Multivariate analysis applied in dataset of Poison Control Center of São Paulo, Brazil.

Multivariate analysis techniques could be used to identify possible intercorrelations in intoxications cases. The statistical analyses used were a multiple logistic regression, multiple correspondence analysis, principal component and hierarchical cluster analysis. Of the 320 samples analysed, 192 samples were positive for some of the investigated toxic agents, of which 100 were positive for ethanol and 131 were positive for other substances. It was possible to group the patients into 3 clusters, which appears 66.5% of this information in the three first factorial axes. On the first axis, the male patients were separated from the female patients. Patients exposed to drugs, between 30 and 39 years old were grouped in the same cluster. On the second factorial axis, patients who were intoxicated with ethanol and who became intoxicated with diazepam were grouped. This work contributed to the mapping of intoxication cases at the Poison Control Centre of the São Paulo city, Brazil (CCI-SP) and serves as an initial study for the creation of a database that could be updated constantly and thus could provide a toxicovigilance system for educational policies.