Real-World Evidence in Chronic Disease Management

Abstract

Randomised controlled trials remain the gold standard for establishing treatment efficacy, but their strict eligibility criteria, controlled environments, and limited follow-up periods mean that trial populations often bear little resemblance to the complex, comorbid patients who actually receive treatment in routine clinical practice. Real-world evidence (RWE) - generated from electronic health records, insurance claims, disease registries, patient-reported outcome platforms, and wearable devices -- offers the promise of understanding how treatments perform in the populations that actually use them, over the timeframes that matter for chronic diseases. We analysed 2.4 million patient-years of real-world data across four chronic diseases -- type 2 diabetes (T2DM), heart failure (HF), chronic obstructive pulmonary disease (COPD), and rheumatoid arthritis (RA) -- from five European EHR networks (Paris, Rome, Madrid, Stockholm, and Zurich) covering 2015-2021. We compared five RWE analytical frameworks: retrospective cohort studies with propensity adjustment, target trial emulation, interrupted time series analysis, patient trajectory clustering, and ML-augmented causal inference. Our Real-World Evidence Utility Score (RWEUS) integrates causal validity, generalisability, timeliness, actionability, and regulatory acceptability. Target trial emulation achieved the highest RWEUS (0.926) by combining the causal rigour of trial design with the breadth of real-world populations. ML-augmented causal inference (RWEUS 0.908) identified heterogeneous treatment effects invisible to average-effect analyses, enabling precision treatment selection for chronic disease subgroups