This audit is for higher specialty trainees, usually ST4 and above, who assumed BMJ OnExamination would carry their SCE Medical Oncology revision. The headline finding, verified on 20 July 2026, is the one that matters most before any dashboard discussion: BMJ OnExamination does not currently publish an SCE Medical Oncology question bank. Its adaptive analytics therefore cannot be your primary readiness signal for this exam, and this audit shows you what to use instead and how to read whatever bank you do adopt.
What BMJ OnExamination actually offers for SCE Medical Oncology right now
The honest starting point of any coverage audit is whether the product exists for your exam. For medical oncology, on the publication date, it does not. BMJ OnExamination's SCE catalogue lists Acute Medicine, Endocrinology and Diabetes, Geriatric Medicine, Respiratory Medicine and Neurology, alongside the European examinations ESEGH, ESENeph and EECC. There is no Medical Oncology product in that catalogue.
| Item | Finding (vendor-reported, 20 July 2026) |
|---|---|
| SCE Medical Oncology bank | Not offered. No Medical Oncology SKU in the BMJ OnExamination SCE catalogue. |
| Nearest related content | General MRCP(UK) Part 1 and Part 2 banks cover oncology only at generalist physician depth, not SCE specialty depth. |
| Analytics engine (on covered products) | "High Impact Questions" ranking; filters by topic, type, difficulty and keyword; performance graphs with peer benchmarking; a vendor pass-likelihood indicator; timed mock tests; a daily personalised question. |
| Price | Tiered by access length on covered specialties (for example SCE Neurology from £69.99 for one month to £189.99 for twelve months, vendor-reported). No oncology price exists. |
| Free access | Ten free questions per day on covered products. |
The practical consequence is simple. If you want the BMJ analytics experience, you cannot get it for oncology today, so your specialty content must come from a dedicated SCE Medical Oncology bank such as StudyPRN or RevisionPro, or from the published single-best-answer revision books. What remains genuinely useful from BMJ is the analytics model itself: the way it frames accuracy, repeats, coverage and difficulty is a good template for reading any adaptive bank, and the rest of this audit teaches that literacy so it transfers to the bank you actually buy.
The exam you are actually preparing for: the SCE Medical Oncology blueprint
Every Specialty Certificate Examination shares one structure. The SCE in Medical Oncology is two papers of 100 best-of-five questions each, 200 questions in total, three hours per paper, sat on a single day at a computer-based Surpass test centre. There is one mark per correct answer and no negative marking. Nothing about that structure is specialty-specific; only the blueprint changes.
The published Medical Oncology blueprint distributes those 200 questions roughly as below. Treat these as indicative, and verify the current version on thefederation.uk, because the curriculum was refreshed in 2021 and the exact counts vary slightly between diets.
| Blueprint domain | Indicative questions (of 200) |
|---|---|
| Chemotherapeutic agents | 28 |
| Breast cancer | 23 |
| Lung cancer | 17 |
| Scientific basis of malignancy | 16 |
| Acute oncology | 15 |
| Colorectal and anal cancer | 11 |
| Urothelial and renal cancer | 10 |
| Oesophagogastric cancer | 7 |
| Sarcoma | 7 |
| Hepatobiliary cancer | 6 |
| Clinical research | 6 |
| Skin cancer | 6 |
| Lymphoma, ovarian, uterine and cervical | 5 each |
| Germ cell tumours, carcinoma of unknown primary | 4 each |
| Other (screening, growth factors, analgesia, anti-emetics, CNS, prostate, head and neck and more) | 25 |
The single most important fact a dashboard cannot tell you is that chemotherapeutic agents and acute oncology together carry more marks than any single tumour site. A bank that feels comprehensive on breast and lung but thin on systemic anticancer therapy toxicity is misaligned with where the marks sit.
The analytics vocabulary every oncology candidate should master
BMJ's dashboard, and every competitor's, reports the same core metrics. Define them precisely before you trust any of them.
First-attempt accuracy is the percentage correct on questions you have never seen. This is the only accuracy figure that behaves like the exam. Repeat accuracy is the percentage correct on questions you have already attempted, and it is inflated by memory of the answer rather than mastery of the concept. Percentile benchmarks you against other users of that product, not against the national cohort sitting the SCE, so it moves when the user base moves. A pass-likelihood indicator, which BMJ describes as showing "how likely you are to pass", is a vendor heuristic, not a validated prediction, and should never be read as a probability of passing the real exam. Coverage is the proportion of a domain's items you have attempted at least once. Difficulty is usually the cohort's average success rate on an item, so a "hard" question is one most users get wrong. Time per item is your mean seconds per question, against a real-exam budget of roughly 108 seconds.
The gap between first-attempt and repeat accuracy is your single most honest number. If your headline percentage is 80% but your first-attempt figure is 64%, the difference is recognition, not readiness.
Why adaptive feeds distort your percentage
Any engine that prioritises your weak areas, as BMJ's High Impact ranking is designed to, will feed you a harder-than-average diet over time. That is pedagogically sensible and statistically distorting at once. As the feed concentrates on chemotherapy toxicity and acute oncology because you keep missing them, your rolling percentage falls even as your knowledge rises, because the denominator is now dominated by your hardest material. The mirror-image error also occurs: a candidate who mostly re-tests mastered breast cancer items will post a flattering percentage that reflects sampling, not competence.
This is why a raw bank percentage is not comparable to a mixed, unseen mock, and why "Your Q-Bank Percentage Is Not Your Exam Score" is the caveat to internalise. The number that transfers is first-attempt accuracy on a blueprint-proportional, timed, unseen block.
Auditing your attempted-question distribution against the blueprint
Do not trust the home-screen average. Export or note your attempted-question count per domain and lay it beside the blueprint proportions above. The audit question is not "what is my percentage" but "have I attempted enough unseen items in each domain, weighted the way the exam weights them". A candidate who has done 400 breast and lung items but only 30 chemotherapeutic-agents items has a coverage hole exactly where the marks are densest, regardless of how good the percentage looks. Completion of a bank is not coverage of a blueprint; build the matrix deliberately, as the completion-is-not-coverage framework sets out.
The readiness test: five conditions for a signal you can trust
A number is only a readiness signal when five conditions hold at once. The items must be unseen, so first-attempt only. The block must be timed at exam pace. The content must be mixed and blueprint-proportional, not a single weak domain. You must sit it with no assistance, no notes and no pausing to look things up. And the sample must be large enough, at least 80 to 100 items, before you read anything into the percentage. Fail any one condition and the number is diagnostic feedback, not a readiness estimate.
Override rules: what to force into your revision
Adaptive and popularity-ranked feeds under-surface low-volume, high-consequence material. For medical oncology, manually force the following even if the engine never prioritises them: immune-related adverse events and their grading and steroid thresholds; neutropenic sepsis and other acute-oncology emergencies such as metastatic spinal cord compression, hypercalcaemia, superior vena cava obstruction and tumour lysis syndrome; drug-specific toxicities where the fact is the answer, such as anthracycline cardiotoxicity, cisplatin nephrotoxicity and ototoxicity, bleomycin pulmonary toxicity and DPD-related fluoropyrimidine toxicity; clinical-trial endpoints and hazard-ratio interpretation; and any image or data-interpretation item. Because medicines facts change, verify drug specifics against the SmPC on the eMC and current NICE technology appraisals rather than a static bank explanation.
Worked example: turning a dashboard into next week's quotas
Suppose a hypothetical candidate's oncology bank shows an overall first-attempt accuracy of 71%, but domain figures of 58% on chemotherapeutic agents, 61% on acute oncology, 79% on breast and 74% on lung, with only 35% coverage of the chemotherapeutic-agents domain and a mean of 95 seconds per item. Read against the blueprint, the two lowest-accuracy domains are also two of the three highest-weighted, and coverage is thinnest exactly there.
Next week's quotas follow mechanically without any pass prediction. Allocate the largest share, say 60 unseen items, to chemotherapeutic agents to close both the accuracy and the coverage gap; 40 to acute oncology; a maintenance 20 across breast and lung to hold gains; and 20 forced image or data-interpretation items. Hold pace deliberately at 100 to 108 seconds so speed does not silently degrade as difficulty rises. Re-measure on a fresh unseen block at the end of the week, and compare first-attempt to first-attempt, never headline to headline.
A seven-day pattern for trainees revising around clinical work
This loop assumes a specialist oncology bank for content, because BMJ does not supply one, and iatroX for unseen transfer measurement. It makes no claim about any vendor's internal algorithm.
- Day 1: 40 unseen specialist-bank items in your weakest blueprint domain; log every miss by mechanism, not just by topic.
- Day 2: Read around the Day 1 misses using primary sources, the SmPC and current NICE appraisals; no new questions.
- Day 3: 40 unseen items in the second weakest domain; repeat the miss log.
- Day 4: Space the Day 1 misses by re-testing them as unseen-adjacent items; add 20 forced image or acute-oncology items.
- Day 5: Read around Day 3 and Day 4 misses.
- Day 6: A timed, mixed, blueprint-proportional 80 to 100 item block. Where you want an unseen measurement that is not contaminated by your specialist bank, sit a fresh timed block in iatroX at UK MRCP-level to test whether the knowledge transfers to items you have never seen.
- Day 7: Rest or light review; set next week's quotas from Day 6, not from any rolling average.
Three mistakes this audit is designed to stop
The first is treating a rising headline percentage as rising readiness when it is really rising recognition of repeated items. The second is trusting a vendor pass-likelihood indicator as if it were a calibrated probability; it is a marketing convenience, not an examiner's judgement. The third is letting the adaptive feed set your blueprint for you, so you over-revise the tumour sites you already know and under-revise systemic anticancer therapy and acute oncology, which is precisely where the exam concentrates its marks.
Decision checklist: continue, supplement, switch or stop
| Situation (measurable) | Action |
|---|---|
| No oncology bank yet | Acquire a specialist SCE Medical Oncology bank first; BMJ cannot serve this exam. |
| First-attempt accuracy rising across all domains, coverage even | Continue; do not add tools for novelty. |
| One or two domains persistently below cohort, thin coverage | Supplement with forced quotas and targeted reading, not a new bank. |
| Bank exhausted, repeats dominating, recognition inflating scores | Add a second, unseen source for measurement; apply the two-Q-bank rule. |
| Unseen, timed, mixed first-attempt consistently comfortable | Stop adding volume; move to timed mixed mocks and rest. |
Bottom line
BMJ OnExamination is a capable adaptive platform, but it does not currently sell an SCE Medical Oncology bank, so for this exam it cannot be your core resource. Buy a specialist oncology bank for content, use the analytics literacy above to read it honestly against the 200-question blueprint, and measure readiness on unseen, timed, blueprint-proportional blocks rather than on a flattering rolling percentage.
Frequently asked questions
Is BMJ OnExamination enough for SCE Medical Oncology on its own? No, and not for the usual reasons. As verified on 20 July 2026, BMJ OnExamination does not publish an SCE Medical Oncology bank at all, so it cannot be a sufficient resource, or even a partial one, for this specific exam. Candidates should base their preparation on a dedicated oncology bank and use the analytics discipline described here to read it.
Which SCE Medical Oncology component does BMJ OnExamination not reproduce well? Since BMJ carries no oncology bank, the honest answer is that it reproduces none of the specialty content. More broadly, the components most poorly served by any generic physician bank are rapidly changing systemic anticancer therapy, immune-related adverse events, molecular biomarkers and trial-endpoint interpretation, because these date quickly and need primary-source verification against the SmPC and current NICE appraisals.
How many BMJ OnExamination questions should I complete per day for SCE Medical Oncology? There is no oncology bank to complete, so this figure does not apply to BMJ. Using a specialist oncology bank, a sustainable target for a trainee working clinically is roughly 30 to 50 unseen items per revision day, always logged by mechanism, with reading days in between rather than an unbroken run of high-volume days.
When should I stop using BMJ OnExamination and move to mixed mocks? For oncology the question is really when to move from single-domain drilling to mixed mocks, and the trigger is measurable: once your unseen first-attempt accuracy is stable and even across the high-weight domains, stop drilling domains in isolation and shift to timed, blueprint-proportional mixed blocks to rehearse pacing and decision-switching under exam conditions.
How should I combine BMJ OnExamination with iatroX without duplicating practice? Because BMJ has no oncology bank, the practical combination is a specialist oncology bank for content plus iatroX as the unseen measurement layer. Keep the two jobs separate: learn and drill in the specialist bank, then measure transfer on fresh, timed, unseen iatroX blocks you have never touched, so the second tool never simply re-tests the first tool's items.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. Platform figures, including question counts, prices and features, are vendor-reported and change without notice; confirm the current details on the product pages before relying on them. Disclosure: iatroX operates a UK question bank and clinical-knowledge platform and therefore competes with the products discussed; iatroX does not publish a specialty-specific SCE Medical Oncology bank, and its role here is confined to cross-specialty knowledge and unseen-MCQ measurement, a job the audited product does not claim for oncology. Corrections are welcome via the feedback route on iatrox.com.
References: Federation of Royal Colleges of Physicians of the UK, SCE in Medical Oncology and blueprint (thefederation.uk); BMJ OnExamination SCE resources and features pages (onexamination.com); iatroX, Your Q-Bank Percentage Is Not Your Exam Score; iatroX comparison hub.
Run a fresh, timed unseen block in iatroX and decide your next move →