The study hypothesis is that the loading dose of intravenous colistin (6 million of international units) is associated with greater clinical and microbiological efficacy, and reduced mortality of critically ill patients infected by multidrug resistant Gram- negative bacilli, compared to a scheme without loading dose.

It is a prospective, multicenter, randomized, controlled study to evaluate a scheme with and without a loading dose of 6 million international units of colistin, followed by a maintenance dose of 3 million international units every 8 hours intravenous. The study should be conducted in 3 hospitals in Chile, in critically patients presenting infection by multidrug Gram-negative bacteria and requiring be treated with colistin for at least 48 hours. The objectives of the study are: to evaluate the clinical and microbiological response, and mortality.

Inclusion Criteria:

• patient hospitalized in critical care units
• patient infected by multi drug resistant Gram negative bacteria susceptibly only to colistin
• source of infection: blood, respiratory, intra abdominal or urinary

Exclusion Criteria:

• pregnant or breastfeeding patients
• patient with a history of hypersensitivity to colistin