Radiographic outcomes of CT-based robotic-assisted versus manual direct anterior total hip arthroplasty with an acetabular-only registration workflow: a retrospective cohort study.

Limited exposure through the direct anterior approach may restrict robotic registration to the acetabulum. We compared postoperative radiographic outcomes of CT-based robotic-assisted DAA THA performed with an acetabular-only bony registration workflow with manual DAA THA. This retrospective cohort included 111 primary direct anterior total hip arthroplasties in 108 patients: 57 robotic-assisted and 54 manual procedures performed from October to December 2025. Both groups targeted 40 degrees of radiographic inclination and 15 degrees of radiographic anteversion. The robotic workflow used 30 registration points exclusively on acetabular bone. Co-primary outcomes were absolute postoperative inclination and anteversion target deviations. Generalized estimating equations accounted for bilateral hips and adjusted for baseline covariates. Robot-reported and postoperative angles were assessed using Bland-Altman analysis and intraclass correlation coefficients. Absolute inclination deviation was lower with robotic assistance (2.89 ± 2.39 degrees vs. 4.83 ± 3.53 degrees; adjusted difference, -1.40 degrees; 95% confidence interval [CI], -2.51 to -0.29; Holm-adjusted P = 0.027). Absolute anteversion deviation did not differ significantly (5.09 ± 3.82 degrees vs. 6.02 ± 4.49 degrees; adjusted difference, -0.92 degrees; 95% CI, -2.44 to 0.60; Holm-adjusted P = 0.234). Combined angular error was lower with robotic assistance (adjusted difference, -1.65 degrees; 95% CI, -3.12 to -0.19; P = 0.027). Both angles were within the Lewinnek zone in 86.0% and 72.2% of hips, respectively (adjusted odds ratio, 2.20; 95% CI, 0.76-6.33; P = 0.145). In the robotic group, intraoperative-to-postoperative bias was - 0.05 degrees for inclination and - 0.38 degrees for anteversion, but 95% limits of agreement were - 9.08 to 8.98 degrees and - 10.14 to 9.38 degrees; corresponding intraclass correlation coefficients were 0.319 and 0.491. Robotic assistance increased operative time by 15.02 min (P < 0.001). CT-based robotic-assisted DAA THA using an acetabular-only registration workflow was associated with a modest reduction in postoperative inclination error and combined angular error compared with manual placement, but not with lower anteversion error or a higher adjusted Lewinnek-zone rate. Because the study did not include a robotic comparator using a conventional registration strategy, the observed differences cannot be attributed specifically to acetabular-only registration. Wide individual limits of agreement further indicate that robot-reported and postoperative angles are not interchangeable, and the clinical relevance of the small radiographic differences remains uncertain.