Drawing on a Cochrane systematic review, this paper examines the relatively limited range of outcomes measured in published evaluations of antimicrobial stewardship interventions (ASIs) in hospitals. We describe a structured framework for considering the range of consequences that ASIs can have, in terms of their desirability and the extent to which they were expected when planning an ASI: expected, desirable consequences (intervention goals); expected, undesirable consequences (intervention trade-offs); unexpected, undesirable consequences (unpleasant surprises); and unexpected, desirable consequences (pleasant surprises). Of 49 randomized controlled trials identified by the Cochrane review, 28 (57%) pre-specified increased length of stay and/or mortality as potential trade-offs of ASI, with measurement intended to provide reassurance about safety. In actuality, some studies found unexpected decreases in length of stay (a pleasant surprise). In contrast, only 11 (10%) of 110 interrupted time series studies included any information about unintended consequences, with 10 examining unexpected, undesirable outcomes (unpleasant surprises) using case-control, qualitative or cohort designs. Overall, a large proportion of the ASIs reported in the literature only assess impact on their targeted process goals-antimicrobial prescribing-with limited examination of other potential outcomes, including microbial and clinical outcomes. Achieving a balanced accounting of the impact of an ASI requires careful consideration of expected undesirable effects (potential trade-offs) from the outset, and more consideration of unexpected effects after implementation (both pleasant and unpleasant surprises, although the latter will often be more important). The proposed framework supports the systematic consideration of all types of consequences of improvement before and after implementation.
© The Author 2017. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please email: email@example.com.