To the Editor—One of the main challenges that face antimicrobial stewardship programs (ASPs) is proving their effect on the incidence of multidrug-resistant (MDR) microorganisms and the related mortality.Reference Davey, Marwick and Scott ¹ It can be especially difficult in centers with a limited incidence of bacterial resistance, where large samples may be required to demonstrate significant changes.Reference Shardell, Harris, El-Kamary, Furuno, Miller and Perencevich ² This might be a reason to assess the ecological impact of interventions through the incidence of all MDR isolations, including colonizations,Reference Davey, Marwick and Scott ¹ but such assessments may not reflect their effect on the incidence of clinical infections.

An educational ASP implemented in our hospital in 2011 proved to be effective in reducing the inappropriate use of antimicrobials as well as global antibiotic consumption in the center.Reference Cisneros, Neth and Gil-Navarro ³ In a recent publication,Reference Molina, Peñalva and Gil-Navarro ⁴ we were able to show that the sustained effect of our program on antibiotic use produced a reduction of the incidence and mortality of nosocomial bloodstream infections (BSIs) by MDR bacteria and Candida spp. To solve the aforementioned drawbacks, we employed an ecologic interrupted time-series study and pooled as a sole indicator the incidence of all MDR bacteria and Candida spp producing BSIs, considering that they all share as a preferential risk factor the previous exposure to antibiotics. Additionally, we measured the changes in crude death rate (deaths per 1,000 occupied bed days) of these infections.

We found several important advantages in these 2 novel indicators: (1) pooling several mechanisms of resistance in the same indicator improved the power of the sample to show the effects of the reduction of antibiotic pressure in the center; (2) assessing infections instead of colonization provided information on the clinical benefits of the intervention; and (3) measuring the absolute reduction in mortality using the crude death rate let us show the burden of mortality avoided by the program. This metric is relevant because reductions in the rate of mortality of infections can be difficult to prove when they depend on multiple factors that require complex patient-level analyses. But if antibiotic pressure (and subsequently bacterial resistance) is reduced in a center, a reduction in mortality in absolute terms should be expected, as shown in our article.Reference Molina, Peñalva and Gil-Navarro ⁴

Researchers interested in this approach should be aware of certain considerations. The pooled analysis of MDR does not replace the surveillance of specific mechanisms of resistance; otherwise, occasional outbreaks could pass unnoticed. It also requires an analysis of the possible influence on results of infection control programs coexisting with the ASP. In this case, if different interventions sequentially occur during the study period, a joinpoint regression analysis may allow researchers to establish mathematically (ie, not subjectively) when the inflection point occurs.Reference Kim, Fay, Feuer and Midthune ⁵ Regarding the assessment of mortality, precautions should also be adopted to prevent ecologic bias. For instance, the total number of cultures and incidence of susceptible bacteria should be measured because a decrease in the incidence of MDR-produced deaths could also be explained by a reduction in diagnostic tests or by general improvements in the prevention of infections. Antimicrobial stewardship programs aiming to reduce antibiotic consumption should also monitor the mortality produced by susceptible bacteria to ensure the safety of the intervention.

In conclusion, we propose 2 novel indicators that, used properly, could enhance the ability of ASP to prove their clinical and ecological impact in a feasible and objective way.