This audit is for UK final-year medical students and Foundation Year 1 doctors who are already paying for PassMedicine and want to know what its analytics can and cannot tell them about the Prescribing Safety Assessment (PSA). The principal limitation to grasp before you read another dashboard: PassMedicine does not sell a dedicated PSA bank. Its prescribing material is bundled inside the finals/UKMLA subscription, and its analytics compare you with other finals candidates on single-best-answer questions, not on PSA-format prescribing, review and calculation tasks. Read that way, a rising percentage is a knowledge check, not a readiness signal.
That distinction is the whole article. PassMedicine is a well-built finals and UKMLA resource, and the therapeutics knowledge it drills is genuinely relevant to prescribing safety. But the PSA is a different assessment with its own task formats and its own standard, and a finals dashboard was never designed to model it. Keep the two jobs separate and PassMedicine earns a place in your revision stack; conflate them and you can walk into the PSA with an 80% home-screen average and no calibrated sense of whether you can prescribe safely under time pressure.
What PassMedicine offers for the PSA right now
The table below records what we could confirm on 19 July 2026. Vendor-reported figures are labelled; where we could not confirm a number, we say so. Treat it as a prompt to check the current product page, not a substitute for it.
| Item | What we found (verify on the product page) |
|---|---|
| Dedicated PSA bank? | No standalone PSA product is listed. PSA-relevant material appears as an "OSCE & Prescribing Skills Assessment resource" bundled inside the finals/UKMLA subscription (vendor wording). |
| Finals/UKMLA bank size | "Over 11,000 SBA questions for medical finals and the UKMLA + high-yield textbook" (vendor-reported). |
| Question format | Single best answer, plus the bundled resource above. Not the PSA's eight distinct task formats. |
| Analytics | Per-topic performance breakdown, first-attempt versus repeat accuracy, and a performance histogram comparing you with other users, updated daily (vendor-reported). |
| Personalisation | A "Knowledge Tutor" that behaves like spaced repetition, plus learner-chosen topic filters. Not advertised as a proprietary adaptive-difficulty engine. Verify. |
| Predicted PSA score / PSA percentile | Not surfaced against a PSA candidate cohort as far as we can confirm; the histogram compares finals users. Verify. |
| Access period | Not confirmed here; check the current tiers on the product page. |
| Price | Not displayed on the summary page we checked; third-party guides commonly cite around £30 for twelve months (third-party-reported). Verify. |
Two honest findings sit inside that table. First, there is no PSA product to audit in the strict sense — only a finals product with prescribing content bundled in, under a name that is not even the official exam title. Second, the analytics are cohort-comparison and spaced-repetition tools built around an SBA finals bank, so they tell you how you compare with other finals users, not whether you are ready for the PSA. Neither makes PassMedicine a poor product; both change how you read its numbers.
The PSA blueprint: what the exam actually measures
The PSA is set by the exam body and its format is fixed. It is 60 items in two hours for 80 marks. It samples eight question types: Prescribing, Prescription Review, Planning Management, Communicating Information, Calculation Skills, Adverse Drug Reactions, Drug Monitoring and Data Interpretation. Prescribing items carry the most weight, marked in two halves, drug choice and the dose, route and frequency. Content is drawn from seven clinical domains: Medicine, Surgery, Elderly Care, Paediatrics, Psychiatry, Obstetrics and Gynaecology, and General Practice. The pass mark is set by Modified Angoff standard-setting, so it is not a fixed percentage you can revise towards, and it is not published in advance.
Two features of the exam matter enormously for how you read practice analytics. The PSA is open-reference: an online formulary is available in the exam interface, so it does not reward memorising doses you can look up. It rewards knowing when to look, what to check, and how to convert that into a safe, complete prescription under time pressure. And it tests prescribing safety, not recall — the failure modes it hunts for are the ones that harm patients: the missed interaction, the un-adjusted dose in renal impairment, the decimal-point slip in a paediatric calculation. For the wider principle of matching practice to the real blueprint rather than a bank's internal averages, the iatroX pillar on why completion is not coverage sets out the method.
Every PassMedicine metric, defined for the PSA
You cannot audit a number you have not defined, so define each metric by what it measures and what it cannot see.
First-attempt accuracy is the percentage correct the first time you meet an item — the closest PassMedicine gets to an unbiased knowledge estimate, because it is not yet contaminated by memory of the answer. For the PSA, it still only reflects SBA-format recall on a finals bank, not prescribing-task performance.
Repeat accuracy is your percentage on items you have seen before. It climbs quickly and reassuringly, and it is the single most misleading figure on the dashboard, because it substantially measures recognition memory of specific stems, not transferable prescribing judgement.
Percentile or peer comparison is the daily histogram placing you against other users of the same bank — a self-selected finals cohort, not a PSA cohort. It shows where you sit among people doing PassMedicine questions, not among PSA candidates.
Predicted score is a metric PassMedicine does not, as far as we can confirm, present for the PSA. Treat the absence as honest: no bank can predict a Modified Angoff pass mark, and you should be suspicious of any product that claims to.
Coverage is how much of the bank you have attempted — completion, not blueprint coverage. Finishing every SBA does not mean you have practised calculation skills, prescription review or drug monitoring, because those PSA formats are barely present in an SBA bank.
Difficulty is the item's historical hardness across users — useful for pacing, but irrelevant to the PSA standard, which is set by expert judgement on real PSA items.
Time per item is the one metric that transfers cleanly. If you are slow on knowledge SBAs, you will be slower still on multi-step prescribing and calculation items, so treat any time pressure you feel here as a floor, not a ceiling.
Why the feed biases the numbers
PassMedicine is not a strongly adaptive engine, but its personalisation still bends the statistics. Spaced repetition resurfaces items you got wrong, and learner filters let you drill weak topics, so each week you disproportionately re-see material you have already partly learned. That inflates repeat accuracy and flatters the average without moving your true first-attempt ability much, and the daily histogram compounds it, because heavy users climb through sheer repeated exposure rather than new competence. The result is a percentage that rises smoothly while your ability to handle an unseen prescribing task barely changes — exactly the trap the iatroX article "Your Q-Bank Percentage Is Not Your Exam Score" is written to prevent, with extra force here because the bank was not built for this exam.
Blueprint audit: map your attempts to the PSA, not the home screen
The home-screen average hides the only comparison that matters: how your attempted questions map onto the PSA blueprint. Build the matrix below yourself from your attempt log, once for the eight question types and once for the seven domains, and mark each cell green, amber or red for volume and for first-attempt accuracy.
| PSA question type | Can a finals SBA bank exercise it? | What you must add elsewhere |
|---|---|---|
| Prescribing (write-up) | Poorly; SBAs test recognition, not writing a safe, complete prescription | Dedicated prescribing-format practice with open-formulary lookup |
| Prescription Review | Partly, via therapeutics SBAs | Full drug-chart review under time pressure |
| Planning Management | Reasonably | Little; this is a genuine strength area |
| Communicating Information | Weakly | Structured counselling and information tasks |
| Calculation Skills | Barely | Deliberate calculation drills, including paediatric weight-based dosing |
| Adverse Drug Reactions | Reasonably | Little; strong overlap with therapeutics SBAs |
| Drug Monitoring | Partly | Monitoring-specific items with target ranges |
| Data Interpretation | Partly | Result-to-action items under time constraint |
The pattern is consistent: PassMedicine covers the knowledge-heavy, recognition-shaped parts of prescribing well and the procedural, calculation and write-up parts poorly, because an SBA cannot reproduce them. The domains split similarly — Medicine, Surgery and Elderly Care are well served, while obstetric, paediatric and psychiatric prescribing are lower-volume and easy to under-practise.
The readiness test: five conditions
A number is a readiness signal only when all five conditions hold at once. The items are unseen, so recognition memory is excluded. The block is timed at PSA pace, roughly two minutes an item. It is mixed across question types and domains, not filtered to your comfort zone. You work with no assistance beyond the online formulary the real exam supplies. And the sample is large enough to be stable — repeated mixed blocks over days, not a single lucky set. PassMedicine's default modes satisfy few of these, so a readiness read has to be constructed deliberately, or taken somewhere built for the exam.
Override rules: what to force into your week
Because the feed will happily let you graze on comfortable therapeutics SBAs, set hard floors that force the under-served material to the surface. Force calculation skills every day, including infusion rates and paediatric weight-based doses, because a single decimal slip is a fail-shaped error. Force prescription review under a timer, reading whole drug charts for interactions, duplications and contraindications. Force drug-monitoring items with real target ranges, and adverse-drug-reaction recognition beyond the obvious. And force the low-volume domains — obstetric, paediatric and psychiatric prescribing — that a finals feed samples thinly. If the platform cannot serve a format at volume, that is your signal to bring in a resource that can.
Worked dashboard example
Suppose your screen shows an overall 82%, repeat accuracy 91%, first-attempt 68%, therapeutics 84%, and no calculation or prescription-review category because the SBA bank does not carry them. Do not read 82% as a pass probability; it is not one. Read the 23-point gap between repeat and first-attempt as your recognition-memory inflation, and the missing categories as unmeasured risk. Next week's quotas then write themselves: 40 unseen mixed prescribing-format items, 30 timed calculation drills, 20 prescription-review charts, 15 drug-monitoring items, and 15 items each in obstetric, paediatric and psychiatric prescribing. Almost none of that can be filled inside the finals SBA bank — which is the point of auditing rather than trusting the average.
A seven-day pattern
Use PassMedicine for one defined job — consolidating therapeutics knowledge — and an exam-format bank for unseen transfer and measurement. We make no claim about any platform's internal algorithm; this is simply a division of labour.
| Day | PassMedicine (one job: knowledge) | iatroX (unseen PSA-format transfer) |
|---|---|---|
| 1 | 30 therapeutics SBAs, first-attempt only | Baseline 20-item mixed PSA block, timed |
| 2 | Review errors, read the linked notes | 15 calculation-skills items |
| 3 | 30 SBAs on a weak domain | 15 prescription-review charts |
| 4 | Spaced-repetition catch-up | 20 mixed items, timed |
| 5 | 30 SBAs, ADRs and monitoring focus | 15 drug-monitoring and data-interpretation items |
| 6 | Light; consolidate notes | 30-item mixed block at full PSA pace |
| 7 | Rest or read | Re-audit: first-attempt trend, coverage, pacing |
The logic follows the two-Q-bank rule: one bank teaches, the other measures on unseen material, and you never burn your measurement stock by pre-reading it.
Three mistakes this audit is designed to stop
The first mistake is treating a finals average as a PSA score; they measure different things on different formats against different standards, and the finals number is systematically optimistic. The second is completing the bank and calling it coverage; completion is volume, not proof that all eight formats and seven domains have been practised at pace. The third is verifying against the wrong source. During revision, ground your prescribing in the SmPC/eMC (the electronic Medicines Compendium) alongside NICE and CKS, and practise navigating the online formulary the exam provides rather than memorising figures you are allowed to look up.
Decision checklist: continue, supplement, switch or stop
Continue with PassMedicine if you value it for therapeutics knowledge and you are honest that it is a finals tool. Supplement — usually the right answer — by adding an exam-format bank for the prescribing, calculation, review, monitoring and data-interpretation tasks the SBA format cannot reproduce. Switch your primary PSA practice to an exam-format resource if the finals bank is crowding out format-specific work. Stop only when your unseen, timed, mixed first-attempt performance across all eight types is stable and you have calibrated against official practice material — never on bank completion alone.
Bottom line
PassMedicine is a strong finals and UKMLA resource and a reasonable place to consolidate the drug knowledge the PSA draws on, so it is worth including in the revision stack for that job. But it is not a PSA bank, its analytics were not built to model the PSA, and its rising percentage is not a readiness signal. Use it for what it is, measure readiness on unseen exam-format blocks, and verify your prescribing against the SmPC/eMC. Do that and the dashboard becomes a useful input; trust it as a score and it becomes a liability.
FAQ
Is PassMedicine enough for the PSA on its own? No. PassMedicine does not sell a dedicated PSA bank; its prescribing content is bundled inside the finals/UKMLA subscription as single-best-answer material, and an SBA bank cannot reproduce the PSA's prescribing write-up, prescription-review, calculation-skills or data-interpretation formats. It is strong for the underlying therapeutics knowledge, but pair it with exam-format practice and the official practice papers before relying on it.
Which PSA component does PassMedicine not reproduce well? The Prescribing and Calculation Skills components are the weakest fit. Writing a safe, complete prescription — the correct drug, dose, route and frequency, checked against the online formulary the exam supplies — and performing multi-step calculations under time pressure are procedural tasks that a recognition-based SBA simply cannot recreate, so these need dedicated exam-format drilling elsewhere.
How many PassMedicine questions should I complete per day for the PSA? There is no magic number, and daily volume is the wrong target. A sustainable rhythm is around 20 to 40 first-attempt SBAs for knowledge, paired with a smaller set of exam-format prescribing and calculation items for transfer. Prioritise first-attempt quality and full error review over raw completion, because repeat accuracy inflates fast and tells you little.
When should I stop using PassMedicine and move to mixed mocks? Move to timed mixed blocks once your first-attempt accuracy on unseen material is stable, rather than when you have finished the bank. If your repeat accuracy is high but your first-attempt figure is stalling, that is the signal that further SBA drilling is producing recognition memory rather than competence, and your time is better spent on mixed, exam-format practice.
How should I combine PassMedicine with iatroX without duplicating practice? Assign each tool one job. Use PassMedicine to consolidate therapeutics knowledge, and use iatroX only for unseen, timed, exam-format PSA blocks you never pre-read, so it stays a clean measurement layer. Keeping the roles separate follows the two-Q-bank rule and stops you burning your only source of a genuine readiness signal.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026. Platform figures described here are vendor-reported or third-party-reported where labelled and change without notice, so confirm the current question count, access period, price and feature set on the product page before you rely on them. Disclosure: iatroX operates a competing question bank; in this article iatroX's role is confined to unseen, exam-format measurement — the job the audited finals product does not claim to do — and not to the therapeutics teaching PassMedicine does well. Corrections are welcome via the feedback route on iatrox.com.
References: the official PSA information at prescribingsafetyassessment.ac.uk; the PassMedicine product pages; and the iatroX pillars on interpreting a Q-bank percentage and building a blueprint-coverage matrix. Compare tools on the iatroX comparison hub and the PSA landing page.