| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 2003;126:1335-1344
© 2003 The American Association for Thoracic Surgery
Surgery for acquired cardiovascular disease |
a Division of Cardiovascular Surgery, Toronto General Hospital, University Health Network, Department of Surgery, and the Heart and Stroke Foundation/Richard Lewar Centre of Excellence, University of Toronto, Toronto, Ontario, Canada
Received for publication February 25, 2003; revisions received March 18, 2003; revisions received May 4, 2003; accepted for publication June 4, 2003.
* Address for reprints: Terrence M. Yau, MD, MSc, Division of Cardiovascular Surgery, Toronto General Hospital, University Health Network, 13EN-239, 200 Elizabeth St, Toronto, Ontario, Canada, M5G 2C4
terry.yau{at}utoronto.ca
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
METHODS: Data on demographic characteristics, preoperative risk factors, operative variables, and hospital outcomes were collected prospectively for all patients undergoing isolated coronary artery bypass grafting at a single institution from January 1, 1990, to December 31, 2001. To examine the effect of time on patient risk profiles and outcomes, we divided patients into three groups according to year of operation (1990-1993 n = 5171, 1994-1997 n = 5977, 1998-2001 n = 6893).
RESULTS: In-hospital mortality declined from 2.4% (1990-1993) to 1.2% (1998-2001, P < .0001). Left ventricular dysfunction, increasing age, female gender, hypertension, diabetes, cardiogenic shock, congestive heart failure, peripheral vascular disease, reoperative coronary artery bypass grafting, left main disease, and urgent surgery independently predicted in-hospital mortality in the entire cohort of 18,041 patients. Severe left ventricular dysfunction was the most significant predictor of in-hospital mortality in the 12-year cohort, but it had a declining influence with time (1990-1993 odds ratio 7.1, 1994-1997 odds ratio 5.1, 1998-2001 not statistically significant) because of improving outcomes. Reoperative coronary artery bypass grafting similarly decreased in significance as a predictor of mortality. Emergency coronary artery bypass grafting was performed less frequently in recent years, but the requirement for emergency surgery carried an increasing odds ratio for mortality.
CONCLUSIONS: Despite increasing patient age and comorbidity, improvements in perioperative management have reduced the significance of severe left ventricular dysfunction and reoperative coronary artery bypass grafting but not emergency surgery as predictors of in-hospital mortality.
There have been tremendous efforts to develop risk stratification models to provide more accurate prediction of mortality and morbidity among patients undergoing CABG, culminating in the development of the Society of Thoracic Surgeons National Database. These predicted risks can then be used in patient counseling, to decide between alternative therapies, and to direct efforts to improve outcomes in high-risk subgroups.13 As perioperative management advances, however, predictive models become outdated and inaccurate. It is therefore essential to analyze evolutionary trends in the patient population with time, as well as changes in the predictors of outcomes, to ensure that current decisions are still based on valid data. We therefore analyzed our experience during a 12-year period with patients undergoing isolated CABG to evaluate changes in their risk profiles, document their outcomes, and determine changes in the independent predictors of in-hospital mortality.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Outcome and explanatory variables
Our primary outcome in this study was in-hospital mortality, which was defined as any postoperative death in hospital. Core baseline explanatory variables collected since 1990 included age, sex, left ventricular grade (by left ventricular ejection fraction [LVEF]: grade 1 LVEF > 60%, grade 2 LVEF 40%-60%, grade 3 LVEF 20%-39%, grade 4 LVEF <20%), previous CABG, urgency of operation (elective vs semiurgent, indicating an operation during the same admission as a cardiac catheterization or a cardiac event, vs urgent, indicating an operation within 72 hours of an event, vs emergency, indicating an operation within 12 hours of an event), number of diseased coronary arteries, presence of significant stenosis (>50%) of the left main coronary artery, and severity of angina. LVEF was determined semiquantitatively by contrast ventriculography. Echocardiography and nuclear ventriculography were carried out in an increasing proportion of patients during the course of this study, and when these additional data were available, the value for LVEF obtained closest to the date of operation was used for subsequent analysis.
Other explanatory variables collected to more fully characterize these patients included recent myocardial infarction, diabetes, peripheral vascular disease, previous coronary angioplasty or stenting, a history of hypertension, and preoperative stroke or transient ischemic attack.
Our data collection also included details of cardioplegia temperature (warm [36°C-37°C], tepid [29°C-30°C], or cold) and continuity (intermittent or continuous), as well as direction of administration (antegrade, retrograde, or both). Details of this database have been published elsewhere.14
Analysis
Data were collected and managed in a Microsoft Access database (Microsoft Corporation, Redmond, Wash). All statistical analyses were performed with SAS version 8.2 software for Windows (SAS Institute, Inc, Cary, NC).15 Univariate analyses were performed with 
2 analysis or Fisher exact test as appropriate for categoric variables. Analysis of continuous variables was carried out by analysis of variance. Variables that had a univariate P value <.25 or those judged to be clinically important were submitted to a logistic regression model by stepwise selection. Multivariate logistic regression methods were used to calculate risk-adjusted mortality and factor-adjusted odds ratios. Model discrimination was evaluated by the area under the receiver-operator characteristic curve,16,17 and calibration was assessed with the Hosmer-Lemeshow goodness-of-fit statistic.18 For goodness of fit, the null hypothesis is that the model fits the data. Therefore a nonsignificant P value is desired, because a lower one would indicate a poor fit between predicted and observed results.
Evaluation of temporal trends
Rather than building a complex model to assess the temporal trends in prevalence, risk profiles, and outcomes, we used a simpler approach based on risk stratification and contingency table analysis. To examine the effect of time on patient risk profiles and outcomes, we divided patients into three groups according to the year of operation (1990-1993, 1994-1997, 1998-2001). Contingency table analysis was then used to evaluate changes with time in prevalences of preoperative risk factors, intraoperative events, and operative outcomes.
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
|
|
65 years) age groups (falling from 1.3% in the first cohort to 0.8% in the last cohort). Female patients had consistently higher mortality than male patients, but in-hospital mortality among female patients fell from 4.1% in the first time cohort to 2.1% in the last cohort.
Timing of surgery
Patients requiring urgent or emergency surgery from the cardiac catheterization laboratory or the emergency department were also at increased risk. Although the number of patients requiring emergency surgical revascularization decreased with time, the odds ratio associated with emergency surgery was increased significantly in the last time cohort (odds ratio 19.0, 95% CI 5.0-72.1). Whereas in-hospital mortality among patients undergoing elective or semiurgent surgery declined significantly during the 12-year span of this study, mortality among the decreasing number of patients undergoing emergency surgery was almost doubled from the first cohort (6.5%) to the last cohort (12.5%).
LVEF
Severe left ventricular dysfunction (LVEF <20%) was the most significant predictor of in-hospital mortality in the entire cohort of 18,041 patients. This was also true of the first and second time cohorts, although the odds ratios associated with a LVEF less than 20% decreased from 7.1 in the first cohort to 5.1 in the second cohort. However, left ventricular dysfunction was no longer a significant predictor of in-hospital mortality in the last time cohort, despite a larger number of patients. When LVEF was forced into the model for the 1998 to 2001 period, the resulting P value for it was .07. Patients with the most severe left ventricular dysfunction (LVEF <20%) had approximately 3 times greater in-hospital mortality than those with LVEF between 20% and 40% in the first two time cohorts. As the results in patients with mild or moderate left ventricular dysfunction improved even further, this ratio increased to approximately 7 in the last cohort. Overall in-hospital mortality for patients with severe left ventricular dysfunction declined, however, from 10.8% in 1990 to 1993 to 8.9% in 1998 to 2001.
Reoperative CABG
Patients undergoing reoperative CABG were also at a higher risk of dying in the first time cohort (odds ratio 4.2, 95% CI 2.5-6.8), but the relative risk fell by almost half from the 1990 to 1993 period to the 1994 to 1997 time cohort (odds ratio 2.3, 95% CI 1.3-3.8) and remained about the same in the 1998 to 2001 period (odds ratio 2.7, 95% CI 1.2-6.1).
Congestive heart failure and cardiogenic shock
Congestive heart failure was a predictor in the last two time cohorts, and cardiogenic shock was a predictor only in the 1998 to 2001 cohort. The number of patients in congestive heart failure undergoing CABG increased significantly during the last decade, and in-hospital mortality for these patients decreased only minimally, from 6% to 5.1%. In contrast, mortality among patients in cardiogenic shock increased from 10.9% in the 1990 to 1993 period to 16.3% in the 1998 to 2001 time cohort.
Left main disease
Although in-hospital mortality among patients with left main disease was consistently higher than that in patients with triple-vessel disease but without left main stenosis, mortality among these patients fell during the 12-year period, from 3.4% to 2.4%.
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Increasing risk profile
As anticipated in this study, we also found that the prevalence of risk factors and comorbidities that have consistently predicted in-hospital mortality in previous studies1,2,6,14,19 increased significantly during the entire period of this study. These factors included advanced age, female gender, diabetes, and hypertension. Patients were increasingly likely to be seen with acute coronary insufficiency, a recent preoperative myocardial infarction, cardiogenic shock, or congestive heart failure. Coronary anatomy in these patients worsened through the decade, with the prevalences of left main disease and triple-vessel disease increasing steadily. Noncardiac comorbidities, such as peripheral vascular disease, also increased with time.
Predictors of mortality
The independent predictors of in-hospital mortality in these 18,041 patients (increasing age, female gender, hypertension, diabetes, cardiogenic shock, congestive heart failure, peripheral vascular disease, reoperative CABG, left main disease, decreasing LVEF, and urgent or emergency surgery) are generally similar to those previously described by Christakis and colleagues,6 who had observed that reoperative CABG, LVEF, and urgency of surgery were the most significant predictors of in-hospital mortality. Jones and associates,25 in a review of seven data sets, reported that seven variables (age, gender, previous heart surgery, left main stenosis, extent of coronary artery disease, LVEF, and urgency of operation) predicted mortality after CABG. The predictive power of these factors was greatest for urgency of surgery, age, and previous CABG. Abramov and colleagues2 also identified urgency of operation (odds ratio 4.5), reoperative procedure (odds ratio 3.7), and LVEF less than 35% (odds ratio 3.2) as the most significant predictors of in-hospital mortality. As anticipated, year of operation was an independent predictor of in-hospital mortality in our series, with an odds ratio of 0.4 (95% CI 0.3-0.6) for the third time cohort relative to the first and an odds ratio of 0.8 (95% CI 0.6-1.1) for the second cohort relative to the first.
Reoperative CABG
Despite the commonly held assumption that the prevalence of reoperative CABG would increase, we observed a steady decrease during the 12-year period of study in the number of patients undergoing second or third CABG operations, from 7.4% in the first time cohort to 4.2% in the last cohort. A similar, but less marked, trend was also observed by Ferguson and coworkers19 in the Society of Thoracic Surgeons National Database (falling from 10.6% in 1990 to 8.4% in 1999). In contrast, Abramov and colleagues2 reported a relatively unchanged prevalence of reoperations from 2.2% in 1990 to 1992 to 2.9% in 1996 to 1998. The decline in reoperations at our institution may have been due to later presentation for initial operation (consistent with the increasing age of patients undergoing surgery and perhaps increasing percutaneous interventions before surgery), more aggressive percutaneous intervention in patients with graft atherosclerosis, or perhaps greater graft longevity with the widespread use of statins.
The in-hospital mortality among patients undergoing reoperative surgery also declined with time, with the odds ratio associated with reoperative CABG declining by approximately 50% from the first to the second time cohorts. Improved perioperative care and the increasing use of retrograde cardioplegia most likely accounted for this fall in in-hospital mortality.26 Stephan and colleagues27 reported a mortality of 7.3% in 164 patients undergoing elective coronary reoperations, which was similar to our early experience. Shimada and coworkers28 reported outstanding results in a series of 200 reoperations, with an in-hospital mortality of only 2.5%, similar to our results in the most recent cohort.
LVEF
The prevalence of patients with LVEF 40% or lower fluctuated, but only slightly, among the three time cohorts, whereas in-hospital mortality in this group fell by 50% during the span of this series. The slight decrease in the prevalence of LVEF less than 20% in the last cohort may have been due to, among other potential explanations, (1) an increasing number of patients with acute myocardial infarction undergoing primary percutaneous transluminal coronary angioplasty and stenting, thus preventing deterioration of left ventricular function, or (2) reduced referral of patients with severe left ventricular dysfunction in favor of percutaneous interventions. The mortality associated with moderate left ventricular dysfunction fell to the point (3.2% in the first cohort, odds ratio 1.9, to 1.4% in the last cohort, odds ratio 0.9) that only an ejection fraction less than 20% is now considered to be a finding of clinical importance in our practice. Whereas severe left ventricular dysfunction was the most significant predictor of in-hospital mortality in the first two time cohorts, with odds ratios of 7.1 and 5.1, respectively, it did not emerge as a statistically significant predictor in the last cohort. This is not to say that left ventricular dysfunction did not influence mortality among patients operated on in 1998 to 2001; rather, its influence had declined to the point that, even in a reasonably large cohort of 6893 patients, its effect was now insufficient to reach statistical significance in multivariable modeling. When LVEF was forced into the model for the last time cohort, the P value for LVEF (.07) was marginally higher than the cutoff threshold of .05, and the odds ratio had declined to 2.8 from its previous high of 7.1 in the first cohort. This represents a significant departure from its previously consistent status as the most significant predictor of mortality in any analysis.
This remarkable decline in the predictive value of left ventricular dysfunction may be partly due to the decreasing prevalence of patients with LVEF less than 20% in the last time cohort. However, the odds ratio for mortality associated with LVEF less than 20% also fell dramatically, indicating that outcomes improved in parallel with the modest reduction in the prevalence of severe left ventricular dysfunction. We believe that the improvement in survival of this group of patients was due to the cumulative beneficial impact of various techniques and strategies used during the last decade. Early extubation anesthetic management provided stable perioperative hemodynamics, reduced catecholamine stress response, improved intrapulmonary shunt fraction, and allowed earlier mobilization of patients. The use of mild systemic hypothermia in all cases and retrograde tepid or warm cardioplegia in selected high-risk cases may also have contributed to improved outcomes. Perioperative usage of antifibrinolytic agents increased, and a strict transfusion protocol was adopted. The use of epiaortic and transesophageal echocardiography in patients with a high probability of aortic atherosclerosis or left ventricular thrombi may have reduced the risk of perioperative stroke.
Left main disease
The prevalence of left main disease increased slightly from the first to the last time cohorts. Although the in-hospital mortality in this group of patients fell during the 12-year period, from 3.4% to 2.4%, left main disease was a significant predictor (odds ratio 1.7) of in-hospital mortality only in the last cohort. This somewhat counterintuitive finding may reflect a persistent but more subtle predictive effect of left main disease (overall odds ratio 1.4 in the 12-year period) that was "unmasked" in the last time cohort as previously more powerful predictors like left ventricular dysfunction and reoperative surgery declined in importance.
Age
The prevalence of elderly patients (
75 years of age) undergoing CABG almost doubled from the first to the last time cohort, but in-hospital mortality in this group fell from 6.8% to 2.5%. Although age was a consistent predictor of in-hospital mortality in all three time cohorts, the odds ratio for age fell from 1.064 in the first cohort to 1.038 in the last group. This difference in numbers seems small, but the compounding effect of this incremental risk per year of increasing age means that a substantial reduction in age-related risk was achieved. A similar reduction in mortality has also been reported by Ivanov and colleagues11 from 1982 to 1996, despite an increasing risk profile among these elderly patients.
Urgency of surgery
The percentage of patients referred for urgent and emergency CABG declined significantly from 20.4% in the first cohort to 8.6% in the last one, whereas the proportion of patients requiring semiurgent surgery (during the same admission) rose steadily. Ferguson and coworkers19 found similar trends, with a reduction in emergency and salvage operations (7.1% in 1990 to 5.5% in 1999) and a rise in the number of urgent operations (17.6% in 1990 to 32.9% in 1999). Abramov and colleagues2 also reported an increase in the number of urgent operations from 26.5% in 1990 to 1992 to 41.5% in 1996 to 1998.
In our series the proportion of patients requiring truly emergency CABG (transferred directly from the cardiac catheterization laboratory) decreased from 2.1% to 0.5%. This decline was likely related to the aggressive use of "bailout" stents and perhaps to the increased use of glycoprotein IIb and IIIa inhibitors. The need for emergency surgery was associated with a significantly increased risk of mortality in all three time cohorts, but the odds ratio increased dramatically with time (rising to 19.0 in the last cohort), even as the number of patients in this subgroup fell from 108 in the first time cohort to 32 in the last. In a trend that paralleled that observed for patients undergoing CABG in general, patients who had complications or became unstable in the cardiac catheterization laboratory but who could be rescued by glycoprotein IIb and IIIa inhibitors or stents were removed from this surgical cohort. The remainder of these patients, referred after these percutaneous attempts at rescue had also failed, increasingly represented a scenario of "salvaging the unsalvageable."
Time
In this series, overall in-hospital mortality among patients undergoing isolated CABG decreased from 2.4% in the 1990 to 1993 time cohort to 1.2% in 1998 to 2001, a reduction of 50%. A similar reduction of 52% in risk-adjusted mortality in cardiac surgical centers in Ontario was observed between 1986 and 1995.29 Abramov and colleagues2 reported a reduction in risk-adjusted in-hospital mortality from 2.5% to 1.8%, and Ferguson and coworkers19 reported a relative decrease of 23% in in-hospital mortality in the Society of Thoracic Surgeons database during the course of a decade. Other studies from New York,8 northern New England,30 and Massachusetts10 have also reported decreasing in-hospital mortality after CABG. In contrast, Warner and associates23 studied the effects of patient characteristics on outcomes from 1981 to 1995 and found that mortality increased from 2.4% to 3.4%, a finding that was attributed to a learning curve associated with the shift toward higher risk patients.
Conclusions
Our current study illustrates the ever-changing nature of CABG. With steadily improving results in progressively sicker patients, it is now routine at many tertiary and quaternary referral centers to operate on, and achieve superb results with, patients who a decade ago would have posed extreme risks. Although we did note a steady increase in the risk profile of patients undergoing surgery during the course of this study, it is not possible to determine whether even higher risk patients with the greatest left ventricular dysfunction or previous surgery were diverted to percutaneous procedures, because our database does not track patients turned down for surgery or those preferentially referred for percutaneous intervention by the cardiologist. The trends that we observed in this study, however, along with the overall in-hospital mortality of 1.2% among the 6893 patients in the most recent cohort, suggest to us that advances in perioperative management, technology, and the increasing experience in managing such patients have dramatically reduced the risk associated with even severe left ventricular dysfunction and reoperative surgery. In contrast, patients requiring truly emergency CABG face a greatly increased risk of death.
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. Nardi, A. Pellegrino, A. Scafuri, D. Colella, C. Bassano, P. Polisca, and L. Chiariello Long-term outcome of coronary artery bypass grafting in patients with left ventricular dysfunction. Ann. Thorac. Surg., May 1, 2009; 87(5): 1401 - 1407. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Soliman Hamad, M. E. S.H. Tan, A. H.M. van Straten, A. A.J. van Zundert, and J. P.A.M. Schonberger Long-Term Results of Coronary Artery Bypass Grafting in Patients With Left Ventricular Dysfunction Ann. Thorac. Surg., February 1, 2008; 85(2): 488 - 493. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. E. Roselli, G. B. Pettersson, E. H. Blackstone, M. E. Brizzio, P. L. Houghtaling, R. Hauck, J. M. Burke, and B. W. Lytle Adverse events during reoperative cardiac surgery: Frequency, characterization, and rescue J. Thorac. Cardiovasc. Surg., February 1, 2008; 135(2): 316 - 323. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Albert, N. Halevy, and E. M. Antman Preoperative Evaluation for Cardiac Surgery Card. Surg. Adult, January 1, 2008; 3(2008): 261 - 280. [Full Text]
|
|
|
|
 |
|
 D. L. Ngaage, M. E. Cowen, S. Griffin, L. Guvendik, and A. R. Cale The impact of symptom severity on cardiac reoperative risk: early referral and reoperation is warranted Eur. J. Cardiothorac. Surg., October 1, 2007; 32(4): 623 - 628. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Filsoufi, P. B. Rahmanian, J. G. Castillo, J. Chikwe, A. S. Kini, and D. H. Adams Results and Predictors of Early and Late Outcome of Coronary Artery Bypass Grafting in Patients With Severely Depressed Left Ventricular Function Ann. Thorac. Surg., September 1, 2007; 84(3): 808 - 816. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Jonsson, N. Hammar, T. Nordquist, and T. Ivert Left main coronary artery stenosis no longer a risk factor for early and late death after coronary artery bypass surgery - an experience covering three decades. Eur. J. Cardiothorac. Surg., August 1, 2006; 30(2): 311 - 317. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. S. Hillis, K. J. Zehr, A. W. Williams, H. V. Schaff, T. A. Orzulak, R. C. Daly, C. J. Mullany, R. J. Rodeheffer, and J. K. Oh Outcome of Patients With Low Ejection Fraction Undergoing Coronary Artery Bypass Grafting: Renal Function and Mortality After 3.8 Years Circulation, July 4, 2006; 114(1_suppl): I-414 - I-419. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. F. Sabik III, E. H. Blackstone, P. L. Houghtaling, P. A. Walts, and B. W. Lytle Is Reoperation Still a Risk Factor in Coronary Artery Bypass Surgery? Ann. Thorac. Surg., November 1, 2005; 80(5): 1719 - 1727. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Shuhaiber Epidemiologic contrast of predictors' trends for outcomes of coronary artery bypass grafting: Heart failure versus ventricular function versus left main disease J. Thorac. Cardiovasc. Surg., June 1, 2004; 127(6): 1854 - 1855. [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 |
