Surgeon Training and Revision Rates after Patellofemoral Arthroplasty

Abstract

Importance: Surgeon training with a specific implant is often not considered in implant registry-based studies, which may lead to unobserved confounding bias. Discrepancies between registry and clinical trial outcomes for patellofemoral arthroplasty (PFA) may originate from differences in surgeon training levels. Objective: To compare revision rates for knees operated on by knee surgeons specifically trained for PFA and knee surgeons who were not. Design, Setting, and Participants: In this population-based cohort study, the framework of a target trial was used to compare outcomes for 2 patient groups: patients who underwent PFA performed by knee surgeons who had (trial surgeons) vs who had not (nontrial surgeons) received focused PFA training as part of a randomized clinical trial. All primary PFA procedures from January 1, 2008, through December 31, 2015, were identified using Danish registries and individual hospital notes with 6 years' follow-up. Data were analyzed from January 24 to March 1, 2024. Exposure: Focused PFA training. Main Outcomes and Measures: The primary outcome was 6-year risk of revision. Analyses were conducted according to a prespecified statistical analysis plan, using multiple logistic regression to estimate marginal risk ratios for 6-year risks of revision, reoperation, and mortality, adjusting for potential confounders. Results: Of 482 eligible knees of patients who had undergone PFA, 274 (57%; 206 female [75%]; mean [SD] age, 61 [13] years) were operated on by trial surgeons, and 208 (43%; 142 female [68%]; mean [SD] age, 57 [12] years) by nontrial surgeons. Trial surgeons operated on knees with higher patellofemoral Kellgren-Lawrence osteoarthritis grade (range 0-4, with 0 indicating no osteoarthritis and 4 indicating severe osteoarthritis) than nontrial surgeons (79% vs 53% with grade 3 to 4) and higher tibiofemoral Kellgren-Lawrence osteoarthritis grades (37% vs 17% with grade 2 to 4). The 6-year revision rate for trial surgeons was 8% (22 of 274 knees) vs 26% (54 of 208 knees) for nontrial surgeons, corresponding to an adjusted relative risk (RR) of 0.35 (95% CI, 0.22-0.56; P