Relationships between patient- and institution-specific variables and decreased antimicrobial susceptibility of Gram-negative pathogens.

The identification of patients infected with antibiotic-resistant strains of bacteria for inclusion in clinical trials remains a serious challenge for the future development of agents for use against such infections. To identify patient- and institution-specific factors predictive of reduced susceptibility of Enterobacter species, Pseudomonas aeruginosa, and Klebsiella pneumoniae to cefepime, ciprofloxacin, and piperacillin-tazobactam, 5 years (1997-2001) of North American surveillance data were analyzed. The relationship between minimum inhibitory concentration (MIC) values for each organism-agent pair and patient- and institution-specific variables was analyzed using multivariable general linear modeling. The variables most commonly associated with decreases in susceptibility were duration of hospital stay before pathogen isolation, hospital size, primary diagnosis, and medical service. Combinations of these variables were associated with increases in observed MIC90 values of as much as 16-32-fold. Our findings demonstrate a relationship between MIC and certain patient- and institution-specific variables. Such data should be considered in the design of clinical trials directed at the study of resistant pathogens.