Managing resistance in the ICU : an evolutionary approach to rational antibiotic deployment

Nosocomial infections account for 5 to 10% of all infections in the United States and act as a continuous reservoir for the maintenance of antibiotic resistance. Development, testing, and implementation of broad antimicrobial deployment strategies are crucial in the proper management of resistance over the long term. We present a population-based model which describes the spread of variably-resistant nosocomial pathogens amongst patients in an intensive care unit of a hospital. Our purpose is to identify treatment strategies which maximize the number of uninfected individuals while maintaining low rates of multi-resistant infections. This was accomplished via the expansion of a previously published model by introducing pharmacodynamics, pharmacokinetics, and cross- resistance tradeoffs. Most importantly, we depart from this model's predecessors by treating the minimization of resistant-infected individuals as secondary to maximizing uninfected. We confirm that the benefit of a random mixing regimen over periodic cycling is minimal, while a hybrid of the two is slightly more effective. Finally, we show that time- and probability-based strategies are inferior to Multi-Drug cocktails in their ability to exploit resistance-associated fitness tradeoffs; thereby selectively favoring susceptible genotypes. These results provide an impetus to identify Multi-Drug cocktails which serve to minimize the incidence of multi-resistance while still maintaining curing efficac