4 ms·

It seems to me that doctors are the only people with the opinion that testing doesn't help people. Is this taught? Either I test positive, and this gives me a

It seems to me that doctors are the only people with the opinion that testing doesn't help people. Is this taught? Either I test positive, and this gives me a treatment program or more knowledge about the situation - and how I can manage it physically and emotionally. Even given a chance of testing false positive, it allows me to do more appropriate research and weigh up the risks and benefits - something I don't have without that test. or I test negative, and I gain reassurance that I don't need to worry about this risk. The worst case (in my view) is a false negative - but that's the same position as I was in without a test.

neaden 7y ago
I'm an epidemiologist and I think excessive testing doesn't help people, so it is definitely not just doctors. False positives can do real harm, including killing people if it leads to unnecessary treatment like surgery. You need to think about a tests Postive Predictive Value and Negative Predictive Value. The PPV is the number of people who get a positive test who have the condition divided by the number of people who get a positive test who do not have the condition. When the PPV is low the test can provide functionally no information. Now the important thing to remember is that the PPV changes based on the prevalence of the disease in the population, so a given test could be good for some people and bad for others, all depending on their base risk.
- learnstats2 7y ago
  Yes - you should always think about the risk of a false positive when considering a treatment (or, equivalently, the PPV). That risk hasn't vanished at the point of testing. If you know the risk of a false positive and the benefit and risk of treatment, then you have the option of reducing risk: choosing an option with increased risk is a mistake, but a mistake in judgement, not a mistake due to the test.
nosianu 7y ago
You assume that given a positive result you can gain certainty, but that is not so. Medicine is a statistics game, even for individuals with test results, and even when adding even more tests on top you don't turn it into certainty. What is abnormal for somebody else may be normal for your body, and trying to do something about it may actually be worse. Weighing risks and benefits sounds easy, but it only really works for doctors (looking to optimize their overall success for all their patients) rather than for individuals - for large numbers. Plus, it's always based only on the "known knowns". There is a huge amount of "known unknowns", and the amount of "unknown unknowns" is unfathomable (thanks Rumsfeld, famous American philosopher, for popularizing those terms). What is "normal" is not based on a complete understanding of every individual as a complete system but the parts are compared among individuals after abstracting away the context that they are in. So knowing what is "normal" or not does not give you nearly as much information as you may like. Sure, your method works for a lot of cases - but it doesn't work or not very well for orders of magnitude more.
toast0 7y ago
I'm not a doctor. I just write software. Not doing unneeded tests has parallels to not doing unneeded logging. The costs may be low, but they're not zero. Sure -- if you have a problem, it might have been nice to have screened for it earlier, or to have logged more data for that flow; but if a problem is rare, the costs for doing all the tests or all the logs adds up. In medicine specifically, we often read about interventions that turn out to have not been the greatest idea. Avoiding or postponing interventions in the absence of symptoms seems more prudent than seeking out more interventions in general. There are some things where widespread screening makes sense, and you shouldn't ignore symptoms or family history, of course.
- learnstats2 7y ago
  The parallel here would be running a test, not writing a test. If the test [likely] takes up more resources to run than the [likely] benefit, then don't run the test, I agree. I wouldn't count that as a diagnostic test at all, though.
michaelbuckbee 7y ago
Here's a concrete example from my own life. I've had a very minor and benign heart condition my whole life. Switching doctors after I moved across the country they're like: "We should do an echocardiogram to just check that it hasn't changed." They do the ECG and find a really crazy high blood pressure in my heart so they think I have pulmonary hypertension. This kicks off: - A sleep study - Lung capacity study - Restricted travel (they didn't want me to fly - which was a problem with my consulting work) - A full-on cardiac catheter procedure - which is outpatient, but still a procedure where they put you under twilight sedation and insert large metal sheath into your groin (femoral artery) Easily a dozen doctors visits, thousands of dollars in additional tests, and multiple weeks of time altogether. Also, non-trivial risks of infection and complication from the procedure. On top of that, it's a mental thing. Every twinge of a muscle in my chest would cause a spike of concern, I'd sit around worrying about my heart inflating and bursting, etc. All to (thankfully) find out that definitively 100% that the ECG was a false positive and I'm perfectly fine. One kicker to this though is that I was initially rejected for health insurance at a new job (pre ACA) as I'd had this procedure. It took patiently explaining over and over again to insurance people the above before they would accept me.
- y-c-o-m-b 7y ago
  I'd like to see some data on the rate of false-positives for this type of scenario. I can only assume your situation was an edge case and that were it indeed a serious problem, you would be singing a different tune. Clearly there is some benefit to using ECG as a diagnostic tool to save lives, so your scenario doesn't necessarily negate the ones that are truly cardiovascular problems. Not that you don't have a valid argument (you do and thanks for sharing btw), but it's a matter of weighing the total outcome across the board I think.
- vageli 7y ago
  I had a hyphema of my retina. Each time I visit a new doctor and they check my pupil dilation, I can tell they are disturbed when they move to that eye and it does not react. I've had doctors express alarm and tell me they were going to run tests on my head. Now I just tell them before they inspect me and avoid the whole thing altogether.
petercooper 7y ago
It's not even just false positives. Real positives can also throw up huge dilemmas. For example, let's say it had been discovered during my scans (which were taking place for unrelated reasons) that I had a minor aneurysm in my brain that had a roughly annual 0.2% chance of rupture. Let's then say I was given the choice of having an operation with a 10% mortality rate. What do I do? This sort of decision is really tough, because I could easily live another 50 years with such an aneurysm, but the terror of having something like that in your head might scare you into taking a risky operation you logically shouldn't take. Sometimes ignorance of minor problems is a good thing. (Note: These stats above are plucked out of thin air to make a point, but I know of real cases where the numbers have fallen this way.)
- learnstats2 7y ago
  If you're over 35 then you have at least 0.2% chance or so of death, sorry to say. Knowing what might kill you might give you the chance to deal with it, or choose how to live with it.

neaden 7y ago

I'm an epidemiologist and I think excessive testing doesn't help people, so it is definitely not just doctors. False positives can do real harm, including killing people if it leads to unnecessary treatment like surgery. You need to think about a tests Postive Predictive Value and Negative Predictive Value. The PPV is the number of people who get a positive test who have the condition divided by the number of people who get a positive test who do not have the condition. When the PPV is low the test can provide functionally no information. Now the important thing to remember is that the PPV changes based on the prevalence of the disease in the population, so a given test could be good for some people and bad for others, all depending on their base risk.

learnstats2 7y ago
Yes - you should always think about the risk of a false positive when considering a treatment (or, equivalently, the PPV). That risk hasn't vanished at the point of testing. If you know the risk of a false positive and the benefit and risk of treatment, then you have the option of reducing risk: choosing an option with increased risk is a mistake, but a mistake in judgement, not a mistake due to the test.

nosianu 7y ago

You assume that given a positive result you can gain certainty, but that is not so. Medicine is a statistics game, even for individuals with test results, and even when adding even more tests on top you don't turn it into certainty. What is abnormal for somebody else may be normal for your body, and trying to do something about it may actually be worse. Weighing risks and benefits sounds easy, but it only really works for doctors (looking to optimize their overall success for all their patients) rather than for individuals - for large numbers. Plus, it's always based only on the "known knowns". There is a huge amount of "known unknowns", and the amount of "unknown unknowns" is unfathomable (thanks Rumsfeld, famous American philosopher, for popularizing those terms). What is "normal" is not based on a complete understanding of every individual as a complete system but the parts are compared among individuals after abstracting away the context that they are in. So knowing what is "normal" or not does not give you nearly as much information as you may like. Sure, your method works for a lot of cases - but it doesn't work or not very well for orders of magnitude more.

toast0 7y ago

I'm not a doctor. I just write software. Not doing unneeded tests has parallels to not doing unneeded logging. The costs may be low, but they're not zero. Sure -- if you have a problem, it might have been nice to have screened for it earlier, or to have logged more data for that flow; but if a problem is rare, the costs for doing all the tests or all the logs adds up. In medicine specifically, we often read about interventions that turn out to have not been the greatest idea. Avoiding or postponing interventions in the absence of symptoms seems more prudent than seeking out more interventions in general. There are some things where widespread screening makes sense, and you shouldn't ignore symptoms or family history, of course.

learnstats2 7y ago
The parallel here would be running a test, not writing a test. If the test [likely] takes up more resources to run than the [likely] benefit, then don't run the test, I agree. I wouldn't count that as a diagnostic test at all, though.

michaelbuckbee 7y ago

Here's a concrete example from my own life. I've had a very minor and benign heart condition my whole life. Switching doctors after I moved across the country they're like: "We should do an echocardiogram to just check that it hasn't changed." They do the ECG and find a really crazy high blood pressure in my heart so they think I have pulmonary hypertension. This kicks off: - A sleep study - Lung capacity study - Restricted travel (they didn't want me to fly - which was a problem with my consulting work) - A full-on cardiac catheter procedure - which is outpatient, but still a procedure where they put you under twilight sedation and insert large metal sheath into your groin (femoral artery) Easily a dozen doctors visits, thousands of dollars in additional tests, and multiple weeks of time altogether. Also, non-trivial risks of infection and complication from the procedure. On top of that, it's a mental thing. Every twinge of a muscle in my chest would cause a spike of concern, I'd sit around worrying about my heart inflating and bursting, etc. All to (thankfully) find out that definitively 100% that the ECG was a false positive and I'm perfectly fine. One kicker to this though is that I was initially rejected for health insurance at a new job (pre ACA) as I'd had this procedure. It took patiently explaining over and over again to insurance people the above before they would accept me.

y-c-o-m-b 7y ago
I'd like to see some data on the rate of false-positives for this type of scenario. I can only assume your situation was an edge case and that were it indeed a serious problem, you would be singing a different tune. Clearly there is some benefit to using ECG as a diagnostic tool to save lives, so your scenario doesn't necessarily negate the ones that are truly cardiovascular problems. Not that you don't have a valid argument (you do and thanks for sharing btw), but it's a matter of weighing the total outcome across the board I think.
vageli 7y ago
I had a hyphema of my retina. Each time I visit a new doctor and they check my pupil dilation, I can tell they are disturbed when they move to that eye and it does not react. I've had doctors express alarm and tell me they were going to run tests on my head. Now I just tell them before they inspect me and avoid the whole thing altogether.

petercooper 7y ago

It's not even just false positives. Real positives can also throw up huge dilemmas. For example, let's say it had been discovered during my scans (which were taking place for unrelated reasons) that I had a minor aneurysm in my brain that had a roughly annual 0.2% chance of rupture. Let's then say I was given the choice of having an operation with a 10% mortality rate. What do I do? This sort of decision is really tough, because I could easily live another 50 years with such an aneurysm, but the terror of having something like that in your head might scare you into taking a risky operation you logically shouldn't take. Sometimes ignorance of minor problems is a good thing. (Note: These stats above are plucked out of thin air to make a point, but I know of real cases where the numbers have fallen this way.)

learnstats2 7y ago
If you're over 35 then you have at least 0.2% chance or so of death, sorry to say. Knowing what might kill you might give you the chance to deal with it, or choose how to live with it.