Real-World Evidence: contribuições metodológicas e estudos de caso na prática clínica

Abstract

The production of scientific evidence in healthcare has undergone a profound transformation over the past two decades, driven by the digitalization of clinical information systems and the widespread availability of administrative, operational, and clinical data. Within this evolving landscape, Real-World Evidence (RWE) has emerged as a fundamental component of contemporary medical research, providing critical insights into effectiveness, safety, utilization patterns, and economic impact in settings that are not accessible through randomized clinical trials (RCTs). This dissertation aims to analyze, integrate, and discuss four original studies based on Real-World Data (RWD), conducted across distinct clinical domains within the Brazilian private healthcare system, in order to demonstrate how RWE enhances the understanding of real-world practice and complements traditional sources of evidence. The first study evaluated more than 549,000 consultations delivered through a digital urgent care platform, identifying real-world utilization patterns, predictors of referral to in-person emergency departments, and factors associated with high-frequency use. The findings revealed a low referral rate, high rates of clinical resolution within the digital environment, and distinct profiles of users more likely to exhibit recurrent utilization. The second study employed Natural Language Processing (NLP) techniques to extract LDL-C values from unstructured laboratory reports, enabling a comprehensive evaluation of lipid control and healthcare costs among individuals with and without atherosclerotic cardiovascular disease (ASCVD). The results demonstrated that only a minority of high-risk patients achieved guideline-recommended LDL-C targets and that ASCVD patients had significantly higher annual healthcare costs, highlighting substantial care gaps and the essential role of digital technologies in scaling RWE analyses. The third study examined real-world treatment patterns, observed survival, and direct medical costs among patients with metastatic non–small cell lung cancer (NSCLC). Based on integrated administrative and clinical data, the study found pronounced discrepancies between RCT-based efficacy and real-world effectiveness, as well as significant variation in costs across therapeutic strategies, underscoring the importance of RWE for economic evaluation and decision-making in oncology. The fourth study described population-level prescribing and utilization patterns of metamizole (dipyrone) in over 380,000 patients, revealing widespread use across diverse age groups, marked seasonality, and frequent co-administration with other analgesics. These findings reinforce the critical role of large-scale RWE analyses for post-marketing surveillance and monitoring of commonly used medications. Collectively, the four studies illustrate that RWE is indispensable for understanding the complexity of real-world clinical practice and for addressing questions that are impractical or infeasible to investigate through RCTs. This dissertation concludes that while RWE does not replace randomized trials, it occupies a complementary and essential position within the broader continuum of evidence, supporting clinical, organizational, and regulatory decision-making in modern healthcare systems.