Abstract
Background:
We tested the hypothesis that a single high sensitivity troponin at limits of detection (LOD HSTnT) (<5 ng/l) combined with a presentation non-ischaemic electrocardiogram is superior to low-risk Global Registry of Acute Coronary Events (GRACE) (<75), Thrombolysis in Myocardial Infarction (TIMI) (≤1) and History, ECG, Age, Risk factors and Troponin (HEART) score (≤3) as an aid to early, safe discharge for suspected acute coronary syndrome.
Methods:
In a prospective cohort study, risk scores were computed in consecutive patients with suspected acute coronary syndrome presenting to the Emergency Room of a large English hospital. Adjudication of myocardial infarction, as per third universal definition, involved a two-physician, blinded, independent review of all biomarker positive chest pain re-presentations to any national hospital. The primary and secondary outcome was a composite of type 1 myocardial infarction, unplanned coronary revascularisation and all cause death (MACE) at six weeks and one year.
Results:
Of 3054 consecutive presentations with chest pain 1642 had suspected acute coronary syndrome (52% male, median age 59 years, 14% diabetic, 20% previous myocardial infarction). Median time from chest pain to presentation was 9.7 h. Re-presentations occurred in eight hospitals with 100% follow-up achieved. Two hundred and eleven (12.9%) and 279 (17%) were adjudicated to suffer MACE at six weeks and one year respectively. Only HEART ≤3 (negative predictive value MACE 99.4%, sensitivity 97.6%, %discharge 53.4) and LOD HSTnT strategy (negative predictive value MACE 99.8%, sensitivity 99.5%, %discharge 36.9) achieved pre-specified negative predictive value of >99% for MACE at six weeks. For type 1 myocardial infarction alone the negative predictive values at six weeks and one year were identical, for both HEART ≤3 and LOD HSTnT at 99.8% and 99.5% respectively.
Conclusion:
HEART ≤3 or LOD HSTnT strategy rules out short and medium term myocardial infarction with ≥99.5% certainty, and short-term MACE with >99% certainty, allowing for early discharge of 53.4% and 36.9% respectively of suspected acute coronary syndrome. Adoption of either strategy has the potential to greatly reduce Emergency Room pressures and minimise follow-up investigations. Very early presenters (<3 h), due to limited numbers, are excluded from these conclusions.
Introduction
Acute chest pain is one of the commonest presentations to the Emergency Room (ER) in the Western world, the vast majority of which are not due to acute myocardial infarction (MI).1–4 However, early differentiation of type 1 MI from non-cardiac causes of troponin elevation remains a challenge despite the emergence of high sensitivity troponins. 5 Early, safe discharge for a substantial proportion of chest pain presentations continues to be the single most important challenge, particularly for health care providers. Discharge at presentation with troponin6,7 at the limits of detection (LOD) or early after a 1-h high sensitivity troponin strategy remains an attractive proposition,8,9 though there remain concerns about the sensitivity for rule-out of MI in the 1-h algorithm, particularly for early presenters.9,10 It is not certain if these high sensitivity troponin values, when combined with clinical assessment, electrocardiogram (ECG) findings and risk factors improve early rule-out. A number of risk scores which incorporate some or all of these factors have been in existence or been developed 11 to address this need but to date a high sensitivity troponin strategy either alone or combined with risk scores as a tool for early discharge has not been tested prospectively in an unselected, consecutive chest pain population in whom there is suspicion of an underlying ischaemic cause.
We compared rule-out strategies, namely a single high sensitivity troponin T (HSTnT) (at limits of detection and at 99th percentile) combined with non-ischaemic ECG versus low score History, ECG, Age, Risk factors and Troponin (HEART) (≤3), Thrombolysis in Myocardial Infarction (TIMI) (≤1) and Global Registry of Acute Coronary Events (GRACE) (<75) in an unselected consecutive chest pain patient population with suspected acute coronary syndrome (ACS). Our goal was to define the optimum rule-out strategy for suspected ACS using a single HSTnT and presentation electrocardiogram.
Methods
This study was conducted in concordance with STARD criteria for diagnostic studies (score sheet included in Supplementary Material online). From June 2011 to November 2011 all patients presenting to University Hospital Aintree ER with a predominant symptom of chest pain, who had both an ECG and at least one blood sample for HSTnT check (suspected ACS population), were prospectively ‘recruited’ into this study. Aintree University Hospital is a large hospital providing secondary care cardiology with annual attendances of 80,000 per annum to Accident and Emergency. Prior to the start of the study there was a three-month ‘roll-in’ phase with conversations between Accident and Emergency clinicians and the development of a chest pain proforma (acute chest pain assessment sheet) that highlights important aspects of history taking in chest pain as well as allowing capture of relevant information important in computation of the risk scores noted. Use of this proforma was not mandatory but was incorporated with casualty sheets as an aide memoire for clinicians for assessment of chest pain. Research personnel prospectively and contemporaneously entered demographic, epidemiological data and results of blood tests together with timing of ECGs and troponins in a dedicated database. All ECGs were stored in an electronic register for subsequent analysis by the research team. Each ECG was read by an experienced clinician/researcher, with a second reading if there was troponin elevation to adjudicate if type 1 MI definition had been met. The presence and absence of specific criteria were noted electronically in a standardised fashion. The definition of a non-ischaemic ECG was the following: sinus rhythm or atrial fibrillation or atrial flutter with ventricular rate <110 together and absence of the following: left bundle branch block (LBBB), paced rhythm, ST segment elevation, ST segment depression, T wave inversion or T wave flattening or biphasic T waves in two contiguous leads. For both HSTnT LOD and non-ischaemic ECG strategy and the three risk scores the first (presentation) ECG alone was used as the ECG input variable. Subsequent ECGs were extracted for adjudication of MI and to determine possibilities of MI in those with a single sampled HSTnT.
TIMI and GRACE scores were determined from online calculators by research staff with the use of the chest pain proforma, casualty sheets and ambulance proforma alone. To score positively for ST segment depression in GRACE and TIMI: planar depression of at least 0.5 mm after the J-point consistent with the Minnesota criteria. 12
There was no specific troponin testing protocol for clinicians during recruitment. Guidelines recommended repeat troponin sampling for patients who presented <6 h from chest pain particularly if presentation HSTnT was <99th percentile (<15 ng/l). Discharge at <6 h from chest pain with ‘negative’ troponin (HSTnT <15 ng/l) was discouraged.
Calculation of HEART score, follow-up, definition and adjudication of major adverse cardiac events (and non-ischaemic diagnoses)
For calculation of HEART score, follow-up, definition and adjudication of major adverse cardiac events (MACEs) (and non-ischaemic diagnoses) see Appendix 1.
HSTnT assay
The assay used was the HSTnT (Roche Elecsys). This has been previously evaluated extensively and found to fulfil the definition of high sensitivity with 10% coefficient variation at the 99th percentile. 16 Performance of this assay across a number of centres has previously been evaluated. 16 Analysis was undertaken in-house of COBAS e601 analysers using a standard 18-min assay. Quality control of assay (in-house) at Liverpool Clinical laboratories revealed a coefficient of variation of 11% at low value troponins (<10 ng/l). External quality assurance provided by the independent United Kingdom National External Quality Assurance scheme (NEQA) revealed inter-hospital coefficient of variation of 10% (standard deviation 0.5) low level troponin values. Assay performance (NEQA) was undertaken monthly throughout the duration of the study.
During recruitment of the study there was a downward shift in the Roche Elecsys HSTnT that was a result of calibration of specific lots. 17 We recomputed HSTnT values by adjusting for the shift by reference to local laboratory calibration and that determined by a number of groups, including the manufacturer.17,18 The adjusted or recomputed values were used in the analysis.
Endpoints
The primary endpoint was type 1 MI, unplanned coronary revascularisation and all cause death (MACE) at 42 days (see Appendix 1 for definitions and adjudication). Unplanned coronary revascularisation was defined as admission for unstable angina or MI necessitating same day or same admission coronary revascularisation by percutaneous or surgical means. Secondary endpoint included MACE at one year.
Ethics
This manuscript conforms to the ICMJE Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals.
This was an all-comers chest pain study with national (English) hospital involvement for any representation with possible MI. The project was registered with the hospital research department and the North-west England Regional Ethics Board, which granted full consent to undertake this study. To allow for complete follow-up special permission was granted, in the absence of individual consent, via a confidential advisory group (UK Government Home Office appointed) for the recruitment of consecutive chest pain population and collection of data from any hospital nationwide to facilitate the retrieval of clinical records and blood results for patients with possible ACS (15/CAG/0171) (http://www.hra.nhs.uk/).
Statistics
All analyses were performed in Stata v.14.
Summary statistics were presented as n (%) if categorical and as median (interquartile range (IQR)) if continuous. For categorical variables, a Pearson χ2 test was used to determine the p-value, except for cases where the number of MACEs in any category was ≤5, when Fisher’s Exact Test was used. For continuous variables, a t-test was performed to determine p.
Confidence intervals for positive predictive value and negative predictive value (NPV) were calculated using exact binomial proportions.
To evaluate differences in NPV in patients who had their first HSTnT at ≤3 h and ≤6 h compared with those >3 h and >6 h, p-values were computed using a two-sided Fisher’s Exact Test.
Receiver operator characteristics (ROC) curves were plotted, and area under curve (AUC) computed, using the rocreg function in Stata to bootstrap and obtain standard errors and confidence intervals. The p-value for differences in the AUC was obtained using a Wald test.
It was not possible to compute an exact sample size at the time of recruitment as we had no information on events rates for HSTnT LOD at the time of starting the study or indeed good evidence of event rates for risk scores with high sensitivity troponins.
Results
Figure 1 charts the global chest pain population and those with suspected ACS (defined as sampling of HSTnT together with presentation ECG). Missing data (loss of casualty cards and incomplete/illegible chest pain proforma) was <1.2% of the global population. Follow-up was 100% with tracking of repeat presentations to ER/admissions to eight national hospitals (Table S1 in Supplementary Material online). Retrieval of all necessary reports, blood results, ECGs and relevant imaging reports for all patients that had biomarker positive presentations was achieved for the purposes of adjudication of MI.
Flowchart of final population.
Table 1 describes the characteristics of all patients (both those admitted and those directly discharged from accident and emergency department). It further subdivides the population into those adjudicated to have suffered type 1 MI and MACE at six weeks. Four hundred and forty-four patients (27%) were discharged directly from Accident and Emergency at a median of 363 min (IQR range 271, 468 min). Four of these patients suffered MACE by six weeks (0.9%) (three were adjudicated to fulfil definition of type 1 MI at index admission (i.e. incorrectly discharged)). The decision to admit or discharge was at the discretion of clinicians (at the time no specific protocol or risk score was in place to guide clinicians). Those adjudicated to have suffered a MACE (type 1 MI, all cause death, unplanned coronary revascularisation) were older with greater risk factors. Risk scores were consistently higher for all patients with type 1 MI or MACE at six weeks. There were 10.7% of patients adjudicated to have suffered a type 1 MI at index admission with another 0.3% between admission and six weeks at a median of 11 days following admission. The percentages of patients undergoing unplanned percutaneous coronary intervention (PCI) and unplanned coronary artery bypass graft (CABG) up to six weeks were 2.7% and 0.8% respectively at a median of seven and 18 days following admission. The number of patients undergoing emergency primary PCI for ST elevation myocardial infarction up to one year who were discharged with a diagnosis other than ACS was four (0.2%) at a median of 80 days following discharge. The number of deaths at six weeks and one year were 25 (1.5%) and 104 (6.3%) respectively, occurring at a median of 15 and 112 days respectively. Of the total deaths at one year 11% were in hospital and 89% out of hospital. The overall type 1 MI and MACE event rates at six weeks were 11.2% and 12.9% respectively. For one year, the equivalent figures for type 1 MI and MACE were 11.6% and 17%. The entire population diagnoses other than type 1 MI were the following: type 2 MI 1.2%, unstable angina 8.5%, atypical chest pain 8.5%, cardiac non-coronary chest pain 7.9%, non-cardiac symptoms 57.2%, chest pain unknown origin 2.8% (see Appendix 1 for prespecified criteria).
Population characteristics.
That is, not known smoking status in 8%.
Not known in 32%.
Timing of chest pain uncertain in 1%.
Ischaemic ECG was defined as the following: rhythm other than sinus rhythm or AF/flutter with heart rate >110 beats/min, left bundle branch block, paced rhythm, ST segment elevation, ST segment depression, T wave inversion or T wave flattening or biphasic T waves in two contiguous leads.
MI; myocardial infarction; MACE: major adverse coronary event; IQR: interquartile range; TIMI: Thrombolysis in Myocardial Infarction; HEART: History, ECG, Age, Risk factors and Troponin; GRACE: Global Registry of Acute Coronary Events; CAD; coronary artery disease; PCI: percutaneous coronary intervention; CABG: coronary artery bypass graft; AF: atrial fibrillation; Hb: haemoglobin; MCV: mean corpuscular volume; fl: femtolitres; BP: blood pressure; CP: chest pain.
Table 2 demonstrates performance of the rule-out protocols at six weeks and one year for MACE and type 1 MI. Only HEART ≤3 and LOD HSTnT (with non-ischaemic ECG) achieved the prespecified NPV of >99% at six weeks for MACE (NPV with prespecified 99% performance target is represented graphically in online Supplementary Material Figure S1). The discharge potential for these strategies were 53.4% and 36.9% for HEART≤3 and LOD HSTnT (with non-ischaemic ECG) respectively. They achieved sensitivities for MACE of 97.6% and 99.5% for HEART ≤3 and LOD HSTnT (with non-ischaemic ECG) respectively. For type 1 MI alone as an endpoint the respective figures for sensitivity were 98.9% (HEART ≤3) and 99.4% (LOD HSTnT and non-ischaemic ECG). The respective NPV at six weeks for type 1 MI for LOD HSTnT and HEART ≤3 was identical at 99.8%. At one year point estimates of NPV (99.5%) for type 1 MI was again identical for HEART ≤3 and HSTnT LOD. In terms of overlap between LOD HSTnT with non-ischaemic ECG and HEART ≤3, 521/606 (86.0%) of patients fulfilling discharge rules with LOD HSTnT + non-ischaemic ECG also had a HEART score of ≤3. The equivalent percentage for the reverse was 59.4%, that is 59.4% with a HEART score ≤3 had a LOD HSTnT.
Performance of rule-out protocols for type 1 myocardial infarction and MACE at six weeks and one year.
MACE: major adverse cardiac event; PPV: positive predictive value; NPV: negative predictive value; CI: confidence interval; TIMI: Thrombolysis in Myocardial Infarction; GRACE: Global Registry of Acute Coronary Events; HEART: History, ECG, Age, Risk factors and Troponin; HSTnT: high sensitivity troponin T; ECG: electrocardiogram
In terms of NPV for MACEs, LOD HSTnT and non-ischaemic ECG was clearly differentiated from HSTnT ≤14 ng/l (99th percentile value) (with non-ischaemic ECG) and TIMI ≤1 but not GRACE <75 or HEART ≤3. GRACE <75, in terms of discharge potential, could deliver only 28.6% discharges (similar to actual direct discharge of 27% in this cohort).
Table S2 in the Supplementary Material details patients with HEART ≤3 and LOD HSTnT and non-ischaemic ECG with MACEs (the false negative population or those fulfilling rules for discharge but who had an adjudicated MACE). All three deaths ≤6 weeks from index event, which represented three of the five false negative events for patients with HEART ≤3, were clear non-cardiac deaths. The performance of HEART ≤3 was considerably improved by not counting non-cardiac deaths as events. Sensitivity improved to 99% with NPV to 99.8% for MACE at six weeks. For patients with a first HSTnT LOD (<5 ng/l) there were two patients with a second troponin >14 ng/l, both of which were adjudicated as suffering type 1 MI (only one of which had a non-ischaemic presentation ECG).
Figure 2 describes ROC curves for continuous values of risk scores and with HSTnT being analysed without ECG interpretation as a continuous variable. Compared with HSTnT (ROC area 0.918) there was no significant difference in performance with HEART (ROC area 0.910, p <0.382) but there was significant separation of curves with TIMI (ROC area 0.855, p <0.001) and GRACE (ROC area 0.791, p <0.001). Figure S2 in the Supplementary Material describes time of chest pain to first HSTnT samples.
Receiver operator characteristics curve for primary outcome (MACE) at six weeks.
Fifty-seven per cent of patients had 1 HSTnT sampled; 51.9% of patients had >1 ECG (median time to first and second ECG 10 min (IQR 5, 15 min) and 181 min (IQR 69, 414 min)). To determine the possibility of missed MI due to single HSTnT <14 ng/l we re-evaluated paired ECGs to determine whether there were dynamic ECG changes (a significant change such as new T wave inversion, ST segment shift, development of LBBB). In those with a single HSTnT only three patients had a dynamic change in ECG but in only one was a type 1 MI adjudicated (second ECG demonstrated ST segment elevation myocardial infarction). His initial ECG, though, was not normal, thus excluding a false negative result for HEART ≤3 and LOD HSTnT (presentation HSTnT <5 ng/l).
Sensitivity analysis
A sensitivity analysis was undertaken with GRACE values of <60 and <90 and TIMI score 0 and ≤2 for MACE at 42 days.
GRACE score <60 did achieve NPV of 99.6% with a sensitivity of 99.5% but percentage discharge dropped to 14.3%. GRACE score <90 did not achieve prespecified thresholds of NPV (NPV 97.2%) with a sensitivity of 90.5%. TIMI score 0 did achieve threshold NPV of 99.7% with a sensitivity of 99.1% and allowed discharge of 39.1% patients. TIMI score ≤2 allowed discharge of 74.1% but at a considerable loss of sensitivity and NPV of 73% and 95.3% respectively.
Inter-observer variability of determination or risk scores
Absolute agreement for risk scores was good for HEART and TIMI and excellent (100%) per risk category (К=1). No patients were categorised as high or intermediate risk that were initially deemed low risk (and vice versa) for HEART and TIMI scores. GRACE scores in terms of absolute agreement performed less well but were still very good for risk category agreement (К=0.82) (Table S3 in Supplementary Material).
Discussion
We confirm the impressive rule-out strategy of HSTnT using LOD (or limit of blank alone <3 ng/l)6,7,19 combined with a non-ischaemic ECG. We have prospectively validated the two cornerstones in acute chest pain assessment (namely troponins and ECG) as a simple and effective strategy for decision making with very low risk of adverse outcome. Rubini Gimenez et al. identified the potential of rule-out value of undetectable HSTnT and high-sensitivity troponin I (HSTnI). 6 Mokhtari et al. further demonstrated incremental value of the combination of HSTnT LOD, non-ischaemic ECG and low-risk history but in a non-consecutive patient series. 20 As far as we are aware this is the only study that has prospectively assessed HSTnT at LOD and non-ischaemic ECG) in an unselected, consecutive chest patient population. Our point estimate of NPV is similar to those of Bandstein et al. 7 and Body et al. 19 even though a lower limit of blank (LOB) (<3 ng/l) was used in the latter study. Sandoval et al. define the concept of LOD and LOB as rule-out strategies. 21 The LOD provide higher NPV and sensitivities than that of Rubini Gimenez et al. 6 Incorporation of a non-ischaemic ECG to the limits of detection of HSTnT could explain this; LOD HSTnT together with non-ischaemic ECG increased sensitivity from 98.1% to 99.5% for MACE and 98.9% to 99.4% for type 1 MI at 6 weeks (Table 2).
This study is the first to demonstrate that HEART score ≤3 in the era of high sensitivity troponins (using conventional cut-offs of HSTnT) also, to a large extent, meets criteria for a rule-out strategy. HEART ≤3 was as effective as LOD HSTnT (with a non-ischaemic ECG) for rule-out of type 1 MI at six weeks with NPV of 99.8% with a sensitivity of 98.9% allowing a discharge of 53.4% of patients (as opposed to 36.9% of patients with LOD HSTnT strategy). The performance of HEART ≤3 for type 1 MI is indicative of an acceptable risk for a discharge/rule-out for type 1 MI. 22 MACE sensitivity was <98% but improved to 99% (at six weeks if only cardiac deaths were counted).
In addition this study has demonstrated that both rule-out strategies (HEART ≤3 and LOD HSTnT) have low event rates to one year for type 1 MI (99.5% at one year), indicating that patients can be safely discharged and reassured without the need for early diagnostic testing in the absence of ongoing symptoms. This study suggests that cardiac imaging for LOD HSTnT or low-risk HEART score cannot seek to refine risk further and argues for a reduction of these requests in low-risk patients.
TIMI score ≤1 and GRACE score <75 did not reach prespecified thresholds for rule-out. Although neither of these risk scores had been originally validated for suspected ACS, they have been evaluated with encouraging results in combination with high sensitivity troponins in a number of different protocols in chest pain populations.23–25 Sensitivity analysis revealed that TIMI score of 0 had acceptable performance but discharged considerably less than HEART ≤3 (39.5% vs. 53.4%).
A consecutive, unselected series of suspected ACS presentation, in a sample size this large, is important in terms of generalisability and therefore potential influence to change practice (as opposed to a selected population, a natural consequence of consent and limited or absent recruitment nocturnally or at weekends). There was 100% follow-up tracked nationally and there was adjudication of any index chest pain or representation with HSTnT that was above the 99th percentile. This allowed us not only to validate those coded for type 1 MI but to evaluate all patients with elevated troponins, including those who were deemed to have non-cardiac pathologies (a distinction from other noted studies). 7 Second, we used all-cause death as an outcome rather than cardiac death, as incorporated in some studies. 3 This is not to say that high sensitivity troponin can predict non-cardiac death but to accept the uncertainty of cause of death, particularly for out of hospital deaths and the probability of misdiagnosis and therefore miscoding of cause of death. 26 Using all-cause death grants a safer estimate of rule-out protocols.
Recently Carlton et al. directly compared risk scores using HSTnT and HSTnI in a select patient population of patients ≤80 years of age. 27 In a post-hoc analysis HEART ≤3 did not achieve pre-specified targets of >99.5% NPV and/or sensitivity of >98% with either HSTnT or HSTnI. Differences in study population could explain this; Carlton et al. had a population with much shorter time of chest pain to presentation, which is likely to have reduced both NPV and sensitivity for HEART ≤3.
The largest study to date in a population of all-comers similar to ours is that by Shah et al. 3 The impressive NPV of 99.6% (HSTnI – Abbot) with both a derivation and validation cohort fulfilled criteria for safety and clinical utility (almost two-thirds could be discharged). However, there was no reported sensitivity as far as our group could discern. NPV is affected by disease prevalence rates thus highlighting the attractiveness of risk scores such as HEART, which to some extent incorporate this aspect, allowing potential widespread applicability of rule-out strategies for suspected ACS. Multicentre validation studies, preferably with studies enriched with early presenters, and with competitor high sensitivity troponin assays incorporated in HEART are important to confirm (or refute) the potential of the HEART score as a tool for rule-out of ACS.
A previous study by Backus et al. investigating the same risk scores in 2338 suspected ACS in a multicentre observational study revealed superiority of HEART compared with TIMI and GRACE across all risk categories, but there was a 1.7% MACE at six weeks with HEART ≤3 with fourth generation (non-high sensitivity) troponins. This event rate would not be acceptable strategy as a rule-out in most health care systems. 3 High sensitivity troponins are likely to have improved sensitivity or rule-out value at the expense of reduced specificity.
Limitations
Figure S2 in the Supplementary Material demonstrates time from chest pain to first HsTnT check. Only 4.4% of patients had a HSTnT check <3 h (very early presenters) from chest pain. The equivalent value for ≤6 h was 240 (14.6%). There was no significant difference in NPV between those with ≤6 h and the global population when incorporated in a troponin and ECG strategy or as low-risk scores (Supplementary Material Table S4). However, these results should be treated with caution due to the low number of events occurring in patients presenting early. Most chest pain presentations to the ER are late, and the study does represent comparable numbers of patients with very early (≤3 h) 3 and early (≤ 6h) 7 presentations to similar ‘all-comer’ studies. This study does not validate discharge for very early presenters, based on a single HSTnT.
Forty-three per cent of patients underwent serial troponins. As a rise or fall in troponins or biomarkers of myocyte necrosis is part of the universal definition of MI we cannot be certain that MIs were not missed, although most presentations are late (median time from chest pain to presentation: 9.6 h), thus reducing the possibility of a later rise of HSTnT beyond the 99th percentile. Furthermore, analysis of paired ECGs in those with a single HSTnT sampled for sequential ECG changes only revealed one definite MI (evolving ECG changes from T wave inversion to ST segment elevation).
There was a downward shift seen with this Roche assay between 2010 and 2012. 28 The actual difference computed between affected lots and correct (re-assayed) values was 1.2 ng/l in a similar population of suspected ACS. 18 Wildi et al. suggested a small miss rate with LOD HSTnT in affected lots compared with corrected values with serial samples and re-adjudication for MI. However, two other analyses suggest similar prevalence rates of MI and no ‘missed’ MI with LOD initial samples when ECG ischaemia was taken into account. 29 Body et al., using LOD, revealed two missed MIs using paired samples in 463 patients with affected lots with retesting, but both patients had ischaemic ECGs, maintaining NPV at 100%. In addition Roche internally assessed the effect of affected lots with readjustment mathematically in 1033 patients presenting to the ER. 28 The identical number of MIs was defined.
Finally, HEART score was not determined by attending physicians but by researchers following presentation raising concerns about scoring the chest pain section of HEART. However, due to both engagement and the production of a tick box chest pain proforma that emphasises character and duration of chest pain it was always possible to score chest pain with reasonable reflection of clinician suspicion.
Conclusion
Heart ≤3 and LOD HSTnT (combined with a non-ischaemic ECG) are optimum rule-out strategies for MACE at six weeks, for a suspected ACS population. Further multicentre studies should test HEART ≤3 as a tool for early, safe discharge of patients with chest pain and suspected ACS, particularly in early presenters. A randomised trial of LOD HSTnT versus a one hour high sensitivity troponin protocol decision rule to assess duration of stay and safety endpoints is urgently needed.
Footnotes
Appendix 1: HEART score,follow-up,definition and adjudication of MACEs and Non-MI diagnoses
Acknowledgements
PGC, FF and AK contributed equally to this manuscript. The funder played no role in the design, execution nor had they any part in the writing up of the study. Aintree Innovation Awards (Dragons Den) undertakes to award via competitive bids innovative works that have the potential to improve health care delivery. The trust board and executive do not oversee research or innovative work. Registration: clinical trials.gov NCT02581540.
Author contribution
AK and JJ conceived the idea of the project, developed conversations with accident and emergency clinicians, designed the database and the chest pain proforma. PC, FF, LM, AR, TO, ST, JJ, AK entered data and computed risk scores from definition tables and relevant charts/proforma. AK, MF, HZ, KA, BP, BK, VV were consultant adjudicators for type 1 and type 2 myocardial infarction. RG undertook analysis and derived FU data from the national linked database for hospital admissions per defined ICD 10 codes. AK, FF, PC, MF wrote the manuscript. AK is the guarantor of the data.
Conflict of interest
AK has the following potential conflicts of interest: free of charge material from ROCHE for research. Submissions pending for research funds from ROCHE and Abbott. Funding for national conferences from Daiichi Sankyo, Bayer. Funding for research from Bayer pharmaceuticals. Speaker and expert consultation fees from Astra Zeneca, Menarini, Bayer. No other authors report potential conflict of interest.
Funding
This work was supported by Aintree Innovation Award (Ref:SG/RH/JW), Northwest Educational Cardiac Group.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
