Prosthesis and Prejudice: Potential Bias, Procurement, and Prescribing Patterns in Canadian Knee Arthroplasty

Abstract

Background Total knee arthroplasty (TKA) implant selection is influenced by patient, surgical, institutional, and implant-specific factors, yet national data describing these patterns in Canada remain limited. This thesis used data from the Canadian Joint Replacement Registry (CJRR) to characterize TKA implant use and examine variation in implant selection across Canada from 2019 to 2023.

Methods An open-source implant reference library and validation pipeline were developed to improve the identification and classification of implant manufacturers and models. These tools were used to characterize national and provincial trends in fixation method, patellar resurfacing, implant stability, and manufacturer and model utilization. Finally, clustering methods were applied to identify latent implant prescribing configurations and assess whether patient characteristics explained cluster membership.

Results The implant reference library substantially improved the usability and granularity of CJRR data and demonstrated potential for retrospective data validation. Cementless fixation increased substantially over the study period and was more common among male and younger patients. Patellar resurfacing and posterior-stabilized design utilization declined as medial-pivot designs increased, while cruciate-retaining designs remained relatively stable. Substantial provincial variation was observed across fixation, resurfacing, stability, and manufacturer- and model-level implant selection. Manufacturer trends indicated that some observed practice changes were likely influenced by market composition. Of note, the Stryker Triathlon system accounted for much of the shift toward cementless fixation. Clustering identified six distinct TKA prescribing configurations, primarily differentiated by fixation, stability, and implant model. Similar configurations emerged when model identity was excluded as a clustering input, demonstrating convergence on broad functional design categories across manufacturers. Patient characteristics did not substantially explain cluster membership.

Conclusions TKA implant selection in Canada converged on broad functional design configurations alongside some model-specific trends (e.g., Stryker Triathlon cementless), with substantial variation across patient groups and provinces. The findings of this thesis suggest that market composition, procurement practices, and unmeasured institutional and surgical factors may contribute to the observed prescribing patterns. Strengthening registry data usability through standardized implant identification and incorporating additional surgical and institutional variables could enable more granular analyses of implant selection. Future research should link these prescribing patterns to clinical outcomes to determine whether observed variation influences patient care and identify opportunities to improve the quality and value of TKA in Canada.