Abstract
BACKGROUND: Owing to the common but reductive practice of aggregating heterogeneous robotic systems without accounting for navigation integration or technological iteration, existing classifications hinder the comparability and generalizability of screw placement accuracy assessments. To address this limitation, the present study proposes a four-type framework comprising Spine-Specific Robot A (SSR-A) and B (SSR-B), Orthopedic-Universal Robot A (OUR-A) and B (OUR-B), where A and B respectively denote the absence and presence of integrated navigation, and systematically assesses the relative performance of different robotic types in screw placement precision to guide the refinement and future development of spinal robotic technologies. METHODS: A systematic search of eight databases from January 2010 to July 2024 was conducted to identify eligible studies reporting screw placement metrics, including perfect (Grade A) and acceptable (Grade A and B) screw placement rates, facet joint violation, insertion and endpoint deviations, and angular deviations in axial and sagittal planes, of which the first three were analyzed using both network and proportional meta-analyses (NMA and PMA), whereas the others were assessed through NMA alone. Outcome measures in NMA included risk ratios (RR), mean differences (MD), and Surface Under the Cumulative Ranking Curve (SUCRA) values, whereas pooled proportions are used in PMA, with all analyses conducted in R, the quality of evidence evaluated using the CINeMA (Confidence in Network Meta-Analysis) framework, and subgroup analyses stratified by surgical site, segment, pathological type, operation methods, and surgery type. The protocol was prospectively registered in PROSPERO (CRD42024499190). RESULTS: A total of 42 studies comprising 3,470 participants—including 22 randomized controlled trials, 5 prospective cohorts, and 15 retrospective cohorts—conducted across China, Germany, the United States, South Korea, Switzerland, and France, were included. Based on these, OUR-B demonstrated the highest performance in both Grade A (NMA: RR = 1.14, SUCRA = 97.35%; PMA: 93.89%) and Grade A + B (NMA: RR = 1.07, SUCRA = 86.95%; PMA: 99.16%) screw placement rates. Notably, SSR-B also showed impressive performance in Grade A + B screw placement rates (NMA: RR = 1.07, SUCRA = 81.93%; PMA: 99.57%), while exhibiting the lowest insertion-point (MD = − 2.76, SUCRA = 94.44%) and endpoint (MD = − 2.70, SUCRA = 78.92%) deviations, along with favorable angular alignment in both the axial (MD = − 0.61, SUCRA = 61.62%) and sagittal (MD = − 1.45, SUCRA = 83.68%) planes. SSR-A was most effective in minimizing facet joint violation (RR = 0.22, SUCRA = 81.48%; PMA: 1.10%), whereas OUR-A consistently demonstrated only moderate performance across all assessed metrics. CONCLUSION: Spine robots demonstrate clear advantages in screw implantation accuracy and different robots may achieve optimal results under specific surgical requirements. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s13018-025-06005-6.