Pages Navigation Menu

Meet 'em, greet 'em, treat 'em and street 'em

SGEM#160: Oh Baby, You’re Too Sensitive – High Sensitivity Troponin

SGEM#160: Oh Baby, You’re Too Sensitive – High Sensitivity Troponin

Podcast Link: SGEM160

Date: September 8th, 2016

Reference: Body et al. The use of very low concentrations of high sensitivity troponin T to rule out acute myocardial infarction using a single blood test. AEM Sept 2016.

Guest Skeptic: Dr. Justin Morgenstern is an emergency physician and the director of simulation education at Markham Stouffville Hospital in Ontario. He is the author of the excellent #FOAMed blog called First10EM.com and is an associate editor of Emergency Medicine Cases.

This is a SGEM Hot Off the Press episode.  Here is a summary of the process but you can click on the SGEMHOP link for more details and previous episodes:

  1. A peer reviewed paper is selected pre-publication from AEM or CJEM that we think will be of interest to the SGEMers.
  2. We do a structured critical review of the paper using the quality check list developed by the Best Evidence in Emergency Medicine (BEEM) group.
  3. The paper is then discussed with one of the paper’s authors to give us a better understanding of the strengths and weaknesses of the paper.
  4. A blog and podcast are posted encouraging the FOAMed world to engage with us and the author over a one week period.
  5. A summary of the critical review and the best social media engagement is then published in AEM or CJEM to help cut that knowledge translation window down.

SGEMHOP from AEM Now has CME Credits


Dr. Corey Heitz

Dr. Corey Heitz

We have some exciting news to start Season# 5 of the SGEM. Hot off the Press papers from AEM will now have continuing medical education (CME) credits attached to the episodes. Dr. Corey Heitz will be writing up the five questions and alternating with Justin as the guest skeptic.

Corey is an associate professor of emergency medicine at the Virginia Tech Carilion School of Medicine in Roanoke Virginia. He has been a member of the executive board of the Clerkship Directors in Emergency Medicine, is co-chair of the National EM M4 Examination committee and sits on the examination development board for the NBME EM-Advanced Clinical Exam. In addition, Corey is the CME editor for Academic Emergency Medicine and the associate editor for emergency medicine simulation at the AAEM MedEdPORTAL.

SGEMers can claim their CME credits by following this process:

  1. Go to the Wiley Health Learning website
  2. Register and create a log in
  3. Search for “Academic Emergency Medicine – September”
  4. Complete the five questions and submit your answers

As this is a new feature there may be some glitches. Please email Corey (coreyheitzmd@gmail.com) with any questions or difficulties.

Case: A 53-year-old woman presents to your emergency department a little less than two hours after her anterior, squeezing chest pain hits its peak. She has a history of hypertension, but is otherwise healthy. Thanks to a medical directive, when you sit down with the patient, she already has a normal ECG and a high sensitivity troponin T (hs-cTnT) test that has come back below the limit of detection. She is feeling a lot better and would like to go home. 

Background: Patients presenting to the emergency department with chest pain are a daily conundrum. There are a number of life threatening conditions that must be efficiently ruled out.

Unfortunately, for the most common acute coronary syndrome, there is no universally agreed on algorithm. In addition to serial ECGs, a variety of different biomarkers are used and repeated at a variety of different intervals.

These biomarkers may or may not be followed by admission or further outpatient provocative testing depending on where you practice.

With the advent of high sensitivity troponin tests, it has been argued that there are probably some patients in whom we can exclude the diagnosis of myocardial infarction with just a single blood test.

The SGEM has done a couple of episodes on high sensitivity troponins in previous seasons:

  • SGEM #43 One is the Loneliest Number: High Sensitivity Troponin
  • SGEM #128 One Hour AMI Rule Out/Rule In (Harder, Better, Faster?)

A recent systematic review determined that a single hs-cTnT below the level of detection resulted in a sensitivity of 97.4% and a specificity of 42.4% (Zhelev 2015).

When a non-ischemic ECG is added to the single negative hs-cTnT, one retrospective study of more than 14,000 patients found a sensitivity for myocardial infarction of 98.3%, and a negative predictive value of 99.8% (Bandstein 2014).

However, many of the existing studies are small, single-center, and/or retrospective.


Clinical Question: Can high sensitivity troponin T (hs-cTnT) levels below the limit of detection on arrival in the emergency department be used to safely exclude acute myocardial infarction in patients with no ECG ischemia?


Reference: Body et al. The use of very low concentrations of high sensitivity troponin T to rule out acute myocardial infarction using a single blood test. AEM Sept 2016.

Dr. Richard Body

Dr. Richard Body

Dr. Rick Body is a Professor of Emergency Medicine who leads the Emergency Medicine & Intensive Care Research Group (EMERGING) in Central Manchester, UK. His PhD was in the derivation of clinical decision rules for suspected cardiac chest pain (2009) and he is currently Chief Investigator for the BEST study and the MACS (Manchester Acute Coronary Syndromes) trial. He developed the Manchester Acute Coronary Syndromes decision rule, the limit of detection rule out strategy with high sensitivity troponin and he was part of the group evaluating a 1-hour troponin-based rule out strategy.  Rick is also a Senior Editor at St. Emlyn’s blog and podcast site and an Associate Editor at the Emergency Medicine Journal.

  • Population: Adult patients presenting to the emergency department with new onset chest pain or symptoms suggestive of acute coronary syndrome that had peaked in the last 6 hours.
    • Exclusions: Acute trauma patients, prehospital thrombolytic or revascularization, prehospital defibrillation or cardioversion, dialysis patients, patients who had undergone cardiac surgery within one month of presentation, patients hospitalized for acute myocardial infarction in the previous three weeks and pregnant or breast-feeding women.
  • Intervention: hs-cTnT on arrival to the emergency department.
  • Comparison/Control: This is a diagnostic cohort study so a control/comparison is not strictly necessary. The aim is to report accuracy of a strategy of interest. There was no zero-hour test comparator here; there was, however, a reference standard to compare to.
  • Outcome: 
    • Primary: Acute myocardial infarction (AMI) at time of initial admission. AMI was defined using the 3rd Universal definition of patients who developed a rise and/or fall of troponin with at least one value above the 99th percentile of a healthy reference population in the appropriate clinical context. Thygesen et al 2012. To diagnose AMI a different troponin assay was used to avoid any potential criticism about incorporation bias, whereby the index test also forms part of the reference standard. The Siemens troponin I Ultra assay was used as a reference standard.
    • Secondary: Major adverse cardiac events (MACE). MACE was defined as AMI, death or re-hospitalization for acute coronary syndrome with coronary revascularization within/determined at 30 days.

Author’s Conclusions: In the absence of ECG ischemia, the detection of very low concentrations of hs-cTnT at admission seems to allow rapid, safe exclusion of AMI in one-third of patients without serial sampling. This could be used alongside careful clinical assessment to help reduce unnecessary hospital admissions.

checklistQuality Checklist for Observational Studies:

  1. Did the study address a clearly focused issue? Yes
  2. Did the authors use an appropriate method to answer their question? Unsure. A prospective observational study is reasonable, but this is a secondary analysis of data collected for a different primary purpose.
  3. Was the cohort recruited in an acceptable way? Yes
  4. Was the exposure accurately measured to minimize bias? Yes
  5. Was the outcome accurately measured to minimize bias? Unsure. They used an international definitions of AMI, but it still relies on the consensus diagnosis of two cardiologists.
  6. Have the authors identified all-important confounding factors? Unsure. These were not consecutive patients.
  7. Was the follow up of subjects complete enough? Yes
  8. How precise are the results? Good, but potentially borderline for clinical practice. The 95% confidence interval for the sensitivity of the primary strategy below the limit of detection (hs-cTnT <5ng/L) was 95.3-99.5%.
  9. Do you believe the results? Yes
  10. Can the results be applied to the local population? Unsure. The rule in rate for acute myocardial infarction on the initial visit in this study was 16.6%. While sensitivity is not impacted by prevalence, negative predictive value (NPV) is dependent on prevalence. So the NPV will depend on your local prevalence but not the sensitivity.
  11. Do the results of this study fit with other available evidence? Yes

Key Results: The analysis includes a total of 1282 patients. 213 (16.6%) were diagnosed with acute myocardial infarction on their initial visit. Mean age was 62 years and about two thirds being male (62.8%).

For the primary outcome of acute myocardial infarction, using the primary strategy of an initial hs-cTnT below the limit of detection (<5ng/L) and no ECG ischemia (no ST-segment deviation, T-wave inversion, left bundle branch block or a paced rhythm), the test characteristics are:

  • Primary Outcome: Acute Myocardial Infarction
    • Sensitivity 99.1% (95% CI 96.7-99.5%)
    • Specificity 43.9% (95% CI 40.9-46.9%)
    • PPV 26.0% (95% CI 23.0-29.2%)
    • NPV 99.6% (95% CI 98.5-100.0%)
    • LR+ 1.76 (95% CI 1.67-1.86)
    • LR – 0.02 (95% CI 0.01-0.09)
  • Secondary Outcome: MACE
    • Total 30 day MACE for primary strategy: 1.3% (95% CI 0.5- 2.8%).
    • The actual numbers were six MACE events, including only one death, no AMI and three revascularizations.

Screen Shot 2015-04-25 at 3.11.12 PM

We asked Rick five questions about his research. Listen to the podcast to hear his responses.

  1. Secondary Analysis: This was a secondary analysis of data collected for the international, multicenter TRAPID-AMI (High sensitivity cardiac troponin T assay for RAPID rule out of Acute Myocardial Infarction) study. Was this a pre-planned analysis and how do you think this impacts your results?
  2. Gold Standard: As always with these studies, there is a questionable gold standard. Acute myocardial infarction was based on two independent cardiologists using all available clinical data. They were blinded to the hs-cTnT results, as a different sensitive troponin I test was used clinically for all patients. There is some degree of subjectivity with this process. Can you comment on this imperfect gold-standard and how many times a third cardiologist was needed to adjudicate?
  3. Acceptable Miss Rate/Medicolegal Implications: You found a miss rate of 0.7% (4/560) with hs-cTnT being used alone and 0.4% (2/471) if used with no ECG ischemia (no ST-segment deviation, T-wave inversion, left bundle branch block or a paced rhythm).  In addition, the MACE rate at 30 days was 1.3% with most of it being revascularization. So what is an acceptable miss? In addition, your paper mentions medicolegal implications of using hs-cTnT. Can you expand on that point?
  4. ECGs: When do we ever evaluate a patient suspected of ACS without getting an ECG? Adding the ECG decreased the miss rate 0.7% to 0.4%. However, one investigator retrospectively evaluated the ECG and decided if it had any evidence of ischemia. Who was that one investigator and did you consider having more than one individual interpret the ECGs so inter-rater reliability could be determined?
  5. Clinical Judgement: The primary outcome in this study was acute myocardial infarction. We really liked that you emphasized in your paper that clinicians must still use their clinical judgment. Can you comment further on the role you think clinical judgement plays in these cases?

This was an industry sponsored study, which does not make the results wrong but should make us more skeptical of the paper and the interpretation. Rick discusses this issue and conflicts of interest.

Comment on authors conclusion compared to SGEM Conclusion: We generally agree with the authors’ conclusions.


SGEM Bottom Line: hs-cTnT is a new reality for some emergency physicians. We need to know how to use this test correctly to safely evaluate patients presenting with symptoms of ACS.


Case Resolution: The patient is offered serial troponins and ECGs but decides to go home knowing there is no “zero risk”.

Clinically Application: Use of hs-cTnT will all depend on your comfort level, patients’ preferences and your current practice environment.

Dr. Justin Morgenstern

Dr. Justin Morgenstern

What do I tell my patient? We cannot find any evidence of a heart attack. Your ECG is normal. Your blood work using a new highly sensitive test is also normal. However, this does not mean you did not have a heart attack. There is a very small risk (less than 1%) that it could be missed on the ECG and blood test. It also does not mean you do not have heart disease. It is possible you could have a major adverse event (including death) in the next 30 days. That possibility is also very low but not zero percent. With your test results, less than 1/500 people die in the next 30 days, but 1% of people have what we call a major event – mostly meaning they have to come back to the hospital to have a heart intervention like a stent or a surgery. Would you like to stay for a further testing in 1-3 hours or go home now?

Keener Contest: The winner from the last episode of Season#4 was Karen Adler a second year medical student from Israel.  She knew the HAL 9000 computer sang Daisy Bell in the movie 2001: A Space Odyssey.

Listen to the podcast for this weeks’ question. If you think you know the answer send it to TheSGEM@gmail.com with “keener” in the subject line. The first correct answer will receive a cool skeptical prize.


Remember to be skeptical of anything you learn, even if you heard it on the Skeptics’ Guide to Emergency Medicine.


References:

  • Bandstein N, Ljung R, Johansson M, Holzmann MJ. Undetectable high-sensitivity cardiac troponin T level in the emergency department and risk of myocardial infarction. Journal of the American College of Cardiology. 63(23):2569-78. 2014.
  • Zhelev Z, Hyde C, Youngman E. Diagnostic accuracy of single baseline measurement of Elecsys Troponin T high-sensitive assay for diagnosis of acute myocardial infarction in emergency department: systematic review and meta-analysis. BMJ (Clinical research ed.). 350:h15. 2015.
  • Mueller C, Giannitsis E, Christ M, et al. Multicenter evaluation of a 0h/1h-algorithm in the diagnosis of myocardial infarction using high-sensitivity cardiac troponin T. Annals of Emergency Medicine 2016;Online first.
  • Thygesen K, Alpert JS, Jaffe AS, et al. Third Universal Definition of Myocardial Infarction. Journal of the American College of Cardiology 2012;60(16):1581–98.

Conference Update:

  • Justin Morgenstern

    A comment was left on https://first10em.com by someone who goes by the handle “evinjan” (update, this is Jan Hansel ‏@VirtueOfNothing):
    Would love to see a test/treatment threshold analysis for hs-TnT and coronary angiography, like Jeff Kline did for pulmonary embolism when deriving PERC. As stated, there is no such thing as 0% risk. The “random off-the-street” risk needs to be considered. And we can’t cath everyone presenting for chest pain, can we?

    • Justin Morgenstern

      Personally – I love the concept of test thresholds and agree we should use them as a tool to discuss the harms and benefits of all the tests that we use. However, distilling the harms and benefits down to a single number is very difficult. For example: how do you weigh the harm of a false positive troponin and carrying a label of disease for the rest of your life against the potential benefit of catching coronary artery disease before a “major adverse cardiac event” occurs? These types types of harms and benefits are qualitatively different, and therefore difficult to distill down into a single quantitative number.

  • Pingback: SGEM#160: Oh Baby, You’re Too Sensitive – High Sensitivity Troponin | Online Medical Assistant Training Programs()

  • Pingback: One high sensitivity troponin test to rule out acute myocardial infarction - St.Emlyn's()

  • John Cronin

    Can I just say this critical review is so well done. Interview and all. Thanks alot
    John

  • Andrew Worster

    As Body et al. demonstrate, the 99th percentile is likely not the optimum cutoff for ruling out MI if such a single cutoff exists. Troponin concentration measurements become less precise than at the 99th percentile as they approach the LoD and the ranges reported are not likely reproducible in many hospital laboratories. Furthermore, this pathway would need to be tested prospectively in the community settings in which it is to be applied using patient important outcomes. To that end, while the direction of this research is important given the burden of chest pain complaints on ED flow, ruling out AMI does not rule out ACS or any other important chest pain diagnosis.

    • Rick Body

      Thanks, Andrew! I guess what you’re alluding to (and perhaps what Dan McCollum was saying too) is that we need to know how this strategy works when implemented in practice. I agree that this will be important, and Edd Carlton has taken an interest in this. Hopefully we’ll see more on this from Edd in the future!

  • John Cronin

    I’m not sure that this paper is going to change my practice though. We have a chest pain evaluation unit (CPEU) adjacent to our ED, run by us and an excellent cardiology ANP with buy-in from the cardiology team. When we put patients in there I usually already know they haven’t had an MI. We use hsTrop, although don’t usually get it done as quickly as suggested in this study; and many of our patients present with a longer history of pain than this study population.
    Having the CPEU allows for a period of observation, a review by the cardiology team to assess need for other investigations such as stress test (another can of worms, I know), angio or need for follow-up in their clinic. It is a safety net I guess. Traditionally serial trop+ECGs is performed on the patients where indicated.
    I guess this paper may reduce the need for the serial troponins, but I don’t think it will hugely alter my use of the CPEU for patients.
    Great work though.

  • Simon Carley

    Great work as always.

    A few things to think about and a little nerdiness.

    It was fascinating to hear the international differences in positive results between the US and the centres in this study. The hit rate of 1% quoted for the US is really surprising and will of course affect how this and any other test is used. It clearly means that troponin tests are being used in an ultra low risk population and that will affect the utility.

    You are right to say that statistically the sensitivity and specificity of tests are independent of prevalance, and mathematically that’s true, but in the real world it usually isn’t. Why you might ask? It’s because prevalance is also associated with disease severity and we know that many tests perform better in patients with more obvious/severe disease than in more general populations. This is why you must always evaluate tests in the population relevant to the practice in which you are going to use it. Clearly in an ultra low risk population testing like this may well perform differently and the burden of false positives will rise.

    So I was really surprised at the idea of using troponin testing in such a low risk population.

    We also should stop and reflect on what we really mean by a false negative in this study. The detection of AMI and or MACE is a really good way to look at the data but they are not exactly the same as harm. It is entirely plausible that patients may come back to hospital and require an intervention (e.g. stents) without coming to any significant harm or different outcome as a result of the first testing episode giving a negative result.

    The section on shared decision making is very interesting and a good example of how we can explain the imperfections of testing to our patients.

    Keep up the good work and stay skeptical.

    vb

    S

    • Corey Heitz

      Thanks for the comments, Simon! That 1% number that I used isn’t data driven, it’s my own estimate from personal experience. Like many centers in the US, we have triage protocols in which a troponin is ordered extremely early in the visit, prior to any physician, PA, or NP evaluating the patient. The vast majority of these patients do not “need” a troponin, and I try – sometimes failingly – to each my residents how to think about it compared to what they see done in real life.

      I joke that the only contraindication to getting a troponin level in a US ED is not having a chest. (On a related note, I was interrupted the other day during an HPI on a gentleman with urinary retention – so the tech could perform an EKG.)

  • Daniel McCollum

    Fantastic discussion of a really interesting article. How might this change my practice once the US finally gets widely available high sensitivity troponins?

    I think the days of binary yes/no decision making for the assessment of chest pain are over. The authors of this study do an admirable job of not just lumping everyone together, but dividing patients based on presence of ECG changes and age. I love that they included data on the 4 patients with missed AMI with hs-cTnT < LOD. Old people aren't to be trusted, and I would be very hesitant in using any single troponin rule out in anyone that is elderly.

    Bayesian thinking will be required to properly use these high sensitivity assays. Patients that are young, have a normal ECG, pain for at least a couple hours, and a history of present illness that are not very concerning can likely be ruled out with a single hs-cTnT with acceptably low risk rates. I personally would prefer at least a 2 hour delta troponin for anyone that doesn't fit into the above group.

    Dr. Body does an excellent job discussing how ruling out AMI does not mean that it is wise to discharge the patient. For better or worse, the EM provider is on the hook not just for the current visit but what happens in the coming weeks. Some patients will go on to have bad outcomes in the coming weeks even if they are not having AMI. It is not clear that admission necessarily prevents these bad outcomes, but some of them may (or may not) derive benefit from admission. The excellent discussion of shared decision making highlights the need to not apply our acceptance of risk to the patient without discussion.

    Was the study perfect? Of course not, as no study is. I think it dodges a lot of the problems with retrospective studies, but I would love to see prospective work on this topic in the future. I don't particularly like the composite outcome, as I am less in favor of revascularization as being combined.

    The main problem I have is that the ECG interpretation was by a single provider. While it may have taken dozens of extra hours to get more interpreters, inter-rater reliability of ECG interpretation would be very helpful. It is clear from this study that the ECG interpretation is needed to make any single troponin assay appropriate to rule out AMI, so this is a noticeable (if understandable) limitation of this study.

    It is admirable that Dr. Body rejected money for this study. I can only imagine the amount of time that he spent on this study, and it shows a great deal of character. Having the pleasure of meeting him and Ken at the SMACC EBM workshop, I can't emphasize enough how great of a teacher he is.

    -Dan McCollum

    • Rick Body

      Dan, many thanks for your kind comments! I read your comments with great interest. I always like to know how other people interpret the study with an external perception and your observations are very considered.

      Of note, although it’s a secondary analysis it is actually still a prospective study: the participants were selected for inclusion (and data collected for the study) before outcomes were known. We published another prospective validation of this strategy at the link below (although it was also a secondary analysis) – it’s open access…

      http://clinchem.aaccjnls.org/content/61/7/983.long

      Rick

      • Daniel McCollum

        Great point, Rick. Since you were blinded to outcomes while selecting reasons for inclusion, this makes your results much stronger. I was unaware of the other article that you linked to, and I greatly appreciate you sharing it.

        It seems likely that the future of rapid rule outs in the ED will be very similar to the strategy discussed here being applied to those with a low enough pre-test probability. This seems far superior to me than provocative testing or coronary CTA for rapid rule out in the ED. Thanks again for all the work that you and your colleagues put into this!

  • Salim R. Rezaie

    Fantastic post, podcast, and discussion on this. Physicians beware…High sensitivity troponin is coming to the US at an ER near you. Question & Bottom Line attached. :

    Salim

  • Andrew Worster

    Rick, I confess, the more I study troponin, the less I understand about it. I really think that we need to move from test performance charateristic studies to pragmatic clinical trials for patients with chest pain of suspected ischemic origin in community settings. I also think that you are the one of the best researchers to lead such studies.

    • Justin Morgenstern

      I would just like the second this point. So much of our diagnostic research stops at test characteristics. The randomized, controlled trial really needs to become a bigger part of diagnostics research. (I am just not sure who is going to fund such studies.)
      In the mean time, keep up the good work, Rick!

    • Rick Body

      Thanks, Andrew! I absolutely love your quote: “The more I study troponin, the less I understand about it”.

      Unless you object, you might see it in big letters behind a lectern some time soon! 🙂

      • Andrew Worster

        Be my guest.

  • SAEM EBM IG

    So glad to see SGEM-HOP back again – and congrats to Ken Milne, Justin Morgenstern, and Corey Heitz for this fabulous episode. Take it from me – Captain Cranium is all hype and SGEM-HOP will thrive without him!!!

    Rick – thanks for the insightful and thought-provoking author interview. Some of these ideas and details are difficult to convey in a written manuscript, hence the value of #SoMe and the opportunity to put a study under the microscope. I really enjoyed the nuanced conversation at the end about Conflicts of Interest. The question is how to convey these nuances in a written manuscript where there are so many reasons to be skeptical of industry-sponsored research (see http://epmonthly.com/article/secrets-to-healthy-skepticism/)?

    The comments from “SGEM Nation” thus far are also informative. My biggest recommendation for future hs-Trop researchers concurs with Andrew Worster: clinicians, medical educators, and policy-makers now need randomized controlled trials in diverse settings (U.S. very low risk population, UK higher risk population) in order to understand how hs-Trop will (or will not) impact admission rates, ED flow, and costs. Accurate tests that do not consistently improve outcomes when tested in RCTs exist (example BNP, see http://pmid.us/22123173 with Andrew Worster and Peter Rosen). During the 2015 Academic Emergency Medicine Consensus Conference (entire issue available open access at http://onlinelibrary.wiley.com/doi/10.1111/acem.2015.22.issue-12/issuetoc), we described a more patient-centric approach for imaging that includes a hierarchy of studies culminating in RCTs (http://onlinelibrary.wiley.com/doi/10.1111/acem.12832/full). The hidden dangers of relying solely upon diagnostic accuracy studies (sensitivity/specificity, likelihood ratios) including the slippery slope of overtesting, overdiagnosis, and overtreatment as discussed with Daniel Fatovich on this month’s issue of Emergency Medical Abstracts and summarized in http://onlinelibrary.wiley.com/doi/10.1111/acem.12832/full. Criteria do exist that define when studies of sensitivity/specificity alone are insufficient to change practice and when RCTs are indicated (http://pmid.us/16754927). I hope that both industry and non-industry funders will support these RCTs before introducing hs-Trop into routine practice.

    Three smaller observations responding to others’ comments on this blog. First, Shared Decision Making will play an increasingly important role in these scenarios to avoid the slippery slope of over-treatment, so stay tuned for the December 2016 issue of Academic EM that will be open access and contains all of the output from the May 2016 Consensus Conference on “Shared Decision Making” (when available, you’ll be able to find it here http://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1553-2712/homepage/consensus_conference_proceedings.htm). Second, when discussing diagnostic accuracy research, I’m hopeful that SGEM commentators will assess for, discuss, and provide estimates of observed skew (i.e. what direction is observed sensitivity and specificity likely deviated, if certain forms of bias are present). Michael Kohn and I summarized these biases in http://onlinelibrary.wiley.com/doi/10.1111/acem.12255/abstract. Third, I always find test/treatment threshold discussions fascinating and agree that they are based on so much uncertainty that these thresholds are not meant to be definitive endpoints but rather opportunities to understand our understandings of potential benefits/harms and their impact when contemplating whether or not to obtain a test on an individual patient. Acute coronary syndrome is a hugely important topic for emergency medicine, so I’m hopeful that Rick, Andrew, and/or others will tackle ACS and hs-Troponin in the near future for the AEM Evidence Based Diagnostics series!!!

    Cheers everyone! Welcome back SGEM-HOP!

    Chris Carpenter
    Washington University in St. Louis

    • Rick Body

      Hi Chris,

      Thanks so much for your comments – really informed and insightful! The SGEM has done an amazing job (as usual) in making these conversations happen.

      I totally agree with your points about diagnostic accuracy research, although I think we’re soon going to learn that the RCT itself has limitations in this field. Sometimes the trial can be as much about the mode of delivery of the intervention as it is about the intervention itself. For example, you could have a great decision rule that falls flat because it wasn’t made sufficiently accessible to clinicians in a busy ED.

      We have a pilot RCT that will hopefully appear in the literature soon. That trial has positive findings but I still learned some important lessons. The Clinical Trials Unit suggested that we stratify randomization by the risk group (using the decision rule being investigated – MACS). That meant that we needed to use their system to enter data.

      However, their research governance required that everyone had a password. To get a password, you need both Good Clinical Practice training and study specific training, which is impossible to achieve in a busy ED with around 50 doctors and high staff turnover.

      There are many less extreme examples but some may have led good diagnostic technology to appear useless: e.g. the ESCAPE trial of Chest Pain Unit care from Steve Goodacre, and the recent HEART score implementation trial, both of which were large (with a cluster design) but had negative findings. The HEART score implementation trial from Simon Mahler, Martin Than’s two RCTs and Martin Moeckel’s BIC-8 RCT involving copeptin (all of which were great examples) have been more successful in demonstrating a difference in early discharge rates – but they were smaller trials and therefore not really fully powered for safety outcomes. The Hi-STEACS trial from Edinburgh will be interesting: it’s a large cluster RCT. Could it be the first of its kind to show a difference? Or will the answer lie in evaluating implementation in a less controlled way?

      I think the conversations we’re having will soon need to move on to how we design the most appropriate RCT to evaluate diagnostic technology. It won’t be without its challenges!

      PS, I’d love to work with Andrew and contribute to the AEM Evidence Based Diagnostics series. It’s a big job (with big returns) though – we’d probably need someone young and keen to drive it!

      Rick

  • Eddy Lang

    I think it’s worth asking ourselves what the most pressing problem is for CP evaluation in the ED. From my vantage point reviewing all the reported complaints and adverse outcomes related to CP for 300K visits/yr x 3 years I would venture that over-investigation and prolonged LOS for CP patients due to repeat assays is of greater concern than the exceptionally rare miss. In our current environment work like this will help us improve the efficiency of CP assessment in the ED. Nice perspective on minimal industry role by Rick Body.

    • Rick Body

      Hi Eddy, Thanks for your comments! I totally agree with you. As physicians we’re naturally concerned with the ‘miss rate’ as missed diagnoses come back to bite us as well as potentially harming patients. However, we have to be realistic: there are risks to being in hospital, especially if we admit so many patients that our hospitals and EDs become crowded. That’s why I think shared decision making has so much potential: patients can, in effect, help us to define what’s an ‘acceptable risk’ for them. If I was a patient with a normal ECG and initial troponin, I know exactly what I’d want: to go home and come back for a second troponin on an outpatient basis. Others might want admission; others might want no further tests – but informing patients and taking account of their preferences surely has to be the way forward.

      Rick.

  • Kirsty Challen

    Another great episode thanks. Glad to hear some good sense regarding our need to minimise, not eliminate, risk.
    #paperinapic is here.

  • Kirsty Challen

    As was pointed out on Twitter, unstable angina is still ACS, so a retitled #paperinapic.

  • TheSGem

    This first episode of Season#5 of the SGEM was picked as some of the best global #FOAMed this week by Life in the Fast Lane! http://lifeinthefastlane.com/litfl-review-249/

  • TheSGem

    This #SGEMHOP episode makes the list from Life in the Fast Lane as some of the very best global #FOAMed this week. http://lifeinthefastlane.com/litfl-review-249/

  • Axel Ellrodt

    Hello,
    I posted this elsewhere, hope this isn’t distasteful.

    1- One first question : couldn’t we actually lower the MACE incidence by prescribing low dose aspirin to a subset of the patient we discharge ? This is what I do. I don’t remember seing this detailed in this and similar studies. I’m quite certain many do that and it may confound the results. Or did I miss something ?

    2- Did you ever hear about a study comparing the incidence of major adverse cardiac events (MACE) within 30 days after a patient is discharged from an ED where (s)he attended for chest pain, vs paired controls not attending for any symptom ?

    Anyway, supposing a post ED visit low MACE such as 1.6% I doubt the man (oops: person) in the street comes close to that, but what do I know ?

    I always wonder about the risk the random man in the street has to “develop” a MACE.

    Any idea ?

  • Pingback: SGEMGlobal: German: #160 – Du bist so sensitiv – hochsensitives Troponin | The SGEM Global()