YouTube is one of the main sources for learning clinical skills. This study aims to assess the educational outcomes of medical students from self-directed learning about knee arthrocentesis through searching and using YouTube videos in comparison to traditional supervisor-led sessions.

Seventy-one medical students were randomly assigned in three groups. Group A had a classic supervisor-led clinical session, where the supervisor demonstrated the procedure. Group B students were provided with links to YouTube videos of knee arthrocentesis that were deemed of high educational quality, while group C searched and learned from any YouTube video they found appropriate based on the learning objectives provided. The students' performance pre- and post-feedback was examined using a checklist that was based on the guidelines of the American Academy of Family Physicians on knee arthrocentesis.

Pre-feedback, statistically significant higher mean scores for group A were noted in identification of an appropriate puncture site (p = 0.015), puncture site sterilization (p = 0.046), wearing sterile gloves (p < 0 .001), and direction of needle insertion (p < 0.001). The overall mean scores before feedback for group A, group B and group C were 17.9 ± 1.9, 14.9 ± 2 and 15.4 ± 1.8, respectively (p < 0 .001). None of these scores was below 60% of the total possible score (total score = 21). The overall mean scores after feedback for group A, group B and group C were 21, 20.9 ± 0.3 and 21, respectively (p = 0.037).

Without appropriate feedback to the learners from an instructor, YouTube videos cannot replace traditional supervisor-led sessions in learning knee arthrocentesis.