Predicting resistant etiology in hospitalized patients with blood cultures positive for Gram-negative bacilli

Objective: To develop a risk-scoring tool to predict multidrug-resistant (MDR) etiology in patients with bloodstream infections (BSI) caused by Gram-negative bacilli (GNB). Methods: A prospective multicenter study analyzed patients with BSI hospitalized in 31 Internal Medicine wards in Italy from March 2012 to December 2012. Patients with BSI caused by MDR-GNB (non-susceptible to at least one agent in three antimicrobial categories) were compared to those with BSI due to susceptible GNB. A logistic regression to identify predictive factors of MDR-GNB was performed and the odds ratio (OR) were calculated. A score to predict the risk of MDR was developed. Results: Of 533 BSI episodes, 253 (47.5%) were caused by GNB. Among GNB-BSI, 122 (48.2%) were caused by MDR-GNB while 131 (51.8%) by non-MDR GNB. At multivariate analysis transfer from long-term care facility (OR 9.013, 95% CI 1.089-74.579, p = 0.041), hospitalization in the last 3 months (OR 2.882, 95% CI 1.580-5.259, p = 0.001), urinary catheter (OR 2.315, 95% CI 1.202-4.459, p = 0.012), antibiotic therapy in the last 3 months (OR 1.882, 95% CI 1.041-3.405, p = 0.036), age >= 75 years (OR 1.866, 95% CI 1.076-3.237, p = 0.026) were factors independently associated with MDR etiology. A score ranging from 0 to 10 was useful to recognize patients at lowest risk (0 points: Negative Likelihood Ratio 0.10) and those at highest risk (> 6 points, Positive Likelihood Ratio 11.8) of GNB bacteremia due to a MDR strain. Conclusions: Specific predictors of MDR etiology are useful to calculate probabilities of MDR etiology among hospitalized patients with blood cultures positive for GNB.