[The prognosis of thrombocytopenic purpura using predictive criteria evaluation by multivariate analysis methods].

The development of predictive systems with the resources of multivariate analysis let us to do gradually a more objective medicine. In this study we correlate a set of characteristics observed in 39 patients with thrombocytopenic purpura during the first week of medical treatment against the score of severity of the whole purpuric disorder, evaluated a the 25 week of evolution. It was accomplished with a model of multiple logistic regression of successive steps which let us to get the weights of each one of the indicants in order to predict the severity of the whole hemorrhagic diathesis. Probably the more significative indicants of severity are intracranial bleeding, profuse nasal and vaginal bleeding, profuse petechial bleeding and the presence of hematomas.