> The idea is that thicker artery walls are associated with heart disease, right? Right, except, if you've every scrubbed in for a carotid endarterectomy, the

> The idea is that thicker artery walls are associated with heart disease, right? Right, except, if you've every scrubbed in for a carotid endarterectomy, the culprit is a lesion full of cholesterol and inflammation, not a diffuse thickening. > Is there reason to believe the relationship is linear? Well, yes and no. Things that cause atherosclerosis tend to have linear relationships with the disease. So if you find a new relationship, you expect it to be linear. Problem is, their analysis has to pile on a lot of other stuff before you find a relationship. And "Have the disease" is a pretty big step function compared to the other things they control for. Notice also that N goes way down in those severe subgroup analyses (tables 2 and 3)