I didn't read the JAMA article, but shouldn't it be possible to test that hypothesis differently? E.g., people get CT scans for fractures in knee or wrist. Cancer in those places is very rare, so of CT scans cause cancer, shouldn't there be noticeable difference between scanned and unscanned people?
It's quite hard to give bones and joints cancer. Cancer prefers dividing cells.
Some blood cells are produced in bone marrow and it's not rare to get "bone cancer".
Yup, but it's significantly less common than most other forms of cancer.
If cancer is also rare in those spots without a CT, that would seem to indicate a major confounding variable at play.
That's my point: if CT scans cause cancer, cancer in the wrist or knee should be much more frequent among people who had a wrist or knee CT scan than among the general population. CT scans are relatively new, so there probably is a record in each patient's history.
Again, not necessarily. Different things cause different cancers. You're unlikely to get melanoma from smoking; you're unlikely to get liver cancer from a sunburn.
CTs may not cause significant amounts of wrist/knee cancer - I can't speak either way on that - but that wouldn't mean they're 100% safe elsewhere. For starters, wrist/knee imaging needs less radiation - they're relatively thin parts of the body, and relatively small regions.
But then the assumptions behind the article is not general, and the conclusions would have to amended to the radiation dosis and/or tissue. Which makes sense, of course.