Daily diagnosis games, one case, clues revealed guess by guess, a shareable result, were built for individual players, and the existing writing about them, ours included, addresses the player: how to build the habit, which game fits which exam. This article addresses the other person in the room: the clinical teaching fellow, the registrar with a Thursday lunchtime slot, the GP trainer, because the same mechanic that makes these games sticky solo makes them unusually effective teaching instruments, and almost nobody is using them that way deliberately.
Why the mechanic teaches
The game format bakes in the three things clinical reasoning teaching struggles to enforce. Commitment before reveal: every learner must name a diagnosis before the next clue appears, which converts passive case-following into active reasoning, the difference between watching a differential and building one. Progressive disclosure: information arrives the way it does clinically, in instalments, so learners practise updating a working diagnosis rather than pattern-matching a completed vignette. And low stakes with visible calibration: a wrong guess costs a clue, not face, while the guess count makes everyone's calibration gently public. These are precisely the ingredients of good case-based teaching; the game simply enforces them by rules rather than by facilitation skill, which is exactly what a nervous new teaching fellow needs.
The five-minute session opener
The simplest use: open every teaching session with the day's case. Project it, take guesses from the room clue by clue, and insist on the discipline that makes it teaching rather than trivia: before each guess, someone must say the discriminating reasoning out loud, what in this clue moves you, and away from what. iatroX Rounds serves this neatly, one free UK-context case a day, six guesses, no setup, and the whole ritual takes five minutes, warms the room up, and trains problem representation as a side effect. Groups develop opinions about the day's case remarkably fast, which is the engagement most session openers never achieve.
Structured small-group teaching from the archive
The archive of past cases upgrades the opener into a curriculum tool. Pick archived cases matching your session topic and run them as commit-before-reveal rounds: learners write a private diagnosis and confidence after each clue, then the group compares, and the teaching happens in the deltas, why did half the room jump to the wrong branch at clue two, what was the discriminating feature, what would each learner have asked for next? This is hypothesis-driven reasoning teaching in a container that runs itself, and because everyone committed in writing, the discussion has honest material instead of retrospective wisdom. Close each case by making the problem representation explicit, the one-sentence version of the patient, which is the skill under the whole exercise and the one that transfers directly to referrals and escalation calls.
Teaching the meta-skills
Two higher-order uses for smaller groups. Calibration training: have learners log confidence with each guess across several cases, then review whose confidence predicts correctness, a revealing, slightly uncomfortable, genuinely useful exercise that maps straight onto safe escalation behaviour. And reasoning-aloud practice: one learner plays a case as the designated thinker, narrating each update, while the group critiques the reasoning rather than the answer, the closest cheap approximation of the think-aloud methods reasoning research actually uses. Both exercises need nothing but the game, a whiteboard and fifteen minutes.
The honest limits
A game is a warm-up and a mechanic, not a syllabus, and the caveats owed to learners are the same ones we give players: single-best-diagnosis games train convergence, so pair them with teaching that practises managing uncertainty; daily cases sample, they do not cover, so structured question practice still carries the curriculum; and the competitive layer should stay light, streaks motivate, leaderboards in a teaching group can quietly punish exactly the learners who most need to guess freely. Used inside those limits, the format earns its slot, and the educators who adopt it tend to keep it, because it solves the perennial teaching problem, getting every learner reasoning within ninety seconds, by design rather than charisma.
Frequently asked questions
Does this work for undergraduate as well as postgraduate teaching?
Well for both, with the clue-by-clue pacing doing more work lower down: students get a scaffolded reasoning structure, while for postgraduates the value shifts to calibration and speed. The same archived case can serve both groups with different debriefs.
How does it fit a teaching fellow's portfolio?
Neatly: a games-based reasoning session, with its design rationale and collected feedback, is exactly the kind of documented educational innovation that teaching applications and PGCert assignments reward; the wider case for those posts is at /blog/clinical-teaching-fellow-jobs-worth-doing.
Where does the solo version fit alongside teaching use?
As the habit you recommend learners keep between sessions: the daily case as personal reasoning practice, unpacked at /blog/five-minute-clinical-reasoning-daily-habit-new-doctors, with the game category itself surveyed at /blog/best-medical-diagnosis-games-doctors-students.