Start with what you’re assessing. A good station assesses one clear skill that a practitioner at this level is expected to perform safely. Decide the skill, then build the situation around it. A station that tries to test history, examination, diagnosis and management in eight minutes usually tests none of them well.
Make the task doable in the time. Run through the station yourself with a stopwatch. If a competent candidate can’t finish in the time, the station is testing speed. Cut the tasks, not the reading time.
Tell candidates what to do, not what you’re looking for. The door instructions should say who the patient is, where the candidate is and exactly what to do. They shouldn’t give away the diagnosis or list the marking points.
Mark what you can see. Checklist items should be actions an examiner can observe and tick, such as “checks the patient’s identity against the wristband”, not “shows good communication”. Keep the items in the order a competent candidate would do them, so examiners can follow along.
Use a global rating as well as the checklist. Checklists reward thoroughness, but experienced candidates often skip steps they don’t need. A global rating lets examiners judge the whole performance. Collecting both also lets you set a defensible pass mark after the exam with the borderline regression or borderline group method.
Flag the few things that must not be missed. Some omissions are unsafe whatever else the candidate does well, such as not checking allergies before giving a drug. Mark them as critical items and decide in advance what happens when one is missed.
Script the patient fully. Your simulated patient needs to know what to say straight away, what to say only if asked, how to act and what not to volunteer. Every fact a candidate needs to earn a mark must be in the script, or candidates will be marked down for information they could never get.
Brief your examiners and patients together. Go through the station, the marking sheet and the script before the OSCE, so everyone agrees on what a pass looks like. Consistency across examiners matters as much as the station itself.
Pilot it before it counts. Try each new station on a colleague or a small group of students first. You’ll find the ambiguous instruction, the missing prop and the checklist item no one can see.
Check it before you use it. The station is written by AI. It’s a strong starting point, but check every clinical detail, value and marking item against current guidelines, your curriculum and your assessment rules. The final station is always yours.