Written by Clay Smith
Spoon Feed
With fair to moderate interrater reliability and uncertain accuracy of the HEART score, we may want to consider alternative ways to risk stratify chest pain patients.
Why does this matter?
I feel like we have been down on the HEART score of late. I promise, I don’t have an agenda here. I just search the journals and try to cover what is relevant, and I keep finding articles that raise concerns. Previous studies, comparing different clinicians computing the HEART score have shown good overall agreement, but agreement on the history component was low. That said, agreement between emergency physicians and cardiology was awful. Implementation studies using the HEART score have been encouraging, although compliance with actually sending low-risk patients home was low. What about comparing clinicians with trained research associates? If clinicians don’t rate things the same way the researchers do, this will lead to different performance of the tool in real life vs the research setting.
HEART trouble
This was a single center prospective study of 336 patients with chest pain. Clinicians in the ED and researchers both independently gathered all the data elements to complete the HEART score, and results were compared for agreement. Clinicians were 100% sensitive for MACE vs 86.7% for researchers. But the most important finding, and primary outcome, was that agreement wasn’t great. When the HEART score was dichotomized – low risk (0-3) or moderate to high risk (4-15) – the ED clinicians and researchers agreed 78% of the time overall; kappa 0.48 (95%CI 0.37 to 0.58), which indicates moderate agreement. Clinicians and researchers had very high agreement on age (97%) and troponin results (98%, though weighted kappa was 0.46 for this variable). Where they differed most was on the subjective aspects of the history, 72% agreement (weighted kappa 0.14) and ECG, 85% (weighted kappa 0.4). By the way, the research associates did not read the ECGs; two physician authors were blinded to outcomes and reviewed ECGs for the research team. Most of the scores differed by just one point between clinicians and researchers. Unfortunately, most of these differences occurred at the critical score of 3-4, which is the cut point between going home and staying in the hospital. One third of patients with low scores were still admitted, indicating that clinicians often disregarded and didn’t adhere to the decision tool. The authors summed it up: “With uncertainties in agreement, accuracy, and adherence, we urge caution in the widespread use of the HEART score in isolation as a standard of care to determine the disposition of ED patients with chest pain.”
Source
A Prospective Evaluation of Clinical HEART Score Agreement, Accuracy, and Adherence in Emergency Department Chest Pain Patients. Ann Emerg Med. 2021 Jun 17;S0196-0644(21)00237-7. doi: 10.1016/j.annemergmed.2021.03.024. Online ahead of print.
This is nonsense. The ED docs had 100% sensitivity. Frankly I don’t CARE what the researchers thought and if a lawyer is looking for a way to sue, they will find it. The real problem is INSTITUTIONS AND DOCS who don’t use the HEART score in the first place and base everything on the troponin. This is the same poor logic that indicted original SIRS criteria and change the definition of Sepsis constantly Just to get a paper published. It serves doctors poorly and patients even worse. By the way, prior analysis has shown that even most 5’s are safe for discharge and follow-up if troponin is not elevated.