To measure the impact of exposure to patients using carbapenem on the acquisition of carbapenem-resistant gram-negative bacilli (CR-GNB) among patients not using carbapenems.

An ecological study and a cohort study.

Two medical surgical intensive care units (ICUs) in inner Brazil.

Patients admitted to 2 ICUs from 2013 through 2018 to whom carbapenem was not prescribed.

In the ecologic study, the monthly use of carbapenems (days of therapy [DOT] per 1,000 patient days) was tested for linear correlation with the 2-month moving average of incidence CR-GNB among patients to whom carbapenem was not prescribed. In the cohort study, those patients were addressed individually for risk factors (demographics, invasive interventions, use of antimicrobials) for acquisition of CR-GNB, including time at risk and the “carbapenem pressure,” described as the aggregate DOT among other ICU patients during time at risk. The analysis was performed in univariate and multivariable Poisson regression models.

The linear regression model revealed an association of total carbapenem use and incidence of CR-GNB (coefficient, 0.04; 95% confidence interval [CI], 0.02–0.06; P = .001). In the cohort model, the adjusted rate ratio (RR) for carbapenem DOT was 1.009 (95% CI, 1.001–1.018; P = .03). Other significant risk factors were mechanical ventilation and the previous use of ceftazidime (with or without avibactam).

Every additional DOT of total carbapenem use increased the risk of CR-GNB acquisition by patients not using carbapenems by nearly 1%. We found evidence for a population (“herd effect”-like) impact of antimicrobial use in the ICUs.