If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Teaching the next generation operative cardiac surgery while maintaining the highest level of patient care is an ever-increasing challenge given the growing proportion of patients with multiple comorbidities, the loss of more straightforward cases to percutaneous interventions, and the pressure of public reporting. No study to date has compared the outcomes of similar cases performed entirely (“skin-to-skin”) by the resident with those performed entirely by the staff to confirm the safety of this practice.
Methods
A total of 100 consecutive cardiac cases performed skin-to-skin by the resident (group R) were matched by procedure 1:1 to nonconsecutive cases performed by a single attending surgeon (group A). Patients were excluded from the analysis if there was overlap in any portion of the procedure by the trainee or the attending.
Results
Patients in group A were similar to those in group R with respect to age, gender, body mass index, American Society of Anesthesiologists classification, left ventricular ejection fraction, and diabetes mellitus. Mean operative times were longer in group R (4.6 vs 2.7 hours, P < .001), as were cardiopulmonary bypass times (96 vs 50 minutes, P < .001) and aortic crossclamp times (78 vs 39 minutes, P < .001). There were no significant differences in red blood cell transfusions, reexplorations, stroke, length of stay, or wound infections. There were no in-hospital or 30-day deaths.
Conclusions
Our data indicate that trainees can be educated in operative surgery under the current paradigm, despite longer operative times, without sacrificing outcome quality. It is reasonable to expect academic programs to continue providing trainees significant experience as primary operating surgeons.
An organized approach to cardiac surgical training can safely produce well-trained surgeons in the current era. Academic programs should continue providing trainees experience as primary surgeons.
Increased patient complexity combined with resident work hour regulations and public reporting of outcomes and complications has compromised the training experience of cardiac surgical residents. This study shows that with a carefully planned approach to surgical education, residents can be appropriately trained while patients can continue receiving excellent medical care.
Despite the introduction and acceptance of less-invasive access techniques such as transcatheter aortic valve replacement (AVR) and mitral clip procedures over the past few years, cardiac surgery has largely remained a traditional “open” surgical specialty. The majority of cardiac operations are still performed through a median sternotomy or a right thoracotomy requiring advanced open surgical skills. Surgical staples and anastomotic devices have found limited applicability in adult cardiac surgery and have been largely abandoned. At the same time, other surgical specialties such as general and vascular surgery have moved toward laparoscopic and endovascular platforms, respectively. As a result of this, in combination with duty hour restrictions, many graduates seek additional training for open operations by means of postgraduate fellowships, while those continuing with cardiac fellowships enter their training with a limited set of open skills.
Although most open cardiac operations performed today were first described and popularized more than 4 decades ago, the complexity of those operations has increased because more elderly patients with multiple comorbidities are routinely referred for surgery compared with years past. This has temporally coincided with the introduction of extensive outcomes data collection and public reporting, with plans to use these data as a determinant for physician and hospital financial compensation. Complications are heavily scrutinized, and their financial cost is shifted more toward the medical institutions and less toward the insurance companies. Meanwhile, many best-selling publications and national mainstream media outlets appear more eager to cover unsubstantiated stories of inadequate resident supervision coupled with concurrent surgery practices and have arbitrarily linked these practices to poor outcomes for the patients on the basis of individual case reports. As a result, many academic surgeons are reluctant to provide their trainees with the extent of operative experience and autonomy that they received during their own years of training.
Finally, many medical students and general surgery residents choose not to pursue a career in cardiothoracic surgery given the negative image created around the field and the training process in particular.
Although some of these studies have looked at specific types of operations such as off-pump revascularization or have documented the progress of resident learning as demonstrated by a decrease in the time it takes a trainee to perform a certain task, none of them have routinely provided robust data about the degree of trainee independence.
The majority of these studies do not comment on attending/resident overlap and have not systematically examined many preoperative, perioperative, and postoperative parameters.
The purpose of this study was to compare postoperative outcomes of 2 similar surgical patient cohorts whose operations were performed entirely by an attending surgeon or by a physician in training. By doing so, we hoped to determine if the inefficiencies associated with resident training affect clinical outcomes and to further drill down on appropriate patient selection for resident training. We think that observing residents operating skin-to-skin is the ultimate assessment of competency and readiness for surgical practice (Video 1).
This study was approved by the Partners Human Research Committee's Institutional Review Board for human research to meet ethical and legal requirements. From July 2014 to December 2016, Dr Tolis completed a total of 642 pump cases, either as teaching surgeon or primary surgeon. The breakdown of these cases is as follows: isolated coronary artery bypass grafting (CABG) (321), isolated AVR (84), isolated mitral valve replacement (MVR) (8), AVR/CABG (64), MVR/CABG (3), AVR/MVR (4), mitral valve repair (12), mitral valve repair/CABG (17), and other (129). A prospective longitudinal database was created capturing comprehensive data for all patients undergoing cardiac operations by Dr Tolis at our institution. Consecutive cases done skin-to-skin by the resident surgeon during the study period with attending supervision were matched by specific operative procedure 1:1 with cases done skin-to-skin by the same attending surgeon. A 1:1 matching was achieved once the resident cases reached a total of 100. Because we collected our data prospectively, in almost all categories there were fewer attending alone cases, necessitating the attending to perform several cases in each category himself to achieve equal numbers in all categories and eventually achieve 1:1 matching with the resident alone cases. All cases were performed at the Massachusetts General Hospital main campus.
Training
The cardiothoracic training program at the Massachusetts General Hospital consists of both rotating general surgery residents and full-time cardiothoracic residents who have completed general surgery training. All cases done skin-to-skin by resident surgeons were done so by the 8 full-time cardiothoracic residents who rotated with Dr Tolis during the study period. Of the 8 residents, 4 performed skin-to-skin cases during both full years of their training, whereas the other 4 only performed skin-to-skin cases during the second year of their training.
Before being allowed to perform an operation skin-to-skin, the residents had displayed competency in performing each individual step of the operation during the earlier part of their training (eg, opening, harvesting conduits, cannulating, constructing distal and proximal anastomoses). This had been significantly enhanced after adoption of the “apprenticeship model,” which we implemented in our program in 2013, where a resident would be assigned to a specific attending for a 2- to 3-month block to increase the level of responsibility allowed to the resident and enhance the resident's experience.
Definitions
Cases were considered skin-to-skin if the operating surgeon completed each of the following steps of the operation from the right side of the operating table:
Common steps
1.
Opening of skin, soft tissues, sternum, and pericardium.
2.
Cannulation for cardiopulmonary bypass (CPB).
3.
Placement of aortic crossclamp (ACC).
4.
Completion of the procedure specific steps (below).
5.
Decannulation and separation from CPB.
6.
Achieving hemostasis and closure of sternum, soft tissues, and skin.
CABG-specific steps
1.
Harvesting the arterial conduit(s).
2.
Identification, dissection, and opening of coronary arteries.
3.
Completion of both proximal and distal anastomoses for every bypass graft.
AVR/MVR-specific steps
1.
Performing the aortotomy/atriotomy.
2.
Resection/reconstruction of native valve.
3.
Placing every annular suture.
4.
Tying every knot.
5.
Closing the aortotomy/atriotomy.
If the attending surgeon had to intervene and complete one of these steps, the case was excluded from our analysis. There were several cases that were intended to be resident cases but were converted to mixed (attending and resident cases) because the attending surgeon had to intervene at some point during the operation. Typical reasons that this scenario would take place were (1) dense pericardial adhesions making cannulation and identification of coronary arteries difficult, (2) excess annular calcium in a mitral or AVR procedure necessitating attending involvement, (3) leaking of a distal or proximal anastomosis requiring taking down the anastomosis and reconstructing it, or (4) distorted anatomy of the heart and great vessels making it technically difficult to perform a standard cannulation. All these cases were considered “mixed” cases (both attending and resident involvement) and were not included in this study. During our original analysis of the data, we did not identify any complications in these cases (eg, death, stroke, wound infection) that we could directly attribute to this “conversion.”
There are cases that become “mixed” or even “attending only” simply because of timing issues (eg, starting late in the day, need to complete >2 cases in 1 day) even though they could be easily done entirely by a trainee. When this happened, these cases were not included in this study even though in many of them the bulk of the case was done by the resident. We decided to keep a strict definition of a “resident” case to eliminate any “judgment” issues as to what constitutes a resident case.
Statistical Analysis
All statistical analysis was performed with Stata/SE 12.1 (StataCorp, LP, College Station, Tex). Continuous data, which were normally distributed, were expressed as means with standard deviations, and non-normally distributed data were expressed as medians with interquartile ranges. Categoric data were expressed as numbers and percentages. Student t test or Wilcoxon rank-sum tests were used where appropriate for continuous variables, and the chi-square test was used where appropriate for categoric variables. All tests were performed 2 sided. Normality of the data was assessed using histograms, skewness, kurtosis, or the Shapiro–Wilk test.
Results
A total of 200 cardiac surgery cases were analyzed, which were performed between July 2014 and December 2016. A total of 100 consecutive cases performed by the resident were matched 1:1 with 100 nonconsecutive cases performed by the attending surgeon (G.T.) during the same time interval. Within that timeframe, the attending surgeon performed a total of 642 cases. One case performed by the resident (isolated tricuspid valve repair) could not be matched with an equivalent attending case and was excluded from the study. Of note, that patient made an uneventful recovery.
Preoperative Characteristics
The preoperative characteristics of the patients in the attending and resident groups are shown in Table 1. There were no statistically significant differences in age, sex, body mass index, American Society of Anesthesiologists score, left ventricular ejection fraction, or diabetes mellitus. Patients in the attending group had higher Society of Thoracic Surgeons (STS) predicted mortality scores when compared with patients in the resident group (1.55% vs 1.22%, P = .04). To determine the cause of this difference, we separately analyzed all variables used to determine the STS score and found it was largely due to differences in 3 predictors: age more than 66 years (64% vs 49%, P = .03), use of intra-aortic balloon pump (10% vs 4%, P = .1), and case status (elective, 23% vs 39%; urgent, 74% vs 60%; and emergency, 3% vs 1%, P = .04). Trainees in the resident group were further along in training than those in the attending group (postgraduate year 7.2 vs 5.7, P < .001). This is not surprising given that the more senior level residents are entrusted with and expected to perform the technical aspects of the operation, albeit under supervision.
Table 1Preoperative characteristics for operations performed skin-to-skin by the attending surgeon or the resident surgeon
Variable
Attending surgeon (n = 100)
Resident surgeons (n = 100)
P value
Age (y) Mean, SD
69.2 (10.4)
67 (9.3)
.1
Range
43-91
47-89
Female no. (%)
20 (20)
19 (19)
.9
BMI (kg/m2) Mean, SD
28.6 (5.3)
29.5 (5.7)
.2
Range
17.3-47.1
18.7-55.2
ASA
Median
3
3
.7
IQR
3-4
3-4
Ejection fraction (mL/min) mean, SD
55.9, 13.9
58, 12.9
.3
Diabetes mellitus
30 (30)
39 (39)
.2
STS predicted mortality (%)
Median
1.55
1.22
.04
IQR
0.9-3.2
0.5-2.2
IABP no. (%)
10 (10)
4 (4)
.1
Inserted preoperatively
9 (90)
1 (25)
.02
Inserted intraoperatively
1 (10)
3 (75)
.02
Inserted postoperatively
0
0
Age >66 (y) no. (%)
64 (64)
49 (49)
.03
Case status no. (%)
Elective
23 (23)
39 (39)
.04
Urgent
74 (74)
60 (60)
.04
Emergency
3 (3)
1 (1)
.04
Postgraduate year of trainee Mean, SD
5.7, 1.4
7.2, 0.7
<.001
SD, Standard deviation; BMI, body mass index; ASA, American Society of Anesthesiologists; IQR, interquartile range; STS, Society of Thoracic Surgeons; IABP, intra-aortic balloon pump.
The operative details and cases are shown in Table 2. The caseload included isolated CABG procedures including 2-vessel CABG (n = 34), 3-vessel CABG (n = 34), 4-vessel CABG (n = 10), and 5-vessel CABG (n = 4) in each group. There were 10 isolated AVRs, 7 AVR/CABG operations, and 1 MVR/CABG × 3 operation in each group. Table 2 demonstrates this breakdown and the differences in total operative time (2.7 vs 4.6 hours, P < .001), CPB times (50 vs 96 minutes, P < .001), and ACC times (39 vs 78 minutes, P < .001).
Table 2Operative details of operations performed skin-to-skin by the attending surgeon or the resident surgeon
As demonstrated in Figure 1 and Table 3, operations performed by the attending surgeon were significantly shorter in total operative time, CPB time, and ACC time than the operations performed by resident surgeons. Intraoperative blood product use was not different between groups. Total operative time in hours were CABG × 2 2.3 versus 3.9 (P < .001), CABG × 3 2.7 versus 4.8 (P < .001), CABG × 4 3.3 versus 5.4 (P < .001), CABG × 5 3.7 versus 5.8 (P = .02), AVR 2.2 versus 3.9 (P < .001), AVR/CABG 3.5 versus 5.8 (P = .04), and MVR/CABG × 3 3.6 versus 5.8 (P = .3). The CPB times for the attending surgeon versus the resident groups were CABG × 2 38 versus 69 minutes (P < .001), CABG × 3 50 versus 97 minutes (P < .001), CABG × 4 63 versus 113 minutes (P < .001), CABG × 5 83 versus 144 minutes (P = .02), AVR 59 versus 104 minutes (P < .001), AVR/CABG 91 versus 122 minutes (P = .003), and MVR/CABG × 3 86 versus 140 minutes (P = .3). A similar trend was seen for ACC times with CABG × 2 28 versus 53 minutes (P < .001), CABG × 3 40 versus 78 minutes (P < .001), CABG × 4 52 versus 95 minutes (P < .001), CABG × 5 70 versus 115 minutes (P = .02), AVR 47 versus 89 minutes (P < .001), AVR/CABG 78 versus 103 minutes (P = .003), and MVR/CABG × 3 68 versus 140 (P = .3).
Figure 1Comparison of CPB time and aortic crossclamp time between resident and attending groups. CABG, Coronary artery bypass grafting; AVR, aortic valve replacement; MVR, mitral valve replacement.
Length of stay was equivalent between the groups at a median of 6 days. There were 4 cerebrovascular accidents, 2 in each group. We had 5 superficial incisional wound infections (requiring antibiotics or incision and drainage), 3 in the attending group and 2 in the resident group (P = .7), and 1 deep sternal wound infection in the attending group (requiring sternal debridement and pectoralis flap) (P = .3). The patient requiring a flap was a 68-year-old woman with uncontrolled diabetes and a nonhealing lower-extremity wound who needed urgent revascularization for unstable angina. Three patients required reoperations within 24 hours for postoperative bleeding, 2 in the attending group and 1 in the resident group (P = .6). There were 9 readmissions within 30 days in the attending group and no readmissions in the resident group (P = .002) (Table 4).
Discussion
The findings of this study demonstrate that cardiothoracic residents in a busy academic center were offered a solid operative experience as primary surgeons without compromising the quality of care delivered to patients. Resident training required a significant time commitment by the attending staff as demonstrated by the near doubling of the CPB and crossclamp times, as well as the overall duration of the operations.
This analysis is unique when compared with previously published reports. This study is the first in the literature to compare postoperative outcomes of 2 case-matched groups of patients undergoing a variety of open cardiac operations performed entirely (“skin-to-skin”) by an attending surgeon or by a trainee under attending supervision. By excluding any case in which even minimal crossover of operative responsibilities between trainee and attending took place, we eliminated any ambiguity about what constitutes “doing a case,” which can be interpreted in many ways and removed a major confounding factor that may bias our conclusions when it comes to comparing outcomes between the 2 groups.
During this 30-month interval, the attending surgeon performed 625 cases. The residents performed 100 cases, 16% of the attending's total operative volume. Another 213 cases (34%) were done by the attending skin-to-skin, with the remaining 312 cases (50%) performed by both. The usual reasons why the attending would perform a case skin-to-skin were case complexity (eg, poor targets, small and heavily calcified aortic root), acuity or presence of comorbidities, lack of an appropriate assistant (eg, physician assistant, nurse practitioner), patient/family request, or late start time.
The preoperative characteristics of the 2 groups demonstrate that the attending group was more likely to include older and sicker patients, as well as patients requiring an urgent/emergency operation or intra-aortic balloon pump support. In addition, the postgraduate year of the trainee had a statistically significant impact on the attending's decision to allow the trainee to perform the operation. It should not be surprising that there was intentional bias built into the process of delegating cases to the trainees. With excellent patient care being our primary goal, we tried to identify a low-risk group in which the inefficiencies related to the residents being the primary operators would not affect the postoperative outcomes.
The bias of directing more complicated cases to the attending is further suggested by the remarkable difference in the number of 30-day readmissions, with the attending group having 9 such events compared with no readmissions in the resident group. Although the 2 groups superficially appear similar, the attending group clearly included more complicated patients whose care required further hospitalizations after the initial discharge.
Despite longer operative, CPB, and ACC times in the resident group, clinical outcomes were similar. Although there is ample evidence in the literature that longer operations lead to more complications and worse outcomes, our analysis suggests that in a carefully selected group of patients, these effects can be mitigated, thus making it possible to provide the trainees with an excellent operative educational experience, without sacrificing clinical outcome quality. However, it should be stressed that the low event rate of some of these negative outcomes may not be detected in our study; it is possible that with a larger patient cohort, differences in complication rates between the 2 groups may have surfaced.
Although there was bias in the case selection, with lower STS score cases directed to the residents, we never attempted to allow a resident to perform an entire high complexity case with no attending involvement. Such cases may still be excellent vehicles for resident education by allowing the resident to perform specific steps (opening a redo sternotomy, sewing in a patch on a postinfarction ventricular septal defect, or finding coronary targets on a hostile epicardium) or general steps that do not prolong pump time or ischemia time (cannulating, harvesting conduits), but they require very active attending surgeon involvement and are generally not safe to be delegated to the resident as a skin-to-skin case.
This study reinforces the fact that there is a cost to educating residents. Academic cardiac surgeons must spend more hours in the hospital when resident training in the operating room is part of their teaching responsibilities. Yet, the government-sponsored insurers do not have any provisions, such as a procedure code modifier, that direct more federal funds to the centers that provide this invaluable service to future generations of surgeons, leaving it up to the individual institutions and departments to recognize the contribution to resident education by internally redirecting funds to the surgeons who provide this service.
We recognize that we could have included more cases in our sample had we relaxed the “skin-to-skin” rule. However, we thought that this would open the door to multiple interpretations as to what it means for a resident to “do the case.” We certainly believe that although a skin-to-skin case is the culmination of the residents’ experience, the bulk of their training comes from many more cases that residents are involved in during their training, which are split between the attending and the resident.
Study Limitations
There are several limitations to this study that may render our conclusions not applicable to other clinical settings. First and foremost, our effort to provide residents with superb training is supported by the institution, with nursing and ancillary staff willing and expected to work late into the day and accept the inefficiencies associated with our training philosophy. One may also argue that we have not looked at the specific issue of resident autonomy, because in all the resident cases the attending surgeon was always present for the critical parts of the operation, as recommended by the Commonwealth of Massachusetts Department of Health and mandated by our institutional policy.
Another limitation is the very nature of our comparison groups: We are comparing outcomes of a group of 8 residents with those of a single attending physician. Evaluating a more heterogeneous group of surgeons would have certainly added more power to our conclusions, but unfortunately we did not have the level of detailed data required for such an analysis from other staff members in our group. In addition, we acknowledge that the statistical power to detect a modest but important difference in poor outcomes is limited by our sample size (n = 200). We further acknowledge that no conclusions can be made about an individual resident's performance given the relatively small number of cases done by each trainee.
Last, but certainly not least, we are very fortunate to have in our ranks a highly selective group of residents who come from our general surgery program and other premier training programs in the United States. Their attitude toward patient care, attention to detail, and drive to excel are second to none. Whether this aggressive attitude toward resident participation in the operating room can be applied across all training programs is certainly up for debate.
Conclusions
Resident training and education are complicated endeavors that have to be approached in a systematic way. They require a significant mental frame and time commitment from the attending staff, which may not translate into commensurate financial compensation. Still, the academic attending physician needs to approach the training strategy for each case with the same intensity and thoroughness as the preoperative clinical strategy (eg, choice of incision, conduits, types of prostheses). Despite all the external factors that are often used as excuses to justify the more limited operative experience that residents receive today when compared with years past, we think that with a systematic approach to training that includes careful preoperative planning, insistence on intraoperative technical excellence, and meticulous myocardial protection and postoperative care, current residents can receive hands-on training that is equal to, if not better than, the training that previous generations of cardiac surgeons have received.
In this brave article by Tolis and colleagues,1 the question is posed (which is faced by every academic cardiac surgeon): Can I train the next generation of surgeons while maintaining excellent clinical outcomes? Not only does it take courage to publish such a study, Tolis and colleagues fearlessly set up training to allow residents to do entire cardiac surgery cases—not many surgeons are patient enough or dedicated enough to do this.
30068-0&id=fx1.jpg){kind=link}
30068-0&id=fx3.jpg){kind=link}
30068-0&id=gr1.jpg){kind=link}
30068-0&id=fx2.jpg){kind=link}