[Use of hierarchical models in nephrology]

The use of hierarchical models in public health research is recently increasing in order to study all the factors explaining health outcomes. Thus, a better understanding of those models is needed first to identify questions that may be answered by using them and also to be aware of there limitations. On the one hand, hierarchical models managed to take into account the hierarchical structure of the data allowing a better estimation of the effects of the explanatory variables and the study of the impact of the "environment" (i.e.: neighborhood, treatment center, same clinical trial…) on health outcomes. They also allow the study of factors that may explain this impact of the "environment". On the other hand, they are more complex and a reflection on which determinant to include and how the environment is supposed to impact patients' health is much needed. This article reviews the rationale for using hierarchical models in public health research and especially in nephrologic research. We attempt to give a simple presentation of these models and to illustrate their results and potential use in the field of nephrology, as well as their limits.