Imageless robotic assistance versus laser-projected mechanical-axis guidance in total knee arthroplasty: a deformity-stratified comparative cohort study.

Whether imageless robotic assistance provides measurable radiographic differences beyond laser-projected mechanical-axis guidance in total knee arthroplasty (TKA) is unclear. We studied 259 consecutive primary TKAs performed from 2023 to 2025 at a single high-volume practice: 110 manual, 99 laser-projected, and 50 imageless robotic-assisted procedures. The primary outcome was absolute postoperative hip-knee-ankle (HKA) angle error. Secondary outcomes were coronal component error, operative time, and observed patient out-of-pocket (OOP) payments. Pairwise 1:1 propensity-score matching compared laser-projected with manual TKA and robotic-assisted with laser-projected TKA; the latter comparison also used a prespecified high-varus stratum (preoperative varus ≥ 10°). Interaction models adjusted for clinical and coronal phenotype covariates. Mean absolute HKA error was 2.20°, 1.19°, and 0.95° in the manual, laser-projected, and robotic-assisted groups (P < .001). After 1:1 caliper matching, laser-projected TKA had lower HKA error than manual TKA (-1.01°, 95% CI -1.38 to -0.64; P < .001). HKA error did not differ between robotic-assisted and laser-projected TKA in the full cohort (-0.03°, P = .612; caliper sensitivity, -0.05°, P = .541). In high-varus matched pairs (n = 16), robotic-assisted TKA had lower femoral coronal error (-0.86°, Holm P = .019) and mean coronal component error (-0.61°, Holm P = .039), but neither finding persisted in the caliper sensitivity analysis. Robotic-assisted TKA had 14.41 min longer adjusted operative time and higher observed OOP payments (ratio, 1.208; 95% CI, 1.146 to 1.274); the CPI-standardized sensitivity ratio was 1.189. After caliper matching, laser-projected TKA had lower whole-limb alignment error than manual TKA. Whole-limb HKA error did not differ detectably between the two assisted workflows. In the exploratory high-varus subgroup, robotic-assisted TKA had lower femoral and mean coronal component error, but the small matched sample and lack of persistence in the caliper sensitivity analysis warrant cautious interpretation.