At the stage of writing the protocol for the review, compile a list of potential confounding factors and justify the choice;
At the stage of writing the protocol for the review, decide how the risk of bias in primary studies will be assessed, including the extent of control for confounding;
For NRS conducted entirely prospectively, apply the methods that the Collaboration recommends for randomized trials;
There is no single recommended instrument, so review authors are likely to need include supplementary risk of bias instruments or items;
Issues such as confounding cannot easily be dealt within the format of the new risk of bias tool and require creation of additional tables for reporting assessments;
Collecting some factual information (for example, the confounders considered and what researchers did about confounding) is useful since such information illustrates the extent of heterogeneity between studies;
Review authors who choose to include case-control studies in a Cochrane review should ensure that they are familiar with common pitfalls that can affect such studies and that they assess their susceptibility to bias using an instrument designed for this purpose;
Review authors may decide that collecting great detail about the risk of confounding and other biases is not warranted. However, if this approach is taken, review authors must acknowledge the potential extent of the heterogeneity between studies with respect to potential residual confounding and other biases and demonstrate that they have considered this source of heterogeneity in their interpretation of the findings of the primary NRS reviewed.