Good epidemiological practice: a narrative review of appropriate scientific methods to evaluate the impact of antimicrobial stewardship interventions.

AIMS: In this narrative review, we provide a framework for assessing the quality of evidence provided by studies investigating antimicrobial stewardship (AMS) interventions, and inform the design and planning stage for future AMS evaluation studies to determine the best strategies to keep antimicrobial resistance at bay. SOURCES: Cochrane/Pubmed. CONTENT: As AMS is mostly applied in a complex, real-world setting, bias and random time effects can jeopardize the validity of causal inference. The most important risks include simultaneously implemented infection prevention strategies and regression to the mean. Inclusion of homogeneous intervention and control arms, through randomization of the intervention, can limit these risks. However, contamination can play an important role for AMS; therefore, randomization at cluster-level, instead of randomization at individual-level, is recommended. It can be challenging to identify enough representative clusters, and implementation of a cluster-RCT (cRCT) can be costly. Controlled interrupted time series (ITS) design has a high validity as well, and is relatively straightforward to implement, although time-varying confounding should be considered. Independent of the study design, it is crucial to include multiple process, clinical outcome, microbiological and financial measures, to be able to detect possible, unintended consequences. IMPLICATIONS: Future studies assessing the impact of new AMS strategies should produce compelling evidence by opting for cRCTs, or ITS including a control arm. Furthermore, a holistic view of intended and unintended consequences should be reported, and a detailed process evaluation should be provided to adequately inform implementation of successful AMS strategies to battle the rising burden of AMR.