Most of us learned to read EKGs the same way: rate, rhythm, axis, intervals, then a hunt for ischemia. We eventually reach the point where the common tracings stop requiring conscious effort. That fluency is useful, and it is also where the trouble starts.

The errors that matter clinically are rarely the result of not knowing something. Ask almost any clinician to define Wellens syndrome and they will give you a reasonable answer. The miss happens at the bedside, at 3 a.m., on a busy shift, when the tracing is one of forty things competing for attention. The failure is one of process, not knowledge.

A few cognitive patterns show up again and again. We anchor on the first abnormality we see. We close prematurely once we have a diagnosis that fits. We skim past subtle findings because the obvious ones are reassuring. We read the tracing without reading the patient. And occasionally we run into an uncommon pattern that carries high risk and does not announce itself loudly.

What follows are ten mistakes I see repeatedly, in trainees and in seasoned attendings alike. None of them are exotic. That is exactly why they persist.

You spot left ventricular hypertrophy and your brain quietly files the tracing under "explained." The problem is that the secondary ST changes of LVH can mask, or mimic, genuine ischemia sitting right underneath them. The same thing happens with a paced rhythm, with bundle branch blocks, with anything that gives you a satisfying answer early.

This is anchoring, and it is probably the most common reading error there is. The first finding becomes the frame, and everything after it gets interpreted to fit.

The defense is boring and effective: read every EKG the same way, every time, all the way through, even when you already think you know the answer. Rate, rhythm, axis, intervals, P waves, QRS morphology, ST segments, T waves, then a deliberate look at the leads people skip, especially aVR and the limb leads in patients with chest pain. The systematic approach is not there to teach you something new. It is there to stop you from quitting early.

Posterior infarction is one of the most frequently delayed diagnoses in the field, and the reason is structural: the standard twelve-lead does not look directly at the posterior wall. So instead of ST elevation, you get the mirror image. Horizontal ST depression in V1 through V3, often with tall, broad R waves and upright T waves.

If you are scanning for elevation, depression in the right precordial leads reads as nonspecific, or as anterior subendocardial ischemia, and the occlusion gets missed. The fix is to treat that anterior ST depression as a question rather than an answer. Place posterior leads, V7 through V9, and look for the millimeter of elevation that confirms it. The threshold there is lower, 0.5 mm, because the signal has further to travel.

Posterior MI also travels with inferior and lateral infarcts, so when you see an inferior STEMI, the posterior wall deserves a deliberate look rather than an afterthought.

This one cuts both ways, and getting it wrong in either direction has consequences. Pericarditis classically gives you diffuse ST elevation that crosses vascular territories, which is the first clue, because a single occluded artery should not light up leads supplied by three different vessels. PR depression, particularly in the limb leads, and PR elevation in aVR, supports pericarditis. The ST elevation tends to be concave rather than the convex or tombstone morphology of occlusion.

But morphology is soft, and plenty of STEMIs start concave. The more reliable separator is reciprocal change. A true occlusion produces reciprocal ST depression in the opposing leads; pericarditis generally does not. And then there is the patient. A young person with pleuritic, positional chest pain that improves leaning forward is telling you something the tracing alone cannot. When the story and the EKG disagree, the story usually wins the tiebreak, but you treat the dangerous diagnosis until you have ruled it out, not the comfortable one.

Benign early repolarization is common, especially in younger patients, and it produces ST elevation that can rattle someone reading the tracing cold. The distinguishing features are worth knowing well because they keep people out of the cath lab unnecessarily.

Look for the notch or slur at the J point, most visible in the mid-precordial leads. The ST elevation is concave and tends to be modest, usually under a couple of millimeters, and it sits in leads with prominent, often slightly asymmetric T waves. The ratio of ST elevation to T wave height stays low. Critically, the pattern is stable. It does not evolve.

That last point is the practical one. Early repolarization looks the same on the EKG from two years ago, while ischemia moves. Which leads directly to the next mistake.

A single EKG is one frame of a film. Ischemia is dynamic, and a tracing that looks unremarkable at the moment of acquisition can change meaningfully over the next twenty minutes, particularly when the patient's symptoms are changing with it.

If someone has ongoing or waxing-and-waning chest pain and a first EKG that is not diagnostic, the answer is rarely to stop. It is to repeat it, and to repeat it again if the pain shifts. Comparison with a prior tracing is just as important. A "new" abnormality that turns out to be years old changes the entire interpretation, and an old EKG that proves a finding is truly new can be the thing that gets a patient to the lab.

The tracing tells you about one instant. The disease unfolds over time. Sample accordingly.

This is the error with the highest stakes on the list. Faced with a regular wide complex tachycardia, there is a pull toward the more benign explanation, SVT with aberrant conduction, especially when the patient looks deceptively stable. That instinct gets people hurt.

In an adult, particularly one with structural heart disease or prior infarction, the prior probability heavily favors ventricular tachycardia. Most regular wide complex tachycardias in adults are VT, and treating VT as SVT, particularly with AV nodal blocking agents, can have serious consequences. The morphologic criteria, the Brugada algorithm, AV dissociation, capture and fusion beats, are genuinely useful, but they are tools for confirming VT, not for talking yourself out of it.

The safe default is straightforward: in the adult patient, a wide complex tachycardia is VT until proven otherwise, and you manage it that way. Age and cardiac history should raise your suspicion, not lower it.

Every textbook draws the same progression: peaked T waves, then PR prolongation and P wave flattening, then QRS widening, then the sine wave and arrest. The trouble is that real patients have not read the textbook. Hyperkalemia does not reliably climb those rungs in order, and severe hyperkalemia can present with a bizarre, slow, wide rhythm that looks like nothing in particular and everything dangerous at once.

The clinical move is to think of potassium whenever you see a QRS that is wide without a clean explanation, a bradycardia that does not fit, or any strange tracing in a patient with renal disease, on dialysis, or on the medication combinations that drive potassium up. Peaked T waves are a helpful early flag when present, but their absence proves nothing. The EKG is a prompt to check a level and, when the rhythm is unstable, to treat empirically while you wait.

These two get grouped together because they share a theme: both point to critical proximal LAD disease, and both are easy to underread because the patient in front of you may look fine.

Wellens is the pattern of deeply inverted or biphasic T waves in V2 and V3, seen characteristically when the patient is pain-free after an episode of chest pain. That is the trap. The tracing during the pain-free window can look almost reassuring, and the T wave changes get dismissed as nonspecific. But Wellens marks a high-grade LAD lesion, and these patients are at real risk of a large anterior infarct in the days that follow. It is a reason to admit and pursue catheterization, not to reassure and discharge.

De Winter T waves are the active counterpart: upsloping ST depression at the J point in the precordial leads, with tall, symmetric T waves, often with slight ST elevation in aVR. There is no frank ST elevation, so it fails the STEMI screen, yet it represents acute LAD occlusion and should be treated with the same urgency. Both patterns are missed for the same reason; most of us are trained to look for ST elevation first, and neither pattern presents that way.

Every mistake above shares a root. The tracing gets interpreted in isolation, divorced from the person it came from.

The same millimeter of ST elevation means something completely different in a 70-year-old with crushing chest pain and a rising troponin than it does in an asymptomatic 25-year-old at a pre-op visit. Symptoms, risk factors, the trajectory of the troponin, and above all the prior EKG are not supplementary details. They are part of the read. A computer interpretation that says "normal" is a starting point, never the conclusion, and the algorithm is wrong often enough that overreliance on it is its own pitfall.

Read the patient and the tracing together. The EKG was never meant to be interpreted alone.

There is a quiet assumption that getting better at EKGs means memorizing more criteria. Criteria matter, but expertise does not actually work that way. Experienced readers are not running through checklists for the patterns they know well; they recognize them, the way you recognize a face, because they have seen the pattern hundreds of times.

That recognition is built, not memorized. It comes from repeated exposure to varied examples, from seeing the same diagnosis in its typical and atypical forms, and from active recall rather than passive review. You get better at de Winter T waves by seeing twenty de Winter tracings, not by rereading the definition.

That principle is what shaped the DUK-C EKG Interpretation Center. The idea was to organize common and high-yield EKGs into searchable categories, each with diagnostic criteria, clinical pearls, and quiz-based recall, so that building pattern recognition becomes something you can do deliberately rather than waiting to accumulate it by chance over years of practice. It is the same problem we keep coming back to in writing like Medicine Has a Retrieval Problem and The Hidden Cost of Clinical Information Friction: the knowledge is rarely the bottleneck, getting to it at the right moment is.

Good EKG interpretation is not about never being wrong. The patterns are subtle, the patients are atypical, and the tracing is only ever one slice of a moving picture. The goal is to know where the common traps are, maintain a systematic approach even when you think you already have the answer, and continue sharpening pattern recognition through deliberate practice.

If you want to work on that recognition in a structured way, the DUK-C EKG Interpretation Center is open through our Early Access program, and it pairs well with the companion piece, 10 EKG Patterns Every Medical Student and Resident Should Recognize. You can read more about the platform and the thinking behind it on our About page, or browse the rest of the writing on the blog.