Study Design
Evaluating indirect antimicrobial effects via a two-stage randomized trial: a simulation of oral vancomycin vs. metronidazole for C. difficile infection Juan Gago* Juan Gago Christopher Boyer Marc Lipsitch
Antimicrobial treatments are often approved on the basis of efficacy results from individually randomized trials. However, these trials are unable to capture the indirect or “spillover” effects of the treatment of one individual on the outcomes of others. These indirect effects are particularly important in the context of antimicrobial resistance as the treatment of an index patient can affect others’ exposure to resistant pathogens through a variety of mechanisms, even when the direct effects on de novo resistance are rare. We simulate a two-stage randomized trial, informed by a prior study of oral vancomycin for C. difficile infection in the Veterans Affair (VA) Health System, to illustrate how this design can better capture both direct and indirect effects of antimicrobial treatment strategies on vancomycin-resistant Enterococci (VRE) infections.
In the first stage, hospitals (clusters) were assigned a treatment strategy that varied the proportion of vancomycin vs. metronidazole use. In the second stage, individual patients within each cluster were randomized to one of the treatments approved by their hospital’s assigned strategy. By comparing clusters with different proportions of vancomycin use over multiple time points (six months, one year, and two years), we estimated how the varying distribution of therapies influences overall and indirect antimicrobial effects.
This design extends beyond the classic trial approach, which can overlook how antimicrobial treatment strategies might affect the broader hospital environment. While our simulations suggest that direct effects of oral vancomycin on VRE risk may remain minimal, the two-stage design highlights subtle differences in population-level impacts that could emerge over time . By explicitly capturing cluster-level variation, two-stage randomization enables a more comprehensive evaluation of antimicrobial strategies, informing decisions that optimize both individual and population outcomes.