Wednesday, February 24, 2010

Computer algorithms are not sensitive for STEMI

This 29 year old presented in February 2010 after drinking with friends. He started vomiting, then had onset of chest pain and bilateral arm numbness. Here is his ECG.

The computer read "Probable LVH" and "borderline ST elevation". It did not even suggest acute MI. But as you can see this is an obvious inferior STEMI, with reciprocal ST depression in I and aVL, and even ST depression in V6. Angiography confirmed an RCA occlusion.

If the computer can miss one so obvious, it can (and I know it does) miss many less obvious.

Studies from the 1990's show computer algorithms of 12-lead ECGs are 60-80% sensitive for coronary occlusion. I know of no study that shows they are any better today. 80-lead systems appear to have better overall sensitivity, but with some loss of specificity.


  1. The relatively low sensitivity of computer algorithms is well documented, but I'm actually surprised it missed this one.

    Are you using the GE-Marquette 12SL interpretive algorithm? I've never seen it miss obvious inferior ST-elevation with reciprocal changes before.

    Forgetting about the limb leads for a moment, the precordial leads are interesting on this ECG!

    Can the T-waves in the right precordial leads be explained by LVH with strain pattern or benign early repolarization?



  2. The algorithm is the Philips 080A. I'm not sure how it compares with Marquette 12SL, but our medics use Marquette and I find it just as bad. I have had several similar inferior MIs that were not diagnosed by the Marquette.

    The T-wave do look worrisome, but they are just early repol (good R-wave amplitude and short QT in the precordial leads).

  3. This is a bit off point, but wouldn't one also suspect a Right ventricular infarct based on the STE in V1, and also the STE in III being greater thean the STE in II?

  4. Call me curious, but beyond the ECG, did this patient have major risk factors? Any history, drug use?

  5. 1) yes to RVMI, except that there is such large STE from early repol that the STE in V1 is not unexpected. So, in fact, no RV MI.
    2) III > II is normal for RCA (vs. LCX) occlusion. This only hints at RV because an RV infarct always comes from RCA infarct.

  6. The patient had no risk factors and was, in fact, an athlete. We have had 2 other such patients in their twenties in the last month (24 F with STEMI due to proximal LAD and 29 M with NSTEMI due to 99% LAD)

    This is less unusual than most think. That's why we do EKGs even on 20-somethings with chest pain.

  7. Can I say hyperacute t wave in v2 &V3


DEAR READER: We welcome your Comments! Unfortunately — due to a recent marked increase in SPAM — we have had to restrict commenting to Users with a GOOGLE Account. If you do not yet have a Google account — it should not take long to register. Comments give US feedback on how well Dr. Smith’s ECG Blog is addressing your needs — and they help to clarify concepts of interest to all readers. THANK YOU for your continued support!

Recommended Resources