Although this is not specifically germane to the current discussion regarding surgery, I would like to point out that that figure for psychiatry is probably not accurate.
Reading through that article, it's pretty ambiguous as to how they're getting the figures for non-EM specialties. They define their study population as residents in an ACGME-accredited EM program. They describe their methods, but it's unclear whether those methods only relate to their study population or whether those are the methods they used to arrive at the data for all specialties. For our purposes here, I'll assume they used the same methods to assess attrition in psychiatry.
The article states the following:
There are two main ways to view resident attrition: There is attrition from the training program the resident initially enrolled in, and there is attrition from the specialty altogether. For several reasons, we chose the most inclusive definition by counting all attrition statuses, including attrition from one EM program to go to another EM program as well as attrition from the specialty altogether.
This leads me to believe that they assessed the attrition in psychiatry by including any resident who switches programs prior to the completion of their full 4 years of categorical residency in the attrition group. They don't explicitly describe the details of how this was determined for psych residents but my strong suspicion is that residents who fast-tracked into child psychiatry got lumped into the attrition group. That is, these residents were part of a 4-year general psychiatry program but left their program in PGY-4 to start a 2-year child psychiatry fellowship where the first year is counts as their last year of general psychiatry training.
This would inflate the attrition numbers for psychiatry. I don't think that fast-tracking into child is reasonably seen as attrition for many meaningful purposes.
Anecdotally, a 6% attrition rate seems extremely high to me for psychiatry, which is why I strongly suspect that something's off about how they're measuring it.