Do you even know what “significantly impacted” means in the first quote? It simply means “there appears to be a difference”. No more, no less. It is not equivalent to common parlance expressions such as “this is incredibly significant” which you used above.
Indeed there does appear to be a difference. One that has even been quantified no less. :eek: :eek: :eek:
That’s really inapposite you know, as most of the comments in this thread. Why is it that so many people here try to pretend to know more than they actually do?
[1]This is getting tiresome. Maybe you fancy yourself some sort of expert but if your claim to expertise in the field is being a practicing physician and that’s it, I’m sorry but I’m just not impressed. And certainly your logic or lack of same has done nothing for me either. If you want to start comparing credentials, I’m game for that.
[2]This simply absurd and undermines any credibility you might have had. Since when is the opinion of any scientist on the weight to be given to a result not worth considering? If they value their reputations they will be circumspect in their claims and representations. I guess you haven’t published much huh?
[3]blah, blah, blah. Don’t try to make this about stupid ignorant me. Nice try but that’s not the issue here. You’ll have to do much better than that.
It’s perfectly appropriate in a discussion of why one should be cautious when reacting to unexpected results in a subgroup of a clinical trial.
BTW, you asked earlier about “outside reviewers” who expressed doubts about the study. Here’s one which was published in March 2013 in the same issue of JAMA as the TACT results:
http://jama.jamanetwork.com/article.aspx?articleid=1672219
Irrelevant since the revised study didn’t come out until later. This has all been covered already.
edit: unless you can point out some flaw that would explain why only diabetics saw a 40-50% drop in cardiovascular events. Then I would love to hear your thoughts.
You keep saying 40-50%. Would you mind actually reading the study to see what the confidence intervals in that subgroup are? Hint: It’s not 40-50%.
Hint, why don’t you tell me. I haven’t studied statistics in a while and I don’t feel like wading through a journal article. If you don’t mind sharing your wisdom that is.
http://jama.jamanetwork.com/Mobile/article.aspx?articleid=1672238
It’s in this graph: http://jama.jamanetwork.com/data/Journals/JAMA/926663/joc130024f3.png
That’s nice but that doesn’t mean anything to me as I’m sure you probably realized it would not.
Oooooo. Ya got me. Score.
Now that we’ve got that out of the way, do you have anything meaningful to contribute?
I’ve contributed a link to a lengthy article pointing out many potentially serious flaws in this study. I’m unsure whether you’ve read it or not.
IOW you have no idea what the significance is either. That’s what I thought. Thanks. 
They found a difference of 17-55% in the diabetic group, if that’s what you meant by significance.
It’s broken out by type of cardiovascular event and that’s covered in the OP. Do try to keep up please.
We’re not even speaking the same language here.
When one says, for example, “this treatment reduces risk of heart attack in group x by 41%”, in actuality it means something like “we are 95% confident that it reduces risk by 15 to 70%”.
You can’t go around trumpeting a reduced risk of 40 to 50% when that’s not even what the results of this possibly flawed study show.
OK fine. So 15 to 70% isn’t significant. Got it. :rolleyes: :rolleyes: :rolleyes:
Maybe you should stop to think about why you “get that a lot.”
Sheesh.
I don’t really care. I love to learn and share what I learn. Some people appreciate that but many don’t for whatever reason. I’ve never been able to figure that out. If I still have some sense of awe and wonder at things like this and that annoys people, too fucking bad. It should serve as reminder of not what a fool I am but how dead and jaded they’ve become.
As I explained earlier, you simply cannot jump from “percentage change” to “significance level” without further statistical analysis. This is why we usually talk about p values instead of % difference. Again - if, in your control group, you have four heart attacks out of seven thousand people, and in your test group, you have two heart attacks out of seven thousand people, that is a 50% reduction, but is almost certainly not significant - it’s just statistical noise. You need to look at the actual numbers. 50%, or 70%, or 300%, or 1.2% may be significant, or it may not.