-
that was leading to increased risk.
-
In fact, it's not the weight itself,
-
it's just that the lower weights are associated with smoking,
-
and smoking is what's driving up the mortality.
-
That's why you need--in these studies
-
you need to look at causation rather than just correlation.
-
The solution to that particular problem was
-
to compare smokers to non-smokers in a sample like this,
-
or you can take statistics and control from the impact of smoking
-
and find out how much it's contributing
-
to that little J part of the curve.
-
This is interesting,