470: PREDICTION SCORING FOR PRIMARY RESISTANCE IN GRAM-NEGATIVE BACTEREMIA WITHIN A LARGE HEALTH SYSTEM

Abstract

Introduction: Resistant Gram-negative organisms complicate successful treatment of infectious diseases in critically ill patients as common antibiotic agents are ineffective. Knowledge of risk factors associated with antimicrobial resistant Gram-negative infections is necessary to select appropriate antibiotic agents as culture results are not available immediately. The primary objective was to identify risk factors associated with primary resistance of selected Gram-negative bacteria isolated from blood and establish a scoring method to predict primary resistance. Methods: A total of 378 cases from 390 isolates were analyzed within the Banner health system. Fifty susceptible cases were identified for each of the following organisms: Pseudomonas aeruginosa (Ps), Escherichia coli (EC), Klebsiella pneumoniae (KP), and Enterobacter cloacae (Ent. c). The number of nonsusceptible cases were 50 for EC and KP, 27 for Ent. c and 50 for Ps. Results: The presence of chronic kidney disease (CKD), peripheral vascular disease (PVD), connective tissue disease (CTD), hemi/paraplegia, ventilator use, central vascular access, transfer from long term care facility, admission from outpatient parenteral antibiotic therapy, receiving chemotherapy, history of cerebral vascular accident, diabetes with complication and history of infection within the previous 6 months were strongly associated with resistance. A logistic regression model was developed for predicting primary resistance. An individual with no risk factors for primary resistance would have a predicted probability of resistance of 27%. Further analysis of these factors is required to remove any correlation bias and to develop a prediction scoring system to determine primary resistance. Conclusions: To conclude, this study was able to identify certain risk factors to be statistically significant (p< 0.05) to determine primary resistance which will help develop a prediction scoring system that will aid in appropriate antibiotic selection for the treatment of selected Gram-negative bacteremia infections.