| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2007;83:2103-2110
© 2007 The Society of Thoracic Surgeons
a University of Florida College of Medicine Jacksonville, Jacksonville, Florida
b Duke Clinical Research Institute, Durham, North Carolina
c The Cleveland Clinic Florida, Weston, Florida
d Eastern Carolina Cardiovascular Institute, Greenville, North Carolina
Accepted for publication January 23, 2007.
* Address correspondence to Dr Haan, University of Florida College of Medicine, 653-1 W 8th St, FC-12, Jacksonville, FL 32209 (Email: connie.haan{at}jax.ufl.edu).
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentReferences |
|---|
Methods: Using the Society of Thoracic Surgeons (STS) National Cardiac Database, we studied 369,906 CABG patients undergoing isolated coronary artery bypass graft (CABG) procedures during January 2002 through June 2005. Participating institutions were stratified by residency versus nonresidency status and by perfusion time categories and analyzed for association with clinical outcomes.
Results: Overall, 57 (10%) of 594 STS participants had a residency training program. Residency programs had longer mean cross-clamp and perfusion times than nonresidency programs, 73.10 versus 67.44 minutes and 104.75 versus 98.00 minutes, respectively (p < 0.0001 for both. Longer perfusion time was significantly associated with higher operative mortality at the patient level. Unadjusted mortality rates were, however, similar for patients at residency and nonresidency programs (2.30% versus 2.27%), with an adjusted odds ratio of 0.96 (95% confidence interval, 0.84 to 1.09). Although perfusion times have not changed significantly over time between residency and nonresidency programs, mortality rates have significantly improved over time at each.
Conclusions: Residency programs have longer CABG perfusion times than nonresidency cardiothoracic surgery programs, but these differences are minor. Adjusted procedural outcomes at residency training programs are similar to those at nonresidency centers; thus, patients do not appear to be adversely impacted by the time costs of surgical training.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Our growing understanding of the physiologic impact of cardiopulmonary bypass has escalated interest in clarifying the impact of the pump on outcomes [9–12]. If prolonged perfusion times are associated with worse patient outcomes, then in this era of public reporting and consumer information, one may consider whether perfusion times can be used as a surrogate marker of provider quality. Further, if an accepted price of training residents in cardiothoracic surgery is often a loss in intraoperative efficiencies, one can justifiably question whether patients undergoing coronary artery bypass graft surgery (CABG) at hospitals with residency training programs have longer perfusion times and have worse outcomes [13–16].
We used the Society of Thoracic Surgery (STS) National Cardiac Database to investigate and establish (1) whether residency training programs have significantly longer perfusion times than nonresidency programs, and (2) after adjusting for baseline clinical case mix, whether clinical outcomes for patients undergoing CABG surgery differ by the residency status of a program. Stating the study questions from the patient perspective, if a patient goes to a teaching hospital, will the perfusion time be longer because of the teaching, and will the outcome be worse?
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Research performed on the STS database at the Duke Clinical Research Institute is approved by the Duke University Institutional Review Board. Principal investigators at participating institutions are responsible for reviewing their data collection efforts with their sites’ Institutional Review Boards to assure that patient privacy and confidentiality is protected.
All clinical definitions are based on specifications of core data elements that can be accessed at http://www.sts.org.
Patient Population
Extracted from the STS National Cardiac Database were all CABG procedures for the period of January 2002 through June 2005 that used cardiopulmonary bypass (CPB). The years were selected for timeframe of maximal consistent data quality.
Cases were excluded that reported extremely short (and questionably valid) perfusion-related times—time on CPB of less than 20 minutes, cross-clamp time of less than 10 minutes—and cases from programs reporting fewer than 20 procedures per year or fewer than 20 total CABG procedures. Also excluded were salvage or resuscitation cases, because perfusion times in these procedures are heavily influenced by patient condition and physiology more than by surgeon-driven processes and decision-making. All off-pump CABG procedures were excluded, including those that required conversion to on-pump procedures, in an effort to minimize, where possible, confounding of the analysis of relationships between perfusion times and outcomes.
Variables for Risk-Adjustment and Confounding
Mortality was risk-adjusted for the preoperative patient and cardiac variables according to the 2004 updates to the STS CABG risk model [20]. In addition to the variables normally contained in the risk models, we adjusted for other potential confounders of outcomes such as body mass index, hypertension, smoking status, diabetes treatment, prior valve surgery, arrhythmia, congestive heart failure, aortic stenosis or insufficiency, tricuspid stenosis or insufficiency, and pulmonic insufficiency.
Statistical Analysis
Baseline demographic characteristics, preoperative risk factors, and outcomes of interest (operative mortality, any reoperation, stroke, renal failure, prolonged ventilation, and postoperative length of stay exceeding 14 days) were compared between patients at STS National Cardiac Database participants with and without a residency training program in cardiothoracic surgery. Continuous variables are described as medians (with interquartile ranges), and categoric variables are described as frequencies. Continuous variables were compared by using Wilcoxon rank sum tests, and categoric variables were compared by using 
2 tests.
The predicted logit from the STS mortality model was used to analyze if there was a significant interaction between a patient’s risk profile and residency status.
The percentage of patients in each of six perfusion time intervals (<1 hour, 1 to 1.5, 1.5 to 2, 2 to 2.5, 2.5 to 3, and >3 hours) were compared for patients at residency versus nonresidency participants. We calculated raw mortality rates within each perfusion time interval and with additional stratification by residency status.
Multivariate logistic regression analysis was used to adjust for the patient characteristics as described. All odds ratios are reported with patients from nonresidency participants as the reference group.
Perfusion trends over time were summarized by calculating median perfusion times and mortality rates by residency status for each 6-month interval during the study period.
Because patients within a participant institution were more likely to be similar, all adjusted analyses were performed using generalized estimating equation models to account for correlations among clustered responses (eg, within-participant correlations). A value of p < 0.05 was established as the level of statistical significance for all tests. All analyses were performed using SAS 8.2 software (SAS Institute, Cary, NC).
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Patient Characteristics
Table 1
displays baseline demographics and clinical characteristics for patients undergoing isolated CABG at residency versus nonresidency STS participant programs. Given the very large sample size of our cohort, most demographic features, risk factors, and other preoperative characteristics differed statistically by residency status, yet the magnitude of these differences was slight.
|
Risk assessment of patients was performed for teaching versus nonteaching settings. The interaction of a patient’s risk profile and residency status was at p = 0.0906, and therefore was not significant.
Perfusion Times and Mortality
For the overall population analyzed, the respective mean and median perfusion times were 98.70 and 94.00 minutes. Mean perfusion times for patients who died perioperatively were higher in both residency and nonresidency participant programs: 129.60 and 120.55 minutes for deaths (residency and nonresidency) versus 104.17 and 97.48 minutes for survivors (residency and nonresidency). The odds ratios (ORs) and confidence intervals (CIs) for perfusion time category on mortality are summarized in Table 2. Unadjusted and adjusted ORs for mortality were incrementally higher for each increasing perfusion time category compared with perfusion time of less than 1 hour. Having demonstrated that our data reaffirmed that perfusion time is associated with mortality, we shifted our focus to the impact of residency status on perfusion time and outcomes.
|
|
|
|
|
|
Trend Over Time in Perfusion Times and Outcomes
Finally, we examined trends in perfusion times by residency status for January 2002 through June 2005. The median perfusion times among patients at residency and nonresidency participants during this period remained fairly steady (Fig 4). More notable is the mortality trend (Fig 4), with mortality rates decreasing over time among patients at both residency and nonresidency participants.
|
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Longer perfusion times were consistently associated with higher mortality. Although the absolute differences were slight between those with perfusion times between less than 1 hour and 3 hours, those with extended perfusion times of more than 3 hours had an exponentially increased associated mortality rate of more than 9%. There are, however, multiple potential explanations for this association, and these data do not distinguish to what extent those long (>3 hours) perfusion times are due to patient selection versus surgeon technique and judgment. Perfusion times tend to be longer in those with more complex disease or greater comorbidity. Further, intraoperative adverse events can prolong perfusion times and are associated with higher risk for mortality and morbidity. Thus, although there is strong association between perfusion times and outcomes, direct causal conclusions should not be assumed.
The average perfusion times were found to be longer at residency programs than at nonresidency programs, but unadjusted and adjusted mortality rates were similar. Other major morbidity event rates were also generally similar among those treated at residency versus nonresidency programs, with the exception of slightly higher rates of reoperation and prolonged requirement of ventilator support.
Prolonged ventilator time could be due to longer perfusion time and increased inflammatory tissue response, but it may also be a side effect of caution by trainees and the team structure. By contrast, the decision on timing of postoperative extubation in nonresidency programs may be protocol-driven or delegated to someone more experienced who can make the decision to extubate earlier.
Our finding that CABG surgery performed at a residency program does not represent increased risk for adverse outcomes is consistent with the findings of Roberts [21], Caputo [22], Baskett [23], Oo [24], and Asimakopoulos [25] and their colleagues. Of note, the studies by Roberts and Oo and their colleagues emphasized selection bias for and proper supervision of trainees, but our data analysis did not necessitate those distinctions.
Finally, our study noted that perfusion times have been generally static at both types of program over time. This may in part reflect an increase in patient severity of coronary artery disease and condition for those referred to surgery. With relatively stable perfusion times and no association between residency program status and mortality, decreasing trends for mortality over time suggest that other balancing factors are responsible for the decrease. Such factors may include improved preoperative and postoperative care and decision-making, as well as improved anesthetic and cardioplegia content and management, and improved temperature and glycemic control—representing the impact of evidence-based practice on patient outcomes.
As noted previously, drawing any causal relationship between perfusion times and outcomes is challenged in an observational database because of several potential confounding factors, including patient demographics, case severity, operative variables, and surgeon preferences. Although we have adjusted our findings for measured differences in patient clinical factors, our results should be viewed as hypothesis generating. And although the STS National Cardiac Database is the nation’s largest cardiothoracic clinical data source, participation is voluntary and results are reflective only of those who participate. Prior studies, however, have supported the generalizability of these findings to US Medicare patients [19].
Another limitation of the study is that it is impossible to extract those situations in which the operation was started by trainees and finished by the trainer because of unanticipated findings or complications ensuing. Similarly, residents are not grouped by operating experience. Residents are not, however, allowed to do cases that are more complex than their skills and are not allowed to do cases without supervision. It also cannot be assumed that a case performed in a training institution was performed by or even involved a resident. It is important, therefore, to restate here that our data evaluate training programs versus nontraining programs, and not the individual case mix done by residents versus staff.
In conclusion, we found that prolonged perfusion times (>3 hours) are associated with higher mortality. At the participant level, however, there is no conclusive evidence of an association between residency status and operative mortality when controlling for available confounding variables. Furthermore, both raw and adjusted mortality rates are similar between residency and nonresidency participants. Of more importance is that patient safety was not found to be in jeopardy simply by virtue of resident participation in the intraoperative processes of care.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  F. G. Bakaeen, J. Huh, S. A. LeMaire, J. S. Coselli, S. Sansgiry, P. V. Atluri, and D. Chu The July effect: impact of the beginning of the academic cycle on cardiac surgical outcomes in a cohort of 70,616 patients. Ann. Thorac. Surg., July 1, 2009; 88(1): 70 - 75. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C.-H. Yap, N. Andrianopoulos, D. T. Dinh, B. Billah, A. Rosalion, J. A. Smith, G. C. Shardey, P. D. Skillington, J. Tatoulis, M. Mohajeri, et al. Short- and midterm outcomes of coronary artery bypass surgery performed by surgeons in training. J. Thorac. Cardiovasc. Surg., May 1, 2009; 137(5): 1088 - 1092. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. G. Bakaeen, A. S. Dhaliwal, D. Chu, B. Bozkurt, P. Tsai, S. A. LeMaire, M. J. Wall Jr, J. S. Coselli, and J. Huh Does the Level of Experience of Residents Affect Outcomes of Coronary Artery Bypass Surgery? Ann. Thorac. Surg., April 1, 2009; 87(4): 1127 - 1134. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. C. Stoica, D. Kalavrouziotis, B.-J. Martin, K. J. Buth, G. M. Hirsch, J. A. Sullivan, and R. J.F. Baskett Long-Term Results of Heart Operations Performed by Surgeons-in-Training Circulation, September 30, 2008; 118(14_suppl_1): S1 - S6. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
