Hacker News Reader

Comment by wrp | Hacker News Reader

I've seen discussion about this. It seems the general opinion among doctors is that full-body scans for everyone would be a bad idea, because most people have some structural anomalies inside that shouldn't be a concern but would make people anxious anyway.

View on HN

It's not just that it makes people anxious; it's that the anomalies trigger follow-up procedures which are often less benign than the MRI.

Would the confidence level necessary to trigger a procedure not quickly be adjusted to keep the false positive rate in check?

How would it be? Again, the base rates of a lot of these conditions are very low, but not low enough to say "this is a zebra, not a horse", so I think a suspicious lump is likely to trigger a CT, PET, or (in the worst case) a biopsy.

I don't understand what "base rates" means in this context.

If you have a test result, that you think is showing something with a certain probability, which is higher then the current necessary confidence level required to trigger further probing, and it turns out afterwards it was a false positive, my naive way to go about this would be to raise the minimum confidence level afterwards (and vice versa if it's a true positive)

Does this fail in horrible ways that I am missing right now?

Base rate: select a person at random from the population; how likely is a true positive?

What confidence level? Isn't this advocating for the patient to make their own choices around elective scans and procedures?

Hard to say. Consider a patient without complaints with a scan that shows an abnormal bleb in or on the head of the pancreas. Physician 1 thinks it could be a pancreatic tumor given patient age and family history. Physician 2 thinks it is more likely to be a minor lifelong development anomaly with the bleb actually being an intrusive bit of the spleen.

Is exploratory surgery needed? Watchful waiting?

An acquaintance of mine with the second harmless condition had the head of his pancreas removed and he has suffered greatly for over a decade. Almost killed by FDR misinterpretation.

I like this example because it doesn’t rely on the US oddities that people often come up with for this topic (eg doctors doing extra, likely negative EV, procedures for fear of litigation).

No because the doctor can be successfully sued for not ordering the followup even when the followup would not have been in the patient's best interests. I.e., juries are irrational.

It's not just the anxiety and psychological distress either. If we scanned everyone all the time, you would find literally millions of conditions that would never cause symptoms or major health issues, but would nevertheless be treated unnecessarily, incurring massive additional costs and needless interventions with their own risks. It would be a nightmare.

This doesn't make any sense to me. It's either a problem or it isn't unless doctors are willing to admit they perform lots of unnecessary procedures already.

Most doctors will freely admit that they sometimes do unnecessary procedures out of an abundance of caution.

>It's either a problem or it isn't

Try looking at a full body scan and make that call.

> It's either a problem or it isn't

That's vacuously true. The problem is that we humans can't always discern the truth of the situation. Not even doctors.

They perform a lot of unnecessary procedures. It's well known and not a conspiracy. Sometimes they do it to make money. Other times it's so they don't get sued. Scan the comments here and you'll see sentiments of people upset that their doctor didn't do a procedure and blaming them for adverse health problems.

Let's say you have a mass at the head of the pancreas that is incidentally detected on a scan. Let's further stipulate that the mass does not have anything that clearly indicates that it's pancreatic cancer - but also does not have signs of being a benign cyst. What do you do about it?

Well, you can do nothing, and just follow it up with repeat scans. Or you can go to surgery. Both cost money and time. But if you have pancreatic cancer, aggressive resection is essentially the only chance of a cure. So both patients and doctors are highly incentivized to go straight to surgery. If it turns out, once the mass has been resected, that it's actually a completely benign tumor, was the procedure really "unnecessary"? Because taking a piece of it and looking at it under a microscope is the only way you'll ever know for sure.

Would you like to try live your life normally with that sword of Damocles hanging over your head? I wouldn't.

This is a better philosophical test than the trolly problem.

Your argument implies that having less information is better. Why?

Because sometimes it is. This is widely understood in health epidemiology. There are no screenings or interventions with zero risk: https://www.bmj.com/content/352/bmj.h6080

Information’s value is based on the value of decisions made from that information.

That’s how procedures can have negative value.

This isn’t even just the cost of storing and retrieving useless info. Sometimes decisions get harder the more irrelevant information you have.

So it’s better to act on less information than to get more bad information.

In the case of body scans, the harm comes from more expensive followups or treatments to rule out false positives.

Most things that have net value from screenings are already recommended so if it saved lives, it’s usually already recommended until new research changes the calculations.

> In the case of body scans, the harm comes from more expensive followups or treatments to rule out false positives.

That's not a problem with body scans, it's a problem with the medical system overreacting to the new information.

Avoiding scans is a distraction from the real problem of the medical system not knowing what to do with the new information of a full-body MRI scans.

It’s a problem with getting body scans.

I think it’s useful to look at the overall behavior change that happens with a thing.

I mean I guess lung cancer isn’t a problem with cigarettes, it’s a problem with people smoking lots of them.

Body scans don’t provide useful information due to their false positives. Maybe eventually this gets better, or it may be a limitation of this type of scan.

But I’ve talked with many doctors who say that MRI just isn’t actionable enough due to the limitation of imaging.

But for our purposes, full body MRI scans, done without a specific cause, are net harmful and should be avoided.

I mean, it’s legal and people can spend their money on whatever they like. But people spend money on all sorts of unproven “health” expenses. I don’t think MRI scans are as bad as homeopathy, but it’s in that same area.

This, of course, is different from diagnostic scans. MRIs are valuable tools and it’s not the scan is bad, just how they are used sometimes.

So a potential problem is that they are too low resolution? and that is something specific of full body MRI scans?

I still disagree with the "net harmful" argument if it relies on patient's psychology. People may overreact at first out of ignorance, but we learn to stop worrying after being generally exposed to a new thing

Agreed, I think people imagine these scans are a magic Star Trek tricorder. These aren't magic machines that diagnose exact things precisely. Most of the time you're looking to diagnose something you already know about, different situation and different uses altogether.

It’s funny how people seem to assume that information is perfect and it’s just about discovery. And people trust numbers even if they don’t know the veracity.

I remember people reading Covid case counts and taking shifts as having meaning at the local level that just doesn’t exist. A shift of cases from 50 to 100 and back to 50 over two weeks is much more likely to just be a reporting backup than some one week spike. The data quality especially in small areas is so variable it’s not actionable. But people would watch as if it mattered.

There’s some things like temperature that are accurately measurable, but most things have a lot of context that only lets you know how to use things.

I know if I could get a report on my body that said something like:

    Death by heart disease
    - 50% chance by age 65
    - 99% chance by age 87
    
    Death by cancer
    - 50% chance by age 67
    - 99% chance by age 78

    Death by stroke
    - 50% chance by age 73
    - 99% chance by age 92

... and so on, and it was fairly accurate: It would make me a nervous wreck for the rest of my life. Nothing good would come of this knowledge. Maybe I'd exercise more.

This is absolutely true of many personality types! But not others. (And there are perhaps some marginal temperaments that could, with more practice & supporting culture, become more sanguine about any worrying-but-non-terminal interim/probabilistic estimates.)

So, rather than denying the info from those who could use it without becoming a 'nervous wreck', maybe the optimal policy would be to offer it based on a psychological pre-screen?

> maybe the optimal policy would be to offer it based on a psychological pre-screen?

or have the other ones seek therapy

You cannot pre-screen for whether or not someone will be affected by receiving a report that they are 68% likely to die by heart disease in the future. That's illogical and quite frankly insane.

Of course you can. It's not a unique tendency uncorrelated with all other aspects of personality/temperament/anxiety. There are expecially emotional people, especially neurotic people, especially gloomy people – and their opposites, who remain cold in the same situations that freak others out. And it's largely, though not perfectly, predictable.

Many people will have enough of a self-conception – like the author of the post to which I replied – that they will have a fairly accurate idea of how they'll react.

Any many "in the middle" might, by considering, rehearsing, & precommitting to moods/strategies beforehand, helpfully prep/prime themselves to more-calmly handle whatever probabilistic results come back.

Those too concerned, by their own self-knowledge or that of experts/pretests, can skip the risk of "knowing too much" via an elective test, while others can proceed, knowing the risks, with informed consent.

Also majority can't handle truth, even it's about their body. Humans dont follow logic, they're emotional.

You can already get numbers like that. Put your cholesterol and blood pressure values into this one and see!

https://www.mayoclinichealthsystem.org/locations/cannon-fall...

The artificially rationed – & frankly abusive – traditions for training & licensing doctors in the US makes me heavily discount "the general opinion among doctors" as a guide to what's systemically-optimal.

They've been strongly filtered for obedience & conformity to a deeply-broken system.

here, take this pill, you'll soon forget your worries (and get new ones)

Presumably if we are doing full body scans regularly on everyone, we would learn about anomalies that shouldn't be a concern.

maybe radiology has so many scans of various types over the last 50+ years that there is already a sense for how weird "normal" can be.

Reply on news.ycombinator.com