False positives, or even the screening itself, if at a large enough scale, causes death.
E.g. I had a gastroscopy recently. Endoscopies (e.g. for gastroscopy, colonoscopy) have incidence rates in the 0.1%-1% range. Most incidents are minor. Minor bleeds etc. Some are major. Perforation is a risk. Some performation leads to death. Over-prescribe endoscopies either as a first test or as a followup and you'll be killing more people than you save, even if the risk to any individual patient is tiny.
It might seem worth it. E.g. you might catch a prostate cancer early from a colonoscopy. You might catch a hernia that could become a problem from a gastroscopy. Or any number of other things.
The problem is that broaden screening too much, and your true positive rate becomes too low, and the true positive rate where the problem would otherwise go undetected and cause problems becomes even lower, and so for many conditions routine screening barely budges survival rates for true positives, but creates a mortality rate for false positives.
Sometimes broadening testing is worth it - a serious enough and frequent enough condition can warrant it by saving more than it kills even if you only marginally improve survival rates, or you might have a condition that is rare but where detection can very significantly affect survival rates.
But often it isn't. E.g. too broad screening for prostate cancer and breast cancer are both cases where the benefit in terms of nudging survival rates from broad screening vs. more targeted efforts (e.g. screening people who have felt something) is low enough that the advice most places is to focus on limited age groups or other risk factors. Hernias - I turned out to have one - is another one where the proportion that ever becomes a serious problem is so limited that the (tiny) risk of more broadly using gastroscopy to look specifically for that (rather than to investigate potentially more serious problems) is rarely warranted, and where the risk of a portion of patients pushing and getting (a lot more risky) surgical intervention in cases where it is not medically necessary is another reason to discourage too broad screening.
There's a continuum here from not allowing patients to choose, via allowing it but discouraging it, to actively encouraging it, to broad screening programs. There are a whole lot of things that'd be fine to allow if a patient has specific fears, that still should be broadly discouraged.
I don’t know if I would advocate for “excessive” endoscopies. You’re right: there’s a not insignificant chance for injury or death and I can imagine an situation where the risk goes up due to the staff being overworked.
However, a non-invasive scan (ultrasound, CT) still has a decent chance of catching prostate cancer and hernias, and as far as I’m aware they carry zero risk of injury or death.
CT scans are not risk free. They're x-rays and while again the individual risk is tiny, doing a large number of them over a large population would be expected to increase incidence of cancer [1] (in fact more so than I thought).
External ultrasound (ultrasounds can also be done endoscopically) itself I think is risk free, but inducing anxiety in patients or exposing patients to a healthcare facility is categorically not risk free even though the risk is very small for each individual patient.
The problem is not that the risk is high for individual screening events, but that even a one in a million even will happen regularly once you start regularly screening a multi-million person population.
There is no such thing as a risk free event once the scale is large enough, so the question becomes whether the risk is low enough relative to the potential benefit, and then you need to consider that often a true positive is rare, and making a difference to the outcome of a true positive is even rarer, so depending on what you're looking for even a one in a million adverse outcome might negate the benefit or worse.
EDIT: To add one more point: If you do external ultrasounds and find a potential cancer, you will be prescribing additional tests. Any false positives from the ultrasound will thus end up contributing to excess further testing, and you need to count the harm from that as well. A 1% false positive rate coupled with a 1:10k cancer risk from a follow up CT scan or a (I don't know what the real rate is for perforation) 1:10k risk of perforation with an endoscopy and you have a one in a million severe adverse effect. Do it enough and you'll have deaths even when the starting point is something as innocuous as an external ultrasound. That doesn't mean it can't be worth it. But it does mean you need a risk analysis before any largescale screening effort.
> Individuals who have had multiple CT scans before the age of 15 were found to have an increased risk of developing leukemia, brain tumors (6), and other cancers (7) in the decade following their first scan. However, the lifetime risk of cancer from a single CT scan was small—about one case of cancer for every 10,000 scans performed on children.
E.g. I had a gastroscopy recently. Endoscopies (e.g. for gastroscopy, colonoscopy) have incidence rates in the 0.1%-1% range. Most incidents are minor. Minor bleeds etc. Some are major. Perforation is a risk. Some performation leads to death. Over-prescribe endoscopies either as a first test or as a followup and you'll be killing more people than you save, even if the risk to any individual patient is tiny.
It might seem worth it. E.g. you might catch a prostate cancer early from a colonoscopy. You might catch a hernia that could become a problem from a gastroscopy. Or any number of other things.
The problem is that broaden screening too much, and your true positive rate becomes too low, and the true positive rate where the problem would otherwise go undetected and cause problems becomes even lower, and so for many conditions routine screening barely budges survival rates for true positives, but creates a mortality rate for false positives.
Sometimes broadening testing is worth it - a serious enough and frequent enough condition can warrant it by saving more than it kills even if you only marginally improve survival rates, or you might have a condition that is rare but where detection can very significantly affect survival rates.
But often it isn't. E.g. too broad screening for prostate cancer and breast cancer are both cases where the benefit in terms of nudging survival rates from broad screening vs. more targeted efforts (e.g. screening people who have felt something) is low enough that the advice most places is to focus on limited age groups or other risk factors. Hernias - I turned out to have one - is another one where the proportion that ever becomes a serious problem is so limited that the (tiny) risk of more broadly using gastroscopy to look specifically for that (rather than to investigate potentially more serious problems) is rarely warranted, and where the risk of a portion of patients pushing and getting (a lot more risky) surgical intervention in cases where it is not medically necessary is another reason to discourage too broad screening.
There's a continuum here from not allowing patients to choose, via allowing it but discouraging it, to actively encouraging it, to broad screening programs. There are a whole lot of things that'd be fine to allow if a patient has specific fears, that still should be broadly discouraged.