Table 6 Components of a risk of bias assessment that should be reported in the methods section of a systematic review protocol

1.The tool(s) that will be employed (whether existing, modified, or newly developed by the review team) for assessing each type of study design and the classes of bias and confounding that are covered

A precise description of each risk of bias tool is necessary to enable the overall systematic review methods to be understood, critiqued, and repeated by other researchers, as well as to ensure that the a priori planned approach is followed, to reduce the risk of the review team themselves introducing bias

2.Why and how each tool was selected or developed

Evidence is required that the review team have considered the availability and fitness-for- purpose of existing risk of bias approaches to avoid inappropriate, suboptimal or superseded methods being used

3.The rationale for how the tool assesses the risk of bias

Evidence is required that, as far as can reasonably be inferred, each tool employed has construct validity—that is, it measures the risks of the types of bias that it claims to (rather than measuring factors unrelated to systematic error)

4.How the different classes of bias covered by the tool are defined

Naming and interpretation of bias types is not always consistent in the scientific literature, so an explicit definition of the types of bias to be covered should be provided to avoid misinterpretation

5.Which outcomes the risk of bias assessment will be applied to and whether the process will differ across outcomes

Different outcomes can be subject to different risks of bias or confounders, so it may not always be appropriate to use the same approach across all outcomes

6.The signalling questions that will be used to establish how the review team will classify the risks of bias in a study (if an existing tool is used, this may be cited for such details rather than repeating them)

Signalling questions help to ensure transparency in how the risks of bias in the included studies are identified and classified

7.A copy of the draft template(s) for recording risk of bias assessments (e.g. in an appendix), including the instructions that will be provided to the review team on how to use the tool

Risk of bias judgements are inherently subjective, and it is therefore necessary to provide as complete information as possible on how judgements will be made so the rationale for decisions is clear

8.The number of reviewers who will conduct each assessment and how any disagreements in judgements will be resolved

Single-reviewer assessment of risks of bias could be influenced by implicit bias (i.e. the reviewer’s perspective), so demonstration that the process is not dependent on a single reviewer is required

9.How the risk of bias classification categories used in the tool(s) will be presented and interpreted to inform the data synthesis, both for narrative synthesis and for meta-analysis where applicable (e.g. sensitivity analysis or subgroup analysis)

A priori specification for how risk of bias assessment will inform data synthesis is required to prevent selective inclusion or exclusion of studies in the analysis