The problem is obviously a type error (not different from the use of a Social Security Number as a National Identification Number).<p>The government will pay for medication “that works”.<p>It determines that medication is one “that works” by seeing if the FDA approves it.<p>The FDA doesn’t want to always be too slow to approve medication so sometimes it wants to approve medication as “doesn’t hurt”.<p>Ultimately, the problem is that American policy relies on categories rather than spectra. My personal belief is that this is because the bureaucrats and policy makers are taking advantage of common American knowledge of categories vs spectra. So it needs some categorization mechanism and since one exists, they use that one. Presumably, out there is a bureaucrat with the same impulse as the engineer reading your PR who says “We already have a function to calculate radians to degrees. Why don’t we use that?”<p>In comparison, the NHS will measure how much to spend on a treatment based on an adjusted QALY measure (that boosts the value of the last few months). X QALYs are worth Y pounds.<p>Almost all American policy, by comparison, is category-based.<p>Other examples:<p>- immigration: America does categories (so either family or employment or refugee) while other nations use scores (if you both have family and employment you get some boost from both rather than the US max(categories)) system<p>- social security: Thresholds on disability and poverty are cliffs resulting in marginal income improvement resulting in marginal post-support income decrease. These should scale down, not drop off based on in-category/out-category