[Pragmatism and realism for public health intervention evaluation]

Forty years ago, Schwartz and Lellouch invented pragmatic clinical trials. Their proposal has not yet been fully espoused. This appears to be the case today also in the domain of public health interventions evaluation, where some still insist on the superiority of experimental methods. Yet evaluations of complex public health interventions are fraught with pitfalls for researchers. Most such interventions take place in natural experimental contexts, where they have no control over the context or the factors that modify implementation and influence the effects. Experimental approaches are, in these cases, not very appropriate, and yet decision makers want to be able to take decisions to improve them. This article presents our experience over the past 5years with evaluative research in two public health interventions. We wish to show how we conduct evaluations in practice using a pragmatic approach. The article is focused on elements that have not, to date, received much attention in the francophone literature: the evaluability assesment and intervention logic, research strategies reinforced particularly by mixed methods and time series, and the analysis of implementation fidelity and mechanisms that foster effectiveness. Because the pragmatic approach to evaluative research stresses the need for good understanding of context and uses reinforced methodological strategies, it allows for rigorous responses to evaluation questions raised by those implementing complex public health interventions. Thus, experimental approaches are not necessarily required to analyze the effectiveness of interventions.