This is the implementation companion to our PassMedicine MRCGP AKT analytics audit — read that for what the dashboard measures; read this for the week-by-week workflow. The job is narrow: a concrete PassMedicine routine for the AKT that stops a clinically-driven practice loop from starving the two non-clinical strands that decide many results, and tells you when to stop drilling and start simulating at the new exam pace.
What you are working with
PassMedicine offers low-cost fixed-term AKT access with a large single-best-answer bank, timed and revision modes, textbook notes and peer-comparison analytics; no AI tutor is advertised, so you configure the loop. Verify current counts and price on the product page. The exam is the MRCGP Applied Knowledge Test — from October 2025, 160 single-best-answer questions in 2 hours 40 minutes, four sittings a year at Pearson VUE, split roughly 80% clinical medicine, 10% evidence-based practice and 10% organisational content, all in a UK general-practice context.
Week structure at a glance
| Phase | Weeks (typical 10-week run) | Primary mode | Exit trigger |
|---|---|---|---|
| Baseline | 1 | Unseen three-strand sample | ~100 items logged across strands |
| First pass | 2–6 | Strand-floored revision + notes | Every strand above its floor |
| Consolidation | 7–8 | Timed mixed blocks + error transfer | First-attempt accuracy stable 2 weeks |
| Simulation | 9–10 | Full mixed timed mocks at one minute/item | Pacing on target; go/no-go |
Baseline all three strands before drilling
Spend week one on an unseen sample that deliberately includes clinical, evidence-based-practice and organisational items, before filtered revision biases the picture. Two of those strands are the ones candidates discover too late they never practised, so baselining them from day one is the whole point: you cannot floor a strand you have not measured.
Three-strand floors — not just clinical
Set a minimum question count per strand and enforce it against a rising overall percentage. The clinical 80% will look after itself because it is where you are comfortable; the floors exist to protect the evidence-based-practice strand (statistics, critical appraisal, study design) and the organisational strand (certification, fitness-to-practise, UK regulatory and practice-management content), which a clinically-driven loop starves. A non-zero weekly floor on each non-clinical strand is the single highest-value override in this workflow.
Error taxonomy that dictates the fix
Sort each clinical miss into the six types — knowledge gap, misread stem, premature closure, guideline error, calculation error, time-pressure error — and match the fix to the type. Add two AKT-specific patterns. Statistics errors (in the evidence strand) need worked-calculation practice with every step checked, not more clinical revision. Organisational errors need the specific UK framework read and dated, because this content is pure knowledge with a right answer. Default to transfer questions over immediate repeats.
The mixed-block switch and exit criteria
Move from strand-filtered practice to timed random blocks when every strand floor is met, first-attempt accuracy on unseen material has held for two weeks, and pace sits at or under the new one-minute-per-item budget. Then invert to mostly-mixed. Exit criteria are five things together: all three strand floors met; first-attempt accuracy stable across two weeks; pacing on target; retention holding on spaced review; and a calibration check against official RCGP sample material — not bank completion, and not a full row of green.
A seven-day pattern for busy trainees
Monday: 40 PassMedicine clinical questions in a weak area, explanations read, misses taxonomised. Tuesday: a dedicated evidence-based-practice block — statistics and appraisal — every calculation checked. Wednesday: a timed, unseen 40-question mixed block across all three strands in iatroX's free MRCGP AKT bank for a first-attempt signal. Thursday: an organisational-content session plus error transfer. Friday: 40 clinical questions, timed at one minute each. Saturday: a full timed three-strand simulation, same-day review by strand. Sunday: rest. PassMedicine drills clinical volume; iatroX measures across all three strands on unseen items; the non-clinical strands get protected time they never get by default.
A worked three-strand example
Six weeks out, suppose PassMedicine shows overall 70%, but split by strand it reads very differently: clinical medicine 76% over 800 attempts, evidence-based practice 52% over 40, and organisational content 15 questions attempted. A candidate reading the overall 70% feels on track. Read by strand, the plan is obvious and uncomfortable: the clinical majority is maintained, not grown, because at broad coverage its marginal value is low; evidence-based practice at 52% over a thin sample is a live liability that takes three dedicated blocks this fortnight, every calculation worked step by step; and organisational content at 15 attempts is essentially unexamined and gets its first proper sessions. The overall 70% was doing exactly what an aggregate does — hiding the two strands that discriminate most at the borderline behind the one that discriminates least.
The statistical reason is worth stating plainly: because the clinical 80% is where every candidate is well-resourced and heavily practised, clinical scores cluster and rarely decide a borderline result, whereas the under-practised evidence and organisational strands vary enormously between candidates and therefore carry disproportionate weight at the margin. Once clinical coverage is broad and stable, an hour moved from clinical drilling to critical appraisal or organisational content is almost always the higher-yield choice — which is exactly the move the strand floors force and the comfort of the clinical feed resists.
A note on the format change
The move from 200 questions in 3 hours 10 minutes to 160 questions in 2 hours 40 minutes (from October 2025) kept the content weightings but tightened the clock slightly and reduced the number of items across which a weak strand can be diluted. With fewer questions, each strand has less room to absorb a bad run, which raises rather than lowers the value of covering the non-clinical strands properly. Confirm the current format for your sitting on the RCGP pages, and set your timed-block pace to the current per-item budget rather than an older one.
Continue, supplement, switch or stop
Continue while all three strands improve on unseen timed material. Supplement the moment your evidence or organisational strands lag — the highest-value intervention available. Switch only for a measurable gap (our AKT hub is here). Stop accumulating clinical volume in the final fortnight once all three strands meet their floors; simulate the whole paper at the new pace.
Three mistakes this workflow is designed to stop
First, letting the overall percentage stand in for three separate readings. The AKT is really three exams — clinical, evidence-based practice and organisational — and a single average is dominated by the clinical strand that discriminates least at the borderline; strand floors force you to read each on its own. Second, treating the clinical majority as where the work is. Once clinical coverage is broad, its marginal value is low, and an hour moved to critical appraisal or organisational content is almost always the higher-yield choice — but it is also the less comfortable one, which is why the floors have to be defended against the pull of the clinical feed. Third, drilling untimed at the old pace. The format change tightened the clock and left less room for a weak strand to be diluted, so every timed block should run at the current per-item budget, and pacing should be an exit criterion in its own right rather than a final-week afterthought.
Frequently asked questions
Is PassMedicine enough for MRCGP AKT on its own? It can carry the clinical majority well, but its analytics are blind to the evidence-based-practice and organisational thirds unless you deliberately filter, so pair it with dedicated practice in those strands and unseen timed mixed blocks.
Which MRCGP AKT component does PassMedicine not reproduce well? The evidence-based-practice and organisational strands — around a fifth of the exam and a disproportionate share of failures — which specialty-clinical drilling systematically under-samples.
How many PassMedicine questions should I complete per day for MRCGP AKT? 40–60 on study days, but composition is the point: protect weekly evidence and organisational blocks regardless of the clinical count, and time practice at the new one-minute-per-item pace.
When should I stop using PassMedicine and move to mixed mocks? When all three strands meet their floors, unseen accuracy is stable and pace is under a minute per item — the final two to three weeks, given to full timed simulation.
How should I combine PassMedicine with iatroX without duplicating practice? PassMedicine for clinical drilling; iatroX (free for MRCGP AKT) for unseen adaptive measurement deliberately sampling all three strands and Socratic repair on misses — so your readiness signal reflects the whole paper, not the clinical majority.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; the AKT format (from October 2025: 160 questions, 2 hours 40 minutes) and weightings are per the RCGP; PassMedicine details are vendor-published — verify before purchase and confirm your attempt limits. Disclosure: iatroX offers a free competing MRCGP AKT bank. Corrections via the feedback route on iatrox.com. References: RCGP Applied Knowledge Test pages (rcgp.org.uk); PassMedicine product pages; related reading: the PassMedicine MRCGP AKT analytics audit and how to pass the AKT first time.