The classic teaching case has a flaw: it is written in the order of the answer. Present the history, reveal the labs, name the diagnosis — the student rides a track someone else laid. Real clinical reasoning is nothing like that. It is hypothesis-driven, sequential, and full of commitments: every question you ask and every test you order closes some doors and opens others.
A branching case study restores those commitments. The student chooses which history to take, whether to order the test or wait, when to commit to a diagnosis — and the case unfolds differently depending on the path. Cognitive traps like premature closure and anchoring stop being vocabulary words and become things students catch themselves doing.
ScenarioSplice lets faculty build these cases without code: a visual decision-tree editor, color-coded feedback tags, and a shareable link students open on any device with no account.
The anchor in urgent care. A 54-year-old presents with chest pain — and a normal stress test from two months ago in the chart. Students choose the history they take and whether that prior result reassures them. One branch anchors on the stress test and discharges; another works the problem fresh. The debrief names the trap only after students have seen where it leads.
The pharmacy consult. A patient stable on warfarin is prescribed trimethoprim-sulfamethoxazole for a UTI. Pharmacy students decide: dispense as written, call the prescriber to suggest an alternative, or dispense with a monitoring plan. Each branch plays out the INR at follow-up, and the partial tag separates a defensible monitoring plan from the stronger interaction-avoiding call.
Fatigue with a fork in it. A 41-year-old reports six months of fatigue and weight gain. Students choose between a broad lab shotgun, a targeted thyroid work-up, or a depression screen first — and the case rewards a reasoned sequence over either extreme.
Multiple-choice questions test whether students recognize the endpoint. Branching cases test whether they can walk the path: gathering, weighting, committing, revising. Because the student chooses before seeing what happens, errors surface as experiences rather than red marks — a discharged patient who bounces back, an interaction that shows up at follow-up. That kind of feedback survives until clerkship.
Branching also makes reasoning visible to the instructor. Writing the tree forces you to articulate exactly why the tempting-but-weaker choice is weaker — which sharpens the debrief for everyone.
Start from the decision-tree template and write the presentation page. Give each decision two or three defensible options — the goal is discrimination between plausible moves, not right-versus-silly. Reveal information only on the pages where it is earned: a lab result the student never ordered stays unseen, because a page is only reachable through the choices that link to it. Tag outcomes correct, incorrect, or partial; use info pages for results, and paste an image URL for the ECG or film. Keep teaching rationale in private facilitator notes, validate for broken links, and share with one Assign link. If drafting stalls, the AI drafting assist can rough out consequence pages — you accept or reject every suggestion, which matters when accuracy is clinical.
Yes. Pages are only reachable through the choices that link to them, so a lab result or imaging finding appears only on branches where the student ordered it.
Write invented patients. Learners play anonymously and no student data is collected, but you should never paste real patient information into any authoring tool — compose fictional composites instead.
Yes — some faculty assign case-writing as the assessment. Authors sign in with Google; a student-built case can then be assigned to the rest of the group for peer play.
Treat AI drafts as prose help, not clinical authority. It is useful for dialogue and structure; dosages, contraindications, and guideline details are yours to verify. Nothing enters your scenario unless you accept it.