BACKGROUND Severe coronal deformity in total knee arthroplasty (TKA) forces a choice between preserving constitutional alignment, correcting toward neutral, and adding stems or constraint. We tested which decision was associated with aseptic revision. METHODS We studied 1,566 primary TKAs in 1,286 patients who had paired long-leg radiographs and a minimum 2-year follow-up, stratified by preoperative Kennedy zone (KZ) into varus (KZ 0 and 1) and valgus (KZ 3 to 5). The nine decisions tested were mechanical hip-knee-ankle (mHKA) angle correction, KZ correction, Coronal Plane Alignment of the Knee (CPAK) phenotype V or II, CPAK preservation, tibial or femoral stem, and posterior-stabilized or constrained bearing. Adjusted Fine-Gray subdistribution hazard ratios (sHRs), with death as a competing event, were clustered by patient. RESULTS Limb-axis correction was associated with lower aseptic revision in both strata, whether defined by an mHKA angle of 177 to 183° (varus sHR 0.43, P = 0.043; valgus sHR 0.25, P = 0.004) or by a central KZ (sHR 0.23, P < 0.001; sHR 0.31, P = 0.012), and was largest in extreme valgus (KZ 5; sHR 0.17, P = 0.026). The 5-year incidence of revision for loosening or instability was 0.98 versus 3.26% in corrected and uncorrected knees (sHR 0.22, P = 0.001). Phenotype targets, stems, and constraints were not statistically significant. CONCLUSIONS In severely deformed TKA, particularly extreme valgus, correcting the mechanical axis toward neutral was associated with lower aseptic revision under two alignment frameworks. Preserving the phenotype, or adding stems or constraints, was not.
