Viewpoint: The Surgical Learning Curve

 

By William A. Hyman, ScD

It should come as no surprise that the ability to do a particular surgery is likely to improve over some number of early attempts. A surgeon’s skill could be evaluated in part by measuring his or her complication rate for a given procedure and watching it decrease to a plateau rate that is somehow deemed “acceptable.” What has seldom been studied, however, is how many surgeries it takes to reach that plateau. A rare example is provided by recent work from Sweden, reporting that for certain cancer surgeries, it takes surgeons 15 times to reach stable results and 60 to reach optimal long-term survival (Markar, Mackenzie, Lagergren, Hanna, & Lagergren, 2016).

This dynamic poses challenges for most if not all surgeries and procedures: Who gets to be—willingly or unwillingly—among the early patients in a surgeon’s learning curve? Should informed consent include telling patients that they are among the first to receive this surgery from this surgeon? In addition, there are actually three separate assessments for many surgeries: 1) how many the surgeon has done, 2) the surgeon’s short-term complication rate, and 3) the surgeon’s long-term complication rate. These questions become even more challenging when patients are among the first anywhere to undergo a procedure or receive a new implant—for example, when there was no clinical trial before a device used in the surgery was marketed, thereby placing patients on the receiving end of that device in a de facto but unidentified trial. In fact, such patients may be worse off than those in real clinical trials because follow-up, multicenter monitoring, and reporting will be ad hoc at best.

What represents sufficient training?

One way to approach the surgical learning curve would be to increase the number of supervised surgeries surgeons must perform before they can conduct aparticular surgery solo. The requirement might include a number of “hands-off” observations, followed by some number of assists and then some number of supervised leads. Simulation, cadaver, and animal surgeries might also be included. Defining the parametersof each phase would prove challenging, with the appropriate numbers likely dependent on the type of surgery. Comprehensive training should also include simulations of responses to known contingencies rather than simply assuming that things will go well.

That type of regimen would far exceed the brief periods of training that now are common. Admittedly there are logistical, cost, and personal considerations, the latter including the surgeon’s acceptance of the need for a high level of supervised training. Moreover, hospitals may have to play a stronger role by better controlling what types of surgeries individual surgeons are allowed to do, as well as what training and experience are required. While hospitals have long had this role through their credentialing process, the rigor of credentialing is inconsistent and depends in part on whether a particular hospital needs the surgeon more than the surgeon needs the hospital.

Surgeons and physicians are not the only personnel who face learning curves, of course. For other healthcare professionals, the learning curvemay include the level of general skill or proficiency in the use of medical devices. Many devices require regular and multiple uses before the user can operate them smoothly and consistently. Here, too, training and supervision should have specific expectations.

How should a patient choose?

From the patient perspective, when surgery is recommended, is it acceptable to ask the surgeon how many of these surgeries he or she has performed and what the surgeon’s complication rate has been? This seemingly straightforward question is likely fraught with interpersonal stress; perhaps there are better ways to disseminate such data.

The question of surgeon skill is closely related to the question of whether high-volume hospitals are inherently better than low-volume hospitals. Although high-volume hospitals may employ some relatively novice surgeons, one could expect that patients in general benefit from the hospital’s greater collective experience and better control of who is doing what. Geography is another factor in choosing providers: Is the high-volume hospital and surgeon located near or far, compared to low-volume alternatives? Having lived for a long time in a relatively small community 100 miles from a major city with major hospitals, this was at times an important personal question for me. For many interventions, I opted to make the longer trip.

Beyond the learning curve, there is the rarely discussed fact that performance of many skills, including surgery, falls into three tiers. There are those that are very good at a given skill, those that are moreorless adequate at it, and those that probably shouldn’t be doing it at all. In the case of implants, the manufacturer’s representatives, who are often in the OR, might have direct knowledge of who is in which group. What they should do with such information is perhaps challenging, but there may be an ethical dilemma in continuing to provide products to those who are known to be not very good at using them.

Other medical device users fall into similar categories. We recognize that some users are highly proficient, others are moderately proficient, and others perhaps shouldn’t be allowed to use a given device at all. The moderately proficient might be restricted to patients and situations that require less skill than more acute patients. When hospital personnel become patients, they know who they want and don’t want caring for them. The public, however, does not have access to that information.

For many skills, people need experience to learn, and their early attempts may yield results that are less than desirable. Surgeons need to learn by experience too, and perhaps some patients would want to make that personal contribution to a surgeon’s progress. However, for those patients who would not, finding out where they are on the surgeon’s experience count is not easy. And for every patient who opts out of being a learning-curve data point, some other patient assumes the role instead.


William Hyman is professor emeritus of biomedical engineering at Texas A&M University. He now lives in New York City, where he is adjunct professor of biomedical engineering at The Cooper Union. Hyman may be contacted at w-hyman@tamu.edu.

Reference

Markar, S. R., Mackenzie, H., Lagergren, P., Hanna, G. B., & Lagergren, J. (2016). Surgical proficiency gain and survival after esophagectomy for cancer. Journal of Clinical Oncology, 34(13), 1528–1536. doi: 10.1200/JCO.2015.65.2875