Clinical trials have formed the cornerstone of Evidence-Based Medicine development over the last decades. Conducted in controlled research settings, they minimize bias through randomization and multicentricity, estimating the potential benefit of a new therapeutic strategy over a conventional one. While these designs provide excellent internal validity, their applicability to everyday practice can be limited. This limitation is primarily attributed to selection bias, a significant factor in study design. Excessive or overly strict exclusion criteria detach the study sample from the target population, resulting in conclusions that reflect laboratory findings rather than clinical research.

In contrast, registries offer evidence with characteristics opposite to those of clinical trials. Retrospective by design, data in registries are typically collected after patient clinical outcomes have occurred. With minimal selection criteria, registries include consecutive patients with few restrictions, resulting in large case volumes and multicentricity that minimize bias, thereby ensuring strong external validity. However, the reliance on strict data collection, which is not as flawless as in clinical trials, and the influence of each country’s healthcare idiosyncrasies hinder the exploitation and relevance of registries in scientific evidence (clinical guidelines).

Various experts have highlighted the need to reevaluate reference sources in Evidence-Based Medicine. A study evaluating 52 clinical trials in Cardiology, Mental Health, and Oncology revealed that 70% of them (37 of 52) did not represent the target population for which their conclusions would be applied. For this reason, the authors conducted an extensive review, analyzing 14 primary clinical trials comparing CABG and PCI, as well as 10 major national registries (4 for CABG and 6 for PCI). They examined 16 pre-procedural variables to define patient characteristics and compared samples from trials and registries for each treatment type, CABG and PCI. This review included 2,252,257 patients, with 11,824 from trials (50% CABG, 50% PCI) and 2,238,623 from registries (423,843 CABG, 1,814,780 PCI). Patients in trials and registries demonstrated notable differences. For CABG, 5 out of 16 variables showed statistically significant differences between registries and trials. Trial patients displayed lower morbidity: reduced smoking rates, better ventricular function, lower prior MI rates, and more single-vessel disease. Early mortality was better in trials (2%) compared to registries (3%), p = 0.005. For PCI, 9 out of 16 variables showed significant differences, indicating lower morbidity among trial participants: they were younger, had lower smoking rates, and better ventricular function. Trials had more anatomical cases characteristic of surgery, LMD and/or three-vessel disease, compared to registries. Early mortality in trials was also better than in registries, 1% vs. 2%, p = 0.003.

The authors emphasized the observed differences and the limited external validity of the clinical trials supporting current clinical guidelines.

COMMENTARY:

Evidence-Based Medicine is currently facing significant scrutiny. The scientific progress achieved over the past decades through clinical trials has been unprecedented across all fields of Medicine. However, the involvement of the pharmaceutical industry with necessary funding, and the demonstrated polarization of results through pre-intervention patient selection or post-procedural variable redefinition, has weakened the credibility and real-world applicability of these studies.

The results identified by the authors confirm that similar trends occur in the evidence base for daily practice. Recommendations for stable coronary artery disease management between PCI and CABG remain controversial, and the new myocardial revascularization guidelines from the ESC/EACTS are expected in the coming months. Although limited validity has been detected in clinical trials within this field, the authors acknowledge that it is less significant than in other pathologies, such as valvular surgery and structural interventionism. A solution must be sought in the coming years. Emerging computational systems and the development of significant national registries, like our RECC, open the door to Big Data as a new evidence source. This potentially more realistic approach may introduce new limitations and biases, which we will need to navigate to identify the best therapeutic strategies for our patients. Until then, multidisciplinary collaboration and common sense should prevail.

REFERENCE

Chadow D, Audisio K, Perezgrovas-Olaria R, Cancelli G, Robinson NB, Rahouma M et al. Differences among clinical trials and registries on surgical and percutaneous coronary interventions. Ann Thorac Surg. 2023 Jan;115(1):79-86. doi: 10.1016/j.athoracsur.2022.05.015.