Before we get into the literature on the test itself, I think we need to think about the patients that we are sending for stress tests. We know that our work-ups for chest pain are imperfect. We know that, even with normal ECGs and negative troponins, we will miss patients with acute coronary syndromes. That is why we order stress tests: to try to catch those patients; to move from a low risk situation to the fictitious no risk situation. But what exactly is the risk for patients with negative chest pain work ups in the emergency department?
Classically, we have been taught that our miss rate is about 2%. However, the study that is always cited for that 2% number paints a very different picture:
Pope JH, Aufderheide TP, Ruthazer R, et al. Missed diagnoses of acute cardiac ischemia in the emergency department. N Engl J Med. 2000;342:(16)1163-70. PMID: 10770981
This is a prospective, multicentre, observational trial that looked at 10,689 patients presenting to the emergency department with symptoms suggestive of myocardial ischemia. 17% of this population ruled in for MI (which is pretty high in comparison to current North American practice). Of the 10,689 patients, 19 (0.17%) were diagnosed with an MI after being discharged from the ED. I will repeat that: there were only 19 misses in over 10,000 patients! 22 more patients were discharged home and ultimately diagnosed with unstable angina (0.20%). In total, this would be a miss rate of 0.38%, but that includes the very subjective, and therefore questionable, diagnosis of unstable angina. How does a fraction of a percent transform in the widely quoted 2%? They use the patients whose final diagnosis is MI as the denominator, rather than the patients we actually see: undifferentiated patients with chest pain. We will miss 2% of patients with MI, but we don’t know who those patients are. The risk to the patient in front of you is tiny (0.2-0.4%). This study is also occasionally cited to indicate that mortality is higher in the group of patients you send home, but that is simply wrong. The mortality was identical whether your MI was diagnosed in the ED or missed (10.5% versus 9.7%, RR 1.1 95% CI 0.3-4.1). Considering that these patients were not treated for their MI, they must be lower risk patients to start with.
Bottom line: The rate of missed MI (or ACS) in undifferentiated chest pain patients VERY low (0.2 – 0.38%). In the context of a discussion about stress testing, it is unlikely that you are going to be able to make that number better, even with an incredibly good test.
Obviously, the miss rate is dependant on a number of variables. It will be higher in a patient population with a higher incidence of disease. Conversely, it will be lower in lower risk populations (and most modern North American chest pain studies have a rule in rate lower than 17%). It will also depend on the rule out strategy used, the type of troponin, and potentially the individual practitioner. There are a huge number of studies that cover this issue, so I can’t cover them all, but let’s do a quick survey to get a general sense of the risk among chest pain patients who we would consider for stress testing (those patients with negative ED work-ups).
A logical place to start is with the studies looking at stress testing after negative emergency department chest pain work-ups. (We will return to these same studies in part 3 to consider the accuracy of stress testing.) Gomez (1996) looked at 74 patients who had stress tests after a negative ED workup and there were 0 MIs and 0 deaths at 30 day follow up. (There are no deaths in any of these studies, so I won’t write it out each time.) Amsterdam (2002) looked at 1000 patients, of whom they had full follow-up data on 862, and there was 1 MI (0.1%) at 30 days. Meyer (2006) looked at 903 patients (871 with follow-up) and there were 2 MIs (0.2%) at 6 months. Manini (2010) looked at 191 patients (171 with follow-up) and there were 2 MIs (1.2%) at 30 day follow-up. Scheuermeyer (2012) looked at 266 patients (254 with follow-up) and there were 0 MIs at 30 day follow-up. In total, this is 2247 patients prospectively studied after negative ED work-ups, and there were a total of 5 MIs between 30 days and 6 months. The result is a 0.2% miss rate (with a range of 0-1.2%), which is very consistent with the Pope paper. (Pope 2000)
That being said, in some places the yield is much much lower. For example, in a prospective cohort that looked at 3543 patients admitted to a chest pain observation unit, there were 0 deaths and 0 MIs during the follow-up period. In other words, these patients had a 0% risk, and yet they were still admitted to hospital! (Napoli 2014)
Another group of studies that can give us insight into our current miss rate is the literature on CT coronary angiography. In general, these studies take patients with symptoms suspicious for ACS, but who have initially negative ECGs and troponins, and use CT angiography to risk stratify them. (These patients will be higher risk than standard chest pain patients, because they usually don’t wait for a second troponin.) Litt (2012) looked at 1370 patients, 908 of who received CCTA and 462 who received standard care. There were no deaths. 10 patients out of the 908 in the CCTA group (1%) had an MI at 30 days. 5 patients out of 462 in the standard care group (1%) had an MI at 30 days. They don’t give us details about these MIs, but based on other literature it is reasonable to guess that a number of these would have been caught with a second troponin during the initial ED visit. Hoffman (2012) looked at 1000 patients randomly assigned to CCTA or standard care. There were 5 total MIs on 28 days follow up (0.5%) plus another 3 diagnoses of unstable angina (0.3%). deFilippi (2001) ran a RCT of CCTA versus stress testing. Of the 248 patients with nonischemic ECGs and negative troponins (more than 10 hours from symptom onset), there was a single MI (0.4%) after a full year of follow up. In all of these studies, we can expect that a reasonable number of these “missed MIs” would have been caught with a repeat troponin during the first ED visit. Therefore, once again, this data illustrates that after a complete emergency department workup, the chance of a missed MI is less than 1%.
It is also worth taking a quick look at the data from the HEART score studies. I can’t do justice to all the data on the HEART score in this post. Salim Rezaie has a great summary here. However, a slightly closer look is required to understand the difference between MACE (major adverse cardiac events) and real misses. In general, patients with a HEART score less than or equal to 3 have a 30 day MACE rate of about 1-2%. (Backus 2010; Backus 2013) That sounds higher than the other studies quoted here, but MACE is a composite outcome that combines truly important things like death and MI with inconsequential things like revascularization. (See part 2 for further discussion of these outcomes and part 4 for a more lengthy discussion about why revascularization is an irrelevant outcome in these patients.) The original HEART score publications don’t report the individual components of the MACE composite for the low risk patients, but Dr. Backus was kind enough to share her data from the 2013 validation study. There were 870 patients in the low risk (HEART score 0-3) group, of whom 15 had a MACE (1.7%). However, only 9 of those 15 patients had an MI, which would make the miss rate for important outcomes 1.0%. (There was one death, but it was a suicide, and the rest of the MACE were revascularization procedures in patients without MI). However, I think the real world miss rate would be even lower than 1%, because of the 9 MIs, 6 of them occurred on day 0, presumably meaning they were caught on the repeat troponin (which is not part of the heart score). So the average emergency physician performing 2 troponins and considering the HEART score would have a miss rate of 0.3%. (Backus 2013)
To round out the HEART data, Poldervaart did a stepped-wedge, cluster randomized trial comparing the HEART score to usual care in 3648 adult emergency department patients with chest pain. (2017) Among the group of patients discharged home with a low risk HEART score (0-3) they report a rate of MACE of 2%. However, out of the 14 patients with MACE, there were only 4 with an MI and 1 who died (from an unknown cause), meaning that the true miss rate was 0.7%. (Poldervaart 2017)
A final source of data to consider is insurance databases. Foy (2015) looked at more than 421,000 emergency department chest pain patients in a large insurance database. Only 464 patients (0.11%) were readmitted with an MI within 7 days, and only 1396 patients (0.33%) were readmitted with an MI at 190 days. These numbers (a miss rate of 1-3 in a thousand) are remarkably similar to the numbers seen in the stress testing studies and in Pope (2000).
The chest pain patients that we discharge after a negative work-up in the emergency department are extremely low risk. Although some of these studies confuse the issue by introducing composite outcomes, the data is actually incredibly consistent. Our miss rate in the emergency department is probably between 0.1% and 0.3%.
This doesn’t mean that you should turn off your brain. These patients were not simply sorted with ECGS and troponins. Judgement matters. There are definitely patients with normal ECGs and negative troponins that still warrant admission. However, the vast majority of patients do not. When you are done your work-up, you can tell your patient that they have a 1-3 in a thousand chance of having an MI in the coming weeks to months. I guarantee you, the vast majority will want to go home.
When it comes to stress testing, this data gives us a pretest probability. The patients we are discharging – the ones we are considering for stress test – have less than a 1% chance of having disease. Unfortunately, although we like to think of our test results as infallible, even an excellent test will fail when applied to a population with such a low risk. And as we will discuss later in this series, the exercise stress test is far from an excellent test.
- This post is part of a series exploring the practice of stress testing.
Amsterdam, E. A., J. D. Kirk, D. B. Diercks, et al. (2002). “Immediate exercise testing to evaluate low-risk patients presenting to the emergency department with chest pain.” Journal of the American College of Cardiology 40(2): 251-6.1
Backus BE, Six AJ, Kelder JC, et al. Chest pain in the emergency room: a multicenter validation of the HEART Score. Critical pathways in cardiology. 2010; 9(3):164-9. [pubmed]
Backus BE, Six AJ, Kelder JC, et al. A prospective validation of the HEART score for chest pain patients at the emergency department. International journal of cardiology. 2013; 168(3):2153-8. [pubmed]
deFilippi CR, Rosanio S, Tocchi M, et al. Randomized comparison of a strategy of predischarge coronary angiography versus exercise testing in low-risk patients in a chest pain unit: in-hospital and long-term outcomes. Journal of the American College of Cardiology. 2001; 37(8):2042-9. [pubmed] [free full text]
Douglas PS, Hoffmann U, Patel MR, et al. Outcomes of anatomical versus functional testing for coronary artery disease. The New England journal of medicine. 2015; 372(14):1291-300. [pubmed] [free full text]
Foy AJ, Liu G, Davidson WR, Sciamanna C, Leslie DL. Comparative effectiveness of diagnostic testing strategies in emergency department patients with chest pain: an analysis of downstream testing, interventions, and outcomes. JAMA internal medicine. 2015; 175(3):428-36. PMID: 25622287
Gomez, M., J. Anderson, L. Karagounis, et al. (1996). “An emergency department-based protocol for rapidly ruling out myocardial ischemia reduces hospital time and expense: results of a randomized study (ROMIO).” Journal of the American College of Cardiology 28(1): 25-33.2
Hoffmann U, Truong QA, Schoenfeld DA, et al. Coronary CT angiography versus standard evaluation in acute chest pain. The New England journal of medicine. 2012; 367(4):299-308. [pubmed] [free full text]
Litt HI, Gatsonis C, Snyder B, et al. CT angiography for safe discharge of patients with possible acute coronary syndromes. The New England journal of medicine. 2012; 366(15):1393-403. [pubmed] [free full text]
Manini, A. F., A. T. McAfee, V. E. Noble, et al. (2010). “Prognostic value of the Duke treadmill score for emergency department patients with chest pain.” The Journal of Emergency Medicine 39(2): 135-43.3
Meyer, M., R. Mooney and A. Sekera (2006). “A critical pathway for patients with acute chest pain and low risk for short-term adverse cardiac events: role of outpatient stress testing.” Annals of Emergency Medicine 47(5): 435e1-3.4
Napoli AM. The association between pretest probability of coronary artery disease and stress test utilization and outcomes in a chest pain observation unit. Academic emergency medicine : official journal of the Society for Academic Emergency Medicine. 2014; 21(4):401-7. PMID: 24730402
Poldervaart JM, Reitsma JB, Backus BE, et al. Effect of Using the HEART Score in Patients With Chest Pain in the Emergency Department: A Stepped-Wedge, Cluster Randomized Trial. Annals of internal medicine. 2017; 166(10):689-697. [pubmed]
Pope JH, Aufderheide TP, Ruthazer R, et al. Missed diagnoses of acute cardiac ischemia in the emergency department. N Engl J Med. 2000;342:(16)1163-70.
Scheuermeyer, F., G. Innes, E. Grafstein, et al. (2012). “Safety and Efficiency of a Chest Pain Diagnostic Algorithm With Selective Outpatient Stress Testing for Emergency Department Patients With Potential Ischemic Chest Pain.” Annals of Emergency Medicine.59(4):256.64.5