NBME CBSE common mistakes
NBME CBSE · Understand the traps and how to avoid them.
When a vignette supplies a test's sensitivity and specificity and asks for the probability that this patient has the disease, the candidate answers with the sensitivity (or with 1 − specificity) instead of computing positive predictive value using the stated prevalence.
Sensitivity and specificity are memorized as 'accuracy' numbers and feel patient-specific, so the candidate never builds a 2×2 table and the prevalence in the stem never enters the arithmetic.
What to do instead
On every test-characteristic question, instantiate 1,000 hypothetical patients, fill the 2×2 from prevalence and sensitivity/specificity, and only then read off PPV or NPV. If no prevalence is given, the item cannot be asking for a predictive value - decide what it is actually asking before calculating.
Assigns a substance-related presentation to intoxication when the stem's timeline places the last use days earlier (or the reverse), missing withdrawal and its characteristic onset window.
Intoxication and withdrawal share autonomic and CNS findings, and candidates pattern-match on symptoms rather than anchoring to time since last use; the onset/duration grids for alcohol, benzodiazepine, and opioid withdrawal are rarely rehearsed as a table.
What to do instead
Box the time since last use in every substance stem, then check it against a memorized onset grid: alcohol tremor/hallucinosis within 6–48 h, seizures 12–48 h, delirium tremens 48–96 h; opioid withdrawal onset 8–24 h and self-limited. Findings that fall outside the window should push you to a different diagnosis, not a different severity of the same one.
Selects the effect of a bedside maneuver on a murmur from a memorized list rather than determining whether the maneuver raises or lowers left ventricular volume and systemic vascular resistance, flipping the answers for hypertrophic cardiomyopathy, aortic stenosis, and mitral regurgitation on Valsalva, squatting, and handgrip.
Maneuvers are learned as linked pairs ('Valsalva → HCM') instead of as physiology, so when the stem uses an unfamiliar combination the direction of change is a coin flip.
What to do instead
For each maneuver, state two things first - change in preload/LV volume and change in afterload - then derive the murmur effect. Anything lowering preload (Valsalva strain, standing) worsens HCM and mitral valve prolapse and softens aortic stenosis; anything raising afterload (handgrip) worsens mitral regurgitation, aortic regurgitation, and VSD. Rebuild the table from those axes instead of recalling it.
Localizes a neurologic lesion by naming the structure most stereotypically associated with the single most dramatic finding (calling a lateral medullary syndrome a cerebellar stroke because of vertigo and ataxia) without checking for crossed findings or cranial nerve involvement.
Stems list three or four findings; the candidate matches the salient one to a remembered association and stops, so the crossed pattern that fixes the lesion's vertical level is never used.
What to do instead
Run a fixed template on every neuro stem: mental status, cranial nerves, long-tract motor and sensory, cerebellum, and side of each. Crossed findings (ipsilateral face with contralateral body) force a brainstem level; commit to a level in writing before reading the answer choices.
Diagnoses a primary versus secondary/tertiary endocrine disorder from the peripheral hormone level alone, treating a mid-range pituitary trophic hormone as 'normal' and therefore as evidence against pituitary disease.
Attention lands on the abnormal cortisol, free T4, or testosterone value; the trophic hormone sits in the reference range and is dismissed rather than interpreted as inappropriately low or high for that degree of end-organ failure.
What to do instead
Interpret every axis as a pair and name the pattern before reading options: gland hormone low with trophic hormone high = primary; both low = secondary/tertiary; both high = resistance. For cortisol, confirm the cortisol is abnormal first, then pair with ACTH; for thyroid, pair TSH with free T4.
Reads hepatitis serologies as a checklist of individually associated markers and misclassifies the window period, resolved infection, or isolated core antibody - for example calling isolated anti-HBc IgM 'immunity'.
Each marker is memorized as belonging to a stage rather than as a curve with an appearance and clearance time, so overlapping or missing markers have no place to sit in the candidate's mental model.
What to do instead
Sketch the HBsAg / anti-HBc IgM / anti-HBs / HBeAg timeline once per study session, then place each patient's exact marker set on it. Anchor three patterns: anti-HBs alone = vaccination; isolated total anti-HBc with negative HBsAg = remote resolved or occult; IgM anti-HBc = acute infection or flare.
Assigns recurrence risks from the generic 25% or 2/3 fractions without updating for phenotypes and test results stated in the stem (using 2/3 for an unaffected sibling when that sibling's carrier test or an affected child is given).
Risk figures are memorized as fixed numbers attached to a mode of inheritance, so the stem details that should condition the probability get read as background color rather than as evidence.
What to do instead
Draw the pedigree and mark every known phenotype and test result on it before computing anything. Then enumerate only the genotypes still possible for that specific person and apply Bayes: an unaffected sibling of an affected child is AA or Aa (2/3 carrier), but an unaffected sibling who already has an affected child is a obligate carrier and the 2/3 no longer applies.
Computes the anion gap, finds it elevated, and stops - never calculating the delta ratio - so a mixed high-anion-gap acidosis with a metabolic alkalosis or a coexisting non-gap acidosis is reported as a single disorder.
The anion gap is taught as the endpoint of the acid-base algorithm rather than as one step among several, and the compensation formulas feel like optional extras under time pressure.
What to do instead
Run four steps in order on every acid-base stem and write the numbers down: (1) pH → acidemia or alkalemia; (2) primary disorder from pCO2 and HCO3; (3) expected compensation (Winter's formula: expected pCO2 = 1.5 × HCO3 + 8 ± 2); (4) if the anion gap is elevated, delta ratio = ΔAG / ΔHCO3 (about 0.8–2.0 appropriate; >2 suggests coexisting alkalosis or chronic respiratory acidosis; <0.4 suggests coexisting non-gap acidosis).
Labels hypoxemia that fails to correct with supplemental oxygen as V/Q mismatch or diffusion impairment, and separates obstructive from restrictive disease using FEV1 alone instead of the FEV1/FVC ratio.
The causes of hypoxemia and the obstructive/restrictive split are remembered as labels; the discriminating feature - response to 100% oxygen and the ratio, respectively - is exactly the part dropped when reading quickly.
What to do instead
Apply two rules. For hypoxemia: a shunt does not correct with 100% O2, while V/Q mismatch improves; a normal A-a gradient points to hypoventilation or low inspired PO2. For spirometry: compute FEV1/FVC first (<0.70 obstructive; normal or high with a low FVC suggests restriction), then confirm restriction with lung volumes rather than inferring it from a low FEV1.
Classifies an anemia by the MCV and jumps straight to iron studies without first computing a corrected reticulocyte index, so a hemolytic or blood-loss anemia is worked up as a production defect.
The MCV is the most visually prominent number on the CBC, so the candidate enters the morphologic branch of the algorithm and never splits hypoproliferative from hyperproliferative disease - the split that actually determines the answer.
What to do instead
Start every anemia stem with the reticulocyte index: (retic % × patient Hct ÷ normal Hct) ÷ maturation factor (1.0 at Hct 45%, 1.5 at 35%, 2.0 at 25%). High (>2–3) means loss or destruction → check LDH, haptoglobin, indirect bilirubin, and smear. Low means hypoproliferative → only then use the MCV and iron studies.