Impact of different methods for describing the extent of prior antibiotic exposure on the association between antibiotic use and antibiotic-resistant infection

Objective. Many studies have investigated the association between prior antibiotic use and antibiotic resistance. However, methods used in past studies to describe the extent of prior antibiotic use (eg, use of the 2 categories exposure versus no exposure and measurement of duration of exposure) have not been reviewed. The impact of the use of different methods for quantifying the use of antibiotics is unknown. The objectives of this study were to characterize past approaches to describing the extent of antibiotic use and to identify the impact of the use of different methods on associations between use of specific antibiotics and infection with an antibiotic-resistant-organism. Methods. We conducted a systematic review of studies that investigated risk factors for extended-spectrum beta-lactamase (ESBL)-producing Escherichia coli and Klebsiella species to identify variability in past approaches to describing the extent of antibiotic use. We then reanalyzed a data set from a prior study of risk factors for infection with ESBL-producing E. coli and Klebsiella species. We developed 2 separate multivariable models: 1 in which prior antibiotic use was described as a categorical variable (eg, exposure or no exposure) and 1 in which antibiotic use was described as a continuous variable (eg, measured in antibiotic-days). These models were compared qualitatively. Setting. Large academic medical center. Results. The 25 articles included in the systematic review revealed a variety of methods used to describe the extent of prior antibiotic exposure. Only 1 study justified its approach. Results from the 2 multivariable models that used different methodologic approaches differed substantially. Specifically, use of third-generation cephalosporins was a risk factor for infection with ESBL-producing E. coli and Klebsiella species when antibiotic use was described as a continuous variable but not when antibiotic use was described as a categorical variable. Conclusions. There has been no consistent method for assessing the extent of prior antibiotic exposure. The use of different methods may substantially alter the identified antimicrobial risk factors, which has important implications for the resultant interventions regarding antimicrobial use.