Compliance with the CONSORT AI Extension in Orthopaedic Randomized Controlled Trials

BACKGROUND: The utilization of artificial intelligence (AI), such as machine learning, in health care has risen exponentially, underscoring the necessity of transparency and traceability in the reporting of AI in randomized controlled trials (RCTs) that involve AI. In 2020, a CONSORT (Consolidated Standards of Reporting Trials) AI extension was published that included 14 additional items that should be reported when AI is used in RCTs. This study aimed to evaluate compliance with these additional guidelines. METHODS: A systematic approach was used to identify all RCTs in orthopaedics that involved AI since the extension was published. Three databases were searched (PubMed, Embase, Cochrane) using the search terms ([orthopaedic] OR [orthopedic]) AND ([artificial intelligence] OR [AI] OR [machine learning] OR [deep learning]). RESULTS: One hundred and seventy-four articles were identified, and 11 articles satisfied the inclusion and exclusion criteria and were assessed for compliance. The average number of CONSORT AI-extension items met per study was 11.4 ± 1.6 (81.2% ± 12.1%). Eight items (1a [i], 1b [ii], 2a [i], 4a [i], 5 [ii], 5 [v], 5 [vi], 19) that were related to standard reporting of AI utilization were reported in all articles. The items with the lowest reporting rates were related to data transparency and included 4a (ii), 5 (iii), and 25, with only 9.1%, 36.4%, and 45.5% of all studies complying, respectively. CONCLUSIONS: Orthopaedic RCTs involving AI showed high compliance with respect to standard trial reporting and broad AI descriptors but low compliance with respect to detailed data transparency, data-quality handling, and code accessibility. Addressing these weaker areas is essential to improve reproducibility, reduce bias, and strengthen trust in AI-driven clinical research. CLINICAL RELEVANCE: Improved compliance with CONSORT-AI reporting standards may help orthopaedic clinicians to critically evaluate AI-based randomized trials, reproduce findings, and determine whether AI interventions are reliable and safe for clinical use.