3 ms·
The answer to those questions is not simple; this is reviewing a bunch of studies and outcomes, so the answer is, it depends. As usual, there is no substitute f
by gwern 12y ago
The answer to those questions is not simple; this is reviewing a bunch of studies and outcomes, so the answer is, it depends. As usual, there is no substitute for looking at the fulltext, eg http://annonc.oxfordjournals.org/content/early/2014/07/30/annonc.mdu225/T1.expansion.html http://annonc.oxfordjournals.org/content/early/2014/07/30/an... (Looks like it omits the metric of most interest to me personally, RCT results for all-cause mortality, where I believe the cumulative n is ~10-20k.)