The reason is the same as why you require higher standards of statistical evidence when you don't have a strictly defined hypothesis before an experiment: if you go looking for something in noisy data, you'll probably find it.
Lab tests have significant margins of error. Just by chance, if you go out and get a bunch of lab tests, you'll probably find something "wrong". Unless you're knowledgable about interpreting the results -- and unless you are a practicing medical doctor, you are highly unlikely to be knowledgable, even if you think you are -- this can/does lead to catastrophically bad outcomes.
(And now, for an ironic aside: the median HN user is extremely quick to attack valid science for the reason that "most results are wrong", while also complaining loudly when the medical establishment wants to restrict access to clinical tests with a large margin of error. Go figure.)
For things like STD tests it raises the barrier and cost of frequent testing. We need a more fine grained policy than all blood tests require a physician.
They don't. You can get cheap/free STD tests from non-physician providers. You can also buy STD tests at the drugstore, over the counter. But like it or not, anyone who collects biological samples from strangers' is going to require some training and licensing.
Moreover, the interpretation of an STD test is not as straightforward as you might think. False positive rates can be high for the cheaper tests, and there's natural variability in things like antibody titers. People also neglect that timing matters -- if you test too soon after infection, even the most sensitive tests won't work. Or they fixate on one particular scary illness (e.g. HIV), and neglect testing for others that are less well-known, but more common (e.g. Chlamydia). Some of these problems can be mitigated. Others can't.
Regardless, having someone trained to interpret tests is a good idea for just about everyone. Even the cheap/free/OTC tests come with that service, at a minimum. It's a bad idea to just allow people to pick and interpret their own tests, even for the "simple" ones, like STD tests.
There is another issue I've read about, which is that outside the context of a medical diagnosis, even a very accurate test would still be relatively unhelpful because it can be measuring natural variability. (E.g., testing some metric associated with prostate cancer in young men/women — prostate cancer only appears in older men, so testing young people is mostly useless.) Whereas if you're already potentially sick and go to a doctor, the relevant tests then have more predictive power.
In reality this doesn't mean much because I can order a blood test online and a doctor's approved receipt will be available in under a minute to take to a local lab. I doubt the doctor looked at anything except his bank account. And this process is legal in most US states.
I used to do this to supplement my doctor's own tests because of my own curiosity and concern. So I just paid for it myself without insurance. For example a standard panel of 26 common tests costs $55.
I don't get it. What is the average person going to do themselves with the results of a simple blood test?
Car analogy: is there any point in taking your car to a garage and having them look it over and just make a list of the status of common components on the car, without any (vehicle specific) interpretations? As in "you have 5/16 inch wear difference between the front left and right brake rotors, the steering toe-in is 1.3 degrees, and the alternator is providing 13.2 volts".
A better question is would you drive a car without a working gas gauge?
Blood tests are flagged as abnormal if they exceed the 95% confidence interval. So there is a possibility of it happening by chance. Or you may get some abnormality due to minor issues like a recent cold or injury. But then you just need to start a conversation with your doctor to get clarity.
Did you know your kidney health could be down to 10-20% of capacity and you would not even feel a thing wrong? And something happened to me in 2 intervening years of tests for that exact scenario to happen. I doubt most people have tested anything other than their cholesterol and BP in the last two years. If anything, we should be checking our body health more than our car because we can always replace our car.
Lab tests have significant margins of error. Just by chance, if you go out and get a bunch of lab tests, you'll probably find something "wrong". Unless you're knowledgable about interpreting the results -- and unless you are a practicing medical doctor, you are highly unlikely to be knowledgable, even if you think you are -- this can/does lead to catastrophically bad outcomes.
(And now, for an ironic aside: the median HN user is extremely quick to attack valid science for the reason that "most results are wrong", while also complaining loudly when the medical establishment wants to restrict access to clinical tests with a large margin of error. Go figure.)