Related Experiment Videos
Visualising exposure-disease association: the Lorenz curve and the Gini index
Javier Llorca1, Miguel Delgado-Rodríguez
1Division of Preventive Medicine and Public Health, University of Cantabria School of Medicine, Santander, Spain. llorcaj@unican.es
Background:
It has been suggested that the summary index of the Lorenz curve, the Gini index, should be used to characterize the exposure-disease association, rather than relative and attributable risks. Nevertheless, the sampling behavior of the Gini index in epidemiological settings, and the relationships between the Lorenz curve and the usual indices of risk, need to be more deeply understood.
Material/Methods:
The present paper analyzes the geometrical relationships between the Lorenz curve and the relative and attributable risks based on two of the main sampling schemes (cohort and case-control designs). Examples for both designs are provided. Gini index confidence intervals are obtained by bootstrap.
Results:
The Gini index is a function of the proportion of the population at each level of exposure, relative risk and attributable risk. If exposure is ordered by increasing levels of risk, the Lorenz curve contains all the information about relative and attributable risks and distribution of exposure in the population. Therefore, the Lorenz curve easily allows both risks and their distribution in the analyzed population to be visualised.
Conclusions:
The Lorenz curve and the Gini index can complement the information provided by relative and attributable risks.