On the one hand, that's the point of the article. That it ceases to be a useful diagnostic indicator.

On the other hand, if there are 100 places in the shoulder where you can have an abnormality, and most people have just one or a couple but the other 98-99 are normal, then each one individually really is abnormal.

So it's complicated, and then it becomes important to figure out which abnormalities are medically relevant, in which combinations, etc.