JOA - 2026-11-06 - Journal Article
Constitutional Restoration versus Mechanical Neutral in the Younger Total Knee Arthroplasty: A Pre-Specified Bayesian Competing-Risk Analysis of Alignment Accuracy and Aseptic Revision.
Hoskins W, Gusho C, Rosser K, Lim M, Dunstall K
Topics
Key Takeaway
In TKA patients under 65, neutralization error (deviation from mechanical neutral) carried 87% posterior probability of harm at 1.08 HR per degree, while restoration error (deviation from constitutional alignment) was uninformative at HR 1.00, across 26 aseptic revisions in 470 knees.
Summary Depth
Choose how much analysis to show on this article page.
Summary
This study asked whether deviation from constitutional (kinematic) alignment or from mechanical neutral more strongly predicts aseptic revision in patients under 65, using a pre-specified Bayesian competing-risk framework with death and infection as competing events. Neutralization error yielded HR 1.08 per degree (87% posterior probability of harm; 50% probability of ≥1.5-fold increase over 5° error), while restoration error was uninformative (HR 1.00, 52% probability of harm). The probability that neutralization error outperformed restoration error as a predictor was 81%, stable under skeptical priors, and neither metric correlated with six-month OKS.
Key Limitation
Only 26 aseptic revision events in 470 knees limits statistical power; the 50% posterior probability of a clinically meaningful neutralization effect (≥1.5-fold at 5° error) reflects substantial residual uncertainty that midterm follow-up cannot resolve.
Original Abstract
BACKGROUND
In younger total knee arthroplasty (TKA) patients, revision rather than death dominates competing risks, so any true alignment effect is least likely to be masked. We asked whether deviation from constitutional alignment or from mechanical neutrality more strongly predicts aseptic revision and whether the effect is clinically meaningful.
METHODS
We retrospectively analyzed TKA patients younger than 65 years who had paired pre- and postoperative long-leg radiographs and minimum two-year follow-up. We used two per-degree exposures to capture the competing philosophies: restoration error (absolute change in arithmetic hip-knee-ankle angle) and neutralization error (absolute postoperative mechanical hip-knee-ankle deviation from 180°). The primary endpoint was aseptic revision, with death and infection as competing events. A pre-specified Bayesian competing-risk model used weakly informative and skeptical priors with posterior-probability inference. Cause-specific and Fine-Gray models and the six-month Oxford Knee Score (OKS) were secondary analyses.
RESULTS
Among 470 TKAs (376 patients) with 26 aseptic revisions, restoration error was uninformative (hazard ratio [HR] 1.00 per degree, 95% credible interval [CrI] 0.90 to 1.10; 52% probability of any harm). Neutralization error was probably harmful (HR 1.08 per degree, CrI 0.93 to 1.22; 87% probability of harm; 50% probability of a clinically meaningful effect [≥ 1.5-fold increase over a 5° error]). The probability that neutralization error mattered more than restoration error was 81% and stable under the skeptical prior. Frequentist cause-specific and Fine-Gray models agreed with these estimates, and neither error was associated with the six-month OKS.
CONCLUSIONS
In younger TKA patients, deviation from mechanical neutral predicted aseptic revision more strongly than deviation from constitutional alignment. Although personalized alignment carried no measurable survivorship penalty at midterm follow-up, the intended target accuracy matters more than the alignment label.