Emergency medicine questions fail in four distinct ways, and they map cleanly onto the physician tasks in the EM Model. You can miss a question because you did not stabilize a patient who needed stabilizing, because you did not reach the diagnosis, because you reached it and chose the wrong intervention, or because you managed the patient impeccably and then sent them to the wrong place. Those are four different failures with four different fixes, and the fourth is the one that residency trains least explicitly and the exam tests most distinctively.
Key takeaways
- Sort every error into one of four tasks: stabilization, diagnosis, treatment, or disposition.
- Stabilization precedes everything, and a critical patient does not wait for your workup.
- Treatment errors are usually sequencing errors: the right action offered at the wrong moment.
- Disposition is a distinct, examinable task and it is the one that defines emergency medicine.
- Tag by task rather than by organ system, because a task failure scatters across every specialty.
Task one: stabilization
The question is whether this patient needed something done before anything else, and whether you did it.
The failure is recognizable: you produced a workup for a patient who was dying. The vignette described a critical patient, you correctly identified what was wrong, and you selected an investigation, or a consultation, or a definitive treatment that takes time the patient does not have.
The discipline is a single question, asked before you look at the options: what is going to kill this person in the next few minutes, and does my chosen option address it?
Airway before breathing, breathing before circulation, and all of them before the diagnosis. A tension pneumothorax is decompressed, not imaged. Anaphylaxis gets epinephrine, not tryptase. Hypoglycemia gets glucose, not a workup.
If your errors cluster here, you are reasoning like an internist in a resuscitation bay, and no amount of additional emergency medicine content will fix it. What will fix it is practicing the question above until it fires automatically.
Task two: diagnosis
The ordinary knowledge gap: you did not recognize the entity, or could not interpret the finding, or missed the discriminator in the vignette.
The tell is straightforward. Could you have answered with what you already knew? If not, this is content, and it is the only one of the four bins that content review actually fixes.
Within emergency medicine, the diagnostic questions cluster around the presentations that must not be missed, and that is where the review should go: the chest pain that is not musculoskeletal, the headache that is not migraine, the back pain that is not mechanical, the abdominal pain in the patient whose presentation is atypical because of age or immunosuppression.
The exam is far more interested in whether you can distinguish the dangerous from the benign than in whether you can name the benign thing precisely.
Task three: treatment
You reached the diagnosis and chose the wrong intervention, and in emergency medicine this is almost always a sequencing failure rather than a knowledge failure.
The distractor is not a wrong treatment. It is a correct treatment offered at the wrong moment: the definitive management chosen when the question wanted the immediate one, the confirmatory test chosen when the patient needed treating, the specialty referral chosen when something had to happen now.
The test to apply: would waiting for this change what I do in the next ten minutes, and can this patient afford the wait? If the answer to the second question is no, the option is wrong however impeccable the medicine.
The review habit that fixes this takes a minute. For each treatment error, write the sentence: "My option would have been correct if..." and complete it precisely. "The CT would have been correct if she were hemodynamically stable." That sentence is the transferable rule, and the exam will test the same rule with a different organ.
Task four: disposition
Here is the task that defines emergency medicine and that candidates train least.
Every emergency patient leaves, and the question is where they go: home, to the observation unit, to the floor, to the ICU, to the operating room, or to another facility entirely. That decision is a clinical act, it is examinable, and it has correct answers.
The disposition question asks several things at once. Has the dangerous diagnosis been adequately excluded, or merely not yet found? Does this patient meet the criteria for discharge, and are there decision rules that govern it? What is their risk of deterioration, and over what timeframe? Do they have the social circumstances to be safe at home? What have they been told, and what will bring them back?
Candidates trained to diagnose and treat find this unsatisfying, because it is not a clinical intervention and it feels administrative. It is not administrative. It is where emergency physicians most commonly harm patients, and the exam knows it.
If your errors cluster in disposition, you have a specific, fixable, high-yield weakness, and it is worth naming because almost nobody trains it deliberately.
Do not forget observation and reassessment
A fifth category worth mentioning, because it sits between treatment and disposition and catches people.
Emergency medicine is a loop. You intervene, you reassess, and you adjust. A significant number of questions present a patient who has already been treated and ask what to do next, and the correct answer is frequently to recheck something rather than to escalate to another intervention.
If you find yourself always reaching for the next drug, ask whether the question is testing whether you know to look before you leap again.
Tag by task, not by system
The practical instruction that makes all of this actionable.
When you get a question wrong, record the task alongside the system. Then, every week or two, count them.
If your errors spread evenly across the four tasks, you have genuine content gaps and system-based review is correct.
If they pile up in one task, you have found a cross-cutting weakness that a system-based dashboard cannot see: a stabilization reflex that is not firing, a sequencing habit that is wrong, or a disposition instinct that is untrained. Each of those is one problem appearing across many specialties, and each is fixable once rather than a dozen times.
Where iatroX fits
iatroX's ABEM bank tracks performance by physician task as well as by condition, so a disposition or a sequencing weakness cannot disguise itself as a scatter of specialty gaps, which is exactly what a conventional dashboard does with it. Missed questions can be opened in the Socratic Tutor, which asks what must happen first and why before it explains, which is precisely the reasoning step that a mistimed or a diagnostically-reasoned answer skips, and the adaptive engine returns the same task in a different condition to test whether the fix transferred. Try it with free sample questions at iatroX. For the three-dimensional structure these tasks sit within, see building a study plan around the EM Model.
Frequently asked questions
How should I classify my ABEM errors? Into four tasks: stabilization, diagnosis, treatment, and disposition. Each has a different cause and a different fix, and only diagnosis errors respond to content review.
Why do I choose the wrong treatment when I know the diagnosis? Because emergency medicine treatment errors are usually sequencing errors. The distractor is a correct action offered at the wrong moment, so ask whether waiting would change the next ten minutes and whether the patient can afford the wait.
Is disposition really examined? Yes, and it is the task that defines the specialty. Where a patient goes, whether the dangerous diagnosis has been excluded rather than merely not found, and what return precautions they were given are all examinable with correct answers.
Why tag errors by task rather than by system? Because a task failure scatters across specialties and looks like several content weaknesses. A stabilization reflex that does not fire, or an untrained disposition instinct, will appear as errors in a dozen different systems when it is one fixable problem.