Dialling back 'impact' claims: researchers should not be compelled to make policy claims based on single studies.