A patient that walks into a doctor's office asking for a certain medicine they heard about on TV distorts the process. After all the patient needs to believe in the treatment and might shop around for the doctor that gives them what they want.
Learning about a product in a medical context can't really happen in an ad. A doctor would have to go to a conference or read journals to figure out the latest science on treatments, and can then go and then choose an appropriate drug. They would probably look for the one that has 20mg of X in it, regardless of brand name or hype.
People need agency in their own medical choices, and knowing there is a drug for your medical condition is very empowering. Especially as we talk more and more about moving to a single-payer system, it's important to make sure patients have the power to question their doctors and explore their own health instead of essentially being dictated to by the state.
There's no way the only possible method of resulting in having informed patients is to put them directly in the middle of a struggle between big companies and doctors. I'd argue that putting doctors in a position of power is terrible (whether single-payer or not), but giving pharma companies' marketing departments the freedom to manipulate people with advertising is also terrible. Two wrongs do not make a right - adversarial systems are poor designs.
Of course the problem is the reliability of the evidence. I recently had to decide whether I wanted to go on a course of some fairly severe drugs with nasty side effects.
The original papers didn't make a lot of sense. The statistical conclusions seemed to be - if not nonsense, then certainly not very consistent.
My doctor said "Well, these are new drugs, so we're still learning about them." I didn't find that reassuring.
The real problem is that big statistical studies don't say much about why side effects happen. There seems to have been far too little research into understanding why/how side effects happen at all, and whether or not it's possible to screen for them.
That would probably require a level of personalised medical screening and dose control that big pharma isn't interested in providing.
Find a natural experiment where two very similar populations, chosen at random, were placed into two different cohorts. I'm sure there's one somewhere in the recent and well documented past.
How can we prove that this is false or that the benefits are smaller than the drawbacks?