Incorrect blinding assessments are common in meta-analyses published in high impact journals.