1. We don't measure the patients metabolism of a drug at all. Individuals bodies process drugs differently and have different side effects. It seems that differences in reactions could be related to different metabolic pathways being used, resulting in more or less toxic byproducts. Could a urine or blood test a few hours after the first dose be an effective mechanism for screening for likelihood of side effects based on the metabolic byproducts detected? Could this test be effective at detecting these byproducts at dosages below what would cause side effects? Could this be incorporated into a "home drug test" style urine dip that patient could do themselves at home without returning to doc?
2. Dosage is too one-size-fits-all. AFAICT, the drug I was given is usually prescribed at the same dose to slightly-underweight me as they would a larger patient double my weight. For these fixed-dosage drugs, it seems like lower-weight patients will have higher risk of side effects and higher-weight patients will have higher risk of dosing at sub-therapeutic levels.
3. Drug half-life is assumed to be nearly constant across the population. Even more than dosage, many drugs have a constant X times per day, but only dosage is adjusted. For people with atypical metabolism of that drug, the effective duration could vary drastically. Patients with slow metabolisms could develop side effects at low dosages if taken over a long enough period.
I'd love to see pharma move towards a model where patients are started at a sub-therapeutic level to test for allergy, then have a urine test to screen for toxic metabolites, then pills are portioned in to small enough increments that the patient can gradually titrate up the dose: start at 1/100th a dose, if no problems go to 1/10, then 1/5, then 1/4, then 1/2, then 3/4, then full dose. If at any point the patient feels like its 'working' they pause dose increases. I really wonder if I had been on half the dosage or frequency if I would have had all of the therapeutic benefits without the gnarly side effects.