Clinical Relevance of Surgical Trials – an analysis of planning, reporting and interpretation of surgical trials

Résumé fourni par la source

Medical research often aims to inform decisions where clinicians face uncertainty in decision-making, for example, when a condition could be treated in different ways and the balance of respective beneficial and harmful outcomes is unclear. Randomized controlled trials (RCTs) are ideally suited for comparing outcomes following different interventions, and to provide high-level evidence intended to meaningfully inform clinical practice. The outcomes they measure primarily and secondly are chosen during planning of the trial. The primary endpoint typically reflects the outcome of greatest interest, guides study design and serves as basis to determine sample size required to detect a predefined difference between interventions. Also, the primary guides interpretation of the comparison of interest, together with secondary outcomes. The secondary endpoint(s) provide complementary measures of the intervention’s effects such as harmful outcomes or additional beneficial outcomes. Together, primary and secondary outcomes contribute to an overall benefit–risk assessment. The direction of expected effects should be reflected in the hypothesis and the aim of a trial followed by the appropriate design choice. Superiority designs aim to assess if a specific intervention leads to better outcomes compared with an alternative. In turn, non-inferiority and equivalence designs assess whether an intervention can be considered a clinically acceptable or equivalent alternative, since the primary outcomes are comparable to that of a standard intervention while safety, cost, or feasibility may be favorable. Latter is often the case in surgical trials where novel surgical techniques, neoadjuvant therapies or perioperative management are compared to some standard intervention. Such novel interventions may not outperform standard treatments (e.g., with respect to survival) but offer less perioperative burden. Therefore, for RCTs to meaningfully inform clinical practice, the aim of the trial and the aim of the primary endpoint must be clearly defined and supported by an appropriate study design, including adequate sample size calculation for a clinically meaningful effect and statistical methods aligned with the study aim. Beyond methodological alignment, clinical relevance must be integrated throughout the trial process, from planning to interpretation. This requires the selection of patient-relevant outcomes, the definition of a minimal important difference (MID) to guide sample size calculation, and interpretation of results considering their clinical rather than only their statistical significance. Failure to integrate clinical meaning at the planning stage may compromise study design, endpoint selection, effect size definition, and ultimately the interpretability and applicability of trial findings. To explore these issues, we previously conducted a pilot meta-research study of surgical RCTs citing the Comprehensive Complication Index (CCI®)[1], the only validated patient-relevant continuous endpoint for postoperative morbidity. We examined how trial endpoints, study design, and sample size calculations were reported in pre-trial documentation and final publications. Among 84 eligible RCTs, sample size calculations were almost universally reported; however, the aim of the trial and/or of the primary endpoint was frequently unspecified in pre-trial documentation (73%), most trials defaulted to superiority designs, and only 20% of statistically significant results were interpreted in a clinically meaningful way. These findings highlighted important shortcomings in the incorporation of a meaningful planning, reporting and interpretation of results in surgical RCTs and motivated the present extension study. In this meta-research study, we aim to assess to what extent surgical contemporary RCTs are designed and reported in a manner that meaningfully informs clinical practice.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.