Impossible to draw any conclusions from such a convoluted and problematic model. No mention of how they determined patients were unique, or whether multiple scans were counted for a single patient. No mention of patient data - seems that covariates were estimated, leading to greater uncertainty. For example, we have no idea if any of these patients already had cancer before getting a scan. And of course, this entire model is incapable of answering the question that patients actually care about: not "will I get cancer from this?" but "will this scan more likely increase or decrease my lifespan and/or quality of life?".
Seems pretty common sense to me that you're gonna have more people with hidden cancers coming in to get CT for phantom pains, etc and later discover they had cancer all along. That doesn't mean CT caused it. Did the study control for this?
> Did the study control for this?
No, because that's not how the study worked, i.e. it wasn't actually tallying cancer rates from people who had CTs. It basically just looked at the amount of radiation that a CT scan gives a person, and then extrapolated the cancer that would be expected from that radiation dose based on other data we have of e.g. people exposed to radiation in their jobs, and nuclear fallout occurrences.
Previous studies I've read did account for this, and put the number at about 3%. That said CT manufacturers have been working on a bunch of technology to make this better, either by doing a better job at not aiming at sensitive areas or by reducing the amount of radiation needed to get the same resolution in the images.
The whole dataset seems like something hard to work with, population that needs CT scan in the first place is very biased, people don't do CT scans for fun but because there is something weird that docs can't explain already.
It's unclear to me whether you read the journal article or just the Ars article? If you're referring to the journal article that gives your assessment much more weight.