Core principles underpinning both methods frameworks for Cochrane intervention reviews

Economic evidence should not be presented alone

In general, BECs and IFSREEs should not be conducted as stand-alone exercises. It is essential that authors place the body of economic evidence that has been assembled for a BEC or an IFSREE into the context of reliable evidence for health and other outcomes assembled in the main review of intervention effectiveness. This can be achieved by comparing effect sizes used as data inputs to identified, eligible economic evaluations with corresponding effect sizes (outcomes) derived from studies of effects included in the main review of intervention effectiveness.

Failure to include this stage of research in a BEC or IFSREE can lead to a biased summary of the assembled bodies of evidence. This is mainly because:

• effect sizes used as data inputs to eligible trial- and/or model-based economic evaluations can (at best) only be derived from a subset of studies of effects included in the main review of intervention effectiveness (i.e. those that were available at some earlier time point, when the economic evaluation was conducted); and

• many economic evaluations are conducted alongside, or based upon, a single study of effects (e.g. a randomised controlled trial).

Economic evaluations may also be susceptible to a specific source of publication bias. Audits of some clinical areas have shown that clinical effect sizes in randomised controlled trials published with a concurrent or ‘piggyback’ economic evaluation (i.e. trial-based economic evaluations) are systematically larger than those in randomised controlled trials without a concurrent or ‘piggyback’ economic evaluation (e.g. see Gilbody 2007).

The source of this potential bias may reflect challenges faced by author teams when attempting to publish planned economic evaluations conducted alongside ‘inconclusive’ trials. It may also reflect decisions made whilst planning trials. For example, a decision not to conduct a concurrent or ‘piggyback’ economic evaluation may be made because it is felt implausible that an effective intervention could be anything other than cost-saving. However, such reasoning may not be made explicit in published trial protocols or final study reports. Both of these issues have the potential to compound known issues of reporting biases in randomised controlled trials, which may also be spread into both trial-based and model-based economic evaluations.

Economics components of reviews should avoid making categorical or general inferential statements or claims concerning costs or cost-effectiveness

BECs and IFSREEs should be reported using careful language that reflects their aims, limitations and target audience. In particular, when summarising the principal findings of identified eligible economic evaluations, Cochrane review authors should avoid making any categorical or general inferential statements or claims concerning the costs or cost-effectiveness of the intervention(s) under investigation (versus comparators).

This is because Cochrane intervention reviews are intended for an international audience of end-users and, as such, BECs and IFSREEs cannot reflect any implicit assumptions about the decision-making contexts and settings in which their findings will be used (such as the availability of resources, the structure of the health system, or the organisation of care pathways), nor draw any inferences about whether or not an intervention is likely to be cost-effective in any or all of those settings. In contrast, economic evaluation is explicitly a decision-informing method of analysis; each economic evaluation that is included in a BEC or an IFSREE was conducted to inform a specific decision, at a specific time, in a specific setting and context. End-users of BECs and IFSREEs in Cochrane reviews will therefore invariably need to consider the extent to which the methods and results of identified eligible economic evaluations may be applicable, or transferable, to their own setting.

For these reasons, the primary aim of a BEC and an IFSREE should be to explain how interventions (versus comparators) impacted on resource use, costs, health outcomes and cost-effectiveness when implemented at specific times in specific settings and contexts (i.e. a focus on ‘what happened?’ (Petticrew 2015)). They should also explain what factors are likely to drive variation in estimates of economic and health outcomes between studies and settings. Adopting this aim will help to ensure that the addition of a BEC or an IFSREE to a Cochrane review can help its international audience to better understand key economic trade-offs between interventions that could be used in practice, in their own setting.

Economics components of reviews should aim to be as useful as possible for de novo model-based economic evaluations

The secondary aim of a BEC and an IFSREE should be to present health economic outcome data parallel to Cochrane intervention reviews in formats that facilitate the re-use of these parameters as inputs to the subsequent, or parallel, development of new model-based economic evaluations. Methods guidance on conducting and reporting BECs and IFSREEs is designed to facilitate this key secondary aim.