The study processed recent official data for the entire nation
and replicated methods previously used in other studies (Faria
et al., 2006; Parron et al., 2011; Bando et al., 2012). The
characterization of exposure up to 2006 and of suicides from
then onwards reduced the temporal ambiguity of the association.
The choice of micro-regions as the unit of analysis meant that we
could avoid oscillations in the rates of smaller municipalities and
problems of population definition regarding the creation of new
municipalities during the study period (Faria et al., 2006).