It is not a good argument. In fact, it makes no sense. If you get a positive with an uncertainty in its accuracy, at the very least, the test is repeated. But even more, you can use the information from the investigation of the reason for the false positive to improve the tests in the first place. If we weren't humans and the uncertainty of the test is known then at the very very very least you could throw a dice to decide whether you discard or not the test result.
If more information leads to worse decisions it just means that the noise level introduced by the test is just too high. A way to reduce the noise is two amplify the signal, and a way to amplify the signal is to look for more information (other tests, other indirect measurements: i.e., look for B if A was positive, etc.).
This assumes that the false positive is caused randomly. That's not the case. False positive tests are usually followed by false positive tests. Then it will take years to find out if it was a false positive or not.
We understand far more than we did say 20 years ago. But the problems are non-trivial on a scale most people don't appreciate.
You can't just repeat the test, that is generally not where the problem lies. You'll get a second positive result and still don't know if it's a false positive or not.
And in many cases there aren't other non-invasive test you can perform. If you can't actually determine reliably if some anomaly will cause trouble before removing it, whole population screening will cause unnecessary operations.
That you want to have some kind of theoretical argument in the face of this evidence might be interesting to you but it isn't to me.
That would be a bit of a paradox in science. I suppose we might take 2 steps forward and 1 step backward in the short term. As medical science advances, hopefully we can address any shortcomings from the additional early knowledge.
I noticed that pancreatic is on the list. This cancer is almost always fatal because we can't detect it early.
This really depends. Information that does not help make a good decision is just noise. It might seem like diagnostic information should always help make a good decision, but that isn’t always the case. If the false positive rate is higher than the base rate, a positive test would be more likely to be wrong than right, even with a very high accuracy.
For example, a large number of unnecessary antibiotic treatments, which fuel resistance, are triggered by doing diagnostic testing on patients with no symptoms.
"Diagnostic stewardship" is a concept that exists for a reason.
And that - no matter who good the test, and no matter how early - leads to a decrease in positive patient outcomes.
This is established medical science, and it pains me to have to continue to point out the same thing over and over again, but since I started with this response I feel obliged to continue to do so.
The outcome of a cancer treatment is not pre-determined, there are a lot of individual factors at play here that will have a huge effect on the outcome, possibly much larger than the effect of that particular cancer itself.
So en-masse screening leading to an increased number of treatments of pre-symptomatic cancers with those current responses is not a choice, we know that this will lead to a worse outcome across a population.
Early detection does not add anything to that. If you could pick out those individuals for which early detection would make a difference then that would be a gamechanger, and here the ball is currently in the genetics court.
The other part where major change can be made is by finding ways to treat cancers in a way that is non-invasive and does not put the patient further at risk (so no surgery, chemotherapy or radiation therapy).
You keep writing comments in absolute terms and talking about evidence, but how do you reconcile your position with the results of successful screening programmes like cervical smear testing? Detecting and treating high risk HPV before it causes changes that can turn into cervical cancer has dramatically reduced the harm caused by cervical cancer itself at a population level. Routine screening of this type isn't normally recommended for young women, but it becomes increasingly effective with age and screening programmes operate accordingly.
Whether this test is one that can serve in that particular role is definitely not something that has already been determined and those advocating for such are ignoring a mountain of established science and are simply jumping the gun.
Policy is set by the overall effect of application, which can be quite different than applying that same tool in an individual setting. This whole thread started with
"This is one of those medical revolutions that I am waiting dearly for.
Facilities that are not hospitals(to avoid the risk of occupying medical devices that sick people need) built to _regulary_ check up otherwise healthy people for preventive care."
And that is not something you do without taking into account the downsides. Whether this is revolutionary or not remains to be seen, it definitely is a useful test based on what I've read about it so far.
This test isn't a 'bad test', though applying it in the wrong way can lead to bad outcomes. This is why tests used absent symptoms have to have a false positive rate that is much lower than the base rate at which the disease occurs. These tests - unlike software tests - should not be thought about in absolute terms but in terms of probability, so a positive test indicates a probability that you have a specific disease, but it is very well possible that you do not have it, and a negative test indicates a probability that you do not have it - but it is very well possible that you in fact do have the disease. And the reasons for a false positive or false negative may have nothing to do with the test itself, but could easily be an environmental factor or some benign aspect of the test subject that was not accounted for when designing the test.
Whether or not a test is suitable for mass screening hinges on the factors above, the base rate for the disease, the age of the group being tested, in some cases gender and so on. In order to have a positive outcome across the population all these factors have to be taken into account and by the time that you have done so there are - unfortunately - at the moment no miracles to be had. But combinations of knowledge, for instance a genetic pre-disposition to a certain disease + a positive test can have much higher signal to noise ratios than either by themselves.
But make no mistake: this is an important development in medical diagnostics and it may very well be that once more evidence has been collected and some of the kinks have been worked out that this particular test or an improved version of it can be applied in a screening setting for one or more of the cancers that we currently have no reliable detection method for and that could be cured if detected early enough given a high enough base rate and a low enough false positive rate.
Note that the scientists behind the project are very careful with their statements and that the reporting on this was actually quite neutral and trends to cautious optimism, which I think is warranted, but until the result of the new studies is known it is way too early to shout 'revolution'.
Maybe we interpreted OP's comment that you quoted differently? I read it as being in favour of preventative medicine as a whole, not necessarily endorsing this specific test at this specific point in time.
Maybe I'm also interpreting some of the other comments in this discussion, including yours, differently to how their authors intended them. My concern is that as written they appear to be criticising all use of screening, regardless of its efficacy, which is extremely dangerous.
The people that have built this are at the forefront of this field, I've been following them for quite a while - since the announcement in March last year - it has direct bearing on some other things that I'm involved with and I'm hopeful that it will at first be a useful diagnostic tool and that in a later stage - after the kinks have been worked out and there is sufficient data - that it might help with more than that.
Preventative medicine obviously has its place and for selected cancers we are now in a phase where early detection leads to improved outcomes. But we should continue to be weary of overselling this - the same has already happened with other cancer tests.
Absent symptoms mass testing has serious risks and these will obviously be taken into account when setting policy, the article is actually reasonably neutral in this respect so I wonder why it leads to an immediate response that is equating this with a medical revolution. It may well be, but there is no evidence right now that this is the case.
This leads to lots of suffering and hardship.
Where can I find the studies that conclude this?
https://www.ncbi.nlm.nih.gov/books/NBK223933/
https://www.theguardian.com/society/2021/may/12/annual-scree...
Those are popular and accessible, the actual studies you can find through Google Scholar, SciHub or various medical publications.
This is not something where the general public - or software developers, who seem to treat cancer as a bug that needs to be fixed - are going to be very helpful, I am more than happy to trust the medical establishment with this.
What would be a game changer would be rather than improved testing something that would destroy tumors in-situ in a non-invasive manner that does not involve radiation or attempts to poison the body just this side of death.
...to be contrasted with the mental health implication of living with the knowledge that you are *not* being tested and that cancer can grow undetected for years.
That, if anything, is a very good reason for anxiety.
If our best treatment for fixing a harddrive was to hit it with a hammer, then maybe we'd also conclude monitoring for minor bit errors in data centers is unwise too.
A positive blood test is a symptom. One which you'll miss if you don't do the blood test regularly.
And looking for symptoms (other than a blood test) is a form of testing... one with much worse accuracy (both positive and negative) than the blood test, especially in the early stages.
https://www.cancer.gov/publications/dictionaries/cancer-term...
"A physical or mental problem that a person experiences that may indicate a disease or condition. Symptoms cannot be seen and do not show up on medical tests. Some examples of symptoms are headache, fatigue, nausea, and pain."
But GP seems to be saying that they hope to see better tests in the future, that are not risky or invasive, don’t have as much false positives, and are less costly to do, so that the equation would change and we could actually meaningfully improve outcomes by doing large-scale preventive testing.
You would still likely have cases where you cannot /improve/ outcome by knowing sooner that you have a disease, but as long as you are not making matters worse and improving chances for a significant subset of people, all while keeping costs the same or even decrease costs, this seems like a great evolution.
Better tests do not automatically lead to better outcomes. They will lead to many more cancers detected, and they will lead to more interventions.
Just one example (there are many more): for many tumors the risk of the operation to excise it already outweighs the risk of the tumor itself leading to damage to the body.
The factors that govern whether an intervention is necessary are determined by the rate of growth, the risk of meta-stasis, the organ(s) affected, the stage the cancer is currently in (and here early detection would at least help to get a grip on that) and so on.
But once detected treatment is going to be the norm, and that's where the problem lies: treatments are not necessarily an improvement over having a mostly dormant cancer.
If you were to autopsy all of the cadavers from any given country for a period of time you would find a correlation with age and the presence of one or more tumors in that cadaver, even if the person never had symptoms and died of a completely unrelated cause. Treating all of these would have resulted in some of those people ending up in the morgue a lot earlier and having a reduced quality of life both from a medical and a mental health perspective.
Deciding to treat - or not - is not a simple matter.
We just need to improve the decision making after getting test results (one of these decisions is to decide to not do anything), and more data make improving it easier.
My argument is not about particular individuals, but about populations as a whole and wholesale screening of those populations. The consensus is that this does not lead to improved patient outcomes across that population, though in individual cases it may very well be the result.
Patient exhibits symptom => perform a not-especially-invasive test Positive test result => invasive test like a biopsy Positive biopsy result => heavy-duty intervention (although I'm not focusing on this part of the chain in what follows)
Both testing and (certain) symptoms have predictive value, and don't completely overlap. So there's something like this going on:
P(actual problem|no additional information) = really really low, which is why they don't scoop out chunks of every organ to test "just in case" every time you go to the doctor
P(actual problem | [symptom AND positive test result]) = generally high enough in at least some cases to justify the risk of the biopsy, which is why it's the standard of care
P(actual problem | just symptom) = probably not super high, which is why the tests are developed
P(actual problem | just a positive test result) = substantially lower than P(actual problem | [symptom AND positive test result]), so in the general case the balance of risk no longer favors the biopsy
In the broadest of strokes, is there anything I've just said that you substantially disagree with?
And that's exactly where the issue with indiscriminate asymptomatic testing lies, that requires much higher quality tests than the ones that can be used in a diagnostic setting once a patient is symptomatic.
To add one more unpopular bit of data to all this: there is some evidence that the indiscriminate testing for certain cancers has gone too far and that it no longer is a net positive. But in the presence of certain mutations those tests are extremely valuable.
https://www.statnews.com/2018/01/01/cancer-screening-misled-...
Biology is messy, and it is quite hard to state up front whether or not a test or a treatment - even if in an experimental setting it is working - will still be a gain if rolled out in a different setting or application. Hence all the trials and studies, that's the only way to really get a grip on this.
I'm quite curious what the outcome of the large scale test the article refers to will be.
This is true today.
But if we could detect cancer at a really early stage, relatively reliably, maybe this means we could develop new and effective treatments that are low risk and low on side-effects ?
And if we had that, early cancer detection will also have a totally different meaning, so that could help with the mental health aspect too.
If we start to test more, and understand the magnitude of the problem better (despite false positives/negatives) we can better allocate capital to solving this problem.
Sure, "cancer is horrible, we should already allocate as much capital as possible" but this just isn't reality. As soon as the addressable market for early-detected cancer treatment goes from X per year to 100X per year (and 1,000X or 10,000X is "even better"), big pharma has more motivation to actually R&D safe treatments for early-detected cancers.
Not testing more to detect cancer early is silly, if only from the perspective of capital allocation.
This is just the first article I ran into.
https://www.sciencealert.com/new-evidence-finds-numerous-can...
This is completely irrational argument. Catching cancers early is crucial.
Furthermore, if better information leads to unnecessary intervention the blame lands squarely on the hospitals being overzealous and greedy.
> For aggressive cancers - the ones that are really problematic - you would have to do such a test too frequently to make any real difference
Because aggressive cancers go from 0 to dead in a week? Please.
"Our tests are bad, so lets not test" is not a thought worthy of respect. This states that you know about a problem but want to continue to ignore that problem. Reprehensible.
No, we should take the exact opposite approach. Test everyone constantly until such methods become both cheap and powerful. So yeah we all have some cancer load but if the diagnostic or treatment systems can't deal with that reality then those systems need to change.
You are treating this like a software problem, but it isn't, it's a medical problem, and medical problems tend to be complex because they have a ton of confounding variables that make it hard to have a one-size-fits-all method for dealing with medical issues.
What needs to change is that people need to realize that they have a field of expertise and that the medical domain has its own experts who typically dedicate a lifetime to their profession, their general consensus is that improved tests are welcome but in and of themselves are not enough to guarantee improved patient outcomes. Yes, this is unfortunate, but it is also a simple reality, you can either accept that or not, that's up to you but if you want to make a change there than you probably should join the medical profession. Most likely by the time that you have completed the requirements you will have shifted your viewpoint away from the software domain's mindset that all bugs can be found and squashed by the next sprint. Which by the way judging by the general quality of software out there is also something that doesn't work out in practice as we believe it should in theory.
Before my software career I spent a few years working in a diagnostic medical field. Specifically osteoporosis testing. I worked both in a research capacity at Stanford looking at osteoporosis in older men (not pretending to be the PI here) and in a day to day testing clinic. So I've seen exactly what happens when you test a cross-sectional asymptomatic sample of the populous and what happens during the normal course of referred testing.
Low bone mass at the spine, hip, heel, and forearm as measured by DXA are correlated with increased risk of fracture, but it's only a correlation. Some people have resilient architecture which looks porous on an x-ray but only leads to serious fracture much later in life.
Because the current diagnostic tests are set up with levels like 'osteopenia' and 'osteoporosis' the reaction to clinical referrals was most often treatment. Some of those treatments have serious side effects like necrotizing impacts on the jaw. However the reaction to testing in a large asymptomatic population was much more likely to be an increase in preventative behavior (exercise) or no treatment except in extreme cases. While our study was exploratory and didn't have a treatment cohort (we cared about the impact of sleep quality on bone mass) we got to see a lot of older men who if they were referred to a clinic might have received treatment because that's what clinics do. Instead we had sufficient data to discuss what's normal and what isn't. For a time we were the leading experts in the world on what "normal" meant.
Because I've conducted these tests myself and seen data from hundreds of experimental and clinical patients I feel comfortable contrasting the two. The problem is the clinical medical field reacting to a lack of data with over-prescription of treatments.
FWIW I too have some experience with medical diagnostic systems (specifically: cancer testing), and one of the main reasons why I'm still skeptical about this test is that for many cancer types tested for the base rate is much lower than the false positive rate.
“We’ve never had blood tests that detect pancreatic cancer early, therefore we should never use them because when we used other means to try to detect early, it didn’t lead to better outcomes.”
If you can detect colon, kidney pancreatic or liver cancer early, you might be able to do surgery or develop treatments at the stage 1 stage. Right now we don’t have anything except MRIs and CT scans that are too hard and expensive to do frequently. And if you are diagnose with pancreatic cancer it’s usually so late that you will basically die in weeks.
You’re basically saying “give up. Even if we detect early you all die anyway” which is frankly stupid. You’re discounting the possibility that early detection by means of a blood test adds whole new layers of possibilities to fight those particularly dangerous cancers.
So no, I'm not 'basically giving up', and no you won't die anyway (well, unless you take that in the most abstract way), in particular likely not from cancer.
Even for those cancers where we do screen (such as for instance breast cancer) it is not a given that the increased frequency of detection has led to better patient outcomes.
But once you know someone's genetic disposition increased frequency of testing might be advantageous.
Even breast cancer needs x-rays to detect lumps, not detect cancer. You need a subsequent biopsy to differentiate. That’s our technology right now. If we had a blood test to detect cancer, not lumps, it’s game changing.
Maybe after a generation of early detection, new outcomes will emerge if we can detect the most deadly cancers early. You’re taking old studies and applying them to new technologies and saying “it won’t work.” It’s ridiculous that you are doing this.
Yes, it is a game changer, but it is not the kind of game changer that this is being made out to be here and mass screening using this method is not going to lead to improved outcomes.
As an extra tool in the toolbox of the diagnostician it is very useful.
I’m generally inclined towards the position of more testing....but there are very real tradeoffs to any of the interventions that a blood test could prompt.
It would also give the person the opportunity to change lifestyle to perhaps prolong the time that the illness will become a problem, or perhaps halt its progress altogether.
Additionally, the person who is aware of such illness, can keep an eye out for symptoms related to it, that might otherwise be ignored as something else. At which point normal cancer treatment can progress.
Data about unnecessary procedures is relatively plentiful, which is of course sad, whether that's driven by commercial incentives or the need for 'something to be done' is not something that I have any grip on but it certainly is problematic.
Another incredibly irrational argument.
If the risk from surgery outweighs the risk from cancer it it should not be performed.
I'm more than willing to be convinced otherwise but with relevant data. FWIW I've been following this particular development closely because it has direct implications for a start-up that I have had a lot of contact with that is also in the early detection space and they were adamant that the combination of factors is such - and so complex - that test accuracy is trumped significantly by absence of symptoms in otherwise healthy patients.
Your input on this would be greatly appreciated.
I am not arguing with most of your points, it's just that you're not being epistemically humble. I find it's far better to just act humble even when you aare schooling people who persist on believing in magic.
I sincerely hope that this test will drive down the cost of testing and that they will get the kinks ironed out, I think it is a major development and I'm actually afraid that those that oversell it will cause it to end up being tainted, as has happened with many other cancer related diagnostics and potential treatments. Cancer seems to bring out an emotional response rather than a rational one, typically because almost all of us have direct knowledge of one or more people in our environment that succumbed to it.
A lot of work has been done since this development was first announced and there are still quite a few kinks that will need to be ironed out. The whole idea - such as promoted in this thread by some - that this makes cancer a matter of 'simply looking at the right organ' is so laughably oversimplifying reality that some balance is required, and I think at some point my irritation got the better of me.
So thank you for pointing that out.
And to add to that: It definitely is not my intention to suggest that this test is without value.
The WHO, for example, disagree.
https://www.who.int/activities/promoting-cancer-early-diagno...
But let me stress this in case it wasn't clear to you: screening and early diagnosis are not the same thing.
If you screen a large population for cancer you will turn up a lot of cancers that may never become a problem, or that may even end up being re-absorbed without the need for intervention, as well as a large number of false positives.
Early diagnosis means that there are already symptoms.
Please don't confuse the two.
This comment is simply a personal attack.
I will write in this deeply arrogant fashion and refuse to link to anything I'm talking about, and when questioned I will infer that the people I'm replying to are utterly moronic (without saying quite such, I will instead say things like 'I'm sorry you don't understand / let me make this clear / you shouldn't be posting things you don't understand').
I will continue to write as if my stance is absolutely correct even though I am not an expert in this field, I will write as if I am.
And then when people call me out for my arrogance I will insist I've done nothing wrong and will be confused.
Baffling.
IMO, that is inflammatory and arrogant.
More broadly speaking, much of the arguments in this post seem almost political. They are dressed up in fancy language, but essentially boil down to "We can't give these peons information that they aren't smart enough to deal with"
That is an extremely arrogant position to take, even though I do believe there is some truth that knowledge can be counter-productive.
Who gets to be the gate-keeper?
I feel like a better conclusion is that we need better tests, that detect things that when acted upon improve patient outcome. Of course, we're nowhere near that yet, but do you really think in 1000 years we will still wait for patients to be responsible for correctly noticing symptoms and going to the doctor? Of course not.
It's good that we have studies that show we should move with caution in this territory, but completely ignoring it forever seems absurd.
Correct. But that's not the issue here.
Current tests can detect benign tumors, and do. People go in for their regular tests, hear "cancer" and of course want to get treatment. But that treatment itself is not idempotent. You want to accept the negative side-effects if the alternative is an aggressive cancer that will end your life soon. But you don't want that if there is no cancer.
As in many fields, it's a matter of trade-offs and risk assessment.
Here's one article detailing some of the issues surrounding overdiagnosis of colorectal cancer:
https://www.gastrojournal.org/article/S0016-5085%2818%293482... (2018)
Take prostate cancer as an example. It's a relatively common cancer, with an incidence of .1% [0].
If you have an _incredible_ screening test with a 99% specificity (ie, 1% false positive rate), then:
If you test 100,000 people, you will correctly identify 100 people with cancer, and misdiagnose one thousand people as false positives.
There are consequences to having a false positive: unnecessary interventions, surgery "just to be safe", stress, etc.
This is why you limit testing to populations that have a higher probability of being ill (namely, show some sort of symptom).
Now imagine being tested for 100 things: the odds of coming up with a false positive for something start to approach 1!
I really hope that false positives become a regular thing. For one it shows that the tests are working according to statistics.
Second, getting comfortable with false positives means that you can more easily hold off treatment, if in your specific scenario the treatment is not beneficial - think an old test that doesn't find the cancer plus a new test that does; and the new test is shown to lead to overtreatment. We need to learn when to hold on and not to treat.
And third, if testing for 10 things will find something that you don't want to know and harm your psyche, when you test for 100 things you will have almost certainty that there is a false positive and this can give you back some of the "blissfull ignorance" I believe (while the doctor can still give you the statistically best treatment).
The enemy is not cancer, the enemy is ill-health and death. You could cure all deaths from cancer by poisoning everyone with cyanide.
We will adjust. Examples: breast, cervical, prostate cancer.
Possible scenario: regular assessments every 2 years starting at age 18. Establishes your personal baseline. When something changes, you'll know.
It's better to know than not know.
That's a wild assumption that every cancer will be acted upon in the same manner as a stage IV even in the nascent state...
Given the quality of healthcare and the desire to not be sued for malpractice, it's a fair concern.
Assuming the United States see here:
https://www.cdc.gov/cancer/dcpc/prevention/screening.htm
and if elsewhere consult your local CDC equivalent.
And maybe out of your garage as well...
Discussion is good. It's ok for people to be wrong and disagree. I'm not even saying they are wrong, but if they were- do you really need to go around telling everyone that some opinion that crossed their mind over a cup of morning coffee would be a disaster if implemented as national policy?
More like “are <practical given available tests> and useful”.
Regarding your last paragraph cancer screening seems very rare except for one or two types (at least where I live), so statistically yes an annual test will miss the worst case you presented, but for the vast majority it will be a huge improvement.
In some cases that 6 months might be very generous, in other cases you are better off to do nothing (especially if you are advanced in age and the cancer is growing slowly). It all depends.
Early detection these days consists of things like CT scans which can’t tell the difference between cancerous tumors and benign tumors. I have a friend with a mass near her liver but they didn’t detect it until quite large. She asked what she should do and their answer was “well if it was liver cancer you would be dead by now, so it must be benign.”
This is the state of detection that you think is such a utopia, that we shouldn’t bother trying to improve, because you are so confident with your answers but you literally have no idea how wrong you are.
Having an accurate blood test that can differentiate a cancerous tumor from benign is ground breaking. Early detection of cancers like pancreatic or liver cancer detection is virtually impossibly today until it’s too late is groundbreaking. It could lead to new treatments that work when the cancer is small vs when it’s too big to operate.
You’re taking studies done using very obtuse, inaccurate and costly detection like MRIs and CT-Scans and conflating them with new technologies. It’s backwards, old thinking and trying to pooh-pooh new ideas and technologies because of poor understanding on your part. It had no place here among people with vision and hope for the future.
With additional asymptomatic testing, there will be false positives. Re-test using a specific test and if still positive, a biopsy. How frequently do false positives lead to unnecessary procedures with the current asymptomatic screening?
What would the rate of unnecessary intervention be, vs the rate of lives saved through early detection?
But I understand your point. How about making it elective? That way, people who prefer to not endure the risk of false positives or the anxiety of awaiting the test results can opt out.
For any particular individual, especially those who end up dying from cancer early detection would have likely mattered. Which is exactly where the problem lies: that is a large number of people, but still (much) smaller than the number of people who will end up with positive cancer test. And policy is set by the outcome for the population as a whole, not for any particular individual.
In most places in the developed world cancer screening is already elective, but not for all types of cancer. Even so, how often are you going to do it? Once every year could easily be too slow to make a difference, and these tests aren't free so say a bi-annual test on all of the population would wreck the ability of the medical world to do much else. This is a tough problem to solve, especially because wetware tends to be finicky to work on and tiny little details will have a huge effect on outcome.
Prostate, Skin, Colon
Those cancers where checkups are useful we already do regular screenings for.
For aggressive cancers - the ones that are really problematic - you would have to do such a test too frequently to make any real difference, for instance, if you were to test annually you'd be on average 6 months away from your next test, plenty of time for such a cancer to develop and kill you.
So this is not the kind of breakthrough that you may think it is.