Elsie Item-Writing Academy
About 45 minutes · Academy module: Distractor Engineering: Shelf / Clerkship — Distractors From the Wards
Unofficial faculty-development resource. Not affiliated with, endorsed by, or representing the AAMC, NBME, USMLE, LCME, or NRMP. Supports faculty-development documentation.
Shelf-exam distractors live close to the wards: the wrong answers should be the wrong turns a real clerk might take — ordering the low-yield test, reaching for the familiar drug, deferring the urgent procedure. Distractor engineering at this level is applied clinical teaching.
Homogeneity of length and detail. Clerkship authors often write the key as the full correct plan and the distractors as stubs. The result is an item where the longest option wins. Force parallel construction: every option the same kind of answer (all next steps, all diagnoses, all drugs), the same specificity, no unique qualifiers. When the options are homogeneous, the student must reason clinically instead of stylistically.
Common misconceptions as distractors. Build from the errors you actually see on the wards: the sulfonylurea reflex in a patient with obesity where a GLP-1 receptor agonist is the better add-on, the CT-first habit when the diagnosis is already clinical, the reassurance that defers needed follow-up. Each distractor should map to one misconception you could name in feedback to the student who chose it.
The "second-best answer" trap. The near-miss is the most instructive distractor on a shelf exam: the right idea at the wrong urgency (outpatient referral for ovarian torsion), the reasonable drug for the wrong patient. It must be clearly wrong given the stem's discriminating detail — and that detail must be in the stem, not in your head. If the second-best option becomes defensible, you have a disputed key, not a trap.
Avoiding the throwaway option. A dead fifth option quietly converts your five-option item into a four-option item and teaches students your exams contain free eliminations. Every option earns its place or the stem gets reframed.
Distractor ordering. Vary key position; avoid patterns. Keep natural clinical orders (escalation ladders) intact where they exist.
Feeding pooled data back into revision. On shelf exams, distractor pull is ward-round intelligence: a distractor that pulls heavily is a misconception your clerkship is currently producing — report it to the rotation, revise the teaching, and keep the item as a sentinel for the next administration. A zero-pull distractor is a candidate for replacement with a live misconception. Track p-values across administrations to catch drift. Negative discrimination is a faculty-review flag only: pull the item, review content for a second defensible answer or miskey, revise with intent — never auto-rekey or auto-delete. The documented review itself supports faculty-development documentation, showing deliberate stewardship of the exam rather than reflexive editing.
---
Flawed item (intended key: E)
A 67-year-old man with a 30 pack-year smoking history presents with painless hematuria. Cystoscopy reveals a papillary bladder lesion. Which of the following is the most likely diagnosis?
A) Renal cell carcinoma B) Prostate adenocarcinoma C) Ureteral stricture D) Nephrolithiasis E) Urothelial carcinoma, the most common bladder malignancy, which arises from the transitional epithelium and classically presents with painless hematuria in older smokers, appearing as a papillary lesion on cystoscopy
Flaw taxonomy label: Heterogeneity of length and detail — the key is the longest, most explained option.
Correction: Option E does not just name the diagnosis; it argues for it, with epidemiology, histology, and the classic presentation spelled out — while every distractor is a bare name. The student does not need to know bladder cancer; they need to know that explained answers are usually correct. The repair is to make every option the same kind of answer at the same specificity: five bare diagnoses, parallel in form, so the cystoscopy finding must do the discriminating.
Model rewrite — cold-solve verified: 2/2 (solver 1: E, solver 2: E)
A 67-year-old man with a 30 pack-year smoking history presents with painless hematuria. Cystoscopy reveals a papillary bladder lesion. Which of the following is the most likely diagnosis?
A) Renal cell carcinoma B) Prostate adenocarcinoma C) Ureteral stricture D) Nephrolithiasis E) Urothelial carcinoma
Key: E
Flawed item (intended key: A)
A 24-year-old woman presents with sudden severe right lower-quadrant pain, nausea, and vomiting. Pelvic ultrasound shows an enlarged right ovary with absent Doppler flow. What is the most appropriate next step?
A) Emergent laparoscopy B) Transvaginal ultrasound with Doppler C) CA-125, CEA, and AFP tumor marker panel D) CT of the abdomen and pelvis E) Outpatient gynecology referral in 2 weeks
Flaw taxonomy label: Missing second-best distractor — no option captures the "right diagnosis, wrong urgency" error.
Correction: The classic clerk error in ovarian torsion is not missing the diagnosis — it is deferring it: the outpatient referral, the "let's watch it" plan. Option E as written ("in 2 weeks") is close, but the set lacks the clean near-miss that isolates urgency as the tested concept, and option B is redundant (the Doppler is already done). The rewrite sharpens the trap: option E becomes the plain outpatient referral — the exact wrong-urgency answer the almost-ready student gives — while the redundant and shotgun options are replaced with genuinely considered alternatives (CT, CA-125). The stem's discriminating detail (absent Doppler flow = torsion until proven otherwise) then does its job against a proper trap.
Model rewrite — cold-solve verified: 2/2 (solver 1: D, solver 2: D)
A 24-year-old woman presents with sudden severe right lower-quadrant pain, nausea, and vomiting. Pelvic ultrasound shows an enlarged right ovary with absent Doppler flow. What is the most appropriate next step?
A) Start oral contraceptives B) Obtain CT of the abdomen and pelvis C) Order serum CA-125 D) Emergent laparoscopy E) Outpatient gynecology referral
Key: D
Flawed item (intended key: B)
A 58-year-old man with type 2 diabetes has an HbA1c of 8.2% on metformin alone. His BMI is 34 kg/m², eGFR is 68 mL/min, and he has no albuminuria. Which of the following is the most appropriate next step in glycemic management?
A) Add basal insulin B) Add a GLP-1 receptor agonist C) Add a thiazolidinedione D) Add an alpha-glucosidase inhibitor E) Add a meglitinide
Flaw taxonomy label: Zero-pull distractors — options students neither know nor choose, missing the live misconception.
Correction: Options C, D, and E are drug classes most students barely encounter; pooled data would show near-zero pull on all three, which means they contribute nothing — no discrimination, no diagnostic signal. Meanwhile the misconceptions students actually hold — the sulfonylurea reflex (cheap, familiar, wrong for this patient with obesity), premature insulin initiation, and clinical inertia ("continue metformin alone") — are absent. The revision rule: replace zero-pull options with the live errors. The rewrite installs all three real misconceptions, so the distractor pull becomes ward-round intelligence about what the clerkship is currently producing.
Model rewrite — cold-solve verified: 2/2 (solver 1: B, solver 2: B)
A 58-year-old man with type 2 diabetes has an HbA1c of 8.2% on metformin alone. His BMI is 34 kg/m², eGFR is 68 mL/min, and he has no albuminuria. Which of the following is the most appropriate next step in glycemic management?
A) Add basal insulin B) Add a GLP-1 receptor agonist C) Stop metformin and start insulin D) Continue metformin alone E) Add a sulfonylurea
Key: B
---
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.