Image by scotth23 on Pixabay
In FDA’s recent announcement on robotic surgery for various cancers, the agency raised the little-discussed issue of surgeon skill and the role of experience in building that skill. In general, we would expect that reasonably talented surgeons would get better as they performed more procedures, even though they might start off a little bit rough. This improvement would be expected to be reflected in better outcomes and fewer complications. We might also expect that the improvement would plateau after some number of procedures, with that plateau preferably being at a decent level of performance. What has only been rarely studied is how many procedures of a particular kind must be done before a surgeon reaches an acceptable level of proficiency. Also little studied is how much variation there is among surgeons. There are the related questions of outcomes for early patients and what kinds of disclosures should occur.
In addressing the robotic-surgery-for-cancer question, FDA recommended that patients ask surgeons about their training, experience, and patient outcomes with robotically assisted surgical device procedures; how many robotically assisted surgical procedures like the patient's a surgeon has performed; and possible complications and how often they occur. The latter question has two parts, one being complication rates across all surgeons and the other being the complication rate for this surgeon in particular. We would all hope that our surgeons were above average. An obvious problem with this question set is that asking these questions can be awkward, and what constitutes an acceptable answer is not known. How much training of what type is necessary to become reasonably proficient? Similarly, there is no best answer for how many surgeries previously done, but if we were the patient, we might want it to be more than zero, unless we were into being a pioneer. But what is a better answer than zero on numbers completed in order to have faith in the surgeon’s ability? Five, ten, 100? A different question is consistency. Does the surgeon have good and bad days, and is there any way to pick a day more likely to be good?
FDA also addressed surgeons. The agency recommended that a surgeon take training for the specific robotically assisted surgical device procedures they plan to perform. The alternative of teaching yourself does not seem attractive. The nature and amount of such training is not elaborated on nor is whether it might include simulations as well as actual patients. Moreover, there is a fundamental problem here. If FDA is warning against such surgeries, how would surgeons get actual patient training unless they and their mentors were ignoring FDA’s warning? FDA further suggested that surgeons talk to their patients about their experience and training. How many surgeons are going to tell a patient that, for example, they have never done this before, but they think they can pull it off? FDA also recommended the establishment of patient registries, which the agency says may help characterize surgeons' learning curves.
This reminds me of a personal experience from the early days of laser surgery. A local (small town) surgeon suggested a fairly extensive cut-and-sew surgery. We sought a second opinion at a major medical center in which a much more superficial laser procedure was recommended. Upon telling the original surgeon that we were going to the big city for the laser surgery, the original surgeon said, “I’m getting a laser next week.” Despite our career commitment to education, we declined to be among the first to be lasered locally. But this does perhaps leave an ethical quandary. I want my surgeon to have a good deal of experience, but I want him or her to get that experience operating on someone else.
It is hardly surprising that in a wide range of endeavors, skills increase with experience. There is no reason to expect surgical skill to be an exception. This might be especially true in robotics for which FDA noted that such systems are novel and complex, and in many cases not yet subjected to clinical trials. What we don’t know about the learning process is how many cases it takes to get good enough.