Validity of the impact factor of journals as a measure of randomised controlled trial quality