How to read vague MRCEM SBA stems

Facts last verified 30 May 2026.

TL;DR — the dissection framework in plain text

When an MRCEM Intermediate SBA stem feels vague, it usually isn’t. The information you need is there; it just hasn’t been labelled with a buzzword. Use this six-step dissection on every stem you can’t answer in under 30 seconds:

1. Read the lead-in first. The final sentence tells you what task you are doing — diagnosis, next investigation, immediate management, definitive management, most likely cause, single best disposition. Re-read the stem through that lens.
2. Lock the demographics. Age, sex, pregnancy status, comorbidities and drugs. These are almost never decoration.
3. Strip the stem to bullet points. One line per fact. Vagueness usually evaporates once you stop reading prose and start reading data.
4. Name the syndrome in your own words before you look at the options — the “cover test”.
5. Find the eliminator. There is always one detail (a number, a drug, a time window, a comorbidity) that kills two or three options. Ask: which fact in this stem changes the answer if you remove it?
6. Re-read the lead-in verb. “Most appropriate” ≠ “next” ≠ “definitive” ≠ “initial” ≠ “most likely”. Mis-mapping the verb is the single most common reason candidates lose marks on stems they actually understood.

If you can do all six in under two minutes, vague stems stop feeling vague. They start feeling under-labelled, which is a different problem — and a solvable one. See also our guide to an MRCEM SBA pacing strategy that actually works.

Why do MRCEM Intermediate SBAs feel so vague in the first place?

The Intermediate SBA tests applied clinical reasoning, not recall. RCEM writes 180 questions blueprinted across the Specialty Learning Outcomes, and a deliberate design choice is to describe a real ED patient rather than hand you a syndrome label. Real patients don’t arrive in resus saying “I have a posterior circulation stroke”. They arrive dizzy, nauseated, with a slightly off gait.

So what reads as “vague” is usually one of three things:

• Under-labelled pattern recognition. The features are there but the diagnosis isn’t named. You’re expected to translate findings into a syndrome.
• A lead-in mismatch. You diagnose the patient correctly but answer a different question to the one being asked (next step vs definitive, for example).
• An incomplete picture by design. The College wants to know what you would do safely with the information available — not what you would do with a full work-up.

In all three cases the cure is the same: stop searching for buzzwords, start dissecting the stem.

What is the dissection framework, step by step?

Step 1 — Read the lead-in before the stem

The lead-in is the final sentence (“What is the single most appropriate next investigation?”). It defines the task. Reading it first means every fact in the stem is filtered through “does this affect the next investigation?” rather than “does this affect the diagnosis?”. The same stem with two different lead-ins has two different correct answers.

Step 2 — Lock the demographics and modifiers

Age, sex, pregnancy, weight, drugs, allergies, comorbidities, returned-traveller status, immunosuppression. A 26-year-old with pleuritic chest pain on the combined oral contraceptive pill is a different question to a 26-year-old with pleuritic chest pain after marathon training. If a modifier is in the stem, it is load-bearing.

Step 3 — Convert prose to bullet points

On scrap paper or in your head, rewrite the stem as a vertical list of facts. Vague stems usually contain six to ten data points hidden in two paragraphs. Once they’re vertical, you can see which ones cluster — tachycardia plus hypotension plus warm peripheries plus raised lactate plus recent urinary symptoms is no longer vague, it’s urosepsis.

Step 4 — Name the syndrome before looking at the options (the cover test)

BMJ Careers recommends the “cover test”: physically cover the five options and answer using only the stem. Your unprompted answer is much more likely to be correct than your second-guessed answer after seeing four plausible distractors. If you can’t name the syndrome here, you have a knowledge gap, not a comprehension problem — useful information for revision.

Step 5 — Find the eliminator

Every well-written SBA contains at least one detail that differentiates the correct answer from the most plausible distractor. The eGFR that rules out metformin, the “30 minutes ago” that rules out thrombolysis being futile, the pregnancy that rules out CTPA as first-line, the platelet count that rules out aspirin. Ask: which single fact, if I deleted it, would change my answer? Vague stems often hide the eliminator inside a clause that looks like background (“he takes warfarin for atrial fibrillation”).

Step 6 — Re-read the lead-in verb

Map the verb before committing:

• “Most likely diagnosis” — pattern recognition; ignore management options entirely.
• “Most appropriate initial investigation” — bedside or first-line, not the gold standard.
• “Most appropriate next step in management” — what you would do right now, before anything else.
• “Most appropriate definitive management” — the disease-modifying treatment, not the holding measure.
• “Most likely underlying cause” — aetiology, not the immediate presenting problem.
• “Most appropriate disposition” — admit, refer, discharge, observe — not what to do clinically.

Which lead-in verbs do candidates most often misread?

• Initial vs definitive. Initial = first thing you do in ED. Definitive = the treatment that fixes the underlying problem. Aspirin is initial in STEMI; primary PCI is definitive.
• Next vs most appropriate. “Next” is sequential — what happens after what’s already been done. “Most appropriate” is judgement-based — the best choice from the options regardless of what came before.
• Most likely diagnosis vs most likely underlying cause. A patient with hyperkalaemic VT has VT as the diagnosis but hyperkalaemia as the cause.
• Investigation vs diagnostic test. ECG is an investigation. Troponin is a diagnostic test for myocardial injury. “Most useful investigation to confirm the diagnosis” means the diagnostic test, not the screening one.
• Best vs only correct. SBA never asks for the only correct answer; it asks for the single best. Two options can both be reasonable and one is still right.

Worked examples: vague stem → reframed interpretation → correct answer logic

Five stems of the kind candidates describe as “vague”. The reframed column shows what the stem is actually testing once dissected.

How do I stop misreading the stem under exam pressure?

Vague-stem errors are almost always pace errors. Candidates who pass tend to do two things differently:

• They time their reading deliberately. 180 questions over four hours is roughly 80 seconds per question. Spending 30–40 seconds on the stem is normal; trying to do it in 10 is where misreads happen.
• They categorise every wrong answer in revision. Label the miss — knowledge gap, stem misread, changed answer, or weak elimination. Most candidates who feel the exam is “full of vague stems” have a stem-misread rate they’ve never measured.

If your stem-misread rate is above 15% of wrong answers, the dissection framework is the highest-yield revision change you can make. If it’s below 5%, your problem is knowledge, not technique. See also our guide to common FRCEM SBA mistakes that cost marks.

How is this different from spotting trick questions?

Trick questions contain a deliberate distractor engineered to mislead — a buzzword that points the wrong way, a number just outside a threshold, a drug contraindicated in a population the candidate didn’t notice. BMJ Careers’ line is that this is rare in well-written SBAs: “rather than suspecting some Machiavellian subterfuge, read the whole question and answer as it is written.”

Vague stems are different. There’s no trick; the information is all there but it hasn’t been pre-digested. Your job is to dissect, not to second-guess:

• Trick-question fix: read every option, beware buzzwords that don’t explain all the stem findings, prefer the answer that is safe in this specific patient.
• Vague-stem fix: dissect with the six-step framework, name the syndrome before looking at options, and map the lead-in verb precisely.

What does a good revision routine for stem comprehension look like?

The single most useful drill is “cover-and-dissect”. Pick a question bank, work through 20 questions, and for every one:

1. Read only the lead-in. Write down what task is being asked.
2. Read the stem. Write down five to eight bullet points of facts.
3. Without looking at the options, write your best answer in two or three words.
4. Identify what you think the eliminator is.
5. Now look at the options and pick one.
6. Mark and review. Label any miss: knowledge gap, stem misread, changed answer, or weak elimination.

Twenty questions like this teaches more than 100 done at normal speed. After two to three weeks, the framework becomes automatic and you can do it in the exam without writing anything down.

FAQ

Are MRCEM Intermediate SBA stems actually vague, or am I just under-prepared?

Both can be true, and the dissection framework helps you tell which. If you dissect a stem and still can’t name a syndrome, that’s a knowledge gap — revise the topic. If you can name the syndrome but pick the wrong answer because you read the lead-in as “definitive” when it said “initial”, that’s a comprehension problem — drill the verbs.

How long should I spend reading a stem?

With 180 questions over 240 minutes, the average is around 80 seconds per question. Spend 30–40 seconds dissecting the stem, 20–30 reviewing options, and the rest committing. Flag anything you can’t answer in under 90 seconds and come back. See also our guide to an FRCEM SBA flagging strategy (mark and return).

Should I read the stem or the lead-in first?

The lead-in first. It tells you what task you’re doing, which then filters how you read the stem. Most candidates who reverse this end up answering a different question to the one being asked.

How do I tell “next step” from “most appropriate management”?

“Next step” is sequential — what should be done immediately after what’s already happened in the stem. “Most appropriate management” is judgement-based — the best management option overall. If the stem says “he has been given 1g paracetamol”, then “next step” means after that dose; “most appropriate management” could still be the paracetamol itself if the alternatives are worse.

What does “most appropriate” really mean in an SBA?

Usually the safest, most evidence-based, most immediately actionable option for the specific patient described. It rarely means the gold standard if the gold standard isn’t available in ED. Prefer the option that explains all the findings in the stem and is appropriate for the patient’s demographics and comorbidities.

Why are some options technically correct but still wrong?

Because SBA asks for the single best answer. Two options can both be defensible and one is still the answer the examiner is testing. Find the eliminator — the fact in the stem that prefers one option over the other. If you can’t find one between two plausible options, re-read the lead-in: the verb usually disambiguates them.

How do I handle pregnancy, paediatric and elderly modifiers?

Treat them as load-bearing. Pregnancy changes imaging (CTPA vs leg dopplers), drug choice (no warfarin, no ACE inhibitors) and physiology baseline. Paediatric stems use weight-based dosing and different thresholds. Elderly stems often hide polypharmacy and renal impairment in the comorbidity list. Whenever you see one of these modifiers, ask explicitly: what does this change about the answer?

Is it worth practising with question banks from outside RCEM?

Mixed. RCEMLearning’s SBAs are closest in style. External banks vary in quality — some have wrong answers, some are written for other exams. For stem-comprehension practice the source matters less than the dissection process; even a poorly worded bank teaches you to extract facts. For final-week revision, prioritise RCEMLearning and any official mocks.

Does going with my first instinct really work?

Yes — first instincts are right more often than reconsidered ones, provided you’ve actually dissected the stem first. The cover test exploits this: name the syndrome before you look at options, and your first answer is usually right. Don’t change it without a concrete reason.

What’s the single highest-yield change I can make in the final two weeks?

Drill the lead-in verbs and run cover-and-dissect on 20 questions a day. Stop chasing new content; consolidate stem technique. Most candidates who fail do so by a small margin (often six marks or fewer) and almost all of those marks are recoverable through better stem comprehension rather than more knowledge.

Next step

If you want to drill these dissection skills on stems written in the exact RCEM SBA style, with the lead-in verbs and demographic eliminators built in, head to emfinalexams.com and work through the MRCEM Intermediate SBA bank.

Facts last verified 30 May 2026.

Ready to build your plan? EMF Premium gives you all 40,000 questions and 20 mocks for £59 — one payment, six months' access.