Cochrane is an international network of health professionals, patients and others who summarize data from research studies in systematic reviews to help users to make informed decisions about disease treatment, prevention and screening (go.nature.com/z5w5ah). The model we use to accomplish this in the Cochrane Eyes and Vision (CEV) US group can also expose errors of the type noted by David Allison and colleagues (Nature 530, 27–29; 2016).

In this group, clinician authors partner with a methodology 'hub' of experts who check data and apply validated methods in conducting the systematic review. These experts are trained to identify potential biases and statistical errors in reviews, and to expose what seem to be deliberate reporting biases or manipulation of outcomes.

The experts can identify errors in primary studies and draw attention to them, but cannot correct existing errors. Content experts need to collaborate with methodologists to eliminate mistakes at the outset, by designing and doing studies at a low risk of bias. Journal editors should insist that authors adhere to reporting standards. And the public must demand that health-care decisions are underpinned by conclusive scientific evidence.