The meeting decided to recommend a change in the emphasis of an Overview. OoRs should be re-defined as reviews that integrate or synthesize (rather than summarize) evidence from existing systematic reviews, and should address a well-defined clinical question. This would involve a change to Handbook guidance, which currently allows OoR to be driven by the question ‘What is in the Library?’, leading to a simple summary of existing reviews without an attempt to integrate their findings. One particular approach to integrating evidence is the use of statistical synthesis across evidence in the included reviews. Authors of OoRs should be encouraged to consider the implementation of indirect comparisons and MTM. Nevertheless, some OoRs will involve no statistical analysis and findings will be presented in a narrative format.
The basic unit of the search strategy for an OoR is the review (Cochrane Reviews and optionally non-Cochrane systematic reviews). As a further evolution in guidance for Overviews, the meeting decided that this may not be sufficient to provide a suitable answer to the clinical question. Broadening the search in an OoR to include individual studies may be appropriate in some cases.
For some OoRs it will be possible to extract all of the required data from the included reviews. Estimated intervention effects from pair-wise meta-analyses may be used to generate indirect comparisons (possibly combined additionally with direct comparisons). Study-specific data, as reported in the individual reviews, may be copied and used to perform MTM. However, in many cases OoR authors will find that there is insufficient information in the published reviews (for example, if the reviews were undertaken before the Cochrane risk of bias tool was in widespread use). An OoR may need to examine the original reports from individual studies and collect data not available in the existing reviews in order to perform an adequate synthesis.
In some cases, the authors of an OoR may decide that the question of interest cannot be adequately answered using the methods designed for OoRs, and decide that an IR is needed. In these cases, the preparation of the OoR will have served an important ‘scoping’ function, and will allow detailed exploration of a number of important issues for the IR, such as the specific framing of the question, considerations of interventions and outcomes to be included and identification of a number of key trials in the area. Strategies for moving from an OoR to an IR are discussed in Section 6.
With the change in emphasis of OoR, the meeting considered future possibilities for providing a ‘friendly front end’ to the CDSR, which is part of the role that OoR have played until now. It was suggested that simple summaries of the contents of existing Cochrane Reviews are best provided in the form of editorials and annotated collections of reviews, as currently prepared by the Cochrane Editorial Unit.
Other sections of this report:
Addressing multiple interventions in Cochrane Intervention Reviews
Clarification of the distinction between Intervention Reviews and Overviews of Reviews
Implications of Overviews for authors and editors of Intervention Reviews
A sequential approach for undertaking reviews that compare multiple-interventions
Role of the Cochrane Comparing Multiple Interventions Methods Group