Abstract 22

Background: Deep learning techniques require an appreciation of the objectives required.

Subjects: 44 paediatric medical students during their final paediatric clinical attachment.

Inverventions: Students asked to develop 10 multiple choice questions[MCQs] (5 stems per question, negative marking) to cover topics they considered important in paediatrics (not just core topics). The MCQs had to pass those that deserved to, reward excellence, but fail those without sufficient knowledge to be a safe doctor with children. Questionnaires were distributed at random to another class member. Students sitting the exam were asked whether (1) questions were too easy/just right/too difficult, (2) questions were an appropriate subject to be examined on. Marks obtained in the student exam (STMCQ) were compared with the MCQ mark obtained in the final MBChB examination (FMCQ).

Results: 99% of questions prepared by classmates were considered an appropriate topic for examination (even though 16% were not core curriculum). 83% of questions were considered an appropriate level of difficulty (4% too easy, 13% too difficult). Marks obtained in the STMCQ were significantly correlated with the FMCQ (p=0.007). All 8 students failing the SMCQ passed at FTMCQ (only 1 vice versa).

Conclusions: The study introduces the concept of medical students producing exam material for their colleagues, thereby encouraging deep learning and a better understanding of course objectives. This area is poorly studied and further work is needed.