Reviewing Questions With a Post-Mortem
Question review is where a question bank becomes a learning system.
The question itself creates retrieval and application.
The review tells you what the attempt revealed.
Students often spend most of their energy on the first part.
They complete a block.
Check the percentage.
Read explanations.
Make notes.
Move on.
That process can teach content.
It can leave the reasoning error intact.
The Post-Mortem step of the 5P Approach™ to Clinical Reasoning is designed to prevent that.
A Post-Mortem asks what happened inside the decision and what should change before the next similar case.
Start before the explanation
When a question is over, resist the urge to read the explanation immediately.
First capture your own process.
What did you think the case was about?
What was your leading diagnosis or management prediction?
Which answer did you choose?
Which other answer tempted you?
How confident were you?
This matters because the explanation changes your memory of the question.
Once the correct answer is visible, your original reasoning becomes harder to reconstruct accurately.
Capture the learner state before it disappears.
Review misses and uncertain correct answers
A missed question obviously deserves attention.
Some correct questions do too.
Review a correct item when:
- You guessed
- You were split between two choices
- The explanation surprised you
- You changed from one answer to another
- You used weak reasoning
- The question took much longer than expected
A correct answer produced by unstable reasoning can fail on the next variant.
The point of review is reproducibility.
Reconstruct the decision
Ask yourself to solve the case again without relying on the explanation.
What are the three highest-value findings?
What is the one-sentence problem representation?
What answer category should follow?
Why does your chosen option fit or fail?
This is the fastest way to see whether the error was in the knowledge or in the structure.
If you still cannot build the case after seeing the answer, the gap is probably deeper than a forgotten fact.
Locate the failure inside the 5P Approach™
The 5P Approach™ gives you a useful diagnostic map.
Prioritize failure
You saw the information but weighted the wrong clue.
Example:
You focused on a dramatic symptom and underweighted the postoperative status and hypoxemia.
Paraphrase failure
Your summary was too vague.
Example:
You reduced the case to “older patient with chest pain” and lost the exertional pattern, risk factors, and recurrence.
Prognose failure
You reached the answer choices without predicting what should fit.
Example:
The options built your differential instead of your clinical model guiding the options.
Pick failure
Your prediction was reasonable, but you selected an answer that did not match the clinical sequence.
Example:
You knew the diagnosis but chose a test that would be appropriate later rather than the next step now.
Post-Mortem failure
You read the explanation but never extracted a transferable lesson.
The same error therefore remains available to happen again.
Separate knowledge from reasoning
This distinction prevents inefficient remediation.
If you did not know that a medication causes a specific adverse effect, learn the fact.
If you knew the adverse effect but failed to connect it to the patient’s timeline, the problem is not primarily memory.
If you knew the diagnosis but selected the wrong next step, study the management threshold or sequence.
If you knew the correct answer and changed it because of an attractive distractor, examine the change behavior.
The treatment should match the error.
Ask why the distractor was attractive
Wrong answers are educational when you understand why they almost worked.
Did the distractor fit one clue but not the full pattern?
Was it the correct diagnosis at a different age?
Would it be appropriate if the patient were unstable?
Was it the right test after another test?
Did it represent a common complication rather than the current problem?
This teaches boundaries.
A diagnosis becomes easier to recognize when you understand both where it fits and where it stops fitting.
Change one feature of the vignette
This is one of the highest-yield Post-Mortem moves.
Ask:
What single change would make another option correct?
Change the age.
Change the time course.
Remove a risk factor.
Make the patient unstable.
Add a focal neurologic deficit.
Change the test result.
Move the patient to a different stage of management.
Now the question becomes a family of cases.
You are learning the relationship between the clue and the decision rather than memorizing one answer.
Use counterfactuals to build illness scripts
Suppose the original case is stable exertional angina.
What change would make acute coronary syndrome more likely?
What change would increase concern for aortic disease?
What change would make pulmonary embolism more plausible?
What test result would change the management pathway?
These counterfactual questions strengthen the borders among illness scripts.
That is exactly what you need when future questions change the wording.
Do not make a flashcard for every miss
A flashcard is appropriate when the missing element is retrievable information that benefits from spaced recall.
It is less useful when the error is:
- Poor clue weighting
- An unbounded differential
- Management sequence
- Timing
- Second-guessing
- Failure to build a problem representation
Those problems need practice, not another isolated fact card.
Your review system should be able to produce different interventions.
Keep the error log compact
An error log should reveal patterns.
It should not become a second textbook.
A useful entry can contain:
Topic
Pulmonary embolism.
Failure mode
Management sequence.
Missed signal
Postoperative high-risk context made D-dimer inappropriate.
Rule to carry forward
Testing strategy follows pretest probability and patient stability.
Changed-case prompt
What if the patient were low risk with a reassuring clinical assessment?
That is enough to make the lesson reusable.
Review the error log weekly
Individual questions are noisy.
Patterns across questions are more informative.
At the end of the week, ask:
Which failure mode appears most often?
Which diagnosis families remain unstable?
Where am I slow?
Which distractor type keeps working on me?
Am I making the same management-sequence error?
Which content gaps are foundational enough to justify deeper review?
Now the error log becomes a study-planning tool.
Retrieval practice still matters
The evidence supporting retrieval practice is strong.
Testing can improve retention and transfer, especially when learners retrieve actively and receive useful feedback.
A Post-Mortem builds on that advantage.
The question creates retrieval.
The review adds diagnostic information about the learner’s thinking.
The changed-case exercise adds transfer.
Those pieces work together.
Keep the Post-Mortem proportional
A review process can become so elaborate that the student stops completing questions.
Do not write an essay for every item.
Most questions need a brief review.
Reserve deeper analysis for:
- Repeated error types
- High-yield concepts
- Questions that exposed a major reasoning failure
- Difficult management thresholds
- Topics that repeatedly appear in assessments
The system should improve learning without making practice impossible to sustain.
Use the Post-Mortem after correct answers selectively
A correct, confident, efficient answer may need very little review.
Check the explanation for anything important you missed.
Then move on.
Deep review should follow educational value.
The goal is not to maximize minutes spent per question.
The goal is to extract the lesson that changes future performance.
Educators can use the same structure
When reviewing a question with a student, do not begin by explaining the answer.
Ask what they saw.
Ask for the problem representation.
Ask what they predicted.
Ask which alternative felt strongest.
Then locate the failure.
This makes the learner’s cognition visible.
The educator can correct the right process rather than deliver another mini-lecture that the student already understands.
The Post-Mortem closes the learning loop
DDQX uses the Exam Performance Flywheel:
Concept learning → Practice → Post-Mortem analysis → Pattern recognition → Improvement
The flywheel depends on the review step.
Practice without analysis can create exposure.
Analysis turns exposure into information about what should change.
That is why review quality can matter more than raw question count once a learner has a substantial knowledge base.
A six-question Post-Mortem
For a meaningful miss or uncertain correct answer, ask:
- What did I think the case was about?
- Which clue should have carried more weight?
- Where did the 5P sequence break?
- Why was my strongest distractor attractive?
- What single change would make that distractor correct?
- What will I do differently on the next case?
If you can answer those questions, you have extracted more than the answer.
You have extracted a reasoning lesson.
The standard
A question should leave you with a better model than you had before it.
Review the decision.
Name the failure mode.
Clarify the boundary between the correct answer and the distractor.
Change one feature.
Carry one rule forward.
Then return to practice.
That is how a question bank becomes a clinical reasoning laboratory rather than a completion counter.
Next step: Use the six-question Post-Mortem on your next five missed or uncertain items and look for the first repeated failure mode.
