Why caution should be applied when interpreting and promoting findings from Mendelian randomisation studies.