This audit is for emergency registrars across Australia and New Zealand using ACEMCQ for the ACEM Fellowship written examination — the Short Answer Question and Single/Select Choice Question papers, not the clinical OSCE. ACEMCQ offers a large written-format bank with domain analytics and AI-marked short-answer feedback. Its principal limitation is that those analytics optimise a visible, rising domain average rather than the blueprint breadth and unseen transfer that actually decide the paper.
A precision point about the title first. ACEMCQ's public materials describe progress tracking by domain, weak-area identification, timed mocks and AI-marked SAQ feedback — but not a verified adaptive-difficulty algorithm that reorders items by your measured ability. So the "engine" doing the optimising is really a reporting layer plus your own decision to chase red bars. That is not a criticism, but it changes what the numbers can mean, and it is the thread that runs through this audit.
What ACEMCQ offers for the ACEM Fellowship right now
The figures below are vendor-reported and last checked on 20 July 2026. Treat them as a starting point and confirm the current numbers on the product page, because they change.
| Item | What ACEMCQ states (vendor-reported, 20 July 2026) |
|---|---|
| Written question volume | "2000+ MCQs" and "700+ SAQs" in the ACEM Fellowship written format. Confirm the live totals before you pay. |
| Curriculum scope | "23 topic domains", advertised as "100% ACEM curriculum". Treat "100%" as a marketing claim to test against your own coverage log, not a guarantee of even sampling. |
| Analytics and "adaptive" features | Progress tracking by domain, weak-area identification, and timed mock exams. AI-marked short-answer feedback gives a mark-by-mark breakdown against model answers (Basic tier limited to roughly three marked SAQs per day; Premium unlimited). The public materials do not describe a verified adaptive-difficulty algorithm. |
| Access and pricing (AUD) | Basic listed at $29/month or $249/year; Premium at $49/month or $499/year; "cancel anytime". Confirm the current tiers and currency at checkout. |
| Components supported | The written papers (SAQ and SCQ/MCQ). No OSCE simulation, and it is not the examination itself. |
The one word doing heavy lifting in searches is "adaptive". ACEMCQ sells analytics and AI-marked feedback, not a machine that silently tunes item difficulty to your ability. That distinction matters because it puts you, not an algorithm, in charge of coverage — which is good news only if you audit yourself deliberately.
The exam you are actually training for
The ACEM Fellowship written examination is two 180-minute papers, six hours in total: a Short Answer Question paper and a Single/Select Choice Question paper of multiple-choice items. The clinical OSCE is a separate hurdle — up to twelve stations, each eleven minutes (four minutes reading, seven minutes assessment), 132 minutes of assessed time spread over two consecutive days, with station types spanning history, examination, communication, resuscitation discussion, teaching and case-based discussion. ACEMCQ addresses the written papers only.
The written papers sample the FACEM Curriculum Framework, which is built around eight domains — Medical Expertise at the core, then Prioritisation and Decision Making, Communication, Teamwork and Collaboration, Leadership and Management, Health Advocacy, Scholarship and Teaching, and Professionalism. The written exam leans heavily on Medical Expertise and Prioritisation and Decision Making; much of Communication, Teamwork and Leadership is assessed in the OSCE. Crucially, ACEM does not publish fixed percentage weightings per clinical topic the way some colleges do, so treat any third-party "blueprint" as an interpretation and audit your own coverage against it. For the fuller treatment, see the ACEM Fellowship content-gap checklist.
A word on where iatroX fits, because it must be exact. iatroX covers the ACEM Primary examination, not the Fellowship. It is a foundation-knowledge and unseen single-best-answer bank pitched at Primary-level anatomy, physiology, pharmacology and pathology — the basic-science floor the Fellowship then assumes. It is not a Fellowship SAQ, SCQ or OSCE product, and nothing here implies otherwise. Its role in a Fellowship stack is narrow: a clean, unseen way to check that the foundational science under your clinical reasoning is still solid.
Every metric on the dashboard, defined
A number you cannot define is a number you cannot act on. Before you trust ACEMCQ's analytics, pin down what each metric is:
- First-attempt accuracy — the proportion correct the first time you ever saw an item. This is the only accuracy figure that approximates unseen performance.
- Repeat accuracy — accuracy on items you have answered before. It rises with familiarity and mostly measures recognition, not exam-day reasoning.
- Percentile — your rank against other ACEMCQ users, not against the ACEM standard. The user pool is self-selected and its size is undisclosed, so read percentiles as motivational, not diagnostic.
- Predicted score — if any predicted or "readiness" figure is shown, treat it with suspicion. No third-party bank knows the live standard-setting for your diet, and this article will not offer a pass prediction.
- Coverage — how much of a domain's item pool you have attempted. High coverage of a bank is not coverage of the blueprint; it is coverage of that bank's sample of it.
- Difficulty — usually the cohort's success rate on an item, not an intrinsic property. A low-difficulty tag means "most users got it", which is not the same as "you will".
- Time per item — seconds elapsed. On the SCQ paper this is the metric that quietly fails candidates, so watch it as closely as accuracy.
Selection bias: why a weak-area feed inflates your average
Here is the structural trap. If you follow ACEMCQ's weak-area prompts, you spend disproportionate time on your softest domains — which is efficient for depth but distorts every headline percentage. As those weak domains improve, your overall average climbs, and it is tempting to read that climb as readiness. It is not. Your average rose partly because you concentrated on beatable terrain, and partly because repeat exposure lifted recognition. Neither is what the exam measures.
The consequence is that a raw ACEMCQ percentage is not comparable with your score on a fresh, mixed, unseen block. The only figure that transfers is first-attempt accuracy on items you have never met, sampled across the whole blueprint rather than the domains the feed kept surfacing. Everything else is context.
Blueprint audit: attempted distribution, not the home-screen average
Once a week, ignore the home-screen average and pull the distribution instead: how many first-attempt items have you actually done in each domain? Lay ACEMCQ's 23 topic domains against the clinical breadth the written papers sample — resuscitation and critical care, cardiovascular, respiratory, neurology, gastroenterology, renal and genitourinary, endocrine and metabolic, haematology and oncology, infectious diseases and sepsis, toxicology and toxinology, environmental, trauma, orthopaedics, ENT and ophthalmology, dermatology, obstetrics and gynaecology, paediatric emergencies, psychiatry and behavioural, pain and procedural sedation, imaging and ECG interpretation, and administration, ethics and medico-legal.
You will almost always find two or three cells with a handful of attempts while your favourite domains carry hundreds. That imbalance — not your average — is your revision plan. A domain you have barely sampled cannot have a trustworthy accuracy figure, however green the bar looks. This is the completion-is-not-coverage problem in miniature.
The readiness test the dashboard cannot fake
A credible readiness signal needs five conditions at once: the items are unseen (not repeats), the block is timed at exam pace, the content is mixed and randomised across domains, you use no assistance (no notes, no lookups), and the sample is large enough to be stable — think a few hundred first-attempt items, not thirty. ACEMCQ's timed mock exams satisfy several of these if you sit them cold and do not peek. Any block that fails one of the five conditions is practice, not measurement — valuable, but not evidence you are ready.
Override rules: what the feed will under-serve
Because you, not an algorithm, control coverage, you must deliberately force the material a self-selected feed under-samples. Override towards a domain whenever you have attempted fewer than a set floor of first-attempt items there, and specifically protect: low-volume but high-stakes topics such as toxinology and Australasian envenomation; paediatric resuscitation and weight-based dosing; ethics, consent, capacity and open disclosure; calculation-heavy items (infusions, corrected values, paediatric doses); and image and ECG interpretation, where a text-only revision habit leaves you slow. If ACEMCQ's mix will not surface these often enough, filter to them directly.
Worked example: turning a dashboard into next week's quotas
Suppose your ACEMCQ analytics read like this after four weeks: cardiology 210 first-attempt items at 78 per cent; toxicology 22 items at 55 per cent; paediatrics 31 items at 61 per cent; obstetrics and gynaecology 14 items at 64 per cent; imaging/ECG 40 items at 58 per cent; overall average a comforting 72 per cent. The average is the least useful number on the screen. The action items are the thin, weak cells.
Convert that directly into quotas rather than a vague intention to "do more tox". Next week: 40 first-attempt toxicology items, 40 paediatrics, 30 obstetrics and gynaecology, and 30 imaging/ECG, each in timed mode with feedback hidden until the block closes. Leave cardiology alone beyond a small maintenance block; it is not where marks are hiding. Notice there is no predicted pass here — just a measured gap turned into a countable task. Re-pull the distribution the following week and let the new thin cells set the next quota.
A seven-day pattern around clinical shifts
Here is a realistic week for a registrar working clinically, using ACEMCQ for one defined job — written-format depth and SAQ marking practice — and iatroX for a separate job, an unseen check on foundational knowledge. No claim is made about either platform's internal routing.
- Day 1 (post-nights): 30 mixed timed SCQ items on ACEMCQ; code every miss by cause (knowledge gap, misread stem, calculation, pacing).
- Day 2: Override to your thinnest domain from the weekly distribution; two AI-marked SAQs, reading the mark-by-mark breakdown against the model answer rather than your gut.
- Day 3: A short iatroX block of unseen Primary-level foundation items in the same weak system — for example the physiology and pharmacology beneath your shaky toxicology — to confirm the basic-science floor is solid.
- Day 4 (clinical): No new bank work; convert one real presentation from your shift into a structured SAQ-style answer from memory, then check it.
- Day 5: 40 mixed timed SCQ items; tally the error codes for the week.
- Day 6: One full timed ACEMCQ mock, cold, no lookups; treat the result as measurement, not morale.
- Day 7: Rest or light review of the mock's misses, sorting them into "re-read the source" versus "drill a new transfer item".
Decision checklist: continue, supplement, switch or stop
- Continue with ACEMCQ if your first-attempt accuracy on unseen mixed mocks is climbing and your domain distribution is filling out evenly.
- Supplement if one component is starving — most commonly SAQ technique or OSCE preparation, neither of which a written bank fully builds — by adding examiner-style SAQ marking, an OSCE course or a study group.
- Switch or add a second source if you are hitting near-duplicate stems, your first-attempt pool is drying up, or coverage of a domain has plateaued below your target. Add a second bank on the two-Q-bank rule so the new bank measures rather than duplicates.
- Stop grinding new items when every domain is above floor, your unseen accuracy is stable rather than still climbing, and your pacing is comfortable — at that point mixed mocks and SAQ timing, not more questions, are the work.
The bottom line
ACEMCQ is a strong, well-scoped written-format bank for the ACEM Fellowship, and its AI-marked SAQ feedback is a genuinely useful feature that many free resources cannot match. What it optimises, though, is a visible domain average and your engagement with weak areas — not blueprint breadth, not unseen transfer, and not the OSCE. Read the distribution, not the average; protect the low-volume domains the feed will skip; and keep one clean, unseen measurement channel that ACEMCQ's own numbers cannot flatter.
Frequently asked questions
Is ACEMCQ enough for ACEM Fellowship on its own? For the written papers it is a credible backbone, but "enough" depends on the OSCE and on your technique. ACEMCQ builds SCQ recall and gives structured SAQ marking, yet it does not simulate the clinical OSCE at all, and no single bank guarantees even coverage of the whole FACEM curriculum. Most candidates pair it with an OSCE preparation route, examiner-style SAQ practice, and a separate unseen measurement channel. Treat ACEMCQ as the largest single component of a written-exam stack, not the entire preparation.
Which ACEM Fellowship component does ACEMCQ not reproduce well? The OSCE, unambiguously. ACEMCQ is a written-format product covering SAQ and SCQ material; it does not stage the twelve-station clinical examination, examiner interaction, timed reading-then-assessment structure, or the communication and resuscitation-discussion stations that decide the OSCE. Its AI-marked SAQ feedback also approximates, rather than replaces, calibration against a live human examiner rubric, so read automated marks as a guide and pressure-test them against real examiner feedback where you can.
How many ACEMCQ questions should I complete per day for ACEM Fellowship? There is no universal number, and daily volume is the wrong target. A sustainable pattern for a working registrar is one or two timed blocks of 30 to 40 first-attempt SCQ items plus one or two marked SAQs, prioritising your under-sampled domains over raw throughput. Doing 150 items in a domain you already know well is busywork; doing 40 in a domain you have barely touched is progress. Let your weekly coverage distribution, not a daily quota, set the pace.
When should I stop using ACEMCQ and move to mixed mocks? Move to predominantly mixed, timed mocks when three things are simultaneously true: every domain in your distribution is above your first-attempt floor, your unseen accuracy has stabilised rather than still climbing steeply, and your SCQ pacing sits comfortably inside the time budget. Until then, keep filtering to fill thin cells. After then, filtered practice should shrink to a small remedial minority and full timed mocks should dominate, because mixed, randomised blocks are the format you actually sit.
How should I combine ACEMCQ with iatroX without duplicating practice? Give each tool one job and never let their items overlap. Use ACEMCQ for Fellowship written-format depth and SAQ marking; use iatroX only as an unseen, timed check on the ACEM Primary-level foundational science beneath your reasoning, since iatroX is a Primary bank, not a Fellowship one. Never re-import an item you have already seen in ACEMCQ into an iatroX session, or your iatroX result stops being an unseen measurement and becomes a memory test. Keep the measurement channel clean.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. All ACEMCQ figures — question counts, topic domains, tiers and prices — are vendor-reported as displayed on the ACEMCQ site on that date and can change without notice; confirm them on the product page, and confirm the current examination format and fees on acem.org.au before relying on any of it. Honesty flag: ACEMCQ's public materials describe domain analytics and AI-marked SAQ feedback but do not describe a verified adaptive-difficulty algorithm, so this article makes no claim about proprietary adaptive routing. Disclosure: iatroX operates a competing question bank, but only at ACEM Primary level; it is not a Fellowship SAQ, SCQ or OSCE product, and it is confined here to the one job ACEMCQ does not claim — a clean, unseen check on foundational knowledge. Corrections are welcome via the feedback route on iatrox.com. References: Australasian College for Emergency Medicine examination and FACEM curriculum pages (acem.org.au); ACEMCQ product and pricing pages; iatroX ACEM Primary bank, the comparison hub, the calibrating AI feedback method, and Your Q-Bank Percentage Is Not Your Exam Score.