Abstract / Summary
Background and purpose: Lateral knee replacement is not commonly performed. Possible explanations are the lower incidence of lateral knee osteoarthritis, relative technical complexity, and different biomechanics, which may affect failure modes and survival rates. We aimed to analyze the difference in risk of revision between lateral and total knee replacements (TKRs) using the Dutch Arthroplasty Register data.
Methods: From 2007 to 2021, procedures were selected from the Dutch Arthroplasty Register. Unique patients based on their first procedure were identified. Procedures performed between 2014 and 2021 were analyzed separately because BMI and Charnley score were collected from 2014 onwards. We used ATT (Average Treatment effect on the Treated) weights based on the propensity score to estimate the average treatment effect in the lateral knee replacement group. The endpoint was revision, censored for death. Kaplan-Meier survival analyses were performed in propensity-based ATT-weighted cohorts of lateral knee replacement and TKR patients. Weighted univariable Cox regression models were used to estimate revision hazard ratios (HR).
Results: In the total cohort, 847 lateral knee replacements and 240,047 TKR patients were included. In the subgroup, 560 lateral knee replacements and 155,621 TKR patients were included. The risk of revision in both the total cohort and the subgroup was higher for lateral knee replacements compared with TKRs with an ATT-adjusted HR for the total cohort of 2.15 (95% confidence interval [CI] 1.74-2.64, P < 0.001) and for the subgroup an ATT-adjusted HR of 1.55 (CI 1.08-2.21, P = 0.02). Comparison of these ATT-weighted effects with the conditional effect estimates from a regular multivariable Cox proportional hazards regression did not show large differences in this study.
Conclusion: This study showed a significantly higher risk of revision for lateral knee replacements compared with TKRs.