Elsie Item-Writing Academy
About 45 minutes · Academy module: Foundations: Writing Step 2 CK–Style Items
The Elsie Item-Writing Academy is an independent faculty-development resource. It is not affiliated with, endorsed by, or sponsored by the AAMC, NBME, USMLE, LCME, or NRMP. Completion of this module supports faculty-development documentation.
Step 2 CK–style items have five options, A through E, and the emphasis shifts from mechanism to decision. The most common lead-in is "most appropriate next step in management" or "most appropriate next step in diagnosis." The vignette is longer than Step 1's — history, examination, laboratories, sometimes imaging — because the decision has to be earned. The tested point is the decision node: given this stability, this severity, and these results, what does the physician do next.
Name the decision explicitly in the lead-in. "What is the best management?" is unfinished — best when? Next step in diagnosis or treatment? In the emergency department or the clinic? The five options must be mutually exclusive actions at the same decision point. The characteristic failure at this level is overlapping answers: two options that both describe acceptable care, or one option that is merely a more detailed version of another, as in this module's third drill. When two options are both defensible, the item has no key — only an argument.
Build the vignette to contain the discriminating data the decision requires, and nothing that points past it. Stability, severity, and time course are the usual discriminators, and the vignette must commit to them. "Chest pain" is not a vignette; "chest pain with hypotension and jugular venous distension" is the beginning of one. When the same presentation resolves to different answers at different severities, the options should force the examinee through the branch explicitly rather than letting the mild and severe paths share an option. Time course does the same work: "for 45 minutes" versus "for 3 days" changes the decision, so state it. If the decision genuinely does not depend on a detail, cut the detail — decorative severity ("the patient appears anxious") is noise that punishes careful readers. State pretest probability or its determinants plainly when the decision turns on it, rather than hoping every examinee reconstructs it identically. Avoid window dressing that mirrors an option's wording — clueing is as fatal here as at Step 1.
Keep the five options homogeneous in form: all diagnostic tests, or all treatments, or all dispositions — never a mixture that lets the examinee eliminate by category instead of by reasoning. If four options are diagnostic tests and the fifth is "reassure and discharge," the examinee can discard it by category — disposition versus workup — without engaging the clinical question. Pick the decision node and populate it with five occupants of that node. The wrong answers should be wrong because the medicine says so, not because the grammar gives them away. Avoid absolute terms, "all of the above," and negative stems, emphasizing the negative word if one is unavoidable. And test the decision the vignette actually supports: if the data only justify a diagnosis, ask for the diagnosis. Do not dress a Step 1 mechanism question in a longer vignette and call it Step 2.
(intended key: B)
Stem: A 45-year-old woman on postoperative day 3 after total hip arthroplasty develops sudden dyspnea and pleuritic chest pain. She is tachycardic at 118/min, tachypneic at 26/min, and her oxygen saturation is 91% on room air. Lung examination is clear. What is the best management?
A. D-dimer assay B. CT pulmonary angiography C. Ventilation–perfusion scan D. Therapeutic heparin E. Reassurance and observation
unfocused stem.
"What is the best management?" names no decision. Is the question asking for the next diagnostic test or the treatment? Now or after further workup? The options span diagnosis (A, B, C), treatment (D), and disposition (E) — three different decision nodes in one option set, so the examinee cannot tell which branch point is being tested. The rewrite names the decision explicitly ("most appropriate next step in management"), states the pretest probability outright via the Wells score so every examinee works from the same branch point, and keeps all five options at the same node: diagnostic actions.
— cold-solve verified: 2/2 (solvers chose B, B)
Stem: A 45-year-old woman on postoperative day 3 after total hip arthroplasty develops sudden dyspnea and pleuritic chest pain. Vital signs: temperature 37.1°C, heart rate 118/min, blood pressure 128/76 mm Hg, respiratory rate 26/min, oxygen saturation 91% on room air. Lung examination is clear. ECG shows sinus tachycardia. Her Wells score is 7.5, indicating high clinical probability of pulmonary embolism. Which of the following is the most appropriate next step in management?
A. D-dimer assay B. CT pulmonary angiography C. Ventilation–perfusion scan D. Bilateral lower-extremity duplex ultrasonography E. Flexible bronchoscopy
Key: B
High pretest probability of pulmonary embolism calls for definitive imaging, and CT pulmonary angiography is the first-line test. D-dimer is inappropriate at high probability — it can only rule out, never rule in, and a negative would be untrustworthy here. V/Q scanning is second-line when CT is unavailable or contraindicated; duplex ultrasonography detects DVT but does not diagnose PE; bronchoscopy has no role. Both independent cold solvers selected B and named no other defensible option.
(intended key: A)
Stem: A 23-year-old man with type 1 diabetes presents with vomiting, Kussmaul respirations, glucose 480 mg/dL, pH 7.21, bicarbonate 12 mEq/L, and positive serum ketones. IV fluids are started. Which is the most appropriate next step?
A. An insulin infusion should always be started immediately regardless of potassium. B. Never give insulin until potassium is completely normal. C. Always give sodium bicarbonate for the acidosis. D. Subcutaneous insulin is never appropriate in DKA. E. Oral potassium alone is always sufficient treatment.
absolute terms.
Every option leans on "always" or "never," so the item tests test-taking folklore instead of DKA management. Worse, the intended key itself relies on "always" — the very word examinees are taught to distrust — which means the item punishes the careful and rewards the careless. Absolutes also flatten real clinical nuance: insulin timing genuinely depends on potassium, and bicarbonate genuinely has a (narrow) role. The rewrite states each option as a plain action without absolutes, so the examinee must decide on content: with potassium 5.8 mEq/L, is insulin safe to start now?
— cold-solve verified: 2/2 (solvers chose A, A)
Stem: A 23-year-old man with type 1 diabetes mellitus presents with 1 day of vomiting and abdominal pain. He is tachypneic with deep, labored respirations and has a fruity odor to his breath. Laboratory studies: glucose 480 mg/dL, pH 7.21, bicarbonate 12 mEq/L, anion gap 24 mEq/L, potassium 5.8 mEq/L, and positive serum ketones. Intravenous isotonic fluids have been started. Which of the following is the most appropriate next step in management?
A. Begin a continuous intravenous infusion of regular insulin B. Withhold insulin until the serum potassium normalizes C. Administer intravenous sodium bicarbonate D. Give subcutaneous long-acting insulin E. Give oral potassium supplementation alone
Key: A
This is diabetic ketoacidosis, and after fluids, continuous IV insulin is the definitive therapy to halt ketogenesis. Potassium of 5.8 mEq/L does not contraindicate insulin (withholding is required only below roughly 3.3); bicarbonate is reserved for pH below about 6.9; subcutaneous insulin cannot match the needed titration; potassium alone treats nothing here. Both independent cold solvers selected A and named no other defensible option.
(intended key: B)
Stem: A 24-year-old man with a known peanut allergy develops lip swelling, urticaria, stridor, and hypotension with a blood pressure of 82/50 mm Hg within minutes of accidental peanut ingestion. Which of the following is the most appropriate next step in management?
A. Epinephrine administration B. Intramuscular epinephrine C. Intravenous diphenhydramine D. Observation with vital-sign monitoring E. Oral prednisone
two correct answers (overlapping options).
Option B is a specific instance of option A, so both are correct and the item cannot be scored — an examinee who chooses A has answered correctly by the item's own logic. Overlap usually creeps in when the author writes the key first as a general concept ("give epi") and then writes the "real" answer as a separate option. At a decision level, options must be mutually exclusive actions: different drugs, different routes, different timing. The rewrite keeps intramuscular epinephrine as the single key and makes every other option a genuinely different action — a different drug, a different route, or a different (inadequate) strategy.
— cold-solve verified: 2/2 (solvers chose A, A)
Stem: A 24-year-old man with a known peanut allergy develops lip swelling, urticaria, stridor, and hypotension with a blood pressure of 82/50 mm Hg within minutes of accidental peanut ingestion. Which of the following is the most appropriate next step in management?
A. Intramuscular epinephrine B. Intravenous diphenhydramine C. Subcutaneous epinephrine D. Nebulized albuterol alone E. Oral prednisone
Key: A
Multisystem involvement with airway compromise and hypotension is anaphylaxis, and intramuscular epinephrine is the first-line, do-not-delay treatment. Diphenhydramine is adjunctive, never first; subcutaneous epinephrine absorbs too slowly in shock; albuterol alone ignores the cardiovascular collapse; oral prednisone is far too slow for an airway emergency. Both independent cold solvers selected A and named no other defensible option.
Recording it adds the module to your Academy completion record on this device — module, track, level, and date, ready to download from your account page for faculty-development documentation.