ECG Interpretation – Never Trust the Computer

April 8, 2022

Written by Clay Smith

Spoon Feed
ECG changes from STEMI-negative occlusion myocardial infarction (OMI) can be subtle, yet lethal, and are easily missed by computerized ECG interpretation.

Why does this matter?
We recently covered an article that indicated that the computer interpretation of an ECG as “normal” could be accepted, thus avoiding the need for emergency physician overread and reducing cognitive load. But is this really a good idea?

Computer ≠ physician
This was a letter in response to the article mentioned above. The authors argue that OMI is common, and 30-50% don’t meet classic STEMI criteria. Yet, these are life-threatening presentations, often with subtle ECG findings. The authors in the original study compared the computer interpretation with the gold standard of cardiologist interpretation. The problem is, computer interpretation is 35% sensitive for OMI, and cardiology interpretation is just 49% sensitive for OMI. The letter authors argue the gold standard must be a final diagnosis of OMI based on coronary angiography, with a culprit lesion – not cardiology interpretation. This matters, because patients without true STEMI criteria, yet with OMI, have similar infarct size and need emergent reperfusion. They also mention that emergency physicians are much better positioned than cardiology to detect ECG changes associated with hyperkalemia. Finally, the authors discuss that they have 30 cases in which the computer read the ECG as normal, yet the patients had confirmed OMI. They conclude that, though burnout is a problem, there is no evidence that ECG interpretation on shift is the culprit. Some “distractions” are essential, and I would agree that ECG interpretation is one distraction that’s worth my attention on shift. They showed this example ECG of a “normal” computer reading with OMI.

Source
Alexander Bracey, H. Pendell Meyers, Stephen W. Smith. Emergency physicians should interpret every triage ECG, including those with a computer interpretation of “normal”. Am J Emerg Med, https://doi.org/10.1016/j.ajem.2022.03.022.