Thoracentesis is a common clinical procedure, but the optimal training method remains unclear.

To investigate whether a novel teaching program combining e-learning and simulation-based self-directed, spaced learning (intervention) is non-inferior to a traditional instructor-led massed training approach (control).

In this multicenter randomized, non-inferiority study, emergency physicians unable to perform thoracentesis were randomized to either the intervention group (e-learning and three self-directed simulation sessions over two weeks) or a control group (single three-hour instructor-led simulation course). Skill acquisition and retention were evaluated at two weeks and three months by blinded assessors. The primary outcome was passing rate at two weeks post training, with a non-inferiority margin of 10 %. Secondary outcome was passing rate at three months.

A total of 63 participants (intervention group: 29, control group: 34) were included. The majority were female (72 % in the intervention group vs. 50 % in the control group). At two weeks, passing rates were 66 % (19/29) in intervention group and 65 % (18/26) control, with a risk difference of 0.8 %, CI-95 %: −30 %;30 %, p = 0.96. At three months, skill retention was significantly higher in the intervention group (92 %) compared to the control group (73 %), with a risk difference of 19 % CI-95 %: 10 %;30 %, p < 0.001.

A novel training approach with self-directed spaced learning for thoracentesis resulted in comparable skill acquisition when comparing to traditional instructor-led massed training although the study was underpowered to demonstrate non-inferiority. Self-directed spaced learning was associated with superior for skill retention compared to instructor-led massed training after three months.