> then reasonable step is to not follow-up in that case.

Then a more reasonable step is not to get the scan. Why do the scan if you don’t follow up.

This is why not all screenings are recommended. People think that insurance won’t pay for scans because insurance is a jerk. They don’t pay because the scans aren’t medically effective at protecting populations.

Because you can check it on the next scan and see if things are growing.

Follow up procedures should be adjusted based on whether the original test was preventative or not, titrating the amount of risk and cost appropriately

The fact that we treat diagnostic results the same regardless of priors just means our medical system is still stuck in the stone age

Saying the screenings aren't recommended is just saying "we're too dumb to use the results of this screening the correct way, so you shouldn't get the screening"

We can only hope that the new developments in AI will mean that modern statistical procedures will finally be applied to medicine as a side effect (though I wouldn't hold my breath)

The premise of the article is that all of these tests were "preventative".

> we're too dumb to use the results of this screening the correct way, so you shouldn't get the screening

I agree with this. One day we might be smart enough. But they say is not today. It’s ok to be dumb.

It’s also dumb to spend thousands of dollars on these diagnostic scans that we’re too dumb to use.

We can plan around our stupidity. We can also research new ways to use these data and, hopefully, one day be less dumb.

You are misunderstanding how ‘incidentalomas’ are reviewed and managed. You don’t know it’s a false positive until you’ve done the invasive diagnostic tests subsequently (with non-zero risk), at which point you can classify it as a false positive

An issue that gets brushed over is you have various possible results to an invasive diagnostic test.

1. Bad 2. Indeterminate 3. Benign 4. Nondiagnostic

One and two are simple enough. Three and four are really problematic. And they are more common that people assume. Worse people confuse the two.

Indeterminate says they have a sample that is sufficient but can't be assigned to Bad or Benign. What do you do then? If you treat some large percentage will be treated for nothing. If you don't some percentage will see progression.

Nondiagnostic says there are issues with the sample that make a determination impossible at all. What do you do then? Repeat the biopsy? What if you get another Nondiagnostic? Forget about it?

The point is that if you know the test has so many false positives that you're going to ignore the results, just don't run the test. What's the point of running a test if you're never going to follow up?

You can follow-up when it's warranted. See the below anecdote. I'm guessing as to the details, but perhaps the thyroid was cancerous early, but it was too indeterminate, so they did nothing. But then when it started growing, they could change the approach. It's a piece of date, to weigh amongst the overall mix. The problem is most people will demand too much action.

"Bob Nelsen says that he has a family history of cancer, so began getting annual Prenuvo scans around 10 years ago.

    - The early scans showed a nodule in his thyroid, but doctors didn't believe it was problematic. Then a scan in 2018 showed that the nodule had changed, which isn't supposed to happen in healthy people.

    - "When I went into Stanford for surgery, I asked how many patients they'd had with five years of longitudinal data on their thyroid, and they said I was the first," Nelsen tells Axios. "The medical profession is trained to be reactive, not proactive."

    - He adds that the scans also discovered a small scar on his brain, which he assumes came from falling off his bike and getting knocked out when he was 11 years-old."

You know whether or not it's warranted before you run the test. That's the whole point of the discussion around when preventative testing is appropriate. The key is that the test is just one additional data point. Before running the test - you want to make sure that the data from the test will be useful for whatever purpose you want the test for (false positives could be ok if the point is to minimize false negatives). That way when you run the test, it's always "warranted" to follow up if needed. We do mammograms for healthy seeming people all the time - it's totally feasible that we'll determine that these new approaches are informative enough to become recommended for the general public, but we're still on that scientific journey