This audit is for GPs and other doctors caring for children who are using BMJ OnExamination to prepare for the DCH theory paper — the RCPCH Foundation of Practice (FOP) exam. It covers the written knowledge layer only. The principal limitation to hold onto is that BMJ OnExamination is a large, well-built question bank with analytics and self-directed filtering, not a black-box adaptive-difficulty engine; its "adaptive" behaviour optimises whatever you tell it to, which is not the same as optimising for the blueprint or for transfer to unseen items.
Current state, last checked 20 July 2026
BMJ OnExamination publishes a dedicated Child Health / DCH revision product. Vendor-reported figures on the product page: 900+ questions written by paediatricians and matched to the DCH syllabus, in single-best-answer, EMQ and MCQ formats. Subscription prices (vendor-reported, GBP): one month £34.99, two months £44.99, three months £54.99, six months £84.99 and twelve months £149.99, with a free ten-questions-a-day trial. Named features include "Select Questions" (choose difficulty and revisit challenging items), "Mock Tests" (curated against recent exam themes), "Group Learning" (a daily leaderboard) and "On the Go" (offline mobile access), plus per-question explanations and a performance area that shows where your weaknesses lie. Verify all of this on onexamination.com before purchase, because counts, prices and feature names change.
Note what is absent from the vendor's own description: there is no claim of a proprietary adaptive algorithm that sets your difficulty for you. That matters for the title question. What people loosely call the "adaptive engine" is a combination of your own difficulty filtering, revision-plan scheduling, and analytics that report the past. It optimises for coverage and confidence on the material you choose to see — not for balanced blueprint sampling and not for performance on questions you have never seen.
Exam anchor: what DCH actually requires
The DCH has two components. The theory exam is the RCPCH Foundation of Practice paper — a computer-based paper of 100 single-best-answer questions in two hours with no negative marking, run at test centres over usually three sittings a year; confirm the current specification on rcpch.ac.uk. Separately, the DCH Clinical is an OSCE-style circuit of eight stations with role-players acting as parents or adolescents, held in two periods a year. BMJ OnExamination, like iatroX, addresses only the FOP knowledge layer. Neither reproduces the DCH Clinical, and no amount of question drilling substitutes for supervised clinical and communication practice. Distinguish the College's official requirements and any free sample material from third-party claims about "what comes up."
Define every metric before you trust it
The dashboard shows several numbers, and each answers a different question:
- First-attempt accuracy — your score on items the first time you see them. This is the closest thing on the platform to a true signal.
- Repeat accuracy — your score on re-seen items. It rises largely through recognition and should never be read as readiness.
- Percentile — your rank against other users who chose to attempt the same items. It reflects the peer group's self-selection, not the exam standard.
- Predicted score — if shown, an extrapolation from your history; treat it as a motivational estimate, not a pass probability.
- Coverage — how much of the bank you have attempted; this is bank completion, not blueprint coverage.
- Difficulty — the bank's own tag or peer-derived facility, useful for filtering but not a personal readiness measure.
- Time per item — the one number most candidates ignore and the one the real paper punishes; the FOP budget is about 72 seconds per question.
Why the adaptive feed distorts your percentage
If you let the platform steer you toward flagged or difficult items, or you keep filtering to weak areas, your attempted set becomes deliberately harder and skewed than the real paper. That is good practice, but it makes your headline percentage incomparable with a mixed, unseen block. A feed that over-samples your weak areas depresses your percentage while genuinely improving you; a feed that lets you replay comfortable, previously seen items inflates it. Either way, the number on the home screen is a function of what you chose to attempt, not of your standing against the FOP blueprint. This is the core of why your Q-bank percentage is not your exam score.
Blueprint audit: check attempts against the DCH domains, not the average
Instead of trusting the home-screen average, tabulate your attempts against the DCH/FOP domains — child development and behaviour, growth and nutrition, community and social paediatrics including safeguarding, acute and chronic paediatric presentations, neonatology, emotional and behavioural problems, prescribing and weight-based calculations for children, and ethics, consent and health promotion. Use the published DCH syllabus and current FOP content on rcpch.ac.uk as your column headings, and count attempts and first-attempt accuracy per domain. Build the blueprint-coverage matrix once and update it weekly. It routinely reveals that an adaptive-feeling feed has left a low-volume domain — often safeguarding, ethics or paediatric prescribing calculations — barely touched.
The readiness test
A credible readiness signal has five conditions, and BMJ OnExamination's default study mode meets only some of them. The block must be: unseen (first-attempt items only), timed (about 72 seconds each), mixed (all domains, not a single filtered topic), taken with no assistance (no pausing to read the explanation mid-block), and of an adequate sample (enough items — a mock-length set — that one lucky run does not fool you). If any condition is missing, you have a study score, not a readiness score. Reserve a protected pool of items you have never opened so that at least one genuinely unseen mixed block is possible in the final fortnight.
Override rules: force what the feed under-samples
Because the feed follows your filters and your history, you must manually force the material it will otherwise starve:
- Low-volume domains — safeguarding, community paediatrics, neonatology — scheduled as their own blocks even when the algorithm would not surface them.
- Image and data items — growth charts, rashes, X-rays — which reward pattern exposure you will not get by reading text explanations.
- Ethics and consent in the paediatric context, easy to skip and heavily represented in a child-health knowledge base.
- Calculations — weight-based paediatric dosing and fluids, checked against the SmPC/eMC and NICE/CKS rather than any single formulary shortcut — practised timed, because they are slow under pressure.
Worked dashboard example
Suppose after two weeks your dashboard reads: overall 71%, first-attempt 63%, 620 of 900 questions attempted, and by domain — development 78%, acute presentations 74%, neonatology 55%, safeguarding 51%, prescribing calculations 49%, with time per item running at 95 seconds. Do not convert this into a pass prediction. Convert it into next week's quotas. The three domains under 60% set your priority: allocate roughly 60% of next week's questions to neonatology, safeguarding and calculations, in short topic blocks to learn, then fold them into mixed blocks to test. The 95-second pace sets a second target: at least two timed blocks run strictly at 72 seconds per item to compress your reading. Leave the 78% domains alone except inside mixed blocks. Nothing here predicts a mark; it just points the week's effort at the measured gaps.
A seven-day plan: one job each for BMJ OnExamination and iatroX
Use BMJ OnExamination for one defined job — learning and reviewing content with explanations — and iatroX for a different job: unseen, timed transfer measurement. Neither side involves a claim about the other's algorithm.
- Monday–Wednesday: 30 BMJ OnExamination questions a day in topic blocks on your two weakest domains, full explanation review, one-line error code per miss.
- Thursday: 40 mixed BMJ OnExamination questions, timed, no explanations mid-block; log the domains that slipped.
- Friday: transfer day — a fresh, unseen, mixed, timed block in iatroX covering the same domains, to test whether Monday–Thursday's learning survives contact with questions you have never seen.
- Saturday: review only — re-test prior misses from earlier in the week (spaced), no new volume.
- Sunday: update the blueprint matrix; set next week's quotas from the gaps.
The point of the iatroX day is measurement on protected, unseen items — the second bank in the two-Q-bank rule — so that your BMJ percentage is never both the thing you train on and the thing you trust.
Decision checklist: continue, supplement, switch or stop
- Continue if first-attempt accuracy is rising, coverage across DCH domains is even, and your unseen transfer blocks track your in-bank scores.
- Supplement (add a measurement bank or a targeted resource) if your in-bank percentage is healthy but unseen transfer lags, or if one or two domains stay below floor no matter how you filter.
- Switch if the bank's coverage of a domain you need is genuinely thin, or the explanations are not teaching you — measured by a flat error pattern, not by boredom.
- Stop generating new volume when domain floors are met, unseen timed accuracy has plateaued, and remaining misses are slips rather than gaps. Move to spaced review and full timed papers.
Base every branch on a measured gap, not on novelty or on the sunk cost of an annual subscription.
Frequently asked questions
Is BMJ OnExamination enough for DCH on its own? For the FOP theory paper, a 900-plus-question bank with explanations can carry most of your knowledge preparation, but "on its own" hides two problems. First, it cannot measure itself, because your percentage climbs through re-seeing items; you need an unseen block to know where you stand. Second, it covers only the theory component — the DCH Clinical OSCE needs supervised communication and examination practice that no bank provides. As the knowledge layer it is a strong option; as your entire preparation it is not enough.
Which DCH component does BMJ OnExamination not reproduce well? It does not reproduce the DCH Clinical examination — the eight-station OSCE circuit with role-players — because that assesses history-taking, examination and communication with children and parents, which a written question cannot rehearse. Within the theory paper, its analytics also under-serve pacing and genuinely unseen performance: the platform is built to help you learn content, not to hand you a clean readiness signal. Treat it as a content-and-explanations engine for FOP, and get your pacing and transfer signal elsewhere.
How many BMJ OnExamination questions should I complete per day for DCH? There is no universal number; anchor it to review quality and weeks remaining rather than a vanity count. A sustainable pattern for a doctor working clinically is 20 to 40 questions a day with full review of every miss, rising to timed mixed blocks of 40 to 60 in the final fortnight. Completing 100 questions you do not review teaches less than 25 you do. Let time-per-item and your error log, not a daily target, set the volume.
When should I stop using BMJ OnExamination and move to mixed mocks? Move to mixed, timed mocks once every DCH domain has been attempted, your first-attempt accuracy has stabilised, and topic-filtered blocks are no longer changing your error pattern — typically the last two to three weeks. Mocks then test the things filtered practice hides: pacing across all domains at once and stamina over a full paper. You can keep BMJ OnExamination open for targeted review of mock misses, but your primary mode should switch from learning to timed rehearsal.
How should I combine BMJ OnExamination with iatroX without duplicating practice? Give each tool one job and never answer the same item twice across them. Use BMJ OnExamination to learn and review with explanations; use iatroX only for unseen, mixed, timed blocks that measure whether that learning transfers, keeping a protected pool of iatroX items you never pre-read. This is the two-Q-bank rule: one bank to train on, one to be graded by. Duplicating questions across both destroys the calibration, because a re-seen item measures memory, not readiness.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. Question counts, prices and feature names for BMJ OnExamination are vendor-reported and change without notice; verify them on onexamination.com and confirm the current DCH and FOP specifications on rcpch.ac.uk. This article makes no claim about any proprietary algorithm inside BMJ OnExamination; it describes the platform's published, user-facing features only. Disclosure: iatroX operates a competing UK question bank, so its role here is confined to the job BMJ OnExamination does not claim — unseen, timed transfer measurement, not teaching and not the DCH Clinical. Corrections are welcome via the feedback route on iatrox.com.
References: RCPCH — About the Diploma in Child Health, Foundation of Practice theory-exam guidance and the DCH syllabus (rcpch.ac.uk); BMJ OnExamination Child Health / DCH product page (onexamination.com); iatroX internal references — the comparison hub and why your Q-bank percentage is not your exam score.