Suspected cardiac chest pain: everyone sees it, everyone has a different clinical pathway, and everyone has a different risk score for it.
This week the Emergency Medicine Journal published our paper “Comparison of four decision aids for the early diagnosis of acute coronary syndromes in the emergency department” headed by Rick Body, directly comparing different risk scores for acute myocardial infarction.
The abstract is below, but also please go and read the full article1 and draw your own conclusions before changing your practice.
What did we do?
We examined the diagnostic accuracy of a new high-sensitivity troponin I (hs-TnI) assay when used with chest pain risk scores.
It’s a nested study within the larger trial, the Bedside Evaluation Sensitive Troponin study (BEST)2. In short BEST was a multicentre prospective observational study that examined the diagnostic accuracy of different troponin assays for acute myocardial infarction.
It included any emergency department patient with suspected cardiac chest pain, the usual exclusions of STEMI and pain greater than 12 hours in duration were applied. A bespoke chest pain proforma was filled in by treating clinicians which collected the data for different risk scores.
The outcome, acute myocardial infarction, was diagnosed by the serial troponin results run on the local assay and interpreted according to the third universal definition of acute myocardial infarction.
14 of the 18 centres had plasma samples available to run the Siemens ADVIA Centaur high-sensitivity troponin I assay and subsequently contributed to this nested study.
What did we find?
From the larger study of 1,487 patients we included 999 patients. The prevalence of AMI was 13.2% which is in keeping with previous work. Unfortunately, we couldn’t calculate all the risk scores for the 999 patients due to missing data. Most of the missing data was for the cardiac risk factors or for the subjective variables (e.g. typicality of the symptoms).
Four risk scores and one troponin strategy were compared; a troponin only strategy (limit of quantification <3ng/L), TMACS1,3, HEART4, TIMI5, and EDACS6.
With the HEART score, patients usually score 1 point if they have a troponin level above the 99th percentile and 2 points if their troponin is at least 3 times above that normal range. To improve its sensitivity, we modified this such that patients with hs-TnI <3ng/L would score 0 points; and patients with hs-TnI above the sex-specific 99th percentile would score 2 points. Those in between would score 1 point. During her recent presentation at EUSEM in Prague, this is how Barbra Backus suggested we might modify the HEART score in the era of high-sensitivity troponin testing.
|Risk Score||Patients included|
|Hs-TnI <3 ng/l||999|
A high sensitivity is achievable for a troponin only strategy (hs-TnI <3ng/L on arrival) and TMACS, both at 99.2%. The other scores (TIMI, EDACS and HEART) all had sensitivities below 99% for AMI.
However, the troponin-only strategy wasn’t as efficient as TMACS. Using the troponin-only strategy would have allowed 28.8% of patients to be reassured with one test, whilst TMACS would have ruled out 46.5% of the total after one test. Impressively EDACS could rule out 48.3% of patients, but this was at the detriment of sensitivity (96.2%).
|Sensitivity||Negative predictive value||Proportion ruled out|
|hsTnI <3ng/L (99.2)||TMACS (99.8)||EDACS (48.3)|
|TMACS (99.2)||hsTnI <3ng/L (99.7)||TMACS (46.5)|
|TIMI (97.5)||EDACS (99.3)||HEART (34.9)|
|EDACS (96.2)||TIMI (98.4)||hsTnI <3ng/L (28.8)|
|HEART (91.8)||HEART (97.0)||TIMI (19.4)|
Risk scores listed in order of results
Enough waffle: what does this mean?
This paper suggests that if you have this high sensitivity troponin I assay and you want:
- High sensitivity, highest negative predictive value and a high rule out rate – then TMACS is for you.
- Highest sensitivity and simplicity: a troponin-only strategy fits your mantra
- Highest rule-out at the cost of sensitivity; EDACS is your match, though TMACS was not so far behind
- This is one study (albeit multi-centre)
- This is only with one assay (Siemens ADVIA Centaur high sensitivity troponin I)
- I’m an author on the paper and involved in the wider implementation of TMACS
- This was a nested study and there is missing data
Global health perspective
It is worth noting that high-sensitivity troponin I is not widely used globally. Sadly there appears to be a good market in low- and middle-income setttings, for the troponin assays that are no longer considered useful in high-income settings. A small, retrospective study performed in such a low- and middle-income setting, revealed a large number of patients with a final diagnosis of unstable angina (almost unheard of within our setting), despite the use of a troponin test in the work-up. (Kabongo D, Kalla M, Allgaier R, et al. Describing suspected non ST-elevation acute coronary syndrome using troponin at a regional, public South African emergency centre with the Roche cardiac reader. SA Heart. 2018;15(2):102-7)
Then there is the issue of resource availability. A small survey conducted at the 2016 International Conference on Emergency Medicine in Cape Town suggests that the management of ACS can be hampered by more than the absence of troponin testing assays: lack of a pre-hospital service, 12-lead ECGs, as well as what we would consider basic treatment (aspirin, anticoagulants and PCI). (Beukes JG, Hendrikse C, Bruijns SR. Lack of Acute Care Resources to Diagnose and Treat Acute Coronary Syndrome in Lower-Income Settings. Global heart. 2018;13(1):35)
Given that international guidance on ACS management almost exclusively refers to practice in high-income settings, it is vital when practicing in low- and middle-income settings to be clear what the accuracy of the assay in use is – on second thought, that is probably good advice no matter where you practice.
For me this study supports strategies using T-MACS, hsTnI <3ng/l strategy or EDACS. It also reaffirms that troponin-only strategies can be enhanced by risk scores.
- 1.Body R, Morris N, Reynard C, Collinson PO. Comparison of four decision aids for the early diagnosis of acute coronary syndromes in the emergency department. Emerg Med J. November 2019:emermed-2019-208898. doi:10.1136/emermed-2019-208898
- 2.HRA U. BEST STUDY. Health Research Agency. https://www.hra.nhs.uk/planning-and-improving-research/application-summaries/research-summaries/the-best-study/. Published 2014. Accessed December 2019.
- 3.Body R, Almashali M, Morris N, et al. Diagnostic accuracy of the T-MACS decision aid with a contemporary point-of-care troponin assay. Heart. January 2019:768-774. doi:10.1136/heartjnl-2018-313825
- 4.Backus BE, Six AJ, Kelder JC, et al. Chest Pain in the Emergency Room. Critical Pathways in Cardiology: A Journal of Evidence-Based Medicine. September 2010:164-169. doi:10.1097/hpc.0b013e3181ec36d8
- 5.Antman EM, Cohen M, Bernink PJLM, et al. The TIMI Risk Score for Unstable Angina/Non–ST Elevation MI. JAMA. August 2000:835. doi:10.1001/jama.284.7.835
- 6.Than M, Flaws D, Sanders S, et al. Development and validation of the Emergency Department Assessment of Chest pain Score and 2 h accelerated diagnostic protocol. Emerg Med Australas. January 2014:34-44. doi:10.1111/1742-6723.12164
5 thoughts on “Risk scores for cardiac chest pain: the first head-to-head comparison!”
Pingback: Tid til Paradigmeskifte: OMI – Del 3: Mellemrisiko patienten (UAP / NSTEMI) og højsensitivitets troponiner – Akutmedicineren.dk
Useful blog information. Thank you for sharing this information.
Hey guys. Great work on this research. Question for you. Were clinicians able to access the Tn value prior to completing their assessments of HEART, TIMI, T-MACS and EDACS? Since there are subjective measures in these (concerning history? crescendo angina?) knowing the Tn before filling out the rest of the assessment could bias providers.
This may also be reflective of real world circumstances and may not actually change much in real world application but curious.
Hey thanks for the interest,
It is always great to get feedback on articles and publications!
The treating clinicians who completed the chest pain proformas did have access to the local troponin result. Pragmatically it would have been difficult to withhold the troponin result from them. You have alluded to, and I agree, that this design reflects the real world use of such risk scores.
Hi, I’m trying to find an online calculator for the T-macs using the Siemens Highly Sensitive Trop I but can only find it using Trop T.
Any chance you could sign post me where to look?
Thanks for the blog, it’s super helpful.