A reproducible framework for monitoring the impact of randomized clinical trials on clinical practice using large-scale real-world data: application to gynaecological surgical trials using the French national healthcare database.

Background: Randomized clinical trials (RCTs) are fundamental to evidence-based medicine, but their real-world impact on clinical practice often remains unmonitored. Leveraging large-scale real-world data can enable systematic monitoring of RCT effects. We aimed to develop a reproducible framework using real-world data to assess how major RCTs influence medical practice, using two pivotal surgical RCTs in gynaecologic oncology as an example-the LACC (Laparoscopic Approach to Cervical Cancer) and LION (Lymphadenectomy in Ovarian Neoplasms) trials. Methods: We utilized data from the French National Health Insurance Database (SNDS), covering 98.8% of France's population. We analysed patients who underwent radical hysterectomy for cervical cancer (2013-2022) and patients who underwent cytoreductive surgery for ovarian cancer (2014-2022). Bayesian structural time series analysis assessed the causal effects of the LACC and LION trials on the discontinuation of minimally invasive surgery (MIS) and lymphadenectomy, respectively. Analyses were stratified by hospital type, academic status, research mission, domain expertise, human resources, and financial condition. Findings: Our nationwide cohorts included 7108 cervical cancer and 23,090 ovarian cancer patients treated across 596 centres. The LACC trial led to a 14.1% reduction in radical hysterectomies by MIS (275 fewer surgeries; 95% CI: -407 to -140), with academic centres showing 27.9% reduction compared to 2.5% increase in nonacademic centres. The LION trial resulted in a 22.6% reduction in lymphadenectomies (2358 fewer surgeries; 95% CI: -2708 to -2003), with academic centres achieving 31.1% reduction versus 15% in nonacademic centres. Significant variation was observed across medical settings. Centres with academic status, high research missions, substantial expertise, and robust resources were more responsive to trial outcomes, highlighting the influence of institutional and human factors on adopting new practices. Interpretation: This study demonstrates that large-scale real-world data can effectively monitor the impact of RCTs on clinical practice. While validated here using surgical trials, this reproducible framework is adaptable to various health domains and can be implemented in any country with national electronic health databases. Systematic monitoring is essential to ensure effective implementation of RCT findings and to address disparities in the adoption of evidence-based practices. Funding: None.