All consecutive Japanese Brazilian out patients in the psychiatric units in Japan and in Brazil from April 1997 to April 2000 were compared. In Japan the psychiatric services were in 2 medical schools (Jichi Medical School, in Tochigi province and Aichi Medical School, in Aichi province) and HClinic in Chiba province (group remaining in Japan). In Brazil the psychiatric unit was Beneficiencia Nipo Brasileira in São Paulo city, a clinic that is a national reference in the assistance of Japanese Brazilians (group returning to Brazil). Diagnosis were based on ICD-10 and done by psychiatrists. The diagnosis of the patients were separated in 4 categories: schizophrenia (including brief psychotic disorder, schizophrenia, delusional disorder , schizoaffective disorder), depression (including major depression, dysthimia, bipolar), anxiety (including generalised anxiety disorder, PTSD, OCD, conversion, panic, somatization, adjustment disorder and social phobia) and other diagnosis. The variable diagnosis generated 3 new variables. For example, in the variable depression the individuals were separated in 2 groups: those with the diagnosis depression and those without this diagnosis (including anxiety, schizophrenia and others). The same procedure was followed for the variable schizophrenia and anxiety.

The sample of patients in Japan and in Brazil were compared in relation to diagnosis by 4 models of logistic regression. Each model used one diagnostic variable described above and sociodemographic variables that showed significance in the univariate analysis.

Results: In the univariate analysis we see that the group in Brazil were mostly male, not married, with past psychiatric history, lived alone and had a short stay period in Japan, and were diagnosed in the schizophrenia group (table 1). Analysing these variables with logistic regression (table 2) we note that the variables associated with the patients in Brazil that continue to be significant were short stay period and living alone (OR= 0.93 and 40.21 respectively). Substituting the variable diagnosis in the regression logistic model we observe that when the depressed were included the variables associated with the group in Brazil were: short stay period (B=-0.78; SE=0.15; p<0.01; Odds=0.92); living alone (B=3.56; SE=0.61; p<0.01; Odds=35.34); and male (B= 1.21; SE=0.56; p<0.05; Odds=3.35); for the group of schizophrenia: short stay period (B=0.72; SE=0.14; p<0.01; Odds=0.93); and living alone (B=3.51; SE=0.60; p<0.01; Odds=33.31); for the anxiety group: short stay period (B=-0.73; SE= 0.014; p<0.01; Odds=0.93, living alone (B=3.67; SE=0.64; p<0.01; Odds=39.38) and to be anxious (B=-1.86; SE=0.76; p<0.05; Odds=0.16)