Chest pain is tricky. And scary. The combination of these two things makes it one of the chief complaints that seems to be difficult to work up in a thoughtful way, which minimizes risk to the patient (and provider) while also not overreaching in one’s diagnostic testing and thereby adding additional harms.
In medical school, we learned about the TIMI score as the best way for evaluating chest pain in our patients– however, this score was developed for inpatients on the cardiology service admitted with NSTEMI/UA, not ED patients presenting with chest pain, and has only really been validated in high-risk ED patients. The GRACE score is another one that seems to slightly outperform the TIMI in terms of predicting certain adverse events, but again was not designed for risk stratification of ED patients with chest pain.
So now we have (and have had for a while, this isn’t exactly new – I am more reviewing it for my own benefit) the HEART score, designed to “identify both low and high risk patients for an acute coronary syndrome” in the emergency department. It was not derived from a database, but from “clinical experience and medical literature”, and was then prospectively validated in 2440 patients at 10 sites. When compared to TIMI and GRACE, the c-statistic (or area under the receiver-operator curve) was 0.83 v. 0.75 and 0.70 respectively, showing that it did a better job discriminating patients with higher risk for major adverse cardiac events (MACE) in these patients. Pertinently for the ED physician, it also did a better job ruling *out* badness, with a lower percentage of “low-risk” scorers having an adverse event. With all this in mind, I plan to try to use the HEART score in my discussions with attendings when presented with chest pain patients, and hope that I will not only catch more (and rule out more) badness, but may also help reduce invasive imaging and stress testing in these low-risk patients.
In a recent single-center study, Mahler et al. looked back at patients who had HEART scores of 0-3 (low risk) admitted to an ED-based observation unit (keep the population in mind) and evaluated the impact of this score on their receipt of further diagnostic testing down the line, as well as the incidence of adverse events in this group. They found, unsurprisingly, that these patients were in fact at low-risk for ACS — only 0.5% of patients in this group had an adverse event in the next 30 days (though to be thorough, they had a LTFU rate of 30% which is pretty significant). The surprising, and meaningful to me finding was that they reduced the rate of further testing by 83%, no doubt saving these patients from unnecessary stress and anxiety, potential harms or complications, and costs to both them and the health care system.
This is a very incomplete treatment of chest pain risk stratification, I know, but I hope to add more as I learn and read more about these scoring systems and others, and grow in my understanding of critical appraisal of the literature.