Effect of antibiotic stewardship on the incidence of infection and colonisation with antibiotic-resistant bacteria and Clostridium difficile infection: a systematic review and meta-analysis
Hospital antimicrobial stewardship: the way forward
Charis A Marwick, Bruce Guthrie, Peter G Davey
We read David Baur and colleagues' systematic review1 of the effect of antimicrobial stewardship programmes on antibiotic resistance and Clostridium difficile infection, and the accompanying Comment,2 with great interest.
With investment in antibiotic stewardship at an all-time high and antimicrobial resistance declared a global public health threat, the large effects reported by Baur and coworkers, such as the 51% (95% CI 0·35–0·68) reduction in infections and colonisation with multidrug-resistant Gram-negative bacteria, are very appealing, but must be interpreted with caution.
No eligible randomised controlled trials were available for inclusion in the meta-analysis; thus 17 (53%) of 32 studies contributing to any meta-analysis, and six of 12 contributing to the multidrug-resistant Gram-negative bacteria meta-analysis, were uncontrolled before-and-after (UBA) studies.1UBA studies are inherently biased and are not considered to provide reliable enough evidence for inclusion in Cochrane Effective Practice and Organisation of Care systematic reviews.3 Cochrane also advises against combining different study designs in meta-analyses because of systematic differences in the influence of various biases.3
Only six of 32 studies included in any meta-analysis by Baur and colleagues (two of 12 for multidrug-resistant Gram-negative bacteria) met study design criteria for inclusion in a Cochrane review.3 All six eligible studies were interrupted time series analyses, five of which were included in our Cochrane systematic review of hospital stewardship interventions.4 Two reported microbiological outcome data as UBA but included interrupted time series prescribing data1 and thus were excluded from microbiological meta-analysis in our review.4 Two studies assessed multiple Gram-negative resistance outcomes, with one reporting only statistically significant results and the other reporting an overall increase in third generation cephalosporin resistant Escherichia coli, but a single positive outcome was selected from each study for inclusion in the meta-analysis by Baur and coworkers.1Observational studies are particularly prone to selection and publication biases. Furthermore, one interrupted time series study and several UBA studies report changes after the implementation of interventions in response to disease outbreaks, thus introducing additional bias.5 On the basis of studies of higher methodological quality, our Cochrane review concluded that the observed effects of stewardship on antibiotic resistance are inconsistent.4
Although Baur and colleagues' study aims are laudable and the results as reported would be worthy of celebration, the conclusions are based on synthesis of unreliable evidence. We welcome the authors' plea for improvements in design and reporting of stewardship interventions to strengthen the evidence base.
CAM and PGD are authors of a Cochrane systematic review. BG declares no competing interests.
Hospital antimicrobial stewardship: the way forward
Céline Pulcini, Peter Collignon
In resource-limited environments, deciding where and how investments should be made to reduce bacterial resistance is challenging.
In their recent systematic review, David Baur and colleagues1 showed that antibiotic stewardship programmes were more effective with infection control measures—especially hand-hygiene interventions —than when implemented alone. Both strategies are essential and must be pursued simultaneously, in accordance with the WHO global action plan.2 Antibiotic use (whether appropriate or not) is the catalyst that initiates and propagates bacterial resistance in the microbiota at the individual level. Resistant microorganisms and genes that confer resistance can then be transmitted to other individuals or the environment.3 The spread of resistant bacteria changes the problem from an individual to a collective one.
Most people would agree that antibiotic stewardship and infection control teams should collaborate closely. This view was highlighted in the 2017 European guidelines4 on antimicrobial resistance, which recommend “coordination and collaboration between antimicrobial stewardship programmes and infection prevention and control programmes”.
However, we feel that these programmes need to go one step further, to radically change the current organisation of care and enable these teams to successfully work together, in all settings (ie, hospitals, nursing homes, and primary care), within the same department. Advantages and disadvantages of this approach are shown in the appendix. Such an approach should involve the use of a consistent department name across all settings because semantics are important for the communication of clear messages and public engagement.5 Names that refer to health-care-associated infections should be avoided, because such terms might suggest that resistance only exists in health-care facilities (mainly hospitals) and only for health-care-associated infections. So-called “superbug fighters” or “infection management” department are names that might be discussed. We believe such a holistic approach deserves further investigation.
We declare no competing interests.
antibiotic stewardship programmes were more effective with infection control measures—especially hand-hygiene interventions —than when implemented alone.
Hospital antimicrobial stewardship: the way forward – Authors' reply
Evelina Tacconelli, Beryl Primrose Gladstone
We thank Charis Marwick and colleagues and Céline Pulcini and Peter Collignon for their comments on our study.1 As stated in the Discussion section, we completely agree that the possibility of uncontrolled studies having observed effects as a result of an underlying secular trend cannot be ignored, and we also recommended the use of controlled interventional designs for future studies. However, we would like to underline that, although some guidance is available on how to design interventions on antimicrobial stewardship,2 specific recommendations about how to select study designs and overcome major bias in assessing antibiotic stewardship programmes (eg, definitions of outcomes or analytical methods) are missing or under development. In the field of public health and health promotion, even the Cochrane handbook suggests that non-randomised studies, including before-and-after studies, might represent the best available evidence provided care is taken in their analysis.3 Studying the effects of in-depth complexities or components of the intervention in real-life settings that are valuable to patients, managers, and policy makers could be achieved by inclusion of non-randomised study designs.4 In our systematic review, to address the biases introduced by observational studies, we did a sensitivity analysis by excluding low-quality studies, which provided estimates supporting our results.
The papers studying cephalosporin resistance were analysed in our meta-analysis, although these data were not presented in the published Article. The effect of antimicrobial stewardship on the rate of infections due to cephalosporin-resistant Gram-negative bacteria (incidence rate ratio 1·06, 95% CI 0·74–1·52) was included in the estimation of the overall effect sizes for Gram-negative, resistant bacteria. As for the concern about outbreak settings, we made the following considerations to reduce the associated biases: incidence rates were based on 1 year or more of follow-up, thus tempering the effects of seasonality, varying length of stays or bed occupancies would be dealt with in the calculation of the incidence rates, and changes in infection control measures were one of the covariates studied.
Although when stratified by study design we found inconsistent effects among interrupted time series studies similar to those reported in the meta-analysis by Davey and colleagues,5 we believe that the added value of a large pool of moderate-level evidence cannot be overlooked and disregarded, particularly if the results could affect the burden of antimicrobial resistance and improve patients' quality of care. This does not mean that we undervalue the limitations of meta-analyses of non-randomised studies. Our results contribute to the understanding of the stewardship benefits and push forward the need to define how to best design antimicrobial stewardship assessments.
We declare no competing interests.