This workflow is for higher specialty trainees in gastroenterology and hepatology — typically ST4 to ST5 — who already use BMJ OnExamination and want to revise "adaptively" for the European Specialty Examination in Gastroenterology and Hepatology (ESEGH) by steering their own feed from performance data. It addresses the written, two-paper knowledge exam only. The principal limitation is honest and worth stating up front: BMJ OnExamination does not advertise a proprietary adaptive-difficulty engine, so the adaptation here is a discipline you impose, not an algorithm you can outsource your blueprint coverage to.
What BMJ OnExamination offers for ESEGH right now
| Attribute | Finding (vendor-reported, checked 20 July 2026) |
|---|---|
| Product | ESEGH (Gastroenterology & Hepatology) bank on onexamination.com |
| Live question count | 360+ best-of-five questions |
| Price / access | Listed as free of charge while the bank is "under full editorial review"; the vendor states this is a temporary period only — verify the current price and access-length options before you plan a budget |
| Format | Best-of-five, aligned to the curriculum areas published by the Royal College of Physicians |
| Assessment mode | Mock Tests curated to recent exam themes; difficulty selection ("Select Questions"); 10 free questions per day without a subscription |
| Personalisation | Group learning with leaderboards, offline mobile app, question review and recap |
| Adaptive / AI | No proprietary adaptive-difficulty or AI algorithm is advertised for this bank |
| ESEGH components covered | The written two-paper knowledge exam (ESEGH has no OSCE or viva) |
These figures are vendor-reported and captured on 20 July 2026; the free-of-charge status is explicitly temporary editorial-review pricing, so treat it as a window and re-check the product page before committing a plan to it. Because ESEGH is entirely a written knowledge exam, no bank "fails" to reproduce a simulation component — the honest question is not whether BMJ OnExamination covers the right format but whether your use of it covers the whole blueprint.
The ESEGH exam you are actually preparing for
ESEGH is the Specialty Certificate Examination for gastroenterology and hepatology, delivered jointly by the Federation of the Royal Colleges of Physicians of the UK and the European Section and Board of Gastroenterology and Hepatology (ESBGH). The format is fixed: two papers of 100 best-of-five questions each (200 in total), three hours per paper, one mark per correct answer and no negative marking. It is computer-based, administered by Surpass at a test centre, and currently held once a year. UK trainees map to the JRCPTB Specialty Training Curriculum for Gastroenterology; European candidates map to the UEMS Blue Book (2017); questions are then selected against the published ESEGH blueprint.
Two facts should shape everything you do in the bank. First, because there is no negative marking, you must answer every item — leaving a stem blank is a strategic error, never a "safe" one. Second, the blueprint is broad, and the domains that carry only a handful of marks each — nutrition and intestinal failure, small-bowel disease, rarer pancreaticobiliary disease, metabolic and vascular liver disease, transplant hepatology, and the gastrointestinal manifestations of systemic disease — are exactly the ones a rising overall percentage will quietly bury. The official blueprint and any example questions the Federation releases are your calibration gold standard; third-party bank size and "high-yield" labels are not the blueprint and should not be treated as if they were.
Baseline week: measure before you personalise
Before you let any difficulty-filtered or "recommended" feed shape your revision, complete a small, blueprint-stratified unseen sample under timed conditions — around 60 to 80 questions spread deliberately across every domain, a handful in each, including the low-volume ones. The point is not the score; it is a map. You want a first-attempt percentage per domain, not a single global figure, so you can see where the thin ice is. Record the baseline outside the platform (a simple spreadsheet), because that record is what you audit against — and it is the honest anchor the phrase "your Q-bank percentage is not your exam score" is pointing at.
First pass: set domain floors the overall score cannot hide
The failure mode this workflow exists to stop is simple: you do 1,200 questions, your overall accuracy climbs to a comfortable-looking number, and it turns out 300 of those questions were IBD and chronic liver disease while transplant hepatology and nutrition were barely touched. The fix is a domain floor — a minimum number of attempted-and-reviewed questions per blueprint area that you commit to before any topic is allowed to "count" as covered.
| Blueprint area (illustrative) | Suggested first-pass floor | Why it is easy to neglect |
|---|---|---|
| Luminal GI, IBD, colonic disease | High volume anyway | You will over-serve this naturally |
| Chronic liver disease and complications | High volume anyway | Familiar from general medicine |
| GI bleeding and emergencies | Moderate | Feels "done" after a few items |
| Pancreaticobiliary (incl. rarer) | Floor it | Autoimmune pancreatitis, cystic lesions get skipped |
| Nutrition and intestinal failure | Floor it | Low mark count, low interest, high skip rate |
| Transplant hepatology | Floor it | Sub-specialised, often under-represented in revision |
| Metabolic / vascular liver disease | Floor it | Wilson, haemochromatosis, Budd–Chiari slip through |
Set the floor as a count, tick each domain off against it, and do not let the overall percentage stand in for coverage. This is the same discipline as a blueprint-coverage matrix, and it is the single most valuable thing you can add to an otherwise-passive bank run.
An error taxonomy that tells you what to do next
A wrong answer is data only if you know why it was wrong. Tag every miss with one of six codes, because each points to a different next action:
| Code | Meaning | Typical fix |
|---|---|---|
| K | Knowledge gap — you did not know it | Short source read, then a fresh transfer item |
| M | Misread stem — knew it, read it wrong | Slow the read; no re-study needed |
| P | Premature closure — anchored early | Practise generating a differential before answering |
| G | Guideline error — outdated or wrong standard | Read the current guideline section |
| C | Calculation error — e.g. correction/clearance sums | Drill the calculation, not the topic |
| T | Time-pressure error — right idea, rushed | A pacing problem, addressed in mixed blocks |
The value is that four of these six do not require you to re-study the topic at all. Treating every miss as a knowledge gap is the commonest way to waste revision time.
Review interval: not every miss deserves an immediate repeat
Repeating the same item soon after getting it wrong mostly trains recognition of that item, which is precisely the memorisation trap. Use the error code to decide the interval instead. A knowledge gap (K) earns a short source read and then a new item on the same principle a few days later. A guideline error (G) earns a read of the current standard now and a spaced re-test later. Misreads (M) and premature closure (P) are process faults, not content faults — flag them and move on. Only reserve the immediate repeat for genuine calculation drills (C), where the mechanic itself is the thing being learned.
The mixed-block switch: objective criteria
Topic-filtered blocks are useful early, when you are building foundations, but they cue you to the answer's territory — if the whole block is hepatology, you are half-way to the diagnosis before you read the stem. Switch towards timed, random, mixed blocks when three conditions are met: every domain floor is ticked; your first-attempt accuracy on unseen items has stopped rising week on week; and you have at least four weeks to the exam. From that point, most of your practice should be mixed and timed at roughly exam pace (180 minutes for 100 questions is 1.8 minutes each — aim to leave review time), because that is the condition the real papers test.
A seven-day plan for a busy trainee
Assume a registrar on a normal clinical week with one theatre-free day. BMJ OnExamination does the structured-coverage job; iatroX supplies unseen transfer practice so you measure understanding rather than recall of items you have already seen. No proprietary-algorithm claims here — the "adaptation" is you reading your own domain data and reallocating time.
| Day | BMJ OnExamination (one job) | iatroX (unseen measurement) |
|---|---|---|
| Mon | 25 questions in two neglected domains; tag misses | — |
| Tue | Source read on Monday's K/G misses | 10 fresh mixed items, timed |
| Wed | 25 questions, next two floor domains | — |
| Thu | Review flagged M/P items (process, not content) | 10 fresh items on Wed's weak principle |
| Fri | 25 questions, remaining floor domains | — |
| Sat | One curated Mock Test, timed | — |
| Sun | Re-test a spaced set of earlier misses | 20-item unseen mixed block; log first-attempt % |
The Sunday unseen block is the honest readout. If your BMJ accuracy is climbing but your unseen first-attempt score is flat, you are learning the bank, not the specialty — which is the exact signal to widen coverage and lean harder on transfer practice.
Three mistakes this plan is designed to stop
First, mistaking a rising overall percentage for readiness while low-volume domains stay unopened — solved by domain floors. Second, re-answering seen items until the numbers look good, which measures memory of the bank rather than knowledge — solved by unseen transfer blocks. Third, treating every error as a knowledge gap and re-reading topics you actually just misread — solved by the six-code taxonomy.
Decision checklist: continue, supplement, switch or stop
- Continue BMJ OnExamination if domain floors are being met and unseen first-attempt accuracy is trending up.
- Supplement with a second, unseen source if your seen-item accuracy is high but unseen accuracy lags by more than about ten points — a classic over-familiarity gap.
- Switch the emphasis to timed mixed mocks once all floors are ticked and unseen accuracy has plateaued.
- Stop adding new banks if you are collecting resources out of anxiety rather than closing a measured gap; more banks with the same blind spots do not fix a coverage problem.
Where iatroX fits (and where it does not)
Be plain about this: iatroX is not a gastroenterology-specific SCE bank. It is a cross-specialty, MRCP-level knowledge and unseen-MCQ measurement layer with spaced retrieval. For ESEGH it earns its place in exactly one role — supplying fresh, unseen questions so you can test whether a principle has transferred, rather than re-testing items you have already memorised in a specialty bank. That is the measurement half of the two-Q-bank rule: let BMJ OnExamination be the specialty-content engine and let iatroX be the neutral yardstick. It does not replace a dedicated ESEGH bank, and it does not claim to reproduce the ESEGH blueprint.
Bottom line: BMJ OnExamination is a reasonable content engine for ESEGH, and while its free editorial-review window lasts it is an easy addition to the stack — but it will not defend your low-volume domains for you. Domain floors, a real error taxonomy and a weekly unseen readout are what turn a passive bank run into an adaptive one.
Frequently asked questions
Is BMJ OnExamination enough for ESEGH on its own? For many well-prepared trainees a single strong bank plus the Federation's blueprint can be enough, but "on its own" is doing a lot of work in that sentence. The bank supplies breadth and explanations; what it does not supply is an unseen measurement that is uncontaminated by the items you have already reviewed, or any guarantee that you covered the thin domains. Used with domain floors and an external unseen check, it can be sufficient; used as an item-memorisation exercise, it is not.
Which ESEGH component does BMJ OnExamination not reproduce well? ESEGH has no OSCE or viva, so there is no clinical-skills component to reproduce — the entire exam is written best-of-five. The genuine gap is not a missing format but a missing condition: the bank cannot easily give you a large, blueprint-stratified pool of questions you have never seen, under timed random conditions, which is what the real two-paper day actually tests.
How many BMJ OnExamination questions should I complete per day for ESEGH? There is no official quota, and the honest answer is that a sustainable 20 to 30 reviewed questions a day around clinical work beats a sporadic 100 that you never review. Quality of review — tagging each miss, reading the source, testing transfer — matters far more than raw volume, and completing the bank is not the goal. Verify your own pace against weeks-to-exam rather than chasing a number.
When should I stop using BMJ OnExamination and move to mixed mocks? Move the emphasis to timed mixed mocks once every domain floor is ticked, your first-attempt accuracy on unseen items has stopped climbing, and you are within about four weeks of the exam. Before those conditions are met, topic-filtered blocks are still building coverage; after them, further filtered practice mostly cues you to answers and stops resembling the real papers.
How should I combine BMJ OnExamination with iatroX without duplicating practice? Give each tool one job and never test the same item twice across them. BMJ OnExamination is your specialty-content and coverage engine; iatroX is your unseen-measurement layer, where you answer fresh questions on the same principle to confirm transfer. If you find yourself re-doing seen items to inflate a percentage, you have duplicated practice and lost the calibration — see why your Q-bank percentage is not your exam score.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. Question counts, prices and features described above are vendor-reported and were correct on the date checked; BMJ OnExamination's free-of-charge ESEGH status is explicitly temporary, so verify current pricing on the product page before planning. Disclosure: iatroX operates a competing question bank; in this article its role is confined to unseen-MCQ measurement and cross-specialty knowledge, a job a dedicated ESEGH bank does not claim, and iatroX is not a gastroenterology-specific SCE product. Corrections are welcome via the feedback route on iatrox.com.
References: the Federation of the Royal Colleges of Physicians of the UK — ESEGH examination pages and blueprint (thefederation.uk); the European Section and Board of Gastroenterology and Hepatology / EUBOGH exam information (eubogh.org); JRCPTB Specialty Training Curriculum for Gastroenterology; BMJ OnExamination ESEGH product page (onexamination.com); and internally, the iatroX comparison hub and completion is not coverage: building a blueprint-coverage matrix.