MRCPI GM Part II Written common mistakes
MRCPI GM Part II Written · Understand the traps and how to avoid them.
Committing to a single diagnosis within the first 30 seconds of a stem and then reading the remaining data only as confirmation, so that a discordant potassium, a normal inflammatory marker, or a drug history that contradicts the working diagnosis is silently discarded.
MRCPI Part II written stems are long and time-pressured, and candidates pattern-match on the first recognisable syndrome (for example 'fever plus murmur equals endocarditis') to buy time. The habit is reinforced because the intuitive answer is often correct, so the cost of premature closure is rarely felt during practice.
What to do instead
Adopt a forced two-diagnosis rule: before answering, write the leading diagnosis and the single best alternative, then identify the one datum in the stem that discriminates between them. If no discriminating datum exists, re-read the last two sentences and the investigation table, which is where examiners usually place the disconfirming finding. Practise on 20 stems by covering the options and generating the differential before looking at them.
Treating the reference ranges printed beside each result as the decision threshold, rather than recognising that the relevant cut-off is context-specific (for example a haemoglobin of 105 g/L being reassuring in a young man with iron deficiency but requiring urgent transfusion in an elderly patient with ischaemic chest pain).
Candidates learn numerical cut-offs in isolation and apply them uniformly; the written paper deliberately presents values that are abnormal but not yet at the classic threshold, or normal but inappropriate for the clinical state.
What to do instead
For every abnormal value, state aloud the patient-specific action it triggers: transfuse, replace, image, withhold, or observe. Build a one-page table of context-dependent thresholds (haemoglobin in acute coronary syndrome, sodium correction rate, potassium in digoxin toxicity, creatinine rise in suspected obstruction) and rehearse the reasoning, not the number.
Managing the elderly patient in the acute medical take using adult-general thresholds and single-organ pathways, ignoring frailty, cognitive impairment, polypharmacy and atypical presentation, so that delirium is attributed to 'old age' and a hip fracture, urinary sepsis or myocardial infarction is missed.
Undergraduate and early postgraduate training is organised around single-system disease in younger patients, and the acute take rewards rapid disposition. Elderly patients present with falls, confusion or reduced mobility rather than classic symptoms, and the candidate lacks a structured screen.
What to do instead
Apply a fixed four-item screen to every patient over 75 in a stem: cognition (baseline and current), mobility and falls, medication review including anticholinergics and sedatives, and hydration/nutrition. When delirium is present, always name and exclude the four commonest precipitants (infection, drugs, pain or retention, metabolic derangement) before attributing it to dementia.
Choosing the most definitive-sounding investigation rather than the next appropriate step in a time-critical sequence, for example requesting magnetic resonance imaging in a patient with suspected acute spinal cord compression who has not yet had steroids, or CT pulmonary angiography before assessing haemodynamic stability.
Written questions reward the correct sequence, and candidates conflate 'best test' with 'next action'. Examination technique that rewards the most sophisticated investigation in other formats carries over inappropriately.
What to do instead
Learn the emergency sequences as ordered lists: suspected cord compression (dexamethasone then urgent imaging within 24 hours), suspected neutropenic sepsis (cultures then antibiotics within one hour), suspected meningitis (antibiotics before imaging unless contraindicated). When answering, ask 'what must happen in the next 60 minutes?' before 'what will confirm the diagnosis?'
Missing the reversible cause in a critically unwell patient because the candidate anchors on the presumed primary diagnosis and does not systematically run through the shock or arrhythmia physiology, so a tension pneumothorax, cardiac tamponade, or opioid toxicity is overlooked in a patient with a plausible alternative explanation.
Critical care questions are embedded in long stems with a dominant narrative, and candidates lack a rehearsed physiological checklist. The written format removes the visual cues that would prompt a bedside clinician.
What to do instead
Memorise and apply the 4 Hs and 4 Ts for every arrhythmia and a shock checklist (hypovolaemic, cardiogenic, distributive, obstructive) to every hypotensive stem. State explicitly which causes the stem excludes and how. Rehearse by writing the checklist from memory before reading each critical care question.
Reading an ECG or chest radiograph description as a single finding and missing the second, decisive abnormality, such as a long QT interval alongside the obvious ischaemic changes, or a widened mediastinum alongside a pneumonic consolidation.
Candidates search for the finding that matches the clinical story and stop when they find it, rather than performing a complete systematic read. In the written paper, the second finding is often what changes management.
What to do instead
Use a fixed reporting sequence every time: for ECG, rate, rhythm, axis, intervals, hypertrophy, ischaemia or infarction, then compare with any previous tracing; for chest radiograph, airway, breathing, circulation, diaphragm, edges, fields, then review areas. In the exam, write the sequence out in the margin and tick each element before selecting an answer.
Selecting drug therapy without checking the contraindication or interaction embedded in the stem, for example prescribing a beta-blocker in a patient with known severe asthma and decompensated heart failure, or adding an NSAID to a patient on lithium and an ACE inhibitor.
Candidates focus on the indication and the guideline first-line agent, treating comorbidity lists as background decoration. The written paper deliberately seeds relative contraindications into the past medical history and medication list.
What to do instead
Before committing to any drug answer, scan the stem for four things: renal function, pregnancy or breastfeeding, interacting medications, and the specific comorbidity that the drug class worsens. Learn the classic exam pairs (beta-blocker and asthma, NSAID and renal impairment or heart failure, metformin and contrast or eGFR below 30, aminoglycoside and myasthenia).
Failing to convert a clinical picture into the correct severity classification or score, so that a patient with community-acquired pneumonia and a CURB-65 of 3, or a upper gastrointestinal bleed with a Glasgow-Blatchford score of 12, is managed in the wrong setting or discharged inappropriately.
Scores are memorised as lists rather than as decision tools, and candidates cannot apply them reliably under time pressure. The written paper tests the management consequence of the score, not the score itself.
What to do instead
For each core score (CURB-65, NEWS2, Glasgow-Blatchford, Child-Pugh, Wells, CHA2DS2-VASc, qSOFA), learn three anchor points and the action each triggers: discharge, ward, high dependency, or immediate intervention. Practise by calculating the score on real or simulated cases in under 60 seconds and stating the disposition.
Misinterpreting the neurological examination in the stem, particularly localising a lesion to the wrong level (peripheral versus central facial weakness, upper versus lower motor neuron signs, cortical versus lacunar stroke syndromes) and therefore selecting the wrong investigation or treatment.
Neurological localisation requires spatial reasoning that is difficult to convey in text, and candidates rely on the named syndrome rather than reconstructing the anatomy from the described signs.
What to do instead
Reconstruct the lesion from the signs: draw a simple neuroaxis (cortex, internal capsule, brainstem, cord, root, plexus, nerve, muscle) and place the findings on it. Learn the discriminators explicitly: forehead sparing in upper motor neuron facial palsy, dissociated sensory loss in cord lesions, and the lacunar syndromes (pure motor, pure sensory, sensorimotor, ataxic hemiparesis).
Answering the question as written in the wrong tense or frame, for example offering long-term management when the stem asks for the immediate next step, or selecting a diagnosis when the question asks for the most likely causative organism, and thereby losing marks on otherwise known content.
Under time pressure, candidates read the clinical vignette closely and the final question line quickly, and the option list often contains a plausible answer to the wrong question. This is the single most recoverable source of lost marks.
What to do instead
Read the final line of the stem first, underline the operative words (next, most likely, initial, contraindicated, single most appropriate), and only then read the vignette. After selecting an answer, check it answers the underlined question and not a related one. Time yourself on 30 stems using this order and compare accuracy with your usual approach.