Re: Validation of the Vancouver Chest Pain Rule


To the Editor:

We are very pleased that Jalili et al. are interested in the early discharge of chest pain patients. We congratulate them on their carefully performed and well-written validation of the Vancouver Chest Pain Rule.[1] We wish to point out two differences in the methodology they used from that used in the original development of the rule.[2]

First is in the definition of acute coronary syndrome (ACS), the primary outcome. They appear to define any patient with symptoms suggestive of ischemia and a level of troponin T (Tn T; 0.014 μg/L) or creatine kinase myocardial band isoenzyme (CK-MB; females 3.77 ng/mL, males 6.73 ng/mL) above the upper limit of normal as either an acute myocardial infarction or unstable angina. The differential diagnosis of patients with elevated troponins is wide[3] and does not necessarily imply ACS. To contrast, in our study, all patients with elevated troponin levels and no other objective findings of ischemia were carefully adjudicated by two blinded cardiologists and if no agreement was reached, categorization was determined by the senior author. It is possible that this difference in assigning outcome is one of the reasons for the modest discrepancy in the two cohorts.

Second, there was a misunderstanding of the definition of the cutoff used for CK-MB. We take full responsibility for the misunderstanding, as the choice of the cutoff for CK-MB could have been more clearly explained.[2] In our rule, continuous variables such as CK-MB were explored and cut points made based on the ability to discriminate between those with ACS and those without. At the time we developed the study, there was no clear consensus on an absolute CK-MB cutoff that defined ACS, but it was considered to be between 5 and 10 ng/mL. Diagnosis of myocardial infarction was based on a rise consistent with the history rather than one absolute value. We entered absolute values into our recursive partitioning program including those in the “normal range.” Surprisingly to us, CK-MB was a powerful discriminator for early discharge using the very conservative cutoff of 3.0 μg/mL. In contrast, Jalili and coauthors have used the cutoff for an upper limit of normal of 3.77 (females) and 6.73 (males) μg/mL, which exceeds our cutoff point. Two of the four patients incorrectly classified as eligible for early discharge in their cohort had elevated initial Tn T levels. In our study, no patient with a CK-MB of <3.0 μg/mL had an initial positive Tn T. In the paper by Jalili et al., the two patients with the initially positive Tn T may have had a CK-MB level of greater than 3.0 μg/mL and, if so, the sensitivity in this validation cohort would be corrected from 95.1% to 97.6% (80 of 82). Specificity would be lowered, but it is impossible to calculate using data from the paper.

It may be of interest to readers that, owing to the decline of CK-MB testing in North America, we have modified the Chest Pain Early Discharge Decision Rule and validated it. Our modified rule does not incorporate CK-MB into the diagnostic algorithm (this manuscript is under publication review).

We think that the information gathered in the paper by Jalili et al., and others,[4, 5] to build decision support for rapidly defining a subset of patients with inconsequential chest pain is very important for individual patients and for managing busy emergency departments. We look forward to more discussion and consensus on the best method for making these early decisions.