Date: September 10th, 2015
Guest Skeptic: Daniel McCollum. Daniel is an Assistant Residency Director at Georgia Regents University in Augusta, GA. He loves craft beer, evidence based medicine, and Batman.
Case: A 53 year-old man presents with left sided chest pain that started 90 minutes ago. The pain radiates to his left arm and he has nausea without vomiting. The pain began while he was doing yard work and he has never had pain like this before. His initial ECG is unremarkable.
Background: Only about 5% of all consecutive patients presenting with acute chest pain will have a ST elevated myocardial infarction (STEMI) (Apple et al). These are the easy ones to diagnose and manage. This leaves the other 95% of chest pain patients. These are the hard ones.
We need to figure out who will rule-in vs. rule-out for acute myocardial infarction (AMI). This is where cardiac biomarkers play a major role.There have been many biomarkers used over the last 60 years to try to identify patients with acute myocardial infarction.
The first practical test utilized, as a cardiac marker was serum glutamic oxaloacetic transaminase (SGOT) that is now called aspartate amino-transferase (AST). Since the late 1990‘s the cardiac marker of choice has changed from CK-MB to Troponin.
A limitation of current troponin assays is that they can take 3-4 hours to rise. This means the diagnosis of Non-STEMI can take many hours of continued monitoring with serial blood sampling.
Ruling out AMI takes time, uses resources, contributes to overcrowding, and causes patient anxiety. High sensitivity troponin assays are all the rage now and used in many emergency departments. They offer very high sensitivity but are less specific than prior troponin assays.
There is limited evidence from studies that the use of high sensitivity troponins may allow for the safe discharge of patients more rapidly from the emergency department than what is possible with traditional troponin assays.
Clinical Question: Is it possible to rule in/rule out acute myocardial infarction in the emergency department with high sensitivity troponin T and ECG in one hour?
Reference: Reichlin et al. Prospective validation of a 1-hour algorithm to rule-out and rule-in acute myocardial infarction using a high-sensitivity cardiac troponin T assay. CMAJ 2015
- Population: Patients presenting to the emergency department within twelve hours of onset of non-trauamtic chest pain or symptoms suggestive of AMI. This was part of an ongoing study in Switzerland, Spain and Italy (Advantageous Predictors of Acute Coronary Syndromes Evaluation [APACE] Study).
- Excluded: Kidney failure requiring dialysis or STEMI
- Intervention: High Sensitivity Troponin T at time zero and one hour with ECG
- Comparison: Two independent cardiologists (ED presentation to 90-day follow-up, including patient history, physical exam, coronary angiography, echocardiography, follow-up data, and serial high sensitivity troponin T measurements).
- Outcome: Acute myocardial infarction (negative predictive value, positive predictive value and area under the curve.
Authors’ Conclusions: “This rapid strategy incorporating high-sensitivity cardiac troponin T baseline values and absolute changes within the first hour substantially accelerated the management of suspected acute MI by allowing safe rule-out as well as accurate rule-in of acute MI in 3 out of 4 patients.”
Quality Checklist for A Diagnostic Study:
- The clinical problem is well defined? Yes. You can rule out myocardial infarction but can’t rule out acute coronary syndrome (ACS)
- The study population represents the target population that would normally be tested for the condition (ie no spectrum bias)? No. The patients may have been more ill and we will discuss later when talking nerdy
- The study population included or focused on those in the ED? Yes
- The study patients were recruited consecutively (ie no selection bias)? Yes
- The diagnostic evaluation was sufficiently comprehensive and applied equally to all patients (ie no evidence of verification bias)? No. Patients that were deemed to be low risk did not always proceed to 6-hour troponin.
- All diagnostic criteria were explicit, valid and reproducible (ie no incorporation bias)? No. The challenge of conducting high-sensitivity troponin studies is that the diagnosis of NSTEMI is a disease entity based on a test without an independent reference standard. Heavy incorporation bias as described above comes into play. Changes in troponin seem to be the main basis for the diagnosis of AMI.
- The reference standard was appropriate (ie no imperfect gold-standard bias)? No. Two independent cardiologists clinical judgement was the gold standard. There is a large degree of subjectivity with the process, hence a third cardiologist had to adjudicate 12% of the time (authors correspondence). This seems quite high considering the diagnosis of AMI would be a common one for cardiologists to make. It is not clear if this disagreement tended toward over-diagnosis versus under-diagnosis.
- All undiagnosed patients underwent sufficiently long and comprehensive follow-up (ie no double gold-standard bias)? Unsure. The follow up was for a very reasonable 24 months. Contact multiple times by phone or written form as well as review of medical records seems very reasonable. Author correspondence revealed a very impressive 99.7% success at 30 day follow up (1316/1320). This is quite impressive for a study of this size and makes it unlikely that the small number of patients missed would greatly influence the results of the study.
- The likelihood ratio(s) of the test(s) in question is presented or can be calculated from the information provided? Yes. LR+ = Sensitivity / (1 – Specificity); LR- = (1 – Sensitivity) / Specificity. The numerator and denominator are the vertical and horizontal axes respectively of the ROC curve.
- The precision of the measure of diagnostic performance is satisfactory? Yes. Seems to be so.
Key Results: 1,320 patients who presented to the emergency department within twelve hours of onset of non-traumatic chest pain or other symptoms suggestive of AMI. The median age was 60 years and 69% were men. Acute MI was the final diagnosis in 17% of patients. They divided patients into three different categories:
- Rule-Out of Acute MI: Base- line high sensitivity troponin T level <12 ng/L and an absolute change within the first hour of less than 3 ng/L. Sixty percent (786/1320) ruled out for AMI.
- Rule-In of Acute MI: Either a baseline high sensitivity troponin T value of 52 ng/L or greater, or an absolute change within the first hour of 5 ng/L or greater. Sixteen percent (216/1320) ruled in for AMI.
- Observation Zone: Patients fulfilling neither criteria for rule-in or rule-out were classified as being in the “observational zone.” Twenty-four percent (318/1320) were in the observation zone.
One-hour Algorithm Test Characteristics for Acute MI:
- Area under the curve 0.96 (95%CI 0.95 to 0.97)
- Rule Out Zone: Sensitivity of 99.6% (95% CI 97.6% to 99.9%) and Negative Predictive Value 99.9% (95% CI 99.3 to 100%)
- Rule In Zone: Specificity of 95.7% (95% CI 94.3% to 96.8%) and Positive Predictive Value 78.2% (72.1% to 83.6%)
One AMI was felt to be missed of 786 patients ruled out with the one-hour algorithm.
This was an attempt by Reichlin et al to investigate whether 1-hour algorithm to rule-out and rule-in acute myocardial infarction using a high-sensitivity troponin T as a biomarker. There were a number of concerns with this manuscript. Dr. Andrew Worster was invited to help us critically appraise this paper. Dr. Worster does research on high sensitivity troponins and recently published a review of this paper. Ann Int Med 2015
Bias: The challenge of conducting high-sensitivity troponin studies is that the diagnosis of NSTEMI is a disease entity based on a test without an independent reference standard. This can lead to a number of biases that can distort the results.
- Incorporation Bias: Occurs when results of the test under study are actually used to make the final diagnosis. This makes the test appear more powerful by falsely raising the sensitivity and specificity. The gold standard appears to be two out of three cardiologists agreeing upon the diagnosis of AMI. This seems to have been driven predominantly by defining AMI as a rise in high sensitivity troponin T. While this is a commonly used definition of AMI, it is inappropriate to judge the accuracy of a test to make a diagnosis that requires the results of the test. The results would be much more robust if there was discussion of other tests such as echocardiography showing wall motion abnormalities, abnormal provocative tests, future ECG changes, or abnormalities seen on heart catheterization.
- Partial Verification Bias: This happens when only a certain set of patients who underwent the index test is verified by the reference standard. This would increases sensitivity but decreases specificity. Patients that were deemed to be low risk did not always proceed to 6-hour high sensitivity troponin T. No additional data is given about which patients received additional testing such as coronary angiography or provocative testing. It is implied that those that were low risk did not undergo additional testing.
Spectrum Bias: Sensitivity depends on the spectrum of disease, while specificity depends on the spectrum of non-disease. So you can falsely raise sensitivity if the clinical practice has lots of very sick people (sicker than who you see in the emergency department). Specificity can look great if you have no sick patients in the cohort (worried well). They included only patients who presented to a cardiac research hospital within 12hrs of pain. These patients could have been potentially more ill. There is also the problem that 20% of eligible patients who failed to complete index testing. We don’t know in which direction the loss of 20% favours the bias but the failure of 20% of those enrolled to complete the first phase of the study raises concerns about study protocol adherence. Correspondence with the author revealed a graph showing the outcomes of those that missed the one-hour troponin. This group had significantly lower mortality for the first year after the study before closely following the mortality curve of the study population. This reveals that this population was fairly different than the study population and appears to have been less critically ill than study patients. The author’s calculated a Kaplan-Meier Curve p value of 0.85 but it is unclear at what time period this was done on.
- Learn more about how bias can impact the diagnostic test accuracy by reading Kohn, Carpenter and Newman in AEM 2013
- Risk of Over-Testing: Another concern with this protocol and others based upon high sensitivity troponin assays is the lack of specificity. The initial high sensitivity troponin T was only 48.4% specific, with more false positives than true positives. If there were inappropriate use of this test in ultra low risk patients, there may be a paradoxical rise in the number of patients being evaluated for chest pain in the emergency department. Similar to the use of d-dimers to evaluate for pulmonary embolism, this tool will need to be carefully applied to avoid over-testing.
- Imprecision of the Assay: The change in high sensitivity troponin T was within the allowable imprecision of the assay. So the change may only be analytical variation and not clinical variation. This is really the keystone because if the change is within the assay’s coefficient of variation, all other issues are moot ( Kavsak PA. High-five for high-sensitivity cardiac troponin T: depends on the precision and analytical platform. JAMA Intern Med 2013)
- Missed AMI: There was one false negative patient in the study. It is not clear that the single false negative patient was having an AMI from the paper. Her peak high sensitivity troponin T was only 17 and there was no discussion of other factors such, as coronary catheterization results were included.
- Conflict of Interest: Roche supplied the test and several authors had conflicts. This does not invalidate or make the conclusions wrong but should make readers more skeptical of the results. However, greater than 90% of research is funded by industry and it’s not necessarily bad. Non-industry sponsors encourage industry-investigator partnerships. In addition, the author was very prompt and generous by sharing data with us.
Author Response to Our Questions:
- Is there data about what happened to those that were excluded due to not having a one-hour repeat high sensitivity troponin T? What were their outcomes?
- See the supplemental Table 1 for the baseline characteristics of these patients (seems that this was lost during the publishing process, sorry). Outcomes in terms of mortality during follow-up were equal (see Kaplan Meier below, p=0.85), the initial difference I would attribute due to chance and the different sample size.
- How complete was follow up by phone or writing? Were any patients lost to follow up?
- 30 days follow-up was complete in 1316/1320 pts (99.7%).
- How often did the independent cardiologists require adjudication by a third cardiologist? Is there a kappa value to assess the inter-rater reliability?
- Adjudication with a third cardiologist was required in 12%. No inter-rater comparisons are available.
- Is there any information about how the cardiologists decided who was and was not having an MI? What percentage of patients had a heart catheterization or other testing?
- The adjudication was made by clinical judgment integrating all medical information available up to 90-days including patient history, physical examination, results of laboratory testing (including serial hs-cTnT levels), radiologic testing, Electrocardiogram, echocardiography, cardiac exercise test, lesion severity and morphology in coronary angiography. Troponin levels were interpreted as described in the supplemental methods of the paper. Only the tests that were considered clinically indicated by the medical teams responsible for the patients care were performed, no additional test for the purpose of the study only (except for serial blood sampling) were performed. Coronary Angiography was performed in 23%, stress testing in 22%. Given that the adjudication of AMI vs. Non-AMI is based mainly on history, symptoms, ECG and troponin levels, additional testing with coronary angiography and stress testing was particularly helpful to distinguish in the non-AMI group between cardiac and non-cardiac causes of non-AMI chest pain.
Comment on Authors’ Conclusion Compared to SGEM Conclusion: The author’s are much more optimistic of their protocol’s current ability to rapidly rule out patient’s in the emergency department that we are based on the data presented. Also, our role in the emergency department is not limited to ruling out AMI but as importantly, to determining if the patient has acute coronary syndrome and if not, how soon do they require outpatient follow-up.
SGEM Bottom Line: Due to aforementioned biases and other factors mentioned, a one-hour protocol utilizing high sensitivity troponin T cannot be recommended at this time. External validation of this protocol along with a more explicit discussion of how the diagnosis of AMI is arrived at might allow for a rapid rule out in the future.
Case Resolution: Shared decision making was done with the patient. After discussion with the patient, he agreed to stay in the emergency department for six hours. His troponin at zero and six hours were negative, and serial ECGs showed no ischemic changes, and his pain had completely resolved. He was discharged home with close follow up with his primary care physician and strict return precautions.
Clinically Application: High sensitivity troponin T may/may not have future clinical application to rapidly rule out low risk patients for acute myocardial infarction. It is unclear at this time what effect the low specificity of the test will have.
What Do I Tell My Patient? There is a new blood test that checks for very early damage to your heart. If you are having a heart attack, this test often picks up early signs of heart injury. However, the test is so sensitive that it may suggest you are having a heart attack even if you aren’t having a heart attack. In order to be as sure as we can be, we will need to get multiple heart tracings and do more than one blood test.
Keener Kontest: Last weeks winner was Dr. Alan Sielaff from the University of Michigan. Alan was able to name more than three super heros who obtained their powers from radiation exposure including the Hulk, Daredevil and the Fantastic Four team.
Listen to the podcast for this weeks’ question. If you know the answer contact me at TheSGEM@gmail.com with “KEENER” in the subject line. The first person to correctly answer the question will receive a cool skeptical prize.
SGEM Hot or Not: If you have a paper you think should have a skeptical review performed then click on the Hot or Not button on the website. Each Friday five new articles are posted for SGEMers to vote on. Those papers registering as HOT will be reviewed.
SGEM Global: The SGEM has gone global. The blog and podcast is now available in four languages (English, French, Spanish and Portuguese). We are trying to cut that KT window down to less than one year by eliminating the language barrier. If you are passionate about emergency medicine, FOAMed, speak another language and have always wanted to podcast then send me an email to TheSGEM@gmail.com to talk about joining the SGEM Global Team.