and benefits of antimicrobial use). This is because of the possible irreversibility of AMR and the potentially severe harm which could be imposed as a result. Yet, given the increasing importance of evidence-based medicine, strategies that have been evaluated using experimental methods and well conducted economic evaluations, may be prioritised above these policies, which are much more difficult to evaluate. This is a danger that should be avoided both by awareness among policy makers of the relative challenges associated with evaluating different types of policy, and by awareness among the research community of the importance of evaluating policies which may potentially be more important, even if the rigour with which they can be evaluated is less than for the potentially less important policies.

Overall, there appears to be no definitive evidence (cost and/or effectiveness) which suggests that one specific control measure (or indeed a combination of measures) is particularly successful in containing the spread of AMR. Although it would seem that surveillance is a basic pre-requisite to tackling AMR, in the absence of evidence it is difficult to go further in making recommendations, or in suggesting priorities for research among those interventions assessed here. Readers are referred to the WHO Global Strategy for the Containment of Antimicrobial Resistance as presenting the most current and complete “best advice” on interventions to tackle AMR, how these should be implemented, and research priorities.

In terms of developing a model to assess the cost-effectiveness of strategies to tackle AMR, the appropriate and desirable model will need to satisfy four broad criteria of feasibility, sensitivity, relevance and flexibility. Considering these criteria, and this review of modelling as applied to AMR, two broad options are outlined:

  1. A “macro-model” approach which attempts to integrate factors within a broad based model aiming to assess strategies on a more “global” level; and

  2. A “suite” of micro sub-models, each “embedded” within a given set of primary parameters, such as country, disease and level of intervention (e.g., hospital or community), which determine which of the “suite” of sub-models is most appropriate for that context.

A definitive recommendation concerning which form of modelling to pursue is not possible at present, as it depends upon both feasibility and relevance to the question and context concerned. However, it is clear that there needs to be further research in to the modelling of AMR. Although such a model will require substantial investment of time and resources, the potential benefits of such a model, if accurately specified and incorporating quality data, could be vast in terms of the potential health benefit to current and future generations.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement