Customize your JAMA Network experience by selecting one or more topics from the list below.
Schuster KM, Hazelton JP, Rattigan D, Perez JM, Bhattacharya B. Association of Acute Care Surgeon Experience With Emergency Surgery Patient Outcomes and Mortality. JAMA Surg. 2021;156(5):472–478. doi:10.1001/jamasurg.2021.0041
Is the posttraining experience of acute care surgeons associated with patient outcomes in emergency general surgery and trauma?
In this cohort study of 772 patients undergoing an emergency surgical procedure at 1 of 5 US academic medical centers, less surgeon experience was associated with increased return to the operating room and postoperative complications. Mortality was similar regardless of surgeon experience.
Acute care surgeons may benefit from additional mentorship during the initial stages of their career.
Previous studies comparing emergency surgery outcomes with surgeon experience have been small or used administrative databases without controlling for patient physiology or operative complexity.
To evaluate the association of acute care surgeon experience with patient morbidity and mortality after emergency surgical procedures.
Design, Setting, and Participants
This cohort study evaluated the association of surgeon experience with emergency surgery outcomes at 5 US academic level 1 trauma centers where the same surgeons provided emergency general surgical care. A total of 772 patients who presented with a traumatic injury and required an emergency surgical procedure or who presented with or developed a condition requiring an emergency general surgical intervention were operated on by 1 of 56 acute care surgeons. Surgeon groups were divided by experience of less than 6 years (early career), 6 to 10 years (early midcareer), 11 to 30 years (late midcareer), and 30 years or more (late career) from the end of training. Surgeons with less than 3 years of experience were also compared with the entire cohort. Hierarchical logistic regression models were constructed controlling for Emergency Surgery Score, case complexity, preoperative transfusion, and trauma or emergency general surgery. Data were collected from May 2015 to July 2017 and analyzed from February to May 2020.
Main Outcomes and Measures
Mortality, complications, length of stay, blood loss, and unplanned return to the operating room.
Of 772 included patients, 469 (60.8%) were male, and the mean (SD) age was 50.1 (20.0) years. Of 772 operations, 618 were by surgeons with less than 10 years of experience. Early- and late-midcareer surgeons generally operated on older patients and patients with more septic shock, acute kidney failure, and higher Emergency Surgery Scores. Patient mortality, complications, postoperative transfusion, organ-space surgical site infection, and length of stay were similar between surgeon groups. Patients operated on by early-career surgeons had higher rates of unplanned return to the operating room compared with those operated on by early-midcareer surgeons (odds ratio [OR], 0.66; 95% CI, 0.40-1.09), late-midcareer surgeons (OR, 0.34; 95% CI, 0.13-0.90), and late-career surgeons (OR, 1.11; 95% CI, 0.45-2.75). Patients operated on by surgeons with less than 3 years of experience had similar mortality compared with the rest of the cohort (OR, 1.97; 95% CI, 0.85-4.57) but higher rates of complications (OR, 2.07; 95% CI, 1.05-4.07).
Conclusions and Relevance
In this study, experienced surgeons generally operated on older patients with more septic shock and kidney failure without affecting risk-adjusted mortality. Increased complications and unplanned return to the operating room may improve with experience. Early-career surgeons’ outcomes may be improved if they are supported while experience is garnered.
Trauma and emergency surgery contribute to morbidity and mortality in the US. Trauma is the leading cause of death for individuals 44 years and younger with societal costs of more than $671 million per year.1 Emergency general surgery (EGS) accounts for half of all surgical mortality cases in the US.2 Acute care surgeons, who manage these 2 large groups of patients, routinely make major treatment decisions with limited data and sometimes limited knowledge of a patient’s disease process. Because of the anatomic variability of trauma and emergency surgery patients, one would expect that junior attending surgeons encounter injuries and disease processes they have not previously managed. This lack of experience may lead to increased morbidity and mortality. Moreover, overall surgical experience may not compensate for a lack of disease-specific experience.3
In trauma care, decisions early in the management of acutely injured patients have a considerable effect on mortality.4-6 Few studies have examined the association between the experience of the admitting trauma surgeon and outcomes and, to our knowledge, none have evaluated the progressive experience of fully trained trauma surgeons. One study identified trauma management by trainees compared with fully trained trauma surgeons as predictive of increased mortality.5 Another study examining trauma mortality at Florida hospitals attributed 9% of mortality to the treating surgeon, although no characteristics of the surgeons other than volume were measured.7 How trauma surgeon volume, experience, and other characteristics interact to affect outcomes remains uncertain.
Emergency laparotomy is often undertaken without a certain diagnosis in patients with complexities or who are at high risk of mortality.2,8,9 These patients often require management of their acute operative condition in addition to perioperative management of underlying disease.10 Surgeon experience may affect both management and the decision to proceed with operative intervention.11 Once operative intervention is undertaken, the most likely reason for errors may be surgeon inexperience.12,13 There is even greater surgeon-level variability in mortality for EGS cases, as high as 32.7% in a 2018 study of Florida hospitals and surgeons.14
There are multiple studies comparing experience with outcomes in a variety of specific procedures. In vascular surgery, small improvements in stroke and mortality were demonstrated over the first 100 carotid endarterectomies performed posttraining.15 In colorectal surgery, conversion of laparoscopic to open surgery, operating time, and cumulative J-pouch survival all improved with surgeon experience.16,17 A study on pancreaticoduodenectomy demonstrated that pancreatic fistula, blood loss, and operative time all decreased with increased surgeon experience.18 Similar findings were identified in orthopedics, where performance of 40 or more cases of periacetabular osteotomy was associated with improved outcomes.19 Taken together, these findings suggest that outcomes in acute care surgery occur at the surgeon level and may be at least partially driven by experience. We hypothesized that outcomes may be improved with increased surgeon experience and undertook this secondary analysis of a previously reported cohort study.20
After approval by the Yale Institutional Review Board and the institutional review boards of all participating institutions, a cohort study of emergency surgery was performed. Two of the institutional review boards at participating institutions required consent on the part of the surgeon; none required consent of the patients. Informed written consent was therefore obtained from all participating surgeons. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Five centers functioning under an acute care model were prospectively identified and data were collected on consecutive emergency surgical procedures by fellowship-trained acute care surgeons. The 5 participating centers included Yale School of Medicine, New Haven, Connecticut (the coordinating center); Cooper University Hospital, Camden, New Jersey; Hackensack University Medical Center, Hackensack, New Jersey; Harbor-UCLA Medical Center, West Carson, California; and Loma Linda Medical Center, Loma Linda, California. All of the centers are American College of Surgeons–verified level I trauma centers. Data were collected from May 2015 to July 2017 and analyzed from February to May 2020. At each institution, the acute care surgery service provided care for all trauma and emergency surgery patients. Both EGS and trauma case data were collected. Patients were considered eligible for inclusion if they went directly from the emergency department to the operating room or if they were transferred from an outside institution and the surgeon documented the need to go emergently to the operating room. Appendectomy, cholecystectomy, and simple incision and drainage procedures were excluded.
All additional data were abstracted from the patients’ medical records, usually at the time of hospital discharge. These data included patient demographic characteristics, laboratory values on admission, physiologic variables at admission, operative time, and operative blood loss. Several intraoperative decisions were also recorded, including the decisions to proceed with definitive laparotomy, create a stoma after colon resection, or create a stoma if definitive laparotomy was performed. To address the heterogeneity of the complex procedures being performed, all operations were classified on an ordinal scale of complexity from 1 to 4.20 Case complexity was assigned independently by 2 surgeons (K.M.S. and B.B.) with discrepancies resolved through discussion. Degree of illness was calculated using the Emergency Surgery Score, a validated scale for degree of illness in emergency surgical conditions.21 Hemorrhagic shock was defined as a preoperative systolic blood pressure of less than 90 mm Hg, later definitively identified as being due to hemorrhage. Estimated blood loss was categorized as less than 200 mL, 200 to 400 mL, 401 to 800 mL, 801 to 1200 mL, 1.2 to 2 L, and greater than 2 L. Surgeons were classified based on years of experience since terminal training, as follows: early career (less than 6 years), early midcareer (6 to 10 years), late midcareer (11 to 30 years), and late career (more than 30 years). These groups were determined based on commonly held beliefs about surgeon experience and on the distribution of patients among surgeons in this cohort.
Outcomes collected were based on standard definitions of the National Surgical Quality Improvement Project Operations Manual.22 Patients were monitored until death or hospital discharge for all occurrences. The primary outcome was mortality and secondary outcomes were total complications, unplanned return to the operating room, surgical site infection, wound dehiscence, estimated procedural blood loss, and hospital length of stay.
Data were securely transmitted to the organizing center and collated in a single database, then compared with descriptive statistics. Differences between surgeon groups were assessed using with analysis of variance, χ2, and log-linear regression for bivariate comparisons. Multivariable hierarchical logistic, ordinal logistic, and log-linear regression models were fit to control for outcomes. The models were designed with patients nested within surgeons. Because of the limited number of participating centers, the centers were not included as a third level. The intercept-only models were initially fit to calculate the intraclass correlation coefficient and determine the amount of variability at the surgeon level. Fixed-effect confounders at the surgeon level added to the models included the Emergency Surgery Score, whether the patient was a trauma patient, whether the patient was in hemorrhagic shock, case complexity, and surgeon level of experience—the main variable of interest. As a sensitivity analysis to examine the least-experienced surgeons, we also compared surgeons with less than 3 years of experience with the entire cohort. All statistical analyses were performed using R version 4.0.0 (The R Foundation) with the following packages installed: lme4, performance, ggplot2, arsenal, DescTools, dplyr. All statistical tests were 2-tailed with P < .05 considered significant.
Among the 5 centers, 883 patients were operated on by 56 uniquely identifiable surgeons. Due to missing surgeon identification, 772 patients were included in the analysis, including the multilevel modeling. Of 772 included patients, 469 (60.8%) were male, and the mean (SD) age was 50.1 (20.0). A total of 618 procedures were performed by surgeons with 10 years of experience or less, a pattern that was consistent across centers (Figure). Center 3 had a mix of surgeon experience levels and also included most of the late-career surgeons. Of 772 operations, 641 (83.0%) were performed by surgeons with 10 years of experience or less, and 574 (74.4%) were performed by surgeons with 8 years of experience or less. There were 376 EGS patients and 396 trauma patients. The mean (SD) age was 56.8 (18.5) years for the EGS patients and 41.9 (18.4) years for the trauma patients. Demographic characteristics, comorbidities, and case complexity, stratified from level 1 (least complex) to level 4 (most complex), varied between EGS and trauma patients (Table 1). Comparing patient factors with surgeon experience, early- and late-midcareer surgeons tended to operate on older patients with more septic shock and kidney failure, modestly higher Emergency Surgery Scores, and who were more likely to be EGS patients. Case complexity was similar among surgeon groups (Table 1 and Table 2).
When considering mortality, the intraclass correlation coefficient of the partially pooled random intercepts model demonstrated that only 1.9% of the variance was at the surgeon level; however, this rose to 8.0% when patients who died the day of admission were excluded. On bivariate analysis, there were no significant differences in outcomes, including mortality, total complications, surgical site infection, wound dehiscence, and postoperative transfusion. There were fewer unplanned returns to the operating room for patients operated on by early- and late-midcareer surgeons but longer length of stay (Table 3). Patients operated on by early-career surgeons (those with less than 6 years of experience) had higher unplanned return to the operating room compared with those operated on by early-midcareer surgeons (odds ratio [OR], 0.66; 95% CI, 0.40-1.09), late-midcareer surgeons (OR, 0.34; 95% CI, 0.13-0.90), and late-career surgeons (OR, 1.11; 95% CI, 0.45-2.75). After controlling for patient degree of illness and case complexity, there was a nonsignificant association of fewer unplanned returns to the operating room with increased surgeon experience followed by reversal of the association with late-career surgeons (Table 4). Patients cared for by the most experienced surgeons also had shorter risk-adjusted lengths of stay. Other findings included worse outcomes with increased degree of illness, as measured by the Emergency Surgery Score, and hemorrhagic shock. Case complexity was associated with length of stay, blood loss, and unplanned returns to the operating room. After stratifying surgeon groups to isolate the least experienced, patients operated on by those with less than 3 years of experience had similar odds of mortality (OR, 1.97; 95% CI, 0.85-4.57) but increased odds of a major complication (OR, 2.07; 95% CI, 1.05-4.07) compared with the rest of the cohort.
In this cohort study, trauma and EGS patients were at high risk of morbidity and mortality.5,23 The study assessed the possibility that surgeon experience may play a role in patient outcomes. We examined several outcomes that might be driven by surgical skill and judgment, both of which should improve with experience. These outcomes included patient mortality, complications, estimated blood loss, need for postoperative transfusion, unplanned return to the operating room, and length of stay. One of the initial findings that affected the interpretation of the data was the level of experience among surgeons at the 5 academic medical centers. Most operations were performed by surgeons with 10 years of experience or less with almost three-fourths quarters of the cases being done by surgeons with 8 years of experience or less. Late-midcareer surgeons operated on patients who were older and those with higher degrees of illness as demonstrated by higher Emergency Surgery Scores and incidence of acute kidney failure, although this finding was not significant. However, the operative interventions were not more complex. With trauma, the decision to operate is typically driven by injury type and patient condition. However, with emergency surgery, such decisions may be driven to some extent by the individual surgeons, some of whom may be more willing than others to operate on higher-risk patients. The tendency in our study is in contrast to prior studies, in which more-experienced surgeons were more likely to decline to operate on sicker patients.11
There were no differences in mortality or blood loss between cohorts. There was a decreased length of stay for patients operated on by late-career surgeons, who were perhaps willing to discharge patients sooner. There were fewer unplanned returns to the operating room for patients operated on by early-midcareer surgeons, and late-midcareer surgeons had a statistically significant lower probability of unplanned return to the operating room. The progressively decreasing association of unplanned returns to the operating room with surgeon experience suggests this may be a direct association. Each of the models were adjusted for potential confounders thought to influence the outcomes. After isolating those surgeons with less than 3 years of experience, there was an increased probability of patient complication, but this did not significantly affect mortality. The fact that mortality was similar despite increased complications and returns to the operating room suggests successful patient rescue. However, an increased sample size may have identified small differences in mortality.
It has been demonstrated that there are differences in outcomes among hospitals performing emergency surgery and caring for trauma patients.7,23 However, the mechanisms that underlie these findings have not been clearly elucidated. The reasons for these differences may exist at the institution level, but they are also found at the surgeon level.7,14,24 Such differences have been demonstrated in both operative cases and in the nonoperative treatment of trauma patients, suggesting clinical decision-making is at least partially responsible for these differences. The specific effect of operative care has not been well examined in emergency surgical interventions. In elective general surgery, however, the technical skill of the surgeon has been identified as a predictor of patient outcome.25,26 Two surgeon-level factors that may contribute to outcome variability are annual volume and total volume of procedures. Both of these seem to play a role in surgical and trauma outcomes.17-19,26,27 Surgical skill may also have a differential effect on complex procedures compared with simple procedures.28 This may be important for improving emergency surgery outcomes where cases vary between many simple procedures, such as appendectomy and cholecystectomy, and fewer, more complex procedures, such as gastric resections. Whether the data from our study point to the often-discussed volume-outcome association, innate surgical skill, or overall surgical experience is not certain. How volume per unit time, total career experience, innate surgical skill, and training interact in the case of acute care surgeons to affect outcomes has not been fully elucidated. However, we did identify an association between surgeon experience and patient outcomes with more unplanned returns to the operating room and increased risk of complication for patients operated on by less-experienced surgeons. These findings were especially apparent in the most junior attending surgeons with less than 3 years of experience, where there were more patient complications and an association with higher patient mortality. These findings were despite a likely higher volume for the less-experienced surgeons.
Unplanned return to the operating room and complications may therefore be potential measures when implementing programs to improve outcomes at the surgeon level. The ultimate goal would be to shrink or eliminate the association between surgeon experience and patient complications. As an initial step, it will be important to determine if unplanned returns to the operating room led to important therapeutic intervention due to complications or if these were unnecessary returns to the operating room based on lack of confidence in inexperienced surgeons. Improvements may be achievable through programs now in use in the elective setting, such as surgical coaching. Additionally, a change in culture may be necessary so that junior surgeons feel more comfortable asking for assistance in the operating room, or are encouraged to debrief with more-experienced surgeons after complex cases. We believe the old surgical adage and approach of “call me if you need me but you shouldn’t need me” is less common than it once was but needs to be eliminated entirely. In addition to required postoperative debriefings, intraoperative consultations might benefit both the junior surgeon learning new skills and the senior surgeon maintaining skills. Senior surgeons who may serve as mentors would rarely be needed and perhaps in urban areas be able to serve as backup to multiple junior surgeons simultaneously. Policy changes to fund additional surgeons are unlikely and any changes to mitigate these issues will need to come from within acute care surgery.
There were several limitations to our study. Although the patients were enrolled prospectively, likely improving the fidelity of abstracted data, some of the data were collected retrospectively. This introduces the potential for bias in patient selection and the expected bias in unmeasured confounding. Although we measured the simple factor of years since the end of training as our primary predictor, factors such as innate ability, recent surgical volume, and prior training may interact in uncertain ways with the measured outcomes. We also had a significant imbalance in experience level in the number of surgeons in each group (Table 2) and the characteristics of patients treated by these surgeons. The consequence of this imbalance is an inability to draw conclusions about later-career surgeon performance. Additionally, in group practice, surgeons may have brought patients back to the operating room due to bias on the part of junior attending surgeons regarding risk of complications.
Given that this is a post hoc analysis of a prior study, no power analysis was performed. However, the prior study assumed an increase in mortality from 5% to 10% with an α of .5 and a power of 80%. A total of 880 patients divided between 2 groups would have been necessary to demonstrate a difference. Although we had fewer total patients, the power would be increased by the greater number of groups and decreased by the imbalance. Taken together with the similar OR for mortality, the probability that this study is underpowered to detect this same difference is small; however, smaller differences in mortality are possible.
Acute care surgeons provide surgical care for a heterogenous patient population with varied surgical complexity. Like other studies, we found that a portion of patient mortality was associated with surgeon experience. Some of that outcome may have been driven by surgeon experience as experience did affect unplanned returns to the operating room and postoperative complications. To improve patient outcomes, interventions that develop surgical experience—whether through simulation, coaching, mentorship, or some other method—may affect patient outcomes.
Accepted for Publication: December 17, 2020.
Published Online: March 10, 2021. doi:10.1001/jamasurg.2021.0041
Corresponding Author: Kevin M. Schuster, MD, MPH, Department of Surgery, Yale School of Medicine, 330 Cedar St, BB310, PO Box 208062, New Haven, CT 06520 (email@example.com).
Author Contributions: Dr Schuster had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Schuster.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Schuster, Bhattacharya.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Schuster.
Administrative, technical, or material support: Schuster, Bhattacharya.
Conflict of Interest Disclosures: Dr Hazelton reports consulting fees from Z-Medica outside the submitted work. No other disclosures were reported.
Meeting Presentation: This article was presented virtually at the 101st Annual Meeting of the New England Surgical Society; October 23 and 24, 2020.