Summary
The authors examined how 66 systematic reviews handled mask studies and found that generic risk-of-bias tools were often used crudely. They argue that excluding observational evidence or pooling mixed designs produced bland summaries that hid real differences, and they call for better-trained review teams and new methods.
Key points
- The authors examined how 66 systematic reviews handled studies of mask efficacy, whose conclusions vary widely.
- Risk-of-bias tools gave unreliable assessments when review teams lacked methods expertise or topic knowledge, especially with large, mixed datasets.
- Excluding observational evidence, ignoring the direction of biases and pooling studies of different designs hid important differences and produced bland summaries.
- The authors call for better training and oversight of reviewers and new review methods for complex evidence.
- Design: methodological analysis of published reviews; it does not estimate how well masks work.
- Funding: UK Research and Innovation.