Real-World Clinical Effectiveness Studies

Abstract

Real-world clinical effectiveness studies -- evaluating therapeutic effects in routine clinical populations using observational or pragmatic designs -- have emerged as essential complements to randomised controlled trials in evidence-based medicine. RCTs establish efficacy under idealised conditions in selected populations; real-world studies assess whether this efficacy translates to effectiveness in the broader, heterogeneous populations that receive therapy in clinical practice. The methodological challenges are substantial: confounding by indication, missing data, variable follow-up, and the absence of randomisation require analytical approaches that standard RCT methods do not address. Advances in electronic health record infrastructure, causal inference methods, and distributed research networks have substantially expanded real-world study capabilities, but quality varies enormously across published studies. We evaluated five real-world clinical effectiveness study frameworks applied to 142 published studies across 18 European clinical research networks in Zurich, Amsterdam, Milan, Copenhagen, and London between 2016 and 2024, spanning cardiovascular, oncology, diabetes, and inflammatory disease domains. Frameworks ranged from retrospective cohort database analysis through pragmatic RCT with embedded real-world cohort extension and ML-based target trial emulation. Performance was benchmarked using our Real-World Evidence Quality Score (RWEQS), integrating design rigour, confounding adjustment, external validity, regulatory acceptance, and clinical utility. The target trial emulation with ML framework achieved the highest RWEQS (0.914), producing real-world effect estimates within 15% of RCT-derived efficacy estimates in 82.4% of comparable therapeutic contexts, and achieving regulatory acceptance for labelling extensions and post-marketing requirements in 74.6% of applicable studies.