High-sensitivity troponin tests for the early rule out of NSTEMI

4.3 There were concerns about the level of clinical evidence available for some of the tests. Most diagnostic accuracy results related to the Elecsys (30 studies) and ARCHITECT (9 studies) tests. Other tests had less evidence available but were mostly still acceptable. The committee noted that there was also limited evidence comparing the diagnostic performance of 1 test with another. This meant that although there may be differences in performance between the tests, it is difficult to estimate these differences with any certainty.

4.4 Results from the diagnostic accuracy studies showed that strategies that test a single sample on presentation using a threshold at or close to the limit of detection gave high sensitivity but low specificity. However, the committee commented that single sample strategies can be useful to rule out non-ST-segment elevation myocardial infarction (NSTEMI) early in emergency departments and, for this purpose, specificity is not a priority. There were concerns about the consistency of different analysers to provide accurate results at these low thresholds. However, clinical experts commented that samples with results close to these low thresholds would be from people considered very low risk and would have a good prognosis regardless of treatment. In contrast, results from diagnostic accuracy studies showed that a single test strategy using the 99th percentile threshold had sensitivity estimates too low to be safely used in clinical practice.

4.5 Results from the diagnostic accuracy studies showed that multiple sample early rule-out strategies, that is, those that included a second rule-out step, had better specificity than single-sample strategies using a very low threshold. Multiple sample strategies also maintained high levels of sensitivity. However, clinical experts commented that it was difficult to make direct comparisons of the different test strategies based on specificity because the number of true negatives would be affected by the prevalence of NSTEMI in each of the different study populations.

4.6 Clinical experts noted that the clinical context in which decisions on discharging or admitting someone to hospital is important, and that decisions should never be based solely on the results of a high-sensitivity troponin test. For example, a person with a negative high-sensitivity troponin test result should not be discharged without further investigations if they look visibly unwell or if the sample was collected too soon after the suspected cardiac event, a practice that could result in a false negative result.

4.7 Clinical experts explained that there was consistent evidence from reference range studies that the 99th percentile threshold differs between men and women. For the Elecsys and ARCHITECT tests, the 99th percentile upper reference limit for women is much lower than the general (mixed) population 99th percentile. The High-STEACS trial was the only strategy included in the cost-effectiveness modelling to use sex-specific 99th percentile thresholds in an early rule-out strategy. But it did not provide a direct comparison with a general population 99th percentile threshold. The committee therefore considered that the evidence on using sex-specific 99th percentile thresholds in early rule-out strategies was unclear. Clinical experts noted that sex-specific 99th percentile thresholds were sometimes used in clinical practice to help diagnose NSTEMI, but that there was no evidence that using them affected clinical outcomes. They noted that the ongoing CODE-MI study aims to evaluate the effect of using the sex-specific 99th percentile threshold for women for high-sensitivity cardiac troponin. This will be compared with the general (mixed) population 99th percentile threshold, for the diagnosis, treatment and outcomes of women presenting to the emergency department with cardiac chest pain. The committee noted that there was a wider equality issue because women with acute myocardial infarction are generally under-diagnosed and under-treated compared with men. The committee considered that using sex-specific 99th percentile thresholds to help diagnose NSTEMI could be a step towards reducing this health inequality. It concluded that, when NSTEMI is not ruled out using early rule-out test strategies, NICE's guideline on recent-onset chest pain of suspected cardiac origin may be used to help diagnose myocardial infarction, and the use of sex-specific thresholds at the 99th percentile should be considered (see section 5.2).

4.8 The committee considered evidence from 2 randomised controlled trials, High-STEACS and HiSTORIC. It noted that the authors of High-STEACS concluded that the implementation of an early rule-out strategy was not associated with any increase in myocardial infarction or cardiovascular death within 1 year of initial presentation. The HiSTORIC trial was submitted to NICE as academic-in-confidence evidence. The committee concluded that evidence from the randomised controlled trials, which reported end clinical outcomes, strongly supported using high-sensitivity troponin tests with early rule-out strategies in clinical practice.

4.9 The committee noted that the TriageTrue point-of-care test had a turnaround time of around 20 minutes and had the potential to be an important development in the field. The rapid time to test results could have benefits because of a reduced length of time spent waiting for a result in the emergency department (see section 4.1). Only 1 study was available on the TriageTrue test. In this study troponin levels were tested in stored plasma samples rather than the whole blood samples used in clinical practice. The committee noted that the evidence did not reflect how the test would be used in clinical practice at the point of care. It concluded that further evidence on TriageTrue's diagnostic performance when used on whole blood at the point of care is needed before the test can be recommended for use in clinical practice.

4.10 The Alinity and Dimension EXL tests were not included in the economic model because there were no direct diagnostic accuracy data for them in the systematic review. However, the committee noted that these tests were based on the same methods and principles, and used the same reagents as other tests that were included in the modelling. The committee heard that the Alinity test was a newer version of the ARCHITECT test and that the Dimension EXL test was a different version of the Vista test, but the tests were all run on different analysers. It noted further that the Elecsys STAT test was the same in terms of technical specification and performance as the Elecsys troponin T-high sensitive test, and is run on the same analysers. The committee concluded that the diagnostic accuracy of these different versions of the tests should be comparable. It also concluded that it was the responsibility of individual laboratories to assess the equivalency of these tests in practice, and to validate their diagnostic performance against their current system. This would be achieved in part by participating in external quality assessment schemes.

4.11 Only early rule-out strategies with a sensitivity of 97% or more were used in the cost-effectiveness modelling, based on expert opinion about the minimum sensitivity acceptable in clinical practice. The committee noted that this approach could mean that some potentially cost-effective strategies were excluded from the economic modelling. But overall it agreed that it was an acceptable approach that was necessary to keep the number of test strategies in the economic model manageable and ensure that those considered were likely to be safe in practice.

4.12 The secondary analysis incorporated prognostic benefits associated with false positive high-sensitivity troponin test results. Clinical experts noted that it is now widely accepted that people with a negative standard troponin test and a positive high-sensitivity troponin test (classified as false positives in the analysis) have an increased risk of reinfarction and mortality compared with people who have a negative result from both tests. The committee concluded that the secondary analysis was most appropriate for decision making.

4.13 Compared with standard troponin testing, high-sensitivity troponin test strategies resulted in incremental cost-effectiveness ratios (ICERs) of less than £7,000 per quality-adjusted life year (QALY) gained. These ICERs are below £20,000 per QALY gained, which NICE would typically consider to be cost effective. The committee noted that there were only small differences in costs and QALYs between the different test strategies included in the economic model, and recalled its previous conclusion that it was not possible to indirectly compare the tests and early rule-out strategies. The committee concluded that it was not possible to differentiate between the different test strategies, and that all early rule-out strategies had the potential to be recommended for clinical practice, provided that each test used with them had enough diagnostic accuracy data.

4.14 The model results were robust to changes in the input parameter values, for example, the number of admissions based on the specificity of the test strategy and the prevalence of NSTEMI. Early rule-out strategies with lower specificities would likely result in fewer people being discharged from hospital, but this does not have a substantial effect on the cost-effectiveness results. In addition, clinical experts noted that people having standard troponin testing and who get a negative test result would probably have a hospital length of stay of 24 hours rather than the 14 hours assumed in the economic model. So the benefits of using early rule-out strategies may have been underestimated. The external assessment group (EAG) commented that changing this assumption would be unlikely to affect the model results.

4.15 Some of the tests had lower levels of clinical evidence than others. But the diagnostic performance and costs of all the tests used in the modelling were comparable (except for the test cost for TriageTrue). There was no strong evidence to differentiate one test over another (see section 4.3), and when used in the early rule-out strategies all were cost effective compared with the standard troponin test (see section 4.13). The committee noted that recommending a range of tests would benefit hospitals by enabling them to operate within existing equipment contracts or investments in a particular platform. It concluded that recommending a range of high-sensitivity troponin tests gives greater flexibility to NHS trusts and enables them to work with any local restrictions or equipment contracts.

4.16 The original NICE diagnostics guidance on troponin tests for myocardial infarction recommended strategies that test at 0 hours and 3 hours, report absolute values and use an upper reference limit at the 99th percentile. The committee noted that much more evidence is now available on different early rule-out strategies. The committee considered the different strategies included in the model:

4.17 The TriageTrue point-of-care test, when used in the European Society of Cardiology 0/1‑hour pathway, was cost effective compared with a standard troponin test strategy, with an ICER of less than £5,000 per QALY gained. The committee considered that the potential benefits associated with the rapid turnaround times of point-of-care tests (see section 4.9) may not have been fully captured in the economic model. But the committee was concerned about the evidence used to provide the diagnostic accuracy inputs in the model, so concluded that further research was needed on the TriageTrue test before it could be recommended for routine clinical use.

4.18 Some evidence supported using sex-specific 99th percentile thresholds in clinical practice in men and women but overall their value in early rule-out is unclear (see section 4.7). The clinical experts said that there are much less data on the 99th percentile in other subgroups, such as older or younger people, people with or without renal disease and black, Asian and minority ethnic groups. The committee considered that it would be helpful to have a better understanding of differences in the 99th percentile between these subgroups.