Lead finding first, because it changes the whole article: as of 20 July 2026, BMJ OnExamination does not publish a dedicated SCE in Rheumatology question bank. Its SCE catalogue lists Acute Medicine, Endocrinology and Diabetes, Geriatric Medicine, Respiratory Medicine and Neurology, but not Rheumatology. So if you are revising SCE Rheumatology, there is no BMJ OnExamination Rheumatology analytics dashboard to optimise around, and the honest question becomes how to read any adaptive-style feed for this exam without being misled.
What BMJ OnExamination offers for SCE Rheumatology right now
Nothing specialty-specific. This is the current-state finding, vendor-checked on 20 July 2026, and you should re-check the catalogue before relying on it because vendors add modules between diets.
| Attribute | Status (checked 20 July 2026) |
|---|---|
| SCE Rheumatology bank | Not offered by BMJ OnExamination |
| SCE specialties BMJ does cover | Acute Medicine, Endocrinology and Diabetes, Geriatric Medicine, Respiratory Medicine, Neurology |
| Nearest overlapping content | General-medicine and MRCP-level items covering musculoskeletal and connective-tissue topics inside other modules |
| Adaptive/AI | No proprietary adaptive-difficulty algorithm is advertised on any BMJ OnExamination module |
Because there is no Rheumatology product, the practical options for a rheumatology candidate are a vendor that does specialise in the SCE (for example StudyPRN, which publishes a dedicated Rheumatology bank), the Federation's own sample questions, and a cross-specialty measurement layer for the general-medicine content. The rest of this article explains what any "adaptive" feed you do use is really optimising, so you can read it correctly whichever bank you choose.
The exam you are actually sitting
The SCE in Rheumatology uses the standard Federation format: two papers of 100 best-of-five questions, 200 total, three hours each, one day, computer-based via Surpass, one mark per correct answer, no negative marking. Only the blueprint differs. The published Rheumatology blueprint is unusually top-heavy, and that shape matters for how any feed behaves.
| Domain | Questions | Share |
|---|---|---|
| Adult inflammatory arthritis (including crystal arthropathies) | 70 | 35% |
| Adult connective tissue disorders | 60 | 30% |
| Adult osteoarthritis, soft tissue, regional and widespread pain | 20 | 10% |
| Adult infection, neoplasia and miscellaneous | 15 | 7.5% |
| Adult metabolic and bone disease | 13 | 6.5% |
| Paediatric and adolescent rheumatology | 12 | 6% |
| Clinical science and pharmacology | 10 | 5% |
Two domains, inflammatory arthritis and connective tissue disease, account for 65% of the paper. The remaining 35% is spread thinly across five areas that candidates routinely under-weight, and those thin domains are where an uncritical adaptive feed will quietly leave you exposed.
What an adaptive engine is actually optimising
Here is the honest mechanics, and it applies to BMJ OnExamination's other modules and to most banks marketed as "adaptive." What candidates experience as an adaptive engine is usually two ordinary things: a difficulty filter you selected, and a recommender that returns you to items you flagged or previously answered wrongly. Neither is a hidden model of your probability of passing the SCE. So the feed is optimising for re-exposure to your self-identified weak items at a difficulty you chose, which is a legitimate and useful revision loop. What it is not doing is guaranteeing blueprint balance, measuring your unseen standing, or predicting your score. On a top-heavy blueprint like Rheumatology, a weakness-chasing feed will happily pour your time into inflammatory arthritis (the 35% domain you already fail most visibly) and starve clinical science, paediatric rheumatology and metabolic bone disease, precisely because you flag fewer items there. The engine is doing what it was built to do; it is just not doing the thing you assumed.
Every metric you might see, defined
Whatever bank you use, name the numbers. First-attempt accuracy is the only accuracy figure close to an unseen estimate. Repeat accuracy is inflated by exposure and tells you little. A percentile or peer comparison ranks you against a self-selected cohort, not the exam population. A predicted score, if shown, is a behavioural model, not a promise, and should be treated as a soft prior at most. Coverage measures completion, not competence. Difficulty reflects your filter or the item's historical facility. Time per item is your pace against roughly 108 seconds per question. The recurring trap is reading a rising blended percentage or a high coverage figure as readiness when only large-sample first-attempt accuracy on unseen items comes close.
Blueprint audit before you trust any feed
Set your attempted distribution beside the blueprint. The point is not your average but whether you have attempted a representative share of every domain. On this blueprint the failure mode is predictable: heavy attempts in inflammatory arthritis and connective tissue disease, and near-nothing in clinical science and pharmacology, metabolic bone disease, paediatric rheumatology, and infection or neoplasia. Those thin domains are still 50 marks combined, and in a tightly standard-set exam 50 marks decide the outcome. Build the two-column check, attempted share versus blueprint share, and treat any domain below half its weight as a blind spot the feed has created rather than closed.
The readiness test
A number is a readiness signal only when the block is unseen, timed at exam pace, mixed across the blueprint, sat with no assistance, and large enough to be a sufficient sample (80 to 100 items). A 90% on a re-seen inflammatory-arthritis drill is not a signal; a 67% on a 100-item mixed unseen timed block is. Because no BMJ OnExamination Rheumatology pool exists, your unseen mixed blocks will come from a specialty bank's fresh items plus a cross-specialty measurement layer for the general-medicine content, and that combination has to be assembled deliberately.
Override rules for any feed
Force blocks in the areas a weakness-chasing feed under-serves. In Rheumatology that means clinical science and pharmacology (immunology, autoantibody patterns, DMARD and biologic mechanisms and monitoring), paediatric and adolescent rheumatology, metabolic and bone disease, infection and neoplasia, imaging, and classification criteria (ACR and EULAR). For biologic and DMARD monitoring, use the SmPC and eMC together with British Society for Rheumatology guidance as the reference source rather than any formulary shorthand. These low-volume domains are exactly what an adaptive feed skips, and they are disproportionately where marginal candidates lose the marks that matter.
Worked example: turning a feed into quotas
Suppose you are working a specialty Rheumatology bank and its feed reports overall 80%, first-attempt 64%, and attempted-by-domain: inflammatory arthritis 90 items, connective tissue 70, osteoarthritis and soft tissue 20, metabolic bone 4, paediatric 3, clinical science and pharmacology 5, infection and neoplasia 4. This is not an 80% candidate; it is a 64% first-attempt candidate who has let the feed over-invest in the two big domains and abandon the thin third of the paper.
Next week's quotas: cap inflammatory arthritis and connective tissue at revision-only, and assign new unseen items to the tail, 12 clinical science and pharmacology, 10 metabolic bone, 10 paediatric rheumatology, 10 infection and neoplasia, 8 imaging and classification criteria. Close with one mixed timed block and record first-attempt accuracy only. No pass prediction is made; the quotas are driven by the blueprint deficit and the first-attempt figure.
A seven-day pattern for busy trainees
One job per tool, no proprietary-algorithm claims. Because BMJ OnExamination has no Rheumatology bank, your specialty coverage comes from an SCE-specialist bank; iatroX is the unseen measurement layer for the substantial cross-specialty content.
- Days 1 to 5 (20 to 30 minutes each): work a specialty Rheumatology bank in the under-served domains, reading every explanation and flagging lucky correct answers.
- Day 6: sit one mixed timed block. Alternate the specialty bank's mock one week with a fresh unseen iatroX block the next. Rheumatology carries a real cross-specialty core (clinical pharmacology, immunology principles, infection, neoplasia, bone metabolism and general-medicine comorbidity), and iatroX's UK/MRCP-level bank is a fair unseen yardstick for that core.
- Day 7: review only. Spaced-queue the misses and open reasoning-gap items in the iatroX Socratic Tutor to be questioned rather than told.
State the boundary: iatroX does not carry rheumatology-specific depth such as classification criteria, characteristic imaging or biologic-monitoring schedules, and it does not claim to. It measures the transferable layer; the specialty bank owns the rheumatology depth.
Continue, supplement, switch or stop
Because there is no BMJ OnExamination Rheumatology product, the first decision is which specialty bank to adopt, not whether to continue an existing one. Continue a chosen specialty bank while first-attempt accuracy is rising and thin domains still hold un-attempted items. Supplement with an unseen measurement layer as soon as you rely on any single bank, so your signal is never taken from seen items. Switch if a domain filter shows the bank cannot populate the thin blueprint areas to depth. Stop drilling a domain once first-attempt accuracy on new items meets target and has stalled. Decide on measured gaps, not novelty or sunk cost.
Bottom line
For SCE Rheumatology, BMJ OnExamination is not currently an option, because it publishes no Rheumatology bank as of 20 July 2026. Use an SCE-specialist bank for coverage, understand that any "adaptive" feed is optimising re-exposure to your flagged items rather than balancing the blueprint or predicting your score, and force the thin two-thirds of domains yourself. Reserve "ready" for a large, mixed, timed, unseen, unassisted block, and take your unseen signal from fresh specialty items plus a cross-specialty measurement layer.
Frequently asked questions
Is BMJ OnExamination enough for SCE Rheumatology on its own? No, and not because of quality but because of coverage: BMJ OnExamination does not publish an SCE Rheumatology bank as of 20 July 2026 (vendor-checked; re-confirm on the product site). A rheumatology candidate needs an SCE-specialist bank for the specialty content and a separate unseen source for measurement, so BMJ OnExamination cannot be the spine here at all.
Which SCE Rheumatology component does BMJ OnExamination not reproduce well? All of it, in the sense that there is no dedicated module. More usefully, no static bank of any vendor fully reproduces raw imaging interpretation, the longitudinal biologic-monitoring workflow, or bedside pattern recognition; those need real cases and unlabelled data alongside best-of-five practice, whichever specialty bank you use.
How many BMJ OnExamination questions should I complete per day for SCE Rheumatology? None specific to Rheumatology are available, so this reduces to your chosen specialty bank. Target blueprint-weighted coverage, 30 to 50 reviewed items a day biased toward the thin domains, and track first-attempt accuracy and attempted share rather than a raw count.
When should I stop using BMJ OnExamination and move to mixed mocks? For Rheumatology there is nothing BMJ-specific to stop using. With your specialty bank, move to mixed, timed, unseen blocks once every blueprint domain is covered at roughly its weight and first-attempt accuracy has plateaued, which is the only stage at which a percentage becomes a trustworthy signal.
How should I combine BMJ OnExamination with iatroX without duplicating practice? Since there is no BMJ Rheumatology bank to combine, pair an SCE-specialist Rheumatology bank with iatroX: the specialty bank owns rheumatology coverage and explanations; iatroX owns unseen measurement of the cross-specialty core (pharmacology, immunology, infection, neoplasia, bone metabolism) and spaced retrieval of misses, under the two-Q-bank rule so the sources never overlap. iatroX is a UK measurement layer, not a rheumatology bank.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. The central finding, that BMJ OnExamination does not publish an SCE Rheumatology bank, was verified against its SCE catalogue on that date and is vendor-dependent; re-check before relying on it, as vendors add modules between diets. Disclosure: iatroX operates a UK question bank; this article confines iatroX's role to jobs no rheumatology SCE product it competes with claims, namely unseen cross-specialty measurement and spaced retrieval, and iatroX is not a specialty SCE bank. Corrections are welcome via the feedback route on iatrox.com.
References: the Federation of Royal Colleges of Physicians SCE Rheumatology specialty and blueprint pages (thefederation.uk); the BMJ OnExamination SCE resources catalogue (onexamination.com); British Society for Rheumatology guidance and the SmPC/eMC for medicines monitoring; iatroX, Your Q-Bank Percentage Is Not Your Exam Score; iatroX, the two-Q-bank rule; iatroX, building a blueprint-coverage matrix; the iatroX comparison hub and Socratic Tutor.
Run a fresh, timed unseen block in iatroX and decide whether to learn, retest, simulate or stop →