Written by Jesse McLaren
Three patients presented with
acute chest pain and ECGs that were labeled by the computer as completely
normal, and which was confirmed by the final cardiology interpretation (which is blinded to patient outcome) also as completely normal.
What do you think?
Case 1:
Case 2:
Case 3:
Triage ECGs labeled ‘normal’
There have been a number of small
studies suggesting that triage ECGs labeled ‘normal’ are unlikely to have
clinical significance, and therefore that emergency physicians should not be interrupted to
interpret them, and that such patients can safely wait to be seen. These have all
been small studies, studying very few patients with ACS, and often used final
cardiology interpretation rather than patient outcome. The most recent study
found a NPV of 100% of triage ECGs labeled ‘normal’ or ‘otherwise normal’ for
final hospital diagnosis of ACS, and concluded that avoiding physician
interruption would “alleviate interruptions in workflow and improve patient
safety.”
Smith: This study had such low risk patients that not a single patient was ultimately diagnosed with ACS. It is well known that NOMI usually has a normal ECG or nonspecific ECG. The fact that not a single one of these patients had ACS shows that the population studied could not possibly support their conclusion. It should never have been published.
According to this data a triage ECG labeled
‘normal’ rules out the possibility of acute coronary occlusion.
This is obviously unreliable
data, as Dr. Smith’s Blog has published 51
cases of OMI with ECGs labeled ‘normal’, 35 of which were identified by the
Queen of Hearts – with 10 examples here.
We also studied 7 years of Code STEMI patients requiring emergent reperfusion,
and found that 4% presented with an ECG labeled ‘normal’, often confirmed by
the final blinded interpretation. This was just published in print in this month's Academic Emergency Medicine:
McLaren,
Meyers, Smith and Chartier. Emergency department Code STEMI patients with
initial electrocardiogram labeled ‘normal’ by computer interpretation: a 7-year
retrospective review. Acad Emerg Med 2024;31:296-300
Many
of these 'normal' ECGs had signs of OMI, and those that were identified
in real time by the treating emergency physician had faster
reperfusion than those that were missed. This study only included patients admitted as Code
STEMI, which
likely underestimates the false ‘normal’ rate because it doesn’t include
those admitted as ‘non-STEMI’ who had delayed reperfusion for OMI. So not interrupting the
physician, or physician reliance on a computer 'normal' ECG will lead to preventable delays to reperfusion that would
threaten
patient safety.
These three cases are from this
study, and this prior
post shows 4 more. For all cases, see the supplement from the online
version of the article.
Now let’s see how these patients
were managed in real time, and the patient outcome. These ECGs were not only
labeled normal by the computer but also the final blinded cardiology
interpretation—which according to some studies would designate these ECGs as
not clinically relevant. We can compare these interpretations with the actual patient outcome, and with the blinded interpretation of
the Queen of Hearts which is expert-trained to identify OMI.
Case 1:
There’s ST elevation in V1-2. The
large S wave in V1 may account for some of the ST elevation in this lead, and there is no reciprocal ST
depression in V6 (swirl pattern). But the convex ST elevation and bulky T wave
in V2 is disproportionate to voltage and indicates OMI until proven otherwise - either LAD or RCA.
The Queen calls this OMI with high confidence:
This was missed, and the patient
was only seen after the first troponin came back at 100 ngL (normal < 26 in
males and <16 in females), and a repeat ECG was done:
Some reperfusion T wave inversion not only in V2 but V1-3,
confirming OMI, but still doesn’t meet STEMI criteria. A stat
cardiology consult led to cath lab activation, with door-to-cath time of 202
minutes. Despite some reperfusion at the time of the repeat ECG, at the time of the angiogram there was 100% mid LAD occlusion, with peak troponin of 19,049 ng/L. Queen of Hearts could have reduced reperfusion delay by 2 hours for this 100%
LAD occlusion that was mislabeled ‘normal.’
The Queen of Hearts PM Cardio App is now available in the European Union (CE approved) the App Store and on Google Play. For Americans, you need to wait for the FDA. But in the meantime:
YOU HAVE THE OPPORTUNITY TO GET EARLY ACCESS TO THE PM Cardio AI BOT!! (THE PM CARDIO OMI AI APP)
If you want this bot to help you make the early diagnosis of OMI and save your patient and his/her myocardium, you can sign up to get an early beta version of the bot here. It is not yet available, but this is your way to get on the list.
Case 2:
There is a subtle biphasic T wave
in aVL, reciprocal to down/up tall T waves inferiorly, suggesting high lateral
reperfusion.
A truly normal or non-OMI ECG
would be labeled "not OMI, high confidence" but instead the Queen calls this
"OMI low confidence", suggesting the ECG is concerning but not yet diagnostic. The
emergency physician who was shown the ECG identified the same concerns and asked
for a repeat ECG, which was done 30 minutes later:
The reperfusion TWI in aVL is now upright (pseudonormalization) with reciprocal ST depression inferiorly. There is also ST elevation and hyperacute T waves V1-2 with reciprocal ST depression V5-6 (precordial swirl). Now the ECG is STEMI(+)OMI,
diagnostic of proximal LAD occlusion, and was identified by the computer. Cath lab was activated, with
door-to-cath time of 118 minutes. There was 95% proximal LAD occlusion, with
first troponin of 31 ng/L and peak of 11,894 ng/L. This infarct would have been much worse
if the physician had not been interrupted to interpret the initial ‘normal’ ECG,
and had not identified the subtle abnormalities.
Case 3:
There’s hyperacute T waves V2-4, with a small Q
in V3 and potentially terminal QRS distortion in V3 (at least by the third beat,
where there is no S wave), indicating LAD occlusion. The Queen calls this OMI with high confidence.
Fortunately this was also
identified by the emergency physician, who asked for a repeat ECG immediately:
Now there’s deWinter waves in
V3-4. Cath lab was activated, with door-to-cath time of only 44 minutes. First
troponin was 4ng/L which is normal and just above the limit of detection of 2.
But peak troponin was greater than 50,000 ng/L despite very rapid reperfusion. This
case could have been a disaster if the emergency physician had not been
interrupted to review the ECG or if they trusted the ‘normal’ interpretation,
and if they waited for and relied on the first troponin which was normal.
None of these were Normal!!
All were diagnostic of OMI!!
Do not pay attention to the conventional algorithm!
Take away
1.
ECGs labeled normal by the conventional computer
algorithm are unreliable, even if confirmed by the final blinded interpretation. The reliability of these ECGs should be based on patient outcome.
2.
Emergency physicians should be interrupted to
review all triage ECGs, even those that labeled ‘normal’, and should
look
beyond STEMI criteria for signs of OMI – including acute Q waves,
terminal QRS distortion, convex ST segments, hyperacute T waves, and
reciprocal change
3.
Expert-trained AI can accurately identify OMI
and lead to faster reperfusion