A review in JAMA shows that journal article findings that a treatment worked were contradicted 16 percent of the time by later studies. And another 16 percent of the time, studies found weaker results than earlier suggested. So nearly a third of original published results did not hold up to further scrutiny.
Dartmouth and VA researchers Steve Woloshin and Lisa Schwartz have warned about news coverage that is "Too Much Too Soon" in coverage of presentations at scientific meetings.
For those journalists who live off weekly journals for their stories, and who fail to follow up on subsequent findings, the same "too much too soon" message is valid.
I was, frankly, surprised that 64% of studies had findings that were upheld!