The exam is tough and always has been. There is no shame in failing, but it would be very unusual to drop from being a good performer on the ITE's to failing the advanced exam. It is rare, but I have seen it.
With regards to the content of the exam, every year, it varies. Some years there is heavy OB and the next year almost none. Some years there are 5 questions on ankle blocks and the next year none. My experience and things I have heard suggest that this is an intentional testing method. It allows separating out those candidates that guessed a question correctly. Chances are, if you guess on a question about an ankle block and get it right, you will likely not guess correctly 5 times when they add in different nuances to the question. Therefore, this allows further separation of the candidates based on how much mastery of the topic exists. Each year, people say, where were all of the OB questions they typically ask (or cardiac, or pain or whatever the topic)? They have 200 questions and they take a sampling of topics. Some (most?) are straightforward general knowledge topics. You either know it or you don't. They are tricky questions though. Often, someone will miss a question multiple times, thinking they have always gotten it correct, because they did not understand the nuance of the question that made their answer incorrect and the other one correct.
Other questions are part of a knowledge set that gets tested 5 different ways to separate out the candidates based on their mastery of certain subject matters. Obviously, they can only do this on a handful of topics, so that is where the "why were there so many questions on topic x, y, or z?" comments come from. I heard this explanation about 15 years ago and it has seemed to hold up over time.
Additionally, a certain subset of the questions are being trialed to check their validity. These are typically thrown out of the final scoring. The questions are also scrutinized closely if they are frequently missed, just to make sure they were not mis-keyed or that their might be two valid answers so that they should be double-keyed. Sometimes, a really bad question gets past all of the vetting and editing and makes its way onto the actual exam. If it is problematic, it may just be thrown out, thereby changing the denominator. If a couple of questions are thrown out, and they happen to be the ones you got correct, it can make a difference if you are right on the line between passing and failing.
They also do standard setting exercises to further vet the questions and the exams as a whole. They will invite diplomates (typically Program Directors) to come to a site and take the exam to determine if they are testing the appropriate topics in a fair manner. Each question gets examined and the group decides if it is fair to ask the question and (for example, the Basic Exam) if it is fair to expect a person of that level of training to possess the knowledge to answer the question correctly. At the end of the exercise, they examine the test as a whole and look for break points where candidates are separated with as little gray area as possible to create a standard for passing the exam. They perform this standard setting exercise every few years, as do all standardized exams, I believe. My experience is that they try to avoid situations where a single question separates someone who failed from someone who passed.
They also seem to maintain a standard performance on the exam as the bar, not just a percentage pass rate such as 90%. Back in the late 90's, the written exam pass rate dipped into the low 70's percent and the oral board pass rate dipped into the high 50's, based on data released. That was the time when the NY Times article suggested that graduating residents could not find work and only 43 US grads matched into anesthesiology (1996, I believe) in the NRMP primary match. The rest of the spots were filled with warm bodies who would not have made the cut in other years. The exam pass rates reflected a lower quality candidate for a few years after, as the bar that had been set was not adjusted.
A given is that there will always be some obscure hyperbaric medicine question and at least a couple of toxicology questions.