The DGM Q-Bank Content-Gap Checklist: What to Verify Before You Stop Doing New Questions

Before you decide you have "covered" the Diploma in Geriatric Medicine (DGM), you should be able to show evidence — not a feeling, and not a bank-completion percentage. This is a checklist of the minimum evidence required to stop doing new questions for the Part 1 Knowledge-Based Assessment, organised as things to verify rather than a study timetable. Work through it honestly and it will usually reveal one or two domains you have been quietly avoiding.

The current DGM exam snapshot

The DGM is awarded by the Royal College of Physicians (RCP), developed in collaboration with the British Geriatrics Society (BGS). The RCP administers the exam and publishes the authoritative syllabus, blueprint and sample questions; the BGS does not run the exam or sell a question bank (RCP; BGS; last checked 21 July 2026). Part 1 is the KBA — an online paper of 100 best-of-five questions over three hours, once a year, with no negative marking (2026 fee £336). Part 2 is a separate Clinical OSCE of four live stations at the RCP in Liverpool, with a four-year window from KBA success. This checklist concerns the KBA only; the Clinical needs a different kind of preparation entirely.

Build a blueprint coverage table

The core tool is a coverage table built from the RCP's own February 2025 blueprint, not from your bank's topic list. Populate it as you revise. The point is to see, at a glance, where practice is thin, where accuracy is low, and where your last review has gone stale.

Domain (RCP blueprint)	Official weight (of 100)	Questions attempted	First-attempt accuracy	Last reviewed	Confidence (L/M/H)
Epidemiological, demographic and social factors	5
Common geriatric problems / syndromes	40
Other illnesses affecting older people	15
Pharmacology and therapeutics	15
Principles of rehabilitation	8
Transfers of care and community care	5
Basic science and biology of ageing	2
Administrative aspects, ethics and law	10

The weighting matters: syndromes, other illnesses and pharmacology together account for 70 of the 100 marks. If your coverage table is full of green on epidemiology and ethics but thin on syndromes, you are well practised on 15 marks and under-practised on 40. Completion of a bank tells you none of this, which is why bank completion is not coverage (blueprint coverage matrix).

Ten domain-level blind spots most likely to stay hidden

Self-selected practice tends to avoid the uncomfortable topics. These ten are the ones most likely to remain blind spots on a DGM coverage table, and each should be reviewed by someone who knows the exam before you sign off. Treat this as a list requiring exam-specific clinician review, not a self-graded quiz.

Frailty identification and scoring — recognising frailty syndromes and poorer prognosis, not just naming frailty.
Delirium versus dementia versus depression — the discrimination, precipitants and management, a perennial high-yield trap.
Mental capacity in practice — the Mental Capacity Act, functional capacity assessment, Lasting Power of Attorney, Court of Protection, advance decisions and testamentary capacity.
Polypharmacy and deprescribing — adverse drug reactions, interactions and structured medication review in the older patient (sourced from the SmPC/eMC).
Falls and bone health — multifactorial falls assessment and fracture prevention.
Continence — assessment and management, a domain candidates routinely under-practise.
Pressure ulcers and tissue viability — risk assessment and prevention.
Movement disorders and Parkinson's — including drug-induced parkinsonism and time-critical medication.
Stroke and rehabilitation principles — comprehensive geriatric assessment and multidisciplinary rehabilitation.
End-of-life and transfers of care — palliative symptom control, and the ethics/law and community-care interface.

Format checklist: deliberate practice for the exam's real content

Verify that you have deliberately practised — not just read about — the four content areas the DGM leans on hardest:

Frailty: comprehensive geriatric assessment reasoning under time pressure.
Capacity: applying a structured capacity test to a specific decision, not reciting the principles.
Medicines: deprescribing and ADR reasoning in a multimorbid patient, sourced from the SmPC/eMC and NICE/CKS.
Multidisciplinary scenarios: questions turning on the roles and limits of the wider team.

If any of these has zero deliberately practised items, it is a gap regardless of your overall accuracy.

Interpretation checklist

Confirm you have practised the item types that test interpretation rather than recall, as applicable to the DGM: laboratory trends in the older patient, medication and monitoring data, basic calculations, and the ethics/law reasoning that carries 10 marks. Where image or ECG interpretation appears, verify you have deliberate practice there too rather than assuming a text-only revision covers it.

Recency checklist

Geriatric medicine is guidance-sensitive: falls, osteoporosis, delirium, continence and anticoagulation guidance all move. For every guidance-linked topic, record the source and its date and jurisdiction — NICE, CKS, SIGN, the SmPC/eMC and NHS clinical content for the UK. A correct answer learned from a superseded guideline is a latent error. Verify that your highest-yield topics are anchored to current UK guidance, with the date noted next to each.

Performance checklist

Coverage is necessary but not sufficient; you also need performance evidence: unseen timed mixed blocks (not topic-cued practice), pacing against the three-hour, 100-item budget, a low and falling rate of high-confidence errors, retention of previously missed items after a spacing interval, and calibration against the RCP's own sample questions. If your only evidence is a high percentage on seen items, you have measured familiarity, not readiness (your Q-bank percentage is not your exam score).

Coverage, confidence and calibration: three separate checks

Candidates routinely collapse three different questions into one. Coverage asks whether you have practised each blueprint domain at adequate volume — a question about breadth. Confidence asks whether you feel ready — a subjective signal, useful only when it agrees with your data. Calibration asks whether your confidence matches your accuracy — whether the topics you feel sure about are the ones you actually get right. The candidate most at risk is well covered and highly confident but poorly calibrated: certain of answers that are wrong. Before you stop doing questions, check all three, because they fail independently. High coverage with poor calibration is not a signal to stop; it is a signal to keep practising until confidence and accuracy line up, domain by domain. The most informative column in your coverage table is therefore the gap between how sure you felt and how often you were right.

High-confidence errors: the quadrant that fails candidates

Sort every reviewed item into a grid of confidence against correctness. Low-confidence wrong answers are honest gaps you already know about. Low-confidence right answers are lucky guesses to firm up. High-confidence right answers are genuine strengths. The quadrant that fails candidates is high-confidence wrong — answers you were sure of and still got wrong, usually because a guideline has moved or a misconception is baked in. These never announce themselves: you will not flag them for review, because you did not know you were wrong. The only way to surface them is to review items you answered confidently, not just the ones you missed, and to re-check your highest-yield topics against current UK guidance. Any content-gap audit that ignores this quadrant will declare a candidate ready who is quietly carrying several wrong certainties into the exam.

The stop/continue decision tree

Measured state	Decision
A domain is floored below the rest	Continue new questions — on that domain only
Coverage even but misses recur	Consolidate: rework the error log, don't add volume
Coverage and accuracy solid, pacing poor	Simulate: timed mixed blocks under exam conditions
A concept resists every attempt	Seek teaching, not more questions
Everything green on unseen timed blocks	Stop new questions; rest and maintain

The decision is driven by the measured gap, not by how many questions remain in a bank or how close the exam is.

One-page checklist (copy this)

Worked example (invented data). A candidate has completed 900 questions and sits on 76% overall — reassuring, until the coverage table is filled in. Syndromes: 210 attempted, 81%. Pharmacology: 40 attempted, 58%. Capacity/ethics: 15 attempted, 60%. Continence: 6 attempted, not reviewed in five weeks. The headline 76% hid two floored, high-weight domains and one stale one. The correct action is not "do 100 more mixed questions" but targeted pharmacology and capacity blocks with error-coded review, a continence refresh against current guidance, and only then a return to timed mixed blocks. The number to trust was never the 76%.

Using the RCP sample questions as your calibration anchor

The RCP publishes official DGM sample questions, and these are the single most valuable calibration tool you have — precisely because they are finite. Use them late rather than early. Once your coverage table is even and your unseen accuracy has stabilised, sit the official samples under timed conditions and treat any miss as a high-priority signal, because these items are the closest available proxy for the real paper's style and standard. If you move smoothly through a commercial bank but stumble on the official samples, trust the samples: your bank may simply be pitched easier or styled differently from the exam. If you have already spent the samples early as learning material, you have used your best calibration check on revision — a reason to protect them for this purpose next time.

Turning the audit into a weekly cadence

A content-gap checklist is only useful if it is run repeatedly, not once at the end. Make it a weekly ritual: update the coverage table with the week's questions, mark any domain that has slipped below your floor, note which guidance-sensitive topics you have re-dated, and scan for new high-confidence-wrong items. Ten minutes a week keeps the audit live and stops the familiar late failure mode in which a candidate discovers an untouched domain the weekend before the exam. The checklist is not a finish line to cross once; it is the instrument panel you glance at every week to decide whether the next block should be new questions, consolidation, simulation, teaching or rest.

FAQ

How do I know whether I have covered the full DGM blueprint? You have covered it when the coverage table above is complete and every domain — weighted by its official share of the 100 marks — has adequate unseen sample volume behind it at a stable accuracy, with none floored below the rest. Coverage is a per-domain judgement against the RCP blueprint, not a total-questions-done figure; a bank can be 100% complete and still leave the 40-mark syndromes domain under-tested relative to a 5-mark domain you happened to enjoy.

Can one question bank be enough for DGM? One bank can carry the bulk of your KBA knowledge work if it maps well to the RCP blueprint, but you should still calibrate against the RCP's official sample questions and measure on unseen items from a second source, so your score reflects reasoning rather than recall of seen stems. And "enough for DGM" is a KBA statement only — no bank prepares you for the Part 2 Clinical, so a single bank is never enough for the diploma as a whole.

What should I measure instead of my overall Q-bank percentage for DGM? Measure per-domain unseen accuracy against the blueprint weighting, pacing against the three-hour budget, your high-confidence error rate, retention of missed items after spacing, and agreement with the RCP sample questions. The overall percentage is an average that hides floored domains, over-confidence and slow pacing — the three things that actually fail candidates — so it is the least useful single number you can watch.

When should I stop doing new DGM questions? Stop when the performance checklist is satisfied: two or more unseen timed mixed blocks at a stable standard, no floored domain, comfortable pacing, a quiet error log and calibration against the official samples. Past that point, new questions mostly reassure rather than inform, and consolidation, spaced review or rest will move your real readiness more than another block will.

Which DGM resource should I use for my weakest component? If your weakest area is KBA knowledge in a specific domain, use targeted topic blocks in a blueprint-mapped bank plus unseen transfer practice to confirm the fix. If it is pacing, use timed mixed blocks. If your weakest component is the Part 2 Clinical, no question bank is the answer — that requires coached OSCE practice and feedback against the RCP station rubric, and the modality-gap article in this series (what MCQ banks cannot prepare you for in DGM) sets out how to train it.

Editorial notes and references

Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 21 July 2026. Exam facts are drawn from the RCP DGM pages and the February 2025 syllabus; any vendor figures elsewhere in this series are vendor-reported as of 21 July 2026 and should be re-verified on the product page. Disclosure: iatroX operates a competing UK question bank that includes DGM-level content; its role here is confined to unseen-MCQ measurement and blueprint auditing, and it does not reproduce the Part 2 Clinical OSCE. Corrections via the feedback route on iatrox.com. References: RCP Diploma in Geriatric Medicine and syllabus (rcp.ac.uk); BGS DGM page (bgs.org.uk); and, internally, the iatroX DGM bank, completion is not coverage, Your Q-Bank Percentage Is Not Your Exam Score and the DGM resource decision tree.

Complete a fresh DGM baseline in iatroX →