[frederickf]

In my totally unauthoritative opinion the diagnostic process has been optimized to find the most likely cause as quickly as possible. This isn't a terrible strategy on the aggregate. It increase the overall likelihood that the most people will get the right diagnosis. But the trade-off is that some people will get the wrong diagnosis. The alternative would be for doctors to spend more time really digging into symptoms and tests to ensure they get it right the first time, every time, but then fewer people would get diagnosed.

This would be okay if doctors would accept the limitations of the system and admit when they don't know the correct diagnosis or are not sure. In that case they would say something like "This is the most likely diagnosis, but it doesn't meet your symptoms, you should seek a specialist" or something to that effect. I think what happens too often is that doctors confuse the most likely diagnosis with the correct diagnosis, possibly because they're unwilling to admit their fallibility.