected heart attack and come up with the right answer. Reducing CCU bed use by 40%, if replicated elsewhere, would appear a pretty stupendous result for hard-pressed hospitals with staffing problems.

Bandolier is conscious that in making a précis of these three papers much detail has been omitted. What is terrific is to have this quality and quantity of information about a laboratory test. It has been too easy in the past to criticise laboratory tests for inadequacy of evidence. This begins to put it right.

What is particularly interesting is that the three studies use different architectures to investigate their questions. One is a classic meta-analysis of clinical trial and cohort studies, but using the characteristic of good diagnostic studies that patients have to be consecutive to avoid bias. Another is a randomised trial to assess health economic outcomes. And the third tests out a diagnostic algorithm, again on consecutive patients.

Bandolier is rather impressed, especially because the diagnostics "industry" has lagged far behind therapeutics in the quality of the evidence it believes necessary to justify its existence.


1 PA Heidenreich et al. The prognostic value of troponin in patients with non-ST elevation acute coronary syndromes: a meta-analysis. Journal of the American College of Cardiology 2001 38: 478-485.
2 S Zarich et al. Impact of troponin T determinations on hospital resource utilization and costs in the evaluation of patients with suspected myocardial ischaemia. American Journal of Cardiology 2001 88: 732-736.
3 S Ming et al. Ninety-minute accelerated critical pathway for chest pain evaluation. American Journal of Cardiology 2001 88: 611-617.
previous or next story in this issue