Risk of bias domains

We’ve already talked about the six main types of bias, but how do you find those types of bias in a study?

To do that, you use risk of bias “domains” or “methodological flags.” Risk of bias domains describe different aspects of how a study was conducted or reported. They help you look for warning signs that a particular type of bias might exist.

Here are the formal titles for the the most common risk of bias domains and the fundamental questions and types of bias that are relevant to each domain.

Domain Fundamental question Relevant type of bias New term for this bias in latest Cochrane risk of bias tool
Random sequence generation
Did the study researchers use a truly random method to assign participants into groups? Selection bias Bias arising from the randomization process
Allocation Concealment
Did the study researchers hide the sequence for group assignments from the person who is doing the assigning? Selection bias Bias arising from the randomization process
Masking/Blinding (of participants and caregivers)
Did the study researchers hide who is receiving the treatment and who is receiving the placebo/alternative treatment from those conducting the study, such as the participants, caregivers, and outcome assessors? Performance bias Bias due to deviations from intended interventions
Masking/Blinding (of outcome assessors)
Did the study researchers hide who is receiving the treatment and who is receiving the placebo/alternative treatment from those assessing outcomes? Detection bias Bias in measurement of the outcomes
Missing participant data
How much data (or information) related to participant outcomes are missing? Does the extent of missing data differ between the groups? Attrition bias Bias due to missing outcome data
Incomplete outcome reporting
Are all the outcomes that the study researchers were interested in reported adequately? Reporting bias Bias in selection of the reported result

We’ll talk more about these domains and how they relate to specific types of bias on the next few pages, starting with random sequence generation.