Does My Patient with Chest Pain Have Acute Coronary Syndrome?

Background: We have already discussed the value of a good history in assessing patients with chest pain on REBEL EM. What is known about chest pain is that it is a common complaint presenting to EDs all over the world, but only a small percentage of these patients will be ultimately diagnosed with Acute Coronary Syndrome (ACS). This complaint leads to prolonged ED length of stays, provocative testing, potentially invasive testing, and stress for the patient and the physician. For simplicity sake, we will say that, looking at the ECG can make the diagnosis of STEMI. What becomes more difficult is making a distinction between non-ST-Elevation ACS (NSTEMI/UA) vs non-cardiac chest pain. ED physicians have different levels of tolerance for missing ACS with many surveys showing that a miss rate of <1% is the acceptable miss rate, but some have an even lower threshold, as low as a 0% miss rate. Over testing however, can lead to false positives, which can lead to increased harms for patients. In November 2015, a new systematic review was published reviewing what factors could help accurately estimate the probability of ACS.

What Study are we Evaluating?

Fanaroff AC et al. Does This Patient With Chest Pain Have Acute Coronary Syndrome? The Rational Clinical Examination Systematic Review. JAMA 2015; 314(18): 1955 – 65. PMID: 26547467

What They Did:

• Systematic review evaluating the accuracy of initial history, physical examination, electrocardiogram, and risk scores in determining the probability of ACS
• Reference standard used was final hospital diagnosis of ACS or occurrence of cardiovascular event within 6 weeks

Outcomes:

• Sensitivity, Specificity and Likelihood Ratio (LR) of findings for determining the diagnosis of ACS

Results:

• 58 articles met inclusion criteria for the study
  • Between clinician reliability was moderate to good for Risk Factors (Kappa > 0.60) & ECG (Kappa > 0.55)
  • Between clinician reliability was fair for chest pain quality, location, radiation, and associated symptoms (Kappa 0.29 – 0.37)
  • Rates of final ACS Diagnosis: 5% – 42% (Median 14%)
  • The estimated incidence of ACS in the 28 studies that evaluated a clinical prediction rule was 13% [95% CI 11 – 16%]
• Most Useful Cardiac Risk Factors most Suggestive of ACS:
  • Prior abnormal stress test – Spec 96%; LR 3.1 [95% CI 2.0 – 4.7]
  • Peripheral Arterial Disease – Spec 97%; LR 2.7 [95% CI 1.5 – 4.8]
  • Prior CAD – Spec 79%; LR 2.0 [95% CI 1.4 – 2.6]
• Most Useful Chest Pain Characteristics most Suggestive of ACS:
  • Pain Radiation to Both Arms – Spec 96%; LR 2.6 [95% CI 1.8 – 3.7]
  • Pain Similar to Prior Ischemia – Spec 79%; LR 2.2 [95% CI 2.0 – 2.6]
  • Change in Pain Pattern Over Prior 24 Hours – Spec 86%; LR 2.0 [95% CI 1.6% – 2.4%]
  • Response to Nitroglycerin NOT Helpful in Prediction of ACS – Spec 35%; LR 1.1 [95% CI 0.93 – 1.3]
• Most Useful Physical Examination Findings Suggestive of ACS:
  • Hypotension (SBP <100mmHg) – Spec 99%; LR 3.1 [95% CI 1.2 – 7.9]
• Most Useful ECG Findings Suggestive of ACS:
  • ST-segment depression – Spec 95%; LR 5.3 [95% CI 2.1 – 8.6]
  • Any Evidence of Ischemia – Spec 91%; LR 3.6 [95% CI 1.6 – 5.7]
• Most Useful Clinical Decision Rules Suggestive of ACS:
  • TIMI Score 5 -7: LR 6.8 [95% CI 5.2 – 8.9]
  • HEART Score 7 – 10: LR 13 [95% CI 7.0 – 24]
• Most Useful Clinical Decision Rules Not Suggestive of ACS
  • HEART Score 0 – 3: LR 0.20 [95% CI 0.13 – 0.30]
  • TIMI Score 0 – 1: LR 0.31 [95% CI0.23 – 0.43]

Strengths:

• Confidence Intervals for many of the endpoints reviewed were narrow enough to inform clinical practice allowing physicians to categorize patients into low, intermediate, and high likelihood of ACS.
• TIMI and HEART Scores were evaluated in 16 and 4 high-quality studies respectively, with more than 30,000 and 13,000 patients. This helps strengthen the statement that these tests have the best diagnostic performance for ACS.

Limitations:

• Studies included in this systematic review were heterogeneous in their diagnosis of ACS. Ultimately, patients diagnosed with ACS do represent a spectrum of pathophysiologic disease making it difficult to utilize a reference standard.
• Use of revascularization to support a diagnosis of ACS has limitations. Specifically that revascularization gives us an anatomic diagnosis of CAD rather than identifying an unstable plaque. Using revascularization as an endpoint potentially will include a subset of patients with true acute coronary artery thrombosis, but also a subset with stable CAD.
• There was incorporation bias of this study. This bias occurred due to the fact that the result of a screening test is a component of the reference standard. In other words if a patient is given a diagnosis of ACS and that is used as the reference standard, then the index test that was used to make that diagnosis may cause positive and negative likelihood ratios appear more useful than they should be (i.e. overinflated).
• There was also verification bias that occurred in this study. This occurred due to the fact that the result of a screening test influences whether or not the reference standard test is performed.  In other words, patients with risk factors or symptoms concerning for ACS are more likely to undergo revascularization or be diagnosed with obstructive CAD. This results in an overestimation of sensitivity making the negative likelihood ratios less useful and underestimation of specificity, which then makes positive likelihood ratios better than what they actually may be.
• Although many studies met the inclusion criteria, there was a wide array of index tests, such that only a minority of the included studies contributed data for any given test. (i.e. chest pain radiation to both arms and history of prior abnormal stress test both were strong predictors of ACS, but both were only evaluated in a single study). Also all physical exam findings were evaluated in only 1 study. This makes it difficult to combine data across studies to help reduce bias from a single study.
• Studies incorporating high-sensitivity troponins were excluded. The reasoning was because this test is not available in the US.
• It is unclear how assessment of chest pain is affected by knowledge of ECG and troponin levels. It is not uncommon due to overcrowding to have the results of both of these tests before ever evaluating a patient with chest pain.

Discussion:

• It is important to remember that cardiac troponin testing can aid in the diagnosis or exclusion of myocardial necrosis, especially when trended over time (i.e. 0 and 3 hours) or maybe even with a single troponin level if you have high sensitivity troponin testing available, but this number alone does not rule out ACS (i.e. Unstable Angina).
• If we use the average pretest probability of 13% for the probability of ACS, then a HEART Score of 0 – 3 would decrease the probability of ACS to 2.9% [95% CI 1.9% – 4.3%] and the probability of ACS with a TIMI Score of 0 -1 would decrease the probability of ACS to 4.4% [95% CI 3.3% – 6%]. Therefore, low risk TIMI Score or HEART Score will not decrease the likelihood of ACS to less than 1%, which is what most EM physicians are seeking.

Author Conclusion: “Among patients with suspected ACS presenting to emergency departments, the initial history, physical examination, and electrocardiogram alone did not confirm or exclude the diagnosis of ACS. Instead, the HEART or TIMI risk scores, which incorporate the first cardiac troponin, provided more diagnostic information.”

Clinical Take Home Point: Although it is nice to know the likelihood ratios of individual components of history and physical, it is the use of all these components, plus ECGs and delta Troponins that helps us include or exclude the diagnosis of ACS in clinical practice.

References:

1. Fanaroff AC et al. Does This Patient With Chest Pain Have Acute Coronary Syndrome? The Rational Clinical Examination Systematic Review. JAMA 2015; 314(18): 1955 – 65. PMID: 26547467

Post Peer Reviewed By: Anand Swaminathan (Twitter: @EMSwami)