Written by Clay Smith
HEART Pathway authors defend the HEART score and its offspring. Let’s see how their arguments hold up.
Why does this matter?
Editors at Annals of EM wrote a critique of the HEART score methodology last year. This is a letter of response from the HEART Pathway authors. We usually don’t cover letters to the editor, but this is so important for our everyday practice, I just had to put this out there for discussion.
Editors Green and Schriger made numerous critiques. Here are just four. 1) HEART score variables were not formally derived or validated. 2) They question whether it is better than gestalt. 3) Some components have poor interrater reliability. 4) Sensitivity is 96-97%, with lower 95%CI 93%.
The response letter fully acknowledges that the HEART score has variables that are intuitive but not statistically derived; it also has subjectivity in some of the components, logical inconsistencies*, and may have too high a miss rate. This, they argue, is why they took HEART to the next level, the HEART Pathway, in which much of the subjectivity is resolved, logical inconsistencies are removed, and diagnostic accuracy is improved. I strongly suggest you get the HEART Pathway app. MDCalc lacks nuance for the HEART Pathway.
Also, the way Green and Schriger used the term “miss rate” may have been confusing, as they stated, “The lower bounds of the 95% confidence intervals for these pooled sensitivities are 93% to 94%, ie, compatible with missing 6% to 7% of major adverse cardiac events. Said another way, these data suggest that the HEART score could, within 95% confidence, miss 1 in 14 occurrences of major adverse cardiac events.” A second letter (see below) pointed out that this statement is easily misunderstood. It could be interpreted that among the whole population of chest pain patients, the HEART score will miss up to 7%, but that’s not actually true. What we really care about is 1-NPV, i.e. the false negative rate, which was actually just 1.6% in a large HEART score meta-analysis. That is the potential miss rate of the patient in front of you with a low HEART score – not 7%. And the HEART Pathway is likely even lower, more like 0.4%.
What’s my take home? I think the plain old HEART score has issues. But I use the HEART Pathway in practice. It’s not perfect, but I think there is enough evidence to use it with confidence. Again – be sure to get the app.
Stopyra J, Ashburn N, Mahler S. A Methodological Appraisal of the HEART Score and Its Variants Response. Ann Emerg Med. 2022 Jan;79(1):84-85. doi: 10.1016/j.annemergmed.2021.09.428.
Moumneh T, Douillet D, Savary D, Roy PM. Do Not Throw Out the Baby With the Bathwater. Ann Emerg Med. 2022 Jan;79(1):85-86. doi: 10.1016/j.annemergmed.2021.09.429.
*i.e. a 64 year old, moderately suspicious story, with a positive troponin could be “low risk” – HEART score 3.