PGIM MD Medicine common mistakes
PGIM MD Medicine · Understand the traps and how to avoid them.
Choosing the single most likely diagnosis and stopping there, rather than committing to the best next management step the stem actually asks for (e.g., answering 'pulmonary embolism' when the question asks what to do first in a haemodynamically unstable patient).
Stems are long and the diagnosis is the most salient feature, so candidates anchor on it and answer the 'what is it' question in their head instead of re-reading the final line, which often asks for investigation, immediate treatment, or contraindication.
What to do instead
Before answering, underline the final interrogative clause and classify it as diagnosis, next investigation, immediate management, or mechanism. If the question asks for management, discard any option that is purely diagnostic even if correct as a diagnosis. Practise by covering the options and writing your own answer to the exact question asked, then matching.
Applying a single-disease heuristic to multi-system stems, ignoring a second diagnosis that explains the discordant findings (e.g., attributing anaemia and renal impairment in an elderly patient to one cause rather than recognising a paraprotein-related picture or a dual pathology).
Candidates trained on single-best-answer pattern recognition default to Occam's razor and dismiss findings that do not fit their first hypothesis instead of treating them as deliberate clues.
What to do instead
For every stem, list each abnormal finding in a column and force yourself to account for all of them in the chosen answer. If one finding is unexplained, the answer is probably wrong. Rehearse 'two-disease' vignettes (e.g., diabetes plus infection, malignancy plus hypercalcaemia, heart failure plus thyrotoxicosis).
Interpreting investigation values in isolation without reference to the clinical context, timing, or the assay's limitations (e.g., calling a raised D-dimer diagnostic of PE, treating a normal anion gap as excluding ketoacidosis, or ignoring that troponin rises in renal failure).
Revision tends to memorise cut-offs and 'diagnostic' tests rather than sensitivity, specificity, and pre-test probability, so candidates treat laboratory numbers as binary truth.
What to do instead
For each common test, write a one-line card stating what it rules in, what it rules out, and situations that produce false positives or negatives. Practise calculating and interpreting derived values (anion gap, corrected calcium, osmolal gap, A-a gradient, Winters' formula) rather than recalling them.
Under-triaging the acutely unwell patient: choosing a definitive diagnostic test or a ward-based plan when the stem describes a peri-arrest or time-critical situation that demands immediate resuscitation or a specific emergency intervention.
Candidates read the detailed history first and build a diagnostic formulation, losing sight of the physiological derangement (hypotension, hypoxia, reduced GCS) that should dominate the answer.
What to do instead
Adopt an ABCDE screen on first read: note vital signs and conscious level before anything else. If any are abnormal, the correct answer is almost always the immediate stabilising step (airway, oxygen, fluids, antibiotics within the hour, decompression, antidote) rather than investigation. Drill time-critical scenarios: sepsis, tension pneumothorax, hyperkalaemia, status epilepticus, DKA, upper GI bleed, anaphylaxis.
Ignoring contraindications and interactions in the chosen drug or procedure, selecting a management option that is effective in principle but unsafe for this patient (e.g., thrombolysis after recent surgery, NSAIDs in renal impairment, beta-blocker in decompensated heart failure or cocaine toxicity).
Options are often written so the 'correct' drug is the one that treats the condition; candidates who focus on efficacy fail to screen the stem for the relative contraindication deliberately planted in the history.
What to do instead
After selecting an answer, run a 10-second safety check against the stem: renal and hepatic function, pregnancy, bleeding risk, allergy, recent surgery, asthma, heart block. Build a personal list of absolute and relative contraindications for the drugs most tested (thrombolytics, NSAIDs, beta-blockers, ACE inhibitors, metformin, contrast).
Misreading the direction of a physiological or pharmacological question, for example confusing cause with consequence in acid-base or endocrine feedback loops, or mixing up which hormone or electrolyte is raised versus lowered in a named syndrome.
Many topics are learned as paired associations without a causal model, so under exam pressure the pair reverses. Long stems with double negatives compound this.
What to do instead
For each major syndrome, draw the causal chain once (e.g., primary hyperaldosteronism: aldosterone up, renin down, sodium up, potassium down, alkalosis) and test yourself by covering parts of the chain and reconstructing it. Rewrite double-negative stems in plain positive form before answering.
Failing to recognise when the question is testing a 'do not do' or 'least likely' concept, and selecting the common or correct-sounding option instead of the exception the examiner wants.
Most questions reward the positive answer, so the habit of picking the most familiar option overrides careful reading of 'which is least likely', 'is not', or 'should be avoided'.
What to do instead
Circle negation words on every stem. When the question is negative, deliberately generate the answer you would give if it were positive, then choose a different option and verify it is genuinely the exception. Practise sets of negative stems until the check becomes automatic.
Managing a condition without addressing the underlying cause or the required parallel step (e.g., treating hypercalcaemia with fluids alone while missing malignancy workup, or treating DKA without potassium replacement planning, or treating a seizure without checking glucose).
Protocols are memorised as single interventions, and candidates stop at the first correct-sounding action rather than the complete immediate package the stem rewards.
What to do instead
For each emergency, memorise a bundled first-line response with three to five components (e.g., DKA: fluids, insulin, potassium, precipitant search). When reviewing mock questions, write out the full bundle and compare with the option chosen to see what was omitted.
Over-relying on pattern-matched 'buzzwords' in stems and options, choosing an answer because it contains a familiar term rather than because the clinical details support it, which fails on stems deliberately written to resemble one condition but describe another.
Question banks and revision courses condition rapid keyword recognition, and candidates stop reading once a keyword appears, missing the discriminating detail later in the stem.
What to do instead
Read the entire stem before looking at options, then summarise it in one sentence in your own words. Identify the two or three features that discriminate between the top differentials and check which option those features support. Deliberately study 'look-alike' pairs (e.g., nephrotic versus nephritic, TIA versus migraine aura, anxiety versus thyrotoxicosis).
Poor time and review discipline: spending disproportionate time on a small number of difficult stems, leaving later questions rushed, and changing correct answers on review without a clear new reason.
Difficult stems create a sunk-cost trap, and anxiety about unanswered questions drives both over-long deliberation and last-minute answer changes based on doubt rather than evidence.
What to do instead
Set a per-question time budget from the total item count, flag and move on after the budget, and return only with remaining time. On review, change an answer only if you can state a specific factual reason; track your changed-answer accuracy in practice to calibrate. Simulate full-length timed papers under exam conditions.