Medical Simulation vs Question Banks

Ask a group of final-year students how they are preparing and most will name a question bank. Ask the same group what they are worried about and most will describe something a question bank cannot rehearse: being the first person to see someone who is genuinely unwell, with incomplete information and no answer options on the screen.

Both observations are correct. Question banks are extremely good at one thing and structurally incapable of another. Knowing which is which turns a vague sense that revision is not quite working into a concrete change to how you spend your evenings.

If you want the second format rather than more of the first, try an evolving clinical case.

What question banks are genuinely good at

Retrieval practice is one of the better-evidenced findings in learning research. Repeatedly pulling information out of memory produces stronger retention than repeatedly putting it in, which is why answering questions beats re-reading notes by a wide margin.

Question banks operationalise that well. They give you high-volume retrieval, immediate feedback, and coverage you would not achieve by chance. They also efficiently surface the specific thing you do not know, which is difficult to discover from a textbook because you tend to skim what feels familiar.

And they are honest about what they measure. A well-written question tests whether you can recognise a condition from a described presentation and select the appropriate next step from a short list. That is a real and necessary skill. It is also, and this is the important part, a recognition task.

What they structurally cannot train

Four things are missing from the format, and no amount of additional questions supplies them.

• Information gathering. The question hands you the relevant findings, pre-selected and pre-summarised. At the bedside, deciding what to ask and what to examine is most of the work, and the findings arrive mixed with irrelevant detail.
• Time and change. A question is a static snapshot with a single correct action. A patient is a trajectory, and the right action at minute two is frequently wrong at minute twenty.
• Commitment under uncertainty. Choosing between five options that someone else generated is not the same cognitive act as generating your own options and committing to one when you are not sure. Recognition is far easier than recall, and multiple choice tests the easier of the two.
• Consequence. Getting a question wrong costs you a percentage point. Getting a decision wrong in a case changes the patient's state, and having to manage the situation you created is where the lesson sticks.

None of that is a criticism of question banks. It is a description of a format boundary. The failure mode is using a recognition tool to prepare for a decision task and then being surprised when the transfer is poor.

What simulation adds, and what it costs

Simulation — in the broad sense of rehearsing a clinical situation, whether in a skills centre, around a table, or on a screen — supplies exactly the four missing elements. Reviews of technology-enhanced simulation consistently find benefits over no intervention, with the largest effects where practice is deliberate, repeated, and followed by feedback.

That last clause is the whole condition. Simulation without a debrief is an anecdote. The evidence base rewards deliberate practice with feedback, not the sophistication of the equipment, which is why a well-run tabletop case with a good discussion afterwards outperforms an expensive manikin used once and not discussed.

The costs are real too. Simulation is slower per unit of content, it usually needs at least one other person, and it is easy to do badly — a case run as a quiz, or a debrief that becomes a critique, teaches very little and can make people less willing to expose uncertainty next time.

A split that works

The two formats are complements, and the ratio should follow what you are short of rather than what you enjoy.

A workable default for someone with an exam in the near future: keep the question bank as the backbone for coverage and retention, and add two or three cases a week for decision practice. Roughly eighty per cent recognition, twenty per cent decision-making.

For someone about to start clinical work, invert the emphasis. Coverage matters less than being able to run a structured assessment, notice deterioration, and escalate clearly. That is decision practice, and it should be the majority of the time you have.

Three rules make either split work:

• Commit before you check. In both formats. Write the answer down, or say the action out loud, before revealing anything. Uncommitted review feels productive and teaches almost nothing.
• Log the reason, not the topic. "Got the arrhythmia question wrong" is not useful. "Did not check the QRS width before reading the ST segments" is a rule you can apply on Monday.
• Debrief the cases, always. Even alone, even briefly. Five questions and four minutes is enough to convert a case into something transferable.

Choosing between them on a given evening

A short decision rule for when you have thirty minutes and no strong preference. If you cannot currently name three conditions you would struggle to recognise, do questions — your coverage has gaps you have not found yet. If you can name them but you froze the last time something urgent happened in front of you, do a case.

And if you are tired, do questions. Decision practice under fatigue tends to produce frustration rather than learning, whereas retrieval practice survives a low-energy evening reasonably well.