Background: Many complex new procedures involve a learning curve, and patients treated by individuals who are new to a procedure may have more complications than those treated by a practitioner who has performed the intervention more frequently. Still, at some point on the learning curve, each individual must decide that he or she is qualified to perform a procedure, presumably on the basis of his or her level of confidence, background, education, and skill. To evaluate the interrelationship of these factors, we designed a study in which we assessed the performance of a simulated knee joint injection.

Methods: Ninety-three practitioners attending a continuing medical education session on knee injection were randomized to receive skills instruction through the use of a printed manual, a video, or hands-on instruction; each performed one injection before and after instruction. The participants completed pre-instruction and post-instruction questionnaires gauging confidence and also provided self-assessments of their performances of injections before and after instruction. Self-assessments were compared with objective performance standards measured by custom-designed knee models with electronic sensors that detected correct needle placement.

Results: Before instruction, the participants' confidence was significantly but inversely related to competent performance (r = —0.253, p = 0.02); that is, greater confidence correlated with poorer performance. Both men and physician-practitioners displayed higher pre-instruction confidence (p < 0.01), which was not correlated with better performance. After instruction, performance improved significantly in all three training groups (p < 0.001), with no significant differences in efficacy detected among the three groups (p = 0.99). After instruction, confidence correlated with objective competence in all groups (r = 0.24, p = 0.04); however, this correlation was weaker than the correlation between the participants' confidence and their self-assessment of performance (r = 0.72, p = 0.001).

Conclusions: Even low-intensity forms of instruction improve individuals' confidence, competence, and self-assessment of their skill in performing the fairly straightforward psychomotor task of simulated knee injection. However, men and physicians disproportionately overestimated their skills both before and after training, a finding that worsened as confidence increased. The inverse relationship between confidence and competence that we observed before the educational intervention as well as the demographic differences that we noted should raise questions about how complex new procedures should be introduced and when self-trained practitioners should begin to perform them.

Level of Evidence: Therapeutic Level I. See Instructions to Authors for a complete description of levels of evidence.