Comment by timr | Hacker News Reader

Comment by timr | Hacker News Reader

> Diagnosis rates in the U.S. rose in 2019 to 107.8 cases per 100,000 people under 50, up 12.8% from 95.6 in 2000, federal data show. A study in BMJ Oncology last year reported a sharp global rise in cancers in people under 50, with the highest rates in North America, Australia and Western Europe.

This is a relatively small increase over ~25 years. How much of it is due to cases like the highlighted one (which is tragic, but clearly driven by symptoms), vs increased screening?

The US "standard" age for colonoscopy, for example, was recently dropped from 50 to 45...without really good evidence (essentially, just "updated models") [1]. This inevitably leads to higher rates of false positives in younger people. Over-treatment is a real risk in places like the US, where there are financial incentives to treat.

[1] old guidelines: https://www.uspreventiveservicestaskforce.org/uspstf/recomme...

new guidelines: https://www.uspreventiveservicestaskforce.org/uspstf/recomme...

> The US "standard" age for colonoscopy, for example, was recently dropped from 50 to 45...without really good evidence

The age was dropped in 2021, in part because of the very numbers from 2019 you quoted. It's also weird to link to the page with evidence and give an unsubstantiated "without really good evidence" summary.

> essentially, just "updated models"

is literally how science is done?

Note also that not all screening is going in this direction, again based on evidence. Prostate cancer screening has been significantly reduced; many men in the US today will likely never get a prostate cancer screening.

> in part because of the very numbers from 2019 you quoted

I gave those links to show that there are no "numbers". They based the change on modeling results. The modeling results are based on assumptions. The assumptions may or may not be valid.

> > essentially, just "updated models"

> is literally how science is done?

My PhD is in statistical modeling of biological systems. To answer your question: no. Modeling, on it's own, is not "science". I would never suggest to anyone that a statistical model is the same thing as evidence. It's a model, which is based on evidence (maybe), after a thick layer of human meddling.

Modeling can be an input to a scientific process, but it isn't "how science is done". If you want to make a big change to all of health care like this, you need to do the damned experiment -- randomize people, and see what the effect of screening at 45 (vs 50) is on a hard endpoint like overall survival.

I can't read the original article, but presumably it treads the same ground as similar articles [1], which in turn generally reference [2]. Past articles talking about screening include [3].

Generally supports the view that early screening and detection are largely responsible

> Archie Bleyer, a clinical research professor at the Knight Cancer Institute of the Oregon Health & Science University, says the surge in cases of thyroid cancer, as well as kidney cancer, is at least partly a result of “overdiagnosis” due to increased screenings, both for cancer and for other health concerns. The screenings have detected tumors and masses in those organs that “look like cancer, so they’ve got to call it carcinoma,” he explains. But they “would never have been a problem if they were never picked up,” because they typically wouldn’t spread or progress.

I think this is rapidly becoming the consensus view, but the subject is complicated. In the long term, these early detections might lead us to the point where they can become more accurate, and we are better able to distinguish between tumors and masses that would progress and those that would not, but I'm not personally very optimistic; there are two many variables -- the same imaging between multiple facets of environment/genetics/diet/metabolism/chance might have wildly different outcomes.

[1] https://thehill.com/curing-cancer/4041032-cancer-rates-are-c...

[2] https://www.nature.com/articles/s41571-022-00672-8

[3] https://www.prospectmagazine.co.uk/essays/41620/the-dark-sid...

It's a longstanding debate in cancer medicine -- google "turtles, rabbits and birds".

What is not a debate is that increased screening inevitably leads to false-positives. It's a mathematical certainty. But yes, you hope that you find more rabbits than turtles.

Yes! That's exactly it. I've seen that metaphor before and forgot where I saw it until I searched around for what ended up being a previous comment of yours [1].

[1] https://news.ycombinator.com/item?id=37463187

The past three years confirm this hypothesis.

They don’t have screening for cholangiocarcinoma. It’s not common in the Midwest.

My partner discovered her cancer as a result of an unrelated condition: kidney stone MRIs showed a spot on her liver. We’ve been disappointed by every western medicine outcome for the last two years.

The Mayo Clinic was an incredibly disappointing experience. Our last interaction there, after giving some shitty news, our oncologist asked if we ate corn, like in tortillas. She had a theory it might be to blame. I pointed out corn is everything and bitterly wished her luck.

Hopefully, my partner will see her kid graduate high school in three years.

I'm sorry! Liver cancer is really tough. Hopefully they found it early and/or find that it's slow growing.

> They don’t have screening for cholangiocarcinoma. It’s not common in the Midwest.

It isn't common for anyone under 70 which is why they don't spend a lot of effort on diagnosis and treatment of it. Friend of mine died from it when he was like 33 though. I hope you beat it.

I'm sorry man. Routine cancer screening that is available to the general population is still super primitive. Right now, our lung cancer blood test is the only of its kind, and you need 20 pack-years of smoking history to get it while in trials. We're expanding to other cancers. Once these tests hit market they'll be cheap and effective, but trials in general take a very long time. I really think something needs to change with the regulatory process to get lifesaving tech to people more quickly, because although it needs to be a regulated marketplace, there is a staggering human cost to how inefficient the current system is.

One alternative to colonoscopy at 45 that is widely used outside the US are FIT tests. They go for $30 on Amazon (and around $80 for Labcorp) and they are much cheaper and incomparably less time consuming than a colonoscopy. The idea is that you only go for a colonoscopy under 50 (and some even say 55) if the FIT test comes positive. Disclaimer: I'm not a doctor and this is not medical advice.

Note that this works well in the case of colon cancer because it has a slowly-progressing prodromal stage in the form of polyps whose weak blood vessels cause tiny amounts of blood loss into the stool.

One side effect (heh) of those, at least in the US: If my doctor orders a colonoscopy, then the insurance covers it. But if he orders one of these tests and it indicates that I need a colonoscopy, then I pay for the colonoscopy.

Note that this may vary, depending on your insurance...

OR... Wait until your doctor orders the colonoscopy and drag your feet until you say, without letting them know of the FIT results, "I'm ready to do it now". Doctors in the US will almost always order a colonoscopy starting at 50, and often starting at 45.

The cost of the FIT tests is negligible compared to the cost of the colonoscopy, so it makes sense that you do these on your own outside the insurance system.

Reply on news.ycombinator.com