False evidence in medical research can endanger lives, so countermeasures must be stepped up — particularly in light of recent cases (see go.nature.com/9ivjnm).

Improving scrutiny and verifying source data would be prohibitively costly and time-consuming, even if it were legally feasible. Instead, journals should demand proof before publication that research has been approved by a recognized body, such as an independent ethics committee or institutional review board, so that fraudulent claims can be identified (see, for example, Nature 477, 384; 2011). This approval is legally required in most countries and has now become an international standard (see go.nature.com/krrr6g), so is easy to obtain.

Authors and their institutions should inform the journal promptly should their results be later invalidated, and permit inspection or auditing of reports if necessary (S. Kleinert and R. Horton Lancet 372, 789–790; 2008). This would speed publication of the formal retraction to correct the scientific record.