> 1. Drug candidates seeking approval must show non-inferiority to gold standard, meaning they are not worse than the top choice for the same treatment. This non-inferiority is quite easy to game, as we all know statistics are fungible and may be manipulated. If all I have to show is no significant difference, and I can use one of 10 different outcomes or surrogates, well my job got a LOT easier than showing superiority on the primary outcome of interest.

Before they to the test they have to declare a primary outcome to avoid this issue. And most diseases just don't have that many outcomes relevant outcomes you can test for.

> 3. Doctors and geneticists should come together and provide genetic testing to ensure that patients and their drugs are appropriately matched. This could be done in cancer, for instance, as you can genotype the cancer and provide a drug cocktail specifically for it. Another instance would be depression, as some drugs do not work for people with certain neurotransmitter genotypes. Physically, this means that each of us may have slightly differently shaped receptors that different drugs bind to differently. This explains a lot of the difference in drug responses, whether it be SSRIs or alcohol or cannabinoids, people simply have different receptors/biochem.

They're trying, and they're failing hard. Precision medicine has been pretty useless outside of cancer.