Failure modes of robotic-assisted arthroplasty platforms in Australia: a descriptive analysis of TGA adverse event reports.

Robotic-assisted arthroplasty is increasingly performed using technically distinct platforms. The types of device problems reported for different robotic arthroplasty platforms have not been characterised using Australian national surveillance data. The TGA Database of Adverse Event Notifications (DAEN) was searched for robotic arthroplasty adverse events reported between July 2012 and December 2025. Failure modes were extracted from the DAEN classification, and qualitative review of event narratives was performed independently by two authors. One hundred and forty reports were identified across five platforms: ROSA Recon (n = 46, 32.9%), MAKO (n = 45, 32.1%), NAVIO/Cori (n = 33, 23.6%), Zimmer CAS/ORTHOsoft (n = 15, 10.7%) and VELYS (n = 1, 0.7%). The retrieved reports for each platform contained different mixtures of DAEN failure-mode categories. MAKO reports commonly described registration-related inaccuracies and resection beyond planned boundaries despite haptic guidance, whereas ROSA Recon reports predominantly described discrepancies between planned and achieved bone resection. Zimmer CAS/ORTHOsoft reports were mostly navigation-related inaccuracies, while NAVIO/Cori demonstrated frequent software-related failures, including an 11-report cluster describing failure of the femoral notch-warning algorithm. Australian adverse event reports contained platform-specific patterns of reported device problems. Qualitative review identified recurring technical and software-related signals that were not always apparent from DAEN failure-mode coding alone. Awareness of these platform-specific failure modes may assist surgeons in recognising device-related problems during surgery and should inform training for surgeons transitioning between robotic platforms.