6.1 The Diagnostics Advisory Committee reviewed the evidence available on the clinical and cost effectiveness of the Elecsys Troponin T high‑sensitive, ARCHITECT STAT High Sensitive Troponin‑I and AccuTnI+3 troponin I assays for the early rule out or diagnosis of acute myocardial infarction (MI) in people presenting to emergency departments with a suspected acute coronary syndrome. The Committee noted that the cost‑effectiveness analysis included both a base‑case analysis, in which it was assumed that people with a false‑positive high‑sensitivity troponin test result had no treatment and have a risk of mortality and re‑infarction equivalent to people with true negative results, and a secondary analysis, in which it was assumed that a proportion of people with a false‑positive high‑sensitivity troponin test result have an increased risk of mortality and adverse events and are treated accordingly.
6.2 The Committee considered the diagnostic accuracy data for each of the high‑sensitivity troponin assays. It noted that 34 publications reported data for the Elecsys Troponin T high‑sensitive assay, 5 reported data for the ARCHITECT STAT High Sensitive Troponin‑I assay and 3 reported data for a test described as a prototype Access high‑sensitivity troponin I assay, which was included because the reported assay characteristics met the definition of high‑sensitivity specified in the scope for the assessment and the prototype assay was developed by Beckman Coulter. The Committee also considered data submitted to the FDA relating to the AccuTnI+3 assay which were provided in confidence by the manufacturer. The Committee considered that the 99th percentile value reported for the prototype assay (9 nanograms/litre) and the commercially available AccuTnI+3 assay (40 nanograms/litre) were substantially different and heard from both clinical specialists and the manufacturer that the prototype assay had not been commercialised. The Committee concluded that the data on the prototype assay could not be considered applicable to the AccuTnI+3 assay. Furthermore, the Committee considered that the data available for the AccuTnI+3 assay itself could not be considered sufficient to determine its diagnostic accuracy.
6.3 The Committee questioned the assumption that serial troponin testing over 10–12 hours with standard troponin assays has perfect accuracy: that is, 100% sensitivity and 100% specificity. The Committee heard from clinical specialists that there is likely to be uncertainty around the accuracy of the high‑sensitivity troponin assays in the clinical‑effectiveness analysis. The Committee heard from clinical specialists that it is plausible that the high‑sensitivity troponin assays may pick up additional people (classified as having false‑positive results in the analysis) who need treatment compared with standard troponin assays, and consequently the accuracy of the high‑sensitivity troponin assays may have been underestimated. The Committee concluded that there is uncertainty around the accuracy of the high‑sensitivity troponin assays in the clinical‑effectiveness analysis, but that it is likely that the Elecsys Troponin T high‑sensitive and ARCHITECT STAT High Sensitive Troponin‑I assay are sufficiently sensitive to allow the early rule out of acute MI in practice.
6.4 The Committee questioned the applicability of the 99th percentile cut‑offs derived from healthy reference populations to populations seen in emergency departments in UK practice. The Committee heard from clinical specialists that there is a wide variability in the composition of reference populations included in validation studies, and that this can impact upon the interpretation of the 99th percentile in routine practice. The Committee discussed the alternative of using receiver‑operator characteristic curve‑derived cut‑offs for the high‑sensitivity troponin assays, but noted that the clinical utility of adopting such strategies is unknown and that the evidence base for this is rapidly developing. The Committee concluded that the instructions for use for each of the assays should be consulted for guidance on the assay specific 99th percentile cut‑off values, and calibrating the high‑sensitivity troponin assays in clinical practice.
6.5 The Committee considered the risk of a false‑positive result in people who have a positive result with a high‑sensitivity troponin assay. The Committee heard from a clinical specialist that treatment for acute MI is not without risk, and heard from a patient expert that incorrect diagnosis of acute MI can have a negative and long‑lasting impact on various aspects of a person's quality of life, including ability to get insurance cover. The Committee heard from clinical specialists that raised troponin levels can be caused by conditions such as heart failure, chronic obstructive pulmonary disease, renal failure and skeletal muscle disease, in which there is no evidence of underlying ischaemic heart disease, and that this would be differentiated from acute coronary syndromes during subsequent clinical investigations. The Committee was advised by clinical specialists that this group of people were likely to have underlying health problems that would need hospital admission regardless of a diagnosis of non‑ST‑segment elevation myocardial infarction (NSTEMI) being confirmed. However, the Committee noted that caution should be advised in interpreting the results of high‑sensitivity troponin tests when they are used outside the suspected acute coronary syndrome clinical setting.
6.6 The Committee considered the impact of a positive troponin test result on clinical decision making and questioned whether it could rule‑in a NSTEMI diagnosis. The Committee noted that there was limited evidence on the use of high‑sensitivity troponin assays to rule‑in NSTEMI within 4 hours of arrival at an emergency department. The Committee concluded that there was insufficient evidence to support the routine use of the high‑sensitivity troponin assays for early rule‑in of NSTEMI and that people in whom NSTEMI could not be ruled‑out within 4 hours should receive further clinical assessment to determine the cause of the suspected acute coronary syndrome.
6.7 The Committee noted that each of the modelled high‑sensitivity troponin test strategies could result in people having a false‑negative result, and being discharged incorrectly. In the modelling the proportion of people incorrectly discharged varied from 1.4% to 3.4% with the single test strategies, and from 0.3% to 1.4% with the sequential test strategies. The Committee heard from clinical specialists that current practice may result in around 2% of people with acute MI being incorrectly discharged and that people who have experienced a very small acute MI might not have elevated troponin levels. On this basis, the Committee considered that the introduction of 'early rule‑out high‑sensitivity troponin test protocols' was unlikely to result in an increased rate of incorrect discharges.
6.8 The Committee considered the applicability of studies included the External Assessment Group's analyses that were done in a mixed population; that is, they included people with any acute MI (both STEMI and NSTEMI). The Committee noted that data restricted to a NSTEMI population were only available for the Elecsys Troponin T high‑sensitive assay, but noted that similar summary estimates were obtained in both the mixed population and the NSTEMI only population analyses, for the 99th percentile presentation test strategies (sensitivity 89% and 88% respectively, specificity 82% and 84%). The Committee concluded that the use of mixed population data for the ARCHITECT STAT High Sensitive Troponin‑I and AccuTnI+3 analyses was unlikely to have a significant impact on results. The Committee also heard from a clinical specialist that the mixed population data, derived from studies where people are most likely to have had a NSTEMI, could be considered applicable to current practice.
6.9 The Committee deliberated about the use and clinical applicability of the 99th percentile presentation and optimal test strategies that were modelled by the External Assessment Group. The Committee noted that the 99th percentile presentation tests strategies were cost effective in both the base‑case and the secondary analysis but questioned the clinical effectiveness of the strategies. The Committee concluded that, on balance, the sensitivity of the modelled Elecsys Troponin T high‑sensitive and ARCHITECT STAT High Sensitive Troponin‑I 99th percentile presentation strategies (88% and 80%) was insufficient to recommend single test strategies for early rule out. The Committee further considered that the sequential test strategies, which included a sample taken at presentation with a repeat sample at 3 hours, showed merit as early rule‑out test strategies, and agreed that the inclusion of a second sample improved the sensitivity of the strategies sufficiently to recommend the use of two‑step early rule‑out test strategies in clinical practice. The Committee concluded that although the results of the External Assessment Group's modelling supported both the clinical and cost effectiveness of two‑step early rule‑out test strategies, there was insufficient evidence to recommend a specific test strategy and agreed that 'early rule‑out protocols' should be chosen according to local preference.
6.10 The Committee considered that the implementation of two‑step early rule‑out strategies would allow the use of delta values to improve the specificity through consideration of the change in high‑sensitivity troponin levels at each step. However, the Committee acknowledged that the use of absolute delta values in the context of early rule‑out was not widely understood and had not been considered in the economic modelling. The Committee concluded that the decision to include delta values in 'early rule‑out protocols' should be made at a local level in conjunction with clinical and laboratory specialists.
6.11 The Committee discussed the use of the limit of blank and limit of detection in the high‑sensitivity troponin optimal test strategies that were modelled by the External Assessment Group. The Committee heard from both clinical specialists and manufacturers that the precision of an assay is generally lower at the lower end of the assay's measuring range, and that consequently cut‑offs employing the limit of blank or limit of detection may be less reproducible in practice. The Committee considered that because of lower precision and reproducibility, the quality assurance of assays using the limit of blank and limit of detection could be problematic, and would need the development of additional quality‑control materials for use in external quality assurance schemes. The Committee also heard from clinical specialists that the limit of blank and limit of detection may become increasingly lower as assays develop, which would have an impact on the use of these thresholds to rule out acute MI. The Committee concluded that, on balance, although the limit of blank and limit of detection showed promise as cut‑off thresholds in early rule‑out test strategies, the clinical and practical implications of introducing limit of blank and limit of detection cut‑off thresholds into practice are currently uncertain.
6.12 The Committee considered the cost‑effectiveness analyses carried out by the External Assessment Group. The Committee noted that, based on the level of clinical evidence available, the Elecsys Troponin T high‑sensitive assay and ARCHITECT STAT High Sensitive Troponin‑I assay were cost effective in both the base‑case and the secondary analysis, but the cost‑effectiveness analysis for the AccuTnI+3 assay was not robust. After consideration of the assumptions applied in each analysis, the Committee concluded that the secondary analysis was the most robust analysis because it is likely that high‑sensitivity troponin testing will detect additional people who would benefit from treatment in practice. The Committee also noted that the true effect of clinical judgement on the interpretation of the high‑sensitivity troponin test results was not fully captured in the model. It heard from clinical specialists that, in practice, a clinical assessment of a person's prior probability of acute MI (that is, whether they are at a low or high risk) is likely to influence the interpretation of high‑sensitivity troponin test results and determine the diagnosis and management recommendations. The Committee concluded that, in routine practice, clinical judgement is likely to reduce the impact of both false‑positive and false‑negative results.
6.13 The Committee considered the subgroup analyses and concluded that there is insufficient evidence to make specific recommendations on the use of the high‑sensitivity troponin tests in any subgroups. The Committee heard from clinical specialists that there is a developing body of evidence that suggests the 99th percentile cut‑off values of high‑sensitivity troponin assays may vary between men and women, which clinicians may need to take into account when interpreting the results of high‑sensitivity troponin tests. The Committee also considered the External Assessment Group's subgroup analyses of acute MI prevalence and noted that when the acute MI prevalence was reduced to 1%, in both the base‑case and the secondary analysis, the no‑testing strategy had an incremental cost‑effectiveness ratio (ICER) of £96,456 saved per QALY lost compared with standard troponin testing. The Committee heard from clinical specialists that inappropriate use of troponin tests in people with acute chest pain and a very low pre‑test probability of acute MI is a significant problem in emergency departments. The Committee concluded that clinicians should assess pre‑test probability of acute MI to determine whether troponin testing is clinically appropriate.
6.14 The Committee considered the logistical barriers to achieving early rule out of acute MI and discharge of patients in routine clinical practice. The Committee considered that the External Assessment Group's base‑case and secondary analysis allowed for a 3‑hour delay between presentation and test results being available. The Committee heard from clinical specialists that the implementation of 'early rule‑out protocols' would need consideration of laboratory turn‑around times, for reporting the results of high‑sensitivity troponin tests. An early rule‑out protocol would also depend on the availability of clinicians within an emergency department to agree patient management options. The Committee concluded that the implementation of protocols that allow earlier rule out of acute MI and consequent earlier discharges should be encouraged.