[The evaluation of pharmacological drugs, medical devices, and non-pharmacological or public health interventions: Experimental design limitations. Moving towards new methods?] / Évaluer les effets d'un médicament, de dispositifs médicaux et d'interventions non médicamenteuses ou de santé publique : limites de l'essai expérimental. Vers de nouvelles méthodes ?

In the field of health, evidence-based medicine and associated methods like randomised controlled trials (RCTs) have become widely used. RCT has become the gold standard for evaluating causal links between interventions and health results. Originating in pharmacology, this method has been progressively expanded to medical devices, non-pharmacological individual interventions, as well as collective public health interventions. Its use in these domains has led to the formulation of several limits, and it has been called into question as an undisputed gold standard. Some of those limits (e.g. confounding biases and external validity) are common to these four different domains, while others are more specific. This paper describes the different limits, as well as several research avenues. Some are methodological reflections aiming at adapting RCT to the complexity of the tested interventions, and at overcoming some of its limits. Others are alternative methods. The objective is not to remove RCT from the range of evaluation methodologies, but to resituate it within this range. The aim is to encourage choosing between different methods according to the features and the level of the intervention to evaluate, thereby calling for methodological pluralism.