Methodologies for generating and evaluating clinical and performance evidence for high-risk and innovative medical devices and in vitro diagnostics: a scoping review

Introduction High-risk and innovative medical devices (MDs) and in vitro diagnostic devices (IVDs) in the European Union are subject to increasingly stringent and evolving regulatory requirements. This scoping review maps methodological approaches to evidence generation across their life cycle. Methods A comprehensive and systematic search of major bibliographic databases and grey literature sources was conducted (January 2025, updated November 2025). Studies were selected through dual independent screening using predefined inclusion criteria. Data were extracted using a structured approach and supported by AI-assisted thematic analysis. Findings were synthesized narratively to map methodological approaches to evidence generation and evaluation. Results A total of 15,679 records were identified, with 14,259 remaining after duplicate removal and 859 full texts assessed for eligibility. Of these, 90 studies met the inclusion criteria. In addition, records identified through other methods were screened, with 75 progressing to full-text assessment; of these, 8 were included . Evidence generation for high-risk and innovative MDs and IVDs requires an integrated, life-cycle approach combining multiple study designs. While randomized controlled trials remain important, observational studies, real-world evidence, registries, and post-market surveillance are essential for assessing long-term safety and effectiveness. However, current practices show substantial variability, methodological limitations, and regulatory inconsistencies. Conclusions An integrated lifecycle approach is needed for robust evaluation of high-risk and innovative MDs and IVDs. Strengthening methodological standards, improving the quality of real-world data, and enhancing regulatory harmonization are essential. Systematic Review Registration Open Science Framework (OSF), doi: 10.17605/OSF.IO/XNBSP.