Using high quality clinical databases to complement the results of randomised controlled trials: the case of recombinant human activated protein C