Hello everyone, thanks for your online (and offline) comments regarding the HEART score. As if on cue, Annals of Emergency Medicine last week published a new evaluation of the score. Soares and co-workers enrolled 336 nonconsecutive Massachusetts patients with potential ischemic chest pain, of whom 9% had a major adverse cardiac event. 78% (!!!) were admitted to hospital. The inter observer agreement was approximately 77% (kappa 0.48) for HEART +/- 3. (This is the standard low-risk cutoff).
If you haven’t reviewed kappa scoring, 0.48 is only moderate agreement. The lowest-performing contributor was –you guessed it–patient history, (the “H” in HEART) with an agreement of 72%.
Furthermore, 4/30 patients who had major adverse cardiac events had a low HEART score.
I had a few offline comments regarding “well, what should we use instead”? This is a great question and it is always easier to criticize someone else’s work than to come up with one’s own. However, we have to recognize that prediction rules from Ottawa (ankle / knee / SAH / Wells) are developed rigorously with populations similar to our own. Development of the HEART score does not stand up to such scrutiny.
As such, Dr Krause pointed out that clinical acumen (and, I will add, high sensitivity troponins) are likely more valuable than the HEART score. Earlier in my career I was fortunate enough to perform a secondary analysis of Dr Jim Christenson’s chest pain cohort: clinicians at St Paul’s evaluated >1100 patients, with 25% admissions, and missed 0 / 120 patients with a 30-day acute coronary syndrome. (Ann Emerg Med, 2012) Not too shabby. Please trust your own judgement!
Add public comment…