I'm always a little hesitant to embrace met-analysis "results."
The farther you get from the actual research, the less faith you can have in your findings.
Summaries of meta-analyses even more so. Once this is published in a journal, peer reviewed, original articles are listed, and the students explain how they made the jump from "early cancer found by mammogram in women with a family history" to "cancer caused by mammogram;" then I'll be glad to read it. Tumors don't wear a little banner saying "I was caused by ---"
Original research is grindingly slow and frustrating work. (2 years down and results presented at an international conference, now the long process of getting published) But at least I know my data, and every single one of the subjects in my study, and exactly what I did, and how I came to my conclusions. I stand behind my results. If instead I had done a re-structuring and re-filtering and re-assessing book report of the published work of 5 other groups... well, I couldn't in good faith stand behind that.
The headline could just as easily be: "mammography more than doubles rates of early cancer detection in high-risk women." As time passes and technology advances it is inevitable that we begin finding cancers earlier in one generation than it was found in the previous generation.
Detection method does not imply cause. Kind of like if my Grandma never went to a dentist until she was 50, and they found a bad cavity and the tooth had to be pulled. She tells me I'm at high risk of cavities, so I go to the dentist every year. He finds a small cavity when I'm 30. Do I turn around and say that seeing the dentist early and regularly caused my cavity? Nope. But I could do a meta-analysis that would "prove" exactly that. And should I stop flossing my teeth and looking at my teeth until I'm 50, and then start going to the dentist? Nope.
Last edited by KnottedYet; 12-02-2009 at 05:18 AM.
"If Americans want to live the American Dream, they should go to Denmark." - Richard Wilkinson