Is Trainee Participation Really Associated With Adverse Outcomes in Emergency General Surgery?
We discussed the article by Kasotakis et al1 in our weekly journal club meeting and received many compelling comments. The authors perused the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database and utilized an impressive coarsened exact matching algorithm to study the impact of trainee participation on patient outcomes in emergency general surgery procedures. The authors conclude that trainee participation is an independent poor prognosticator for 30-day adverse outcomes after emergency general surgical procedures, on the basis of multivariate regression analysis after adjusting for risk factors and operating time. Although the study is interesting and has potential implications on resident training, there are several limitations that must be considered on the background of a provocative title.
Many of the limitations stem from the fact that the data were secondary and not collected to answer the study hypothesis. The lack of information on the level of resident supervision in the operating room and the experience of the trainer may account for a significant bias and confound the study results. In fact, Borowski et al2 reported no significant difference in operative mortality or short-long term survival rates after colorectal cancer resection in a risk-adjusted analysis on more than 7000 patients, who were either operated by unsupervised trainees or those supervised by consultants.
During our meeting, we also explored potential confounding factors contributing to extended operative time when trainees operate. They are often pressurized while under supervision and likely to work more meticulously and with less freedom, which can slow down the thought process and eventually the operative procedure. In addition, trainees may be more prone to being interrupted when consultants have an alternative preferred method of performing a particular procedure. Team members in theaters may be less receptive to trainees than the consultant, for example, delay in providing the instrument required or connecting the diathermy. When trainees operate independently, they are more likely to be assisted by more junior residents or nurse practitioners, thereby inadvertently increasing the operating time. Furthermore, although the authors acknowledge that the increased operative time with seniority was likely a reflection of self-selection of more complex cases among senior trainees, this was not objectively quantified.
Similarly, seniority of anesthetic staff was not included in the analysis. It is likely that when procedures were performed by surgical trainees, anesthetic trainees rather than their senior counterparts were involved. As discussed in the article, it is difficult to standardize intraoperative management of fluids and the decision for blood transfusions. The study design failed to account for the potential bias when including figures such as the apparent increased intraoperative blood transfusions when trainees operate. These numbers could have been skewed because of the presence of anesthetic trainees, who might be more cautious than their senior counterparts. Perhaps a better indicator would be to evaluate intraoperative blood loss.3
The article also reports longer intraoperative time (from having trainee involvement) with poor postoperative outcomes. We feel that this was not a simple linear relationship. Data were collected from more than 400 institutions across the United States, with undoubtedly varying quality of postoperative patient care, hospital facilities, and resident training. In the midst of ever-growing popularity for enhanced recovery programs,4 the study makes no mention of whether or not such programs or any form of standardization was incorporated in the patients’ postoperative care. To adjust for this significant source of bias, the article could have compared surgical outcomes in patients operated in a teaching versus nonteaching institution.
The ramifications of this study could be immense.