This study has some limitations. the definitions of hospital and community acquired CRB infections are not based on the duration of hospitalisation. moreover some strains were classified in another group. therefore the classification are objectionable. however these definition allowed us to classify 93.9 of the 4847 CRB strain and 94.8 of the 23436 E.coli strain in hospital and community acquired infection group. moreover, we were not able to collect data on colistin consumption prior to 2008, which may have prevented us