Table 5 Examples of study-level risk of bias classifications
Risk of bias classification for each class of bias | Possible study-level risk of bias classification for a given outcome of interest |
|---|---|
All classes of bias are judged to have low risk of bias, definitely low risk of bias, or probably low risk of bias for the outcome of interest | Low risk |
At least one class of bias is judged to have high risk of bias, definitely high risk of bias, or probably high risk of bias for the outcome of interest | High risk |
Some classes of bias are judged to have low risk, others unclear risk, but no classes are judged to have high risk for the outcome of interest | Unclear risk (to avoid an unclear judgement for the overall study it is preferable for each domain to reach a probably low risk or probably high risk judgement instead of an “unclear” judgement, where possible) |
No information available for any classes of bias for the outcome of interest | Unclear risk (or no information) |
Classes of bias are judged as having combinations of “moderate”, “serious” or “critical” risks of bias (or other terminology) for the outcome of interest | Summarising judgements other than high/low/unclear may not be intuitively straightforward. A clear rationale should be provided, based on logic (i.e. the criteria should not be arbitrary). See recent Cochrane tools [58, 59, 81] for examples |