No one is an expert on this because everyone has emotional biases and different experiences. Even if mid-levels aren't performing the critical parts of the operation in surgical subspecialities, they can reduce the number of physicians needed.
Let's say a hospital has 3 neurosurgeons each making $800k. Replace one of those neurosurgeons with 4 PA's each making $125k. Each PA gets their own OR with the physician rotating in during "critical" parts of the operation. 2:1 supervision, half of what many (most?) anesthesia practices use. Even if these 4 mid-level OR's only function at 80% of the efficiency of a physician: 3 physician OR's at 100% = 300% billed versus 4 PA OR's at 80% = 320% billed. The hospital is also saving $300k in salary. Instead of taking Q3 first call, the two physicians remaining can now take Q2 second call (PA's on first call).
This sounds crazy, is currently illegal, and would be insanely unsafe, but it would make administrators a lot of money. Laws can change, especially if it means a lot of money can be made. Just because it is unsafe and unethical, doesn't mean it won't happen in the future. Just look at anesthesia or EM.
Also, the more this type of stuff happens, the easier is becomes. The more physicians that are replaced by mid-levels, the worse the job market gets for physicians, and the more BS they will put up with to stay employed.
Inb4 "BuT tHeY wON't bE AbLE tO AFforD MalIprACtIcE!"
Hospitals have so far been able to eat any increased malpractice costs from using mid-levels in the place of physicians. And not in a small sample size or short time scale. Tens of thousands of mid-levels being hired over 10+ years has proven that cost savings > malpractice costs.