Megaprostheses, initially developed for oncological indications, have increasingly found application in traumatology and hip and knee revision surgery. However, their frequent complications highlight the importance of learning the procedure. This study aims to provide original data on the impact of a surgeon’s learning curve on key perioperative outcomes of these surgeries.

The experience of surgeons performing massive arthroplasties influences key perioperative outcomes.

We conducted a monocentric retrospective study between January 2005 and August 2022, involving 93 megaprostheses surgeries performed for non-oncological indications.

Learning curves were constructed using the LOESS (Locally Estimated Scatterplot Smoothing) method, based on intraoperative data such as total blood loss (TBL) and blood volume transfused (BVT). These curves were used to categorize surgeons into the learning phase (SLP) and mastery phase (SMP). We also compared the cumulative incidence of reinterventions for complications between the two groups.

Linear regression analysis revealed a learning effect for the TBL with thresholds of five procedures (R² = 0.139; p = 0.008), and five procedures for BVT (R² = 0.079; p = 0.002). Based on these findings, we defined the threshold for the learning phase as up to four procedures, and the mastery phase from the fifth procedure onwards.

Multivariate analysis showed a slightly higher cumulative incidence of re-interventions for complications in the SLP group (33%; n = 14) compared to the SMP group (22%; n = 11), with a Hazard Ratio of 2.07 (CI 95% [0.94–4.59]; p = 0.071) without reaching statistical significance.

This study describes a learning curve for performing proximal and distal femur replacement with megaprostheses in non-oncological indications.

Despite it is well established that all surgical procedures benefit from supervision during the early learning phase, we did not find a learning effect on the risk of reoperation for complications. Nevertheless, there is an improvement on variables such as bleeding along the learning curve, leading to suggest that these interventions should be performed by experienced surgeons or under supervision.

IV.