This workflow is for FFICM candidates using QMedica, a dedicated FFICM question bank, to prepare for the written MCQ around clinical work. It gives a first-pass structure, an error-review method and exit criteria that do not depend on reaching 100% completion. The principal limitation is transparency: at the last check, QMedica's live question count, price and features were not published in a form we could verify, so treat vendor claims cautiously and confirm before you buy. It does not reproduce the OSCE or SOE.
What QMedica offers for FFICM right now
| Item | What we found (last checked 20 July 2026) |
|---|---|
| Product type | A dedicated FFICM exam question bank (qmedica.co.uk) |
| Question volume | Not published in a form we could verify at the last check — confirm the live count on qmedica.co.uk |
| Content type | FFICM-style single-best-answer questions |
| FFICM components supported | The MCQ knowledge layer only; no OSCE or SOE simulation |
| Adaptive / AI features | None confirmed at the last check |
| Price / access period | Not confirmed — verify on qmedica.co.uk |
Lead with the honest finding: on 20 July 2026 QMedica's public pages did not expose a verifiable question count, price or feature list — the site returned little beyond basic metadata. That does not mean the bank is weak; it means you should treat any number in marketing as vendor-reported and confirm the live figures, price and access length directly with the vendor before committing. The workflow below is deliberately robust to an unknown bank size: it is built on coverage and unseen performance, not on finishing a fixed number of questions.
The exam you are actually preparing for
The Final FFICM has three components and the MCQ gates the rest; the MCQ pass lasts three years. The MCQ is 130 single-best-answer questions in three hours — 80 short one-mark items and 50 longer two-mark scenarios, maximum 180 marks — and has been entirely single-best-answer since June 2022, so MTF-era material is off-format. The OSCE is 13 seven-minute stations across Data, Equipment, Professionalism and Resuscitation. The SOE is four 14-minute stations from eight questions, double-marked. QMedica and iatroX both operate at the MCQ knowledge layer; neither is a substitute for the live OSCE or SOE. Judge QMedica's coverage against the FICM curriculum and official sample material, not against its own marketing.
Create a blueprint inventory and protect a pool of unseen questions
Before drilling, map QMedica's topics onto the FICM curriculum domains and note where the bank looks thin. This inventory is your coverage map; it tells you what the bank cannot teach you, which is as important as what it can. Then ring-fence a protected pool of unseen questions — and any full block you plan to use as a mock — and do not touch it during ordinary study. Unseen items are assessment assets; once seen, they only measure memory.
First pass: topic blocks only where foundations are weak
Do not default to topic-filtered blocks across the whole bank. Use topic blocks only where your foundations are genuinely weak, because a filtered block quietly cues the answer — you know the theme, so you guess within it. Everywhere your foundations are reasonable, start mixed from the outset to avoid that topic cueing. The rule of thumb: filter to learn a weak area, mix to test everything else.
Review each miss with an error code and one corrective action
Reviewing questions by re-reading the whole explanation is slow and teaches little. Instead, assign every miss a short error code — knowledge gap, misread stem, wrong discrimination, pacing, or careless — and write one corrective action per miss. Do not transcribe the explanation; capture the single rule you will re-test. A page of coded errors with one action each is worth more than a notebook of copied explanations, because it converts each miss into a specific, retrievable target.
A worked miss, coded and corrected
Suppose you miss a QMedica item on vasopressor choice in septic shock. The lazy response is to re-read the whole explanation and move on. The coded response is faster and sticks better. First, label the error: was it a knowledge gap (you did not know the first-line agent), a misread stem (you missed that the patient was already receiving it), a discrimination failure (you could not separate the two plausible agents), or pacing (you rushed)? Say it was a discrimination failure. Your single corrective action is one prompt — "first-line versus second-line vasopressor, and the trigger to add the second" — with drug specifics checked against the SmPC/eMC rather than a secondary summary. You do not transcribe the explanation. Two days later you re-test that prompt from memory and answer a fresh item on the same principle from a different source. Get the transfer item right and the rule has stuck; get it wrong and the prompt was poorly framed, so you rewrite it. One miss becomes one code, one action and one transfer check — and that is the whole loop.
Use transfer practice before repeating an item
When you get something wrong, do not immediately re-answer the same QMedica question — you will simply learn that item. Instead, answer a new item that tests the same principle, ideally from a different source, before you revisit the original. Transfer to a fresh stem is the thing the exam actually rewards, and it is the only honest test that the underlying rule has stuck rather than the specific question.
Switch to mixed timed blocks when domain floors are met
You do not need to finish the bank before going mixed. Once a domain's first-attempt accuracy clears your floor, fold it into mixed timed blocks even if the first pass is incomplete. Completion is not the trigger; stable domain performance is. Running mixed timed blocks early also trains pacing near 80 seconds per mark, which pure topic practice never does.
Exit criteria
Stop new-question study and move to predominantly mock practice when the measurable conditions are met: adequate blueprint coverage, stable unseen timed performance, correct pacing, retention across a spaced gap, and calibration against the official FICM sample material. Notice that "100% completion" is not on the list — a fully completed bank with unstable unseen accuracy is not ready; a partially completed bank with stable unseen accuracy and good pacing is closer than it looks.
A seven-day plan around clinical work
Use QMedica for one defined job — first-pass coverage and coded error review — and iatroX for unseen, timed transfer measurement. This is a routine, not a claim about any proprietary algorithm.
| Day | Core task (around clinical duties) |
|---|---|
| 1 | Blueprint inventory; map QMedica topics to FICM domains; ring-fence a protected pool |
| 2 | Topic block on the weakest domain; code every miss with one corrective action |
| 3 | Mixed timed block (40–50 questions) on stronger domains; log pacing |
| 4 | 40-question unseen block on iatroX; classify errors by code |
| 5 | Transfer practice on Day 4 errors — new items on the same principles, not repeats |
| 6 | Weekly mixed timed block (45–60 questions), exam pace, no filter |
| 7 | Review by error code; check exit criteria per domain; adjust next week's filters |
Decision checklist: continue, supplement, switch or stop
Continue with QMedica if its coverage matches the blueprint and your unseen accuracy is rising. Supplement with a second unseen source if the blueprint inventory reveals thin domains, or simply because you could not verify the bank's size and want assured volume for measurement. Switch from topic blocks to mixed timed blocks as each domain clears its floor. Stop any strand that has stopped changing your error pattern; repeating a bank you can recognise is sunk-cost study.
Three mistakes this workflow is designed to stop
The first is chasing completion — treating "questions remaining" as the metric and finishing the bank regardless of whether unseen accuracy has stabilised; coverage and transfer are the metrics, not the progress bar. The second is filtering everything by topic — comfortable and quietly corrosive, because a filtered block hands you the theme and lets you guess within it, so your accuracy flatters you; mix by default and filter only to shore up a genuinely weak foundation. The third is transcribing explanations — copying long answers into notes feels like work but produces a document you never re-test, whereas a coded error with one corrective action is smaller, uglier and far more useful. With an unverified bank size, these disciplines matter more, not less.
Frequently asked questions
Is QMedica enough for FFICM on its own? Possibly for a large share of the MCQ if its coverage and volume are adequate — but that is exactly what could not be verified at the last check, so do not assume it. No single bank covers the OSCE and SOE, and prudent candidates measure their knowledge on at least one unseen second source rather than trusting one bank's percentage.
Which FFICM component does QMedica not reproduce well? The OSCE and SOE. As a written question bank it addresses the MCQ knowledge layer only; it does not simulate the 13-station OSCE or the examiner-led structured oral. Rehearse those with colleagues, a course or simulation.
How many QMedica questions should I complete per day for FFICM? A workable figure for trainees revising around clinical work is roughly 40–60 well-reviewed questions per study day, but with an unverified bank size, prioritise coverage and coded review over raw throughput. It is better to review 30 questions into corrective actions than to skim 100.
When should I stop using QMedica and move to mixed mocks? When your weak domains have cleared their accuracy floors and your pacing is near target, make mixed timed mocks the main activity — you do not need to finish the bank first. Keep QMedica for targeted first-pass coverage of any remaining thin domains.
How should I combine QMedica with iatroX without duplicating practice? Give them distinct jobs: QMedica for first-pass coverage and error review, iatroX for the unseen, timed blocks that measure transfer. Never run the same items through both — that is the two-Q-bank rule, one bank teaches and the other measures. Reading why a Q-bank percentage is not an exam score will keep both platforms' numbers in proportion.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. At that date QMedica's public pages did not expose a verifiable question count, price or feature list; any figure you encounter should be treated as vendor-reported and confirmed on qmedica.co.uk before you buy. Disclosure: iatroX operates a competing FFICM MCQ bank; this article confines the iatroX role to unseen, timed measurement — a job QMedica does not claim — and does not present iatroX as an OSCE or SOE product. Corrections are welcome via the feedback route on iatrox.com. References: the Faculty of Intensive Care Medicine (ficm.ac.uk) FFICM Final MCQ, OSCE and SOE pages, curriculum and regulations; qmedica.co.uk; iatroX's FFICM bank, the comparison hub, the Q-bank-percentage explainer and the FFICM article on what MCQ banks cannot prepare you for.