RESUMO
OBJECTIVE: To analyze the relationship between antimicrobial use and susceptibility in gram-negative and gram-positive bacteria in a general hospital during a 13-year period. METHODS: Normalized antimicrobial consumption (defined daily dose per 100 bed-days) was determined for 58 antibiotics for the period of 1993 to 2005, and susceptibility percentages were calculated for all possible antibiotic-microbial combinations for the same period. Both simple and multiple relationships were considered in 2 different settings: the intensive care unit and the remaining medical-surgical departments. Simple linear regression models for sensitivity-usage were employed, with delays of 0, 1, and 2 years; relationships with determination coefficients (r2) higher than 0.5 and negative correlation coefficients (r) were selected. These selected relationships were further analyzed using both autoregressive models to account for autocorrelation in the error term, and polynomial distributed lag regression models that allow distribution in time of the usage effect on sensitivity, considering all delays simultaneously. RESULTS: The increase in consumption of some antimicrobials has negatively influenced sensitivity to other antimicrobials, with an immediate influence in time for some of them (imipenem, 3rd generation cephalosporins, piperacillin-tazobactam), whereas for others a lag of 1 year (ciprofloxacin, amoxicillin-clavulanate, aminoglycosides) or even 2 years (consumption of macrolides-lincosamides on decreased susceptibility to cloxacillin in S. aureus) was observed. CONCLUSIONS: The significant increment of antimicrobial use observed for many agents seems to produce a negative effect on the sensitivity to other antimicrobials, which is immediate in some relationships, but shows a time lag of 1 or 2 years in others.