Every time I’ve defended Southern California’s use of computerized interpretive algorithms in their STEMI system I’ve taken some flack for it.
I agree that in a perfect world, paramedics would receive extensive 12 lead ECG training as part of their core education (including how to identify the STEMI mimics and how to identify STEMIÂ in the presence of theÂ baseline abnormalities).
Unfortunately, the vast majority of paramedics in the country are not receiving this level of training in school.
That being the case, if you want to regionalize STEMI care, then there are only three options.
- Provide specialized training to paramedics (in my opinion it should be at least 24 hours) and allow them to bypass non-PCI hospitals based on their own interpretation.
- Provide less specialized training and use a computerized interpretive algorithm (i.e., when the computer says ***ACUTE MI SUSPECTED*** and the paramedic agrees the patient is taken to a PCI hospital).
- Provide less specialized training and transmit the ECG off-site for physician evaluation.
None of these solutions is perfect and some locations are using a combination of methods. I applaud Southern California for building a functional regional STEMI system. Is it perfect? No. Is there room for improvement? Of course.
It’s easy to criticize.
While it’s true that the GE-Marquette 12SL interpretive algorithm has a high specificity when it gives the ***ACUTE MI SUSPECTED*** message or ***ACUTE MI***, it’s not perfect. In fact, Southern California has a problem with false positives ECGs.
(The Philips MRx uses a different interpretive algorithm but the Physio-Control LP12 and ZOLL M and E series use the GE-Marquette 12SL interpretive algorithm, as do the majority of systems inside the hospital. To my knowledge, one is not been proven superior to the other.)
Since some EMS systems are using the interpretive algorithms to influence whether or not patients are taken to PCI hospitals, I thought I would devote some time to discussing how to get the most out of them.
Interpretive algorithms are a tool like any other. They have limits and they require understanding. For example, you can get a false positive from the computer when you capture an ECG with poor data quality (which is one of the reasons I spend a lot of time talking about data quality).
Capturing a 12 lead ECG with good data quality is a sign of professionalism. Conversely, handing over (or transmitting) an ECG to the hospital with poor data quality shows a lack of professionalism.
There have been a couple of times in my career that no matter what I did, I couldn’t get a good tracing (Parkinson’s disease, bad electrodes, broken leads, combative patient) but it’s rare.
In addition to poor data quality, sometimes tachycardias can fool the interpretive algorithm, especially atrial flutter.
The specificity of the computerized interpretive algorithm is maximized when you capture the ECG with excellent data quality, the chief complaint is chest pain, and the heart rate is less than 100.
Even with these caveats, you will occasionally run into false positives.
Consider the following ECG.
The computer is giving the ***ACUTE MI SUSPECTED*** message. Do you see why?