Revista Brasileira de Epidemiologia
versão impressa ISSN 1415-790X
LUPPI, Irene e BOGGIO, Gabriela. Multilevel models: an analysis strategy for the study of health problems in society. Rev. bras. epidemiol. [online]. 2006, vol.9, n.1, pp. 42-55. ISSN 1415-790X. http://dx.doi.org/10.1590/S1415-790X2006000100007.
This paper presents the theoretical-methodological discussion about the applicability of multiple level models in the study of the health/sickness process, its determinants and conditioning factors, as a function of the stratification of society and the living conditions of its inhabitants. It goes back to the concept of population according to the theory of hierarchical complex systems, which seeks not to reduce reality, but rather to build the problem trying to identify different levels of abstraction in its approach. These models are options to overcome prior experiences, with the application of conventional statistical techniques, given that they make it possible to simultaneously analyze different levels of aggregation, while keeping its hierarchical structure. They consider the influence of the variables taking into account their belonging to lager units and the potential association existing between the units of a same level, that is, the intraclass correlation among variables relative to individuals, families and groups, close amongst themselves, which share similar conditions. In this manner, it tries to avoid oversizing the effect of macro level variables. The multiple level models are particularly appropriate to evaluate inequalities in the health/sickness/care process of the population groups and to analyze how social contexts affect the results and health risks of people. It highlights the need to develop information production strategies and analyses that make it possible to recognize levels of explanation and intervention to provide inputs and trigger actions suited to local specificities, at the level of micro-areas, so as to have more equity in healthcare.
Palavras-chave : Multiple level models; Hierarchical systems; Epidemiological study methods; Studies of the inequalities and equalities in healthcare; Local space and micro area..