Andrew McRae: Personalized risk prediction for low risk chest pain patients


Patients with chest pain and symptoms of acute coronary syndromes (ACS) account for over 600,000 emergency department (ED) visits annually in Canada. 85% of these patients do not have a myocardial infarction (MI), and the vast majority of these patients are discharged from the ED. After ED discharge, many of these patients (almost 200,000 annually) undergo outpatient  cardiac testing (exercise stress tests, myocardial perfusion scans, coronary CT angiography), even though the risk of major adverse cardiac events such as death, new myocardial infarction or need for revascularization within 1 year of ED discharge is very small. Conversely, some patients at high risk of adverse events do not receive adequate follow-up and experience worse outcomes. This mismatch between the patients who receive appropriate follow-up, and the patients who actually need it, stems from a lack of dedicated, individualized tools to predict patients’ risk of adverse cardiac events after a myocardial infarction has been ruled out in the ED.

Existing risk prediction tools for ED chest pain patients perform poorly. They sacrifice specificity for the sake of sensitivity, committing an excess of patients to unnecessary testing. Moreover, they are not sufficiently granular to guide decisions around the best outpatient testing approach–let alone to identify which patients are likely to benefit from additional testing at all.


Individualized risk prediction tools, derived in patients who have had MI ruled out, and which incorporate multiple clinical and laboratory parameters, hold the key to better prediction of both the presence of coronary artery disease and risk of short-term adverse events in patients discharged from the ED. The objective of this research program is to develop individualized risk prediction tools for patients who have had an MI ruled out in the emergency department, to identify patients who are likely to benefit from additional cardiac testing, to guide decision-making around what type of cardiac testing is best for each patient, and to guide the appropriate timing of testing. In other words, the objective is to provide personalized risk estimates to get the Right Patient the Right Test at the Right Time.

We will conduct a multicenter prospective cohort study of patients with chest pain who have had MI ruled out in the ED, to derive and validate personalized risk prediction scores to predict risk of major adverse cardiac events. The risk prediction scores derived in this project will then be validated in another prospective cohort study to ensure adequate performance and clinical utility.

Impact for Emergency Medicine

The risk prediction tools that we will develop will improve on current tools in the following ways:

  • We will recruit patients AFTER MI has been ruled out to ensure the prediction tools are derived in the most relevant population;
  • We will provide improved accounting for the effect of sex and pre-existing coronary disease compared to previous tools;
  • We will use new-generation, high-sensitivity troponin assays;
  • We will integrate multiple, commonly-measured laboratory markers to enhance personalized risk prediction;
  • We will use a contemporary, Canadian patient cohort;
  • We will provide more granular risk estimates than current tools, and will use time-to event analyses to identify ideal time-to-testing.

The knowledge product of this work will improve patient outcomes while also optimizing the appropriate utilization of objective cardiac testing after ED discharge.

comment_form(Protocol Feedback)

1 Comment on Andrew McRae: Personalized risk prediction for low risk chest pain patients

Natalie Le Sage said : Guest Report 6 years ago

Congrats Andrew! Probably a systematic/scoping literature review about utility of existing biomarkers for prediction of MACE be pertinent/necessary here. About collaborating, I don't say yes because we know that it is too late for this year...

  • Universite Laval
  • Maybe

Leave a Reply

Your email address will not be published. Required fields are marked