Imagine there is no regulatory agency saying which drugs are efficacious, and consequently, which drugs could be marketed. (Assume that safety is still regulated.) Pharmaceutical companies would develop and manufacturer medicines and sell them to patients. If the patient felt better or got better, he/she would keep taking that medicine. If the patient didn't get better, had a tolerability issue, or didn't see the value, he/she would stop taking that medicine.
The problem with this, as we all know, is the placebo effect. The patient's perceived net benefit is equal to the real benefit plus the placebo effect. There could be a real effect like reduced blood pressure or increased CD4 cell counts--or there could be none at all. The placebo effect could convince the patient that there's a real benefit even when there is none.
The FDA's approach to get around this problem is to separate the real effect from the placebo effect via clinical trials with large numbers of patients to determine whether the medicine is really working or not. Sometimes this divide-and-conquer solution isn't possible, or even necessary, when the entire patient experience is holistic and subjective.
Consider pain. In clinical trials, the visual analog scale is used frequently to assess a patient's perceived pain level. This is merely an objective measurement of nothing other than a subjective assessment. A patient's assessment of pain is highly individualized and situationally dependent.
"Already, neuroscientists know that how people perceive pain is highly individual, involving heredity, stress, anxiety, fear, depression, previous experience and general health. Motivation also plays a huge role--and helps explain why a gravely wounded soldier can ignore his own pain to save his buddies while someone who is depressed may feel incapacitated by a minor sprain." (Rewiring the Brain to Ease Pain, Wall Street Journal, 15-Nov-11).
The FDA's approach is as follows: (1) See if a drug is "really" efficacious. (2) If so, approve it. (3) Once approved, let patients and physicians experiment with what works the best for them.
For medicines with an entirely subjective response, say, for pain, we could just as easily skip steps 1 and 2 and go right to 3. We could use the huge and efficient laboratory known as the American medical system to let patients and physicians experiment to find what works for them. Doing so would be easier, cheaper, and intellectually honest.