[Factors predicting metallo-beta-lactamase-producing Acinetobacter baumannii in Iwate Medical University Hospital]

We conducted a decision tree analysis in which 6 isolates of metallo-beta-lactamase (MBL) -producing Acinetobacter baumannii detected by a sodium mercaptoacetic acid disk method and 38 isolates of MBL nonproducing A. baumannii were investigated from January 2003 to December 2003 in Iwate Medical University Hospital. Factors predicting MBL-producing A. baumannii were analyzed from the clinical backgrounds such as laboratory data, therapeutic histories and bacterial circumstances. In the decision tree analysis, seven significant rules were found for predicting MBL-producing A. baumannii; previous isolation of P. aeruginosa, administration period of 3rd generation cephem antibiotics > 11 days, both TP < or = 6.9 g/dl and Hb < or = 6.3 g/dl, both administration of anticancer drugs and a use of respirator, TP > 6.9 g/dl, CRP > 22.7 mg/dl, both CRP 8.7 mg/dl and body temperature > 36.6 degrees C. Two cases, which were matched with the rule of TP > 6.9 g/dl, were hospitalized in the same division, and their pulsed-field gel electrophoresis patterns were identical. In comparison with conventional statistical analysis, the decision tree analysis has advantages in explaining the tendencies and the relationships of the data. It is possible that, based on the rules extracted in this study, we could issue precautions in preventing nosocomial MBL-producing strains even in the earlier stages when the number of these isolates are still small.