Reading an ECG: learn a method, not a pattern

Pattern recognition fails on the tracings that matter most. A repeatable eight-step method you can apply to every ECG you are handed.

Almost everyone learns ECGs backwards. You are shown a wall of classic tracings, you memorise their shapes, and then you are handed a real one from a real patient at two in the morning and none of the shapes quite fit. Pattern recognition works beautifully on the tracings you have already seen and fails on the ones that matter.

The alternative is a method. It is slower for the first fifty tracings and faster forever afterwards, because it does not depend on having met this particular abnormality before.

The eight steps

• Check the details: right patient, right date, correct calibration and paper speed.
• Rate. Count it rather than estimating, at least until estimating becomes reliable.
• Rhythm. Is it regular, and is there a P wave before every QRS and a QRS after every P?
• Axis. Leads I and aVF answer this in a couple of seconds.
• P wave morphology and the PR interval.
• QRS: width, height, and any pathological Q waves.
• ST segments and T waves, compared against the baseline rather than against your expectation.
• QT interval, corrected for rate.

Why the order matters

The sequence is deliberate. Rate and rhythm change what counts as normal for everything that follows, and QRS width changes how you are permitted to interpret ST segments at all. Reading ST changes before you have established the QRS width is the single most common way to talk yourself into a diagnosis that is not there.

Compare with the old tracing, every time

A borderline finding on a single ECG is an unanswered question. The same finding next to a tracing from six months ago is either reassuring or urgent. Asking for the previous ECG is the cheapest diagnostic test available on any ward, and it is skipped far more often than it should be.

Say what you see before you say what it is

When you present a tracing, describe it before you label it. "Sinus rhythm at ninety-eight, normal axis, two millimetre ST elevation in the inferior leads with reciprocal change in aVL" is a finding a senior colleague can act on and, if necessary, correct. "This looks like an inferior MI" is a conclusion with the evidence hidden inside it.

Practise on the boring ones

The habit is built on normal tracings, not dramatic ones. Run the full method on every ECG you touch, including the obviously unremarkable ones, until the sequence takes fifteen seconds and runs whether you are paying attention or not. That is the point at which it will hold up at two in the morning.

One thing the method does not replace: an ECG is interpreted in a patient, not in isolation. A tracing that looks unremarkable in a patient who is unwell in front of you deserves a repeat, not reassurance.