> “Cancer screening was never really designed to increase longevity. Screenings are really designed to decrease premature deaths from cancer.” Explained another way, Dahut said, if a person’s life expectancy at birth was 80, a cancer screening may prevent their premature death at 65, but it wouldn’t necessarily mean they’d live to be 90 instead of the predicted 80.

Personally I think this is just a matter of terminology in public health not necessarily aligning with our intuitive understanding. I presume most people would think that preventing a shortening of lifespan is prolonging your life, but the article makes clear that they are different.

Just take as given that the analysis is correct, and screening for rare Disease A on net has no effect on life expectancy. Almost no one actually gets Disease A, but everyone is screened for it, and that has some diffuse cost to life expectancy: Screen enough people enough times and someone will die in a car accident on the way to or from the doctor's office. More likely the screening crowds out other more net-beneficial medical testing or is taken as some false comfort to continue an unhealthy lifestyle.

Modern cancer treatment, especially for the most common types (i.e. the most likely to be screened for) is very good, even if the cancer is caught later due to lack of screening. So even the folks who catch it early due to screening don't incur a benefit in many cases, further pushing down the life-expectancy win on average.

Still: This is like saying home insurance is a bad deal because on average the insurance companies make money. Screening is an insurance policy (not a free one, to be sure) against a catastrophic outcome.

If you're a public health authority in a utilitarian and budget-constrained mindset, sure, don't encourage screenings by the logic and findings of this analysis. But I don't think individuals should consider on-average-LE-negative screenings as something to avoid.

Open: https://simplecast.econtalk.org/episodes/vinay-prasad-on-can...

Apple: https://podcasts.apple.com/us/podcast/econtalk/id135066958?i...

Highly recommend the episode and show. The gist is that cancer is only 4% of deaths, and screening only reduces those 4% by 20% (so it does not help whatsoever 80% of cancers). But there is a lot more to the episode.

The most important thing you need to do to prevent early death is reduce heart attack and stroke, and that requires better diet and exercise. Really the best drug of all is diet and exercise.

https://jamanetwork.com/journals/jamainternalmedicine/fullar...

and I do not like it. It suffers from what I consider an extremely common problem in statistics: if you define the question poorly, then your output is garbage no matter how fancy or careful your analysis is.

We can start with the beginning of the abstract:

> Importance Cancer screening tests are promoted to save life by increasing longevity, but it is unknown whether people will live longer with commonly used cancer screening tests.

I have never heard of a doctor suggesting a cancer screening by saying "this might save your life by increasing your longevity." What does that even mean?

So let's try to figure it out. The paper uses the terms "lifetime" and "longevity" somewhat interchangeably, and it does not define either term. The best I can figure out is that, for an individual deceased person, they have a certain lifetime in days from when they were screened to when they because deceased. (Or a certain lifetime in days from birth to death, and I'm not sure this distinction matters.)

Great, but this is only for one patient. What about for a sample of patients or for a population in general or for the probability distribution of lifetimes of a given patient conditioned on whether they do or do not get screened? The article does not say, and a single "lifetime" number is not a probability distribution. Is it an expected value or a mean? A median? A mode? No comment.

One of the headline conclusions is:

> Based on the observed relative risks for all-cause mortality and the reported follow-up time in the trials, the only screening test that significantly increased longevity was sigmoidoscopy, by 110 days (95% CI, 0-274 days) (Table 2, Figure 2)

Figure 2 is useless. Table 2 is somewhat informative, and it has a column for relative risk of all-cause mortality and a column for lifetime gained and its 95% CI. But WAIT A MOMENT! The only way you can know the lifetime of an individual patient is if they're dead. If they're dead, their risk of all-cause mortality by the time they died is 100%. That's not 100% plus or minus something with some relative risk thrown in -- they are dead enough to have a date of death so that someone could compute their lifetime! Or maybe "lifetime" means something else, and the authors didn't bother to figure it out and say what they meant. So what exactly is this paper even analyzing?

So I suspect this is a meta-analysis of studies, of which some may or may not have been high enough quality to define their terms, and probably several of which used "lifetime" to mean some estimated property of a distribution, and this meta-analysis completely failed to figure out what the included studies were talking about.

So I rate this meta-analysis as almost entirely useless, on account of it failing to actually analyze anything that makes sense.

So I don't think any conclusions can be drawn. Although... ACS puts the lifetime risk of colorectal cancer at 1:25 or so. So one might naively translate a 110-day lifetime extension for everyone to a 110 day · 25 = 2750 day = ~7.5 year expected lifetime extension for people who actually get colorectal cancer. Sign me up -- 7.5 more expected years of life and presumably more than that of quality life years in the event I contract a not-particularly-rare disease sounds like a pretty good deal. (Colorectal cancer screening is not all that unpleasant, and I apparently only have a 96% chance of the screening being unnecessary.)

This statement makes it clear what's going on. In ML terms: class imbalance. 99.9% of people won't ever get colon cancer, and therefore won't ever benefit from a colonoscopy. It won't make any statistical difference in overall population survival. But for the 0.1%? It will save their lives.

So, while I see the "statistically" part, asking everyone to get zero screenings until you start coughing up blood (or whatever happens when the cancer starts showing very obvious signs)... it just seems weird on the individual level. Nobody knows (nor can they know) if you're the person who would be cured with treatment, or if you're the person whose outcome wouldn't have changed a bit with treatment (for better or for worse). That question matters individually, and while "statistically you're slightly more likely to be in the second group" according to these studies, that doesn't make me feel great about just declining all screenings.

In colonoscopy, the entire colon is examined by using a long flexible tube. In sigmoidoscopy, only the lower portion of the colon is checked, again by using a flexible tube.

The things that come to mind is that sigmoidoscopy may be better tolerated by patients or have fewer complications.

If you assume all the costs are fungible, then the analysis is straightforward (the costs are not actually fungible -- some things cost money, other things cost patient time, or use resources that are in finite supply -- so it turns into a linear programming problem, and we've been able to solve those since before computers existed).

With fungible costs, for each of the available health care services, you the expected increase of life expectancy and divide it by the cost, and prioritize the things that have the highest benefit/cost ratio.

The article doesn't talk at all about the cost of the screens, which is fairly low (vs. spending time exercising, or a year of exotic chemo). It also doesn't look at patient quality of life.

That is shockingly bad. An extra three months of life is not nothing, but not exactly a clear win to push people towards the added stress and effort to be screened.

The real cost of a preventive health scan goes well beyond the price tag - https://news.ycombinator.com/item?id=37266189 - Aug 2023 (254 comments)

That can only be that these experts disagree with the study, because if it were properly conducted such that its findings are true, then in fact its conclusions do add up to skipping colonoscopies and mammograms.

The study is saying that the diagnostic procedures have no effect on outcome. You should not waste time on procedures that don't change outcome.

If anyone's interested the "Wilson criteria" is the basis of decisions about screening programmes and many run programmes fail it.

I found this comment useful: - https://news.ycombinator.com/item?id=37297963

Several comments noted that a 4 month improvement in life expectancy (for sigmoidoscopy) over the whole population is actually pretty good for a low-incidence cancer. That's several years for the people who actually get it.

But the comments seem naive about the downsides of broad screening: over-treatment, iatrogenic disease, false positives, opportunity cost, etc.

Positive screening result usually leads to treatment of one kind or another. Without screening, treatment would start later, if at all. Is that a corollary of this study?

Why? Imagine 1 million people get tested. Well we know exactly 1 person (on average) in that group is going to have the disease. But our 99% accurate test will ring a positive 1%, or 10,000 times. So the odds that you really have the disease are the odds that you're that 1 in 10,000 which is 99.99% against! Well just run the test again. Oh no! It turns up positive again! What are the odds it's two false positives? 99%! Same math. Now we know that 1 person has the disease, but our test will show 1%, 100 people, in the 10,000 as being positive. So your odds of having it are 1 in 100, or 99% against.

I'm not especially interested in being tested for rare conditions.

I'd say this depends on the nature of the diffuse negative effects you mention - if it's car accidents on the way to the doctor's office that's one thing, but if it's people dying during surgery they actually didn't need that's another

The risks are not 'car crash from doctors visit' they are 'got cancer from imaging', 'unnecessary surgery and complications', etc - the risks are directly related to the screening.

This is an active topic in medical ethics as well which you seem ignored of (no offense intended), given that you are framing this in terms of insurance or public health from the perspective of a beaurocrat - the bottom line is that if your screening is more likely to kill or maim you than the thing being screened for then that screening shouldn't be standard practice, and when it is less clear cut than that you still have to make a determination about which screenings make sense to perform on a population level.

That is something that a caring doctor has to think about as part of their duty, there is the very real potential to do much more harm than good by being thoughtless about the interventions you perform.

Also cancer screenings happen in the US. Thus 750k deaths without screenings and 80% of that = 600k deaths with screenings and 150k deaths postponed.

Are they using some unusual definition or something?

Also not that harmless as you think. Only you suggest to perform it on 24 other patients who won't ever have a colon cancer. And that might be you, right? With the real risk (albeit small, but you need to multiply it by 25) of having a serious complication which eventually may result in a premature death.

A diagnostic approach to healthcare can be easily bureaucratized (come collect your stamp after your colonoscopy!). A touchy-feely "million factors" approach cannot.

Quote:

> "Two recent data reviews deserve further mention. The United Kingdom commissioned an independent review after dissenting voices swelled, for the purpose of better informing shared decision making and educational materials about the harms and benefits of screening.6 Unfortunately the review concluded that screening mammography trials were inadequately powered to detect an impact on all-cause mortality, and therefore used breast cancer mortality as a primary outcome. They concluded a 20% reduction and used this in their discussion of harms and benefits. As noted above, cause-specific mortality is both scientifically unstable (conclusion reversal is common when all-cause mortality is considered)7 and disease-centered rather than patient centered (patients would prefer to avoid death altogether). Thus, either screening mammography does not save lives or else we have inadequate data to say whether it does or does not. In neither case can a benefit be scientifically claimed."

> "They concluded a 20% reduction and used this in their discussion of harms and benefits."

We want to do things that will help people. On an individual level, people have the right to make decisions for themselves. At the population level, we should do things that can be proven to be helpful. This leads to decisions that feel heartless - denying people a vaccine during a deadly pandemic while it is being tested; delaying or denying introduction of a new medicine that shows only marginal improvements over an old one.

So their independent review showed benefit by changing the criteria for comparison. We want to help people, and medical interventions feel like helping. Unfortunately, sometimes they can be the equivalent of The Politician's Fallacy: "There was a problem. I did something about the problem. Therefore the problem is fixed"

> "As noted above, cause-specific mortality is both scientifically unstable (conclusion reversal is common when all-cause mortality is considered)7 and disease-centered rather than patient centered (patients would prefer to avoid death altogether)."

People die of /something/. That thing becomes a target to fix; and we deploy resources to fix it. We become emotionally connected to some interventions. (Speaking personally, I'm participating in a bike ride in a few weeks to raise money for breast cancer screenings.) Some of those resources are beneficial, some are not, some may be harmful.

> "Thus, either screening mammography does not save lives or else we have inadequate data to say whether it does or does not. In neither case can a benefit be scientifically claimed."