Impact of microbial ecology on accuracy of surveillance cultures to predict multidrug resistant microorganisms causing ventilator-associated pneumonia

Summary Objectives The objective of this study was to assess surveillance cultures (SC) prediction accuracy in two periods and settings of the same Department with a different microbiological epidemiology (high and low prevalence of multi-drug resistant microorganisms (MDRM)). Methods Prospective and observational study. SC were obtained twice a week in consecutive mechanically ventilated patients. Patients fulfilling VAP criteria were analyzed. Results 440 patients were followed up, 71 patients had VAP (50 in period I and 21 in period II). MDRM causing VAP were more prevalent in the first period (48% vs. 19%; p  = 0.033). The rate of empirical appropriate treatment in period I was lower than in period II (52% vs. 76%; p  = 0.031). SC prediction accuracy was similar in the two periods (80% vs. 81%; p  = 0.744). However, if antibiotic treatment had been guided by SC, the percentage of appropriate treatment would have increased by 28% in the first period but only by 5% in the second; p  = 0.024. Conclusions SC were able to predict VAP etiology in 80% of cases regardless the prevalence of MDRM. However, the potential benefit of SC in terms of appropriate empirical treatment could be only observed when MDRM were prevalent.