"Epidemiologists call this the problem of 'surrogate endpoints,'"
and I'm very glad the submitted article gets into this problem of many clinical trials. Setting up an observable endpoint, rather than a "hard endpoint" (such as preventing death or disease) that really matters, robs many medical studies of real-world meaninfulness, even if they have statistical significance and reasonably large effect sizes. As usual, my suggestion is to check the article by Peter Norvig on how to read research articles
Jerry Avorn's proposed a two-phased drug evaluation process would be a great improvement. By shortening the first phase and lengthening the second we'd get more drugs and more drug discovery and in the long term better trial results.
and I'm very glad the submitted article gets into this problem of many clinical trials. Setting up an observable endpoint, rather than a "hard endpoint" (such as preventing death or disease) that really matters, robs many medical studies of real-world meaninfulness, even if they have statistical significance and reasonably large effect sizes. As usual, my suggestion is to check the article by Peter Norvig on how to read research articles
http://norvig.com/experiment-design.html
for more about this important subject.