The NPIS model: a consensus-based framework for evaluating non-pharmacological interventions

The term non-pharmacological intervention (NPI) refers to health prevention and care protocols supervised by healthcare professionals. However, no precise and widely adopted definition currently exists. Additionally, unlike drugs, NPIs do not have a global evaluation framework because of a lack of consensus among stakeholders, stemming from heterogeneity in intervention content and study protocols. This heterogeneity limits scientific impact, hinders dissemination and continuous improvement of practices, and contributes to significant mistrust among professionals and users. From 2022 to 2023, we conducted a consensus study, collaborating with over 500 stakeholders—including researchers, healthcare users, healthcare practitioners, health operators, scientific societies, and health authorities—to co-construct an NPI definition and a framework for evaluating NPIs that align with their specific characteristics and international health research standards. Through a structured, hybrid consensus process, primarily inspired by the Nominal Group Technique and supplemented with elements of a modified Delphi approach, we developed the NPIS Model under the guidance of a multidisciplinary committee of 22 experts through iterative, open, and documented exchanges across four stages: (i) creation of an initial list of ethical and methodological items by a committee of 70 members, (ii) refinement of this list by a committee of 300 members, (iii) open voting on each item by 503 voters, and (iv) consultation with 36 scientific societies and 14 health authorities. We reached a consensus defining an NPI as an “evidence-based, effective, personalized, non-invasive health prevention or care protocol, registered and supervised by a qualified professional”. Based on this definition, we developed the NPIS Model, an evaluation framework structured as a set of 77 recommendations for evaluating NPIs—14 ethical and 63 methodological. Each recommendation received at least 80% agreement among voters. The recommendations are categorized into five types of studies: mechanistic, observational, prototypical, intervention, and implementation. The NPI definition was established through consensus but, unlike the 77 recommendation items, was not submitted to the collegial vote. To date, 31 scientific societies and three health authorities have provided letters of support for the NPIS Model. The NPIS Model is intended to promote transparency, methodological rigour, ethical standards, and transferability for NPI research, potentially increasing the value of NPI evaluation for researchers, healthcare practitioners, healthcare users, and health authorities. Because the framework was developed by francophone contributors, with voting restricted to residents of France, its applicability beyond this setting remains to be established through future international consultation and validation.