| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Clinical Research Unit, Division of Cardiothoracic Surgery, Emory University School of Medicine, Atlanta, Georgia
b Department of Biostatistics, Rollins School of Public Health, Emory University School of Medicine, Atlanta, Georgia
Accepted for publication April 1, 2008.
* Address correspondence to Dr Thourani, Crawford Long Hospital, 6th Floor, Medical Office Tower, Cardiothoracic Surgery, 550 Peachtree St, Atlanta, GA 30308 (Email: vinod.thourani{at}emoryhealthcare.org).
Presented at the Fifty-fourth Annual Meeting of the Southern Thoracic Surgical Association, Bonita Springs, FL, Nov 7–10, 2007.
| Dr Puskas discloses that he has a financial relationship with Medtronic and Boston Scientific; Dr Guyton with Medtronic Inc, Boston Scientific Corp, and Guidant.
|
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Methods: From January 1996 to June 2006, 14,278 patients underwent primary, isolated CABG (5641 OPCAB and 8637 CCAB) at Emory Healthcare Hospitals and were prospectively entered into a computerized database. In a multiple logistic regression analysis using propensity score adjustment, we evaluated the impact of preoperative neurologic events on in-hospital mortality and postoperative neurologic events in patients undergoing OPCAB vs CCAB using adjusted odds ratios (AOR).
Results: Overall, postoperative neurologic events occurred in 1.9% (274 of 14,278) of patients and were associated with an in-hospital mortality of 13.5% (37 of 274). In the logistic regression analysis, patients with preoperative neurologic events had a higher incidence of postoperative neurologic events (AOR, 3.07; p < 0.0001) and in-hospital mortality (AOR, 2.19; p < 0.0001). The OPCAB patients were less likely to have postoperative neurologic events compared with CCAB patients (AOR, 0.49; p < 0.001). However, no significant interaction existed between preoperative neurologic events and operation type for any of the postoperative comparisons.
Conclusions: Preoperative neurologic events are risk factors for postoperative neurologic events and in-hospital mortality. OPCAB is associated with a reduced risk of postoperative neurologic events compared with CCAB. However, OPCAB does not confer additional risk reduction in patients with preoperative neurologic events.
Despite advances in surgical techniques and perioperative care, postoperative neurologic events (NEs), which include permanent stroke and transient ischemic attack, continue to be a major source of morbidity and mortality after coronary artery bypass grafting (CABG). In addition to known risk factors for postoperative NEs [1–3], a history of preoperative NEs has also been reported to be associated with an increased risk of postoperative NEs [4–7]. Previously, we have reported a stroke incidence of 2.2% after primary coronary operations from 1988 to 1996 [8]. However, that study did not specifically evaluate the effects of a preoperative NE on postoperative outcomes.
The theoretic advantages of reduced postoperative NEs in patients undergoing off-pump coronary artery bypass (OPCAB) has not been proven in clinical trials, including recent meta-analyses of OPCAB vs conventional coronary artery bypass (CCAB) [9, 10] or in observational analyses [11]. Potential limitations of these studies include the lack of patients at high risk for neurologic complications and the relatively small sample sizes that may underpower these studies to detect significant differences in this rare complication.
The purpose of the current study was to evaluate 14,278 consecutive patients undergoing primary, isolated CABG from 1996 to 2006 and to assess the impact of a history of preoperative NEs on postoperative outcomes. Specifically, we wanted to evaluate whether patients with a history of preoperative NEs had a higher risk of postoperative NEs or in-hospital death, or both. In addition, we wanted to assess whether patients undergoing OPCAB had a lower risk of postoperative NEs compared with patients undergoing CCAB.
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Exclusion Criteria
Patients were excluded from this analysis if they (1) presented with a permanent stroke within 2 weeks before their operation, (2) required cardiopulmonary resuscitation before operation (salvage), (3) underwent a redo sternotomy, (4) underwent concomitant CABG/carotid endarterectomy, and (5) underwent other concomitant cardiothoracic procedures.
Definitions
Definitions for preoperative variables were according to the definitions used by the STS National Database (www.sts.org): "stroke" indicates whether a patient has a history of any confirmed neurologic deficit of abrupt onset caused by a disturbance of cerebral blood supply that did not resolve within 24 hours; and "transient ischemic attack" (TIA) indicates whether a patient has a history of loss of neurologic function that was abrupt in onset but with complete return of function within 24 hours. For the purpose of this study, patients with a preoperative history of stroke or TIA were classified as having preoperative NEs. Patients with preoperative neurocognitive deficits or those with other neurologic disorders (seizure disorder) were not captured in the database and were not included in the definition of preoperative NEs. Similarly, the functional status of those with preoperative NEs was not categorized.
Postoperative definitions and outcomes were also determined according to the STS National Database criteria. Postoperative stroke and TIA criteria are defined above. Patients with either postoperative stroke or TIA were classified as having postoperative NEs. Neither prolonged coma not associated with focal neurologic findings nor neurocognitive changes was included in the definition of postoperative NEs. Patients suspected of having postoperative NEs were evaluated by a neurologist and underwent brain imaging with computed tomography (CT) or magnetic resonance imaging (MRI), or both. However, the classification of postoperative NEs did not require positive findings on imaging studies. Instead, focal deficits on neurologic examination and the STS database classifications defined the postoperative NEs according to the duration of findings as stated above.
Demographic and Preoperative Data
Before analysis, 25 preoperative risk factors for neurologic outcomes were identified and harvested from the STS database (Table 1). These same risk factors were adjusted for when estimating a propensity score (see below). Comparisons of preoperative characteristics were made using t tests for continuous variables and 
2 tests for categoric comparisons. Race was dichotomized to Caucasian or non-Caucasian according to physician assessment. Cerebrovascular disease type was segregated into four dichotomous variables: stroke, TIA, previous carotid endarterectomy, or carotid disease with 75% or more stenosis or occlusion. Similarly, status of the operation level was parameterized as separate dichotomous variables of elective, urgent, or emergency, as was internal mammary artery use (right, left, both, none).
|
A secondary goal was to analyze the potential association between the presence of postoperative atrial fibrillation and postoperative NEs in patients with no previous NEs. This was achieved by sub-setting only patients with no previous atrial fibrillation and fitting a logistic model that predicts postoperative NEs as a function of operation type, preoperative NEs, and postoperative atrial fibrillation. The adjusted odds ratio (AOR) of postoperative atrial fibrillation term was recorded. In addition, a logistic model was created to predict the impact of urgent/emergency status on postoperative NEs and in-hospital mortality, regardless of operation type.
Surgical Technique
OPCAB or CCAB was performed at the discretion of the attending surgeons. Patients were categorized into OPCAB or CCAB according to the operation that they ultimately received. OPCAB was performed with one of several commercially available suction-based cardiac positioning and coronary artery stabilizing devices. CCAB was performed with standard techniques, using roller head pumps, membrane oxygenators, cardiotomy suction, arterial filters, cold antegrade and retrograde blood cardioplegia, and moderate systemic hypothermia (32° to 34°C).
Statistical Methods
The retrospective nature of this study and the associated risk of selection bias indicated the use of propensity scoring methods [12, 13]. This method nonparsimoniously "balances" patients with respect to the effect of their preoperative risk factors on their probability of group assignment (OPCAB vs CCAB). The propensity score in this study was the probability of assignment to OPCAB based on 25 preoperative risk factors (see Table 1). Importantly, preoperative NE status was not included in the propensity score calculation because its direct effect was of primary interest. Date of operation and surgeon identity were included in the propensity score to account for the differences in these variables. Once formulated, the propensity score was used as a single covariate in a multiple logistic regression model that also included the primary variables of interest—operation type (OPCAB vs CCAB), preoperative NE status, and their interaction. A statistically significant interaction would indicate that the effect of operation type is different for patients with a history of preoperative NEs than it is for patients with no history of preoperative NEs.
Data were complete for the critical risk factors such as preoperative stroke and operation type as well as for each postoperative outcome. Data were missing on the following variables: Caucasian, 1248 (8.7%); Canadian Cardiovascular Society (CCS) classification, 914 (10.6%); ejection fraction, 1387 (16.1%); and number of diseased vessels, 1028 (11.9%). To prevent patients with missing data from being excluded from the analysis, multiple imputation methods described by Molenberghs and colleagues [14] were used to impute missing values. The goal of multiple imputations is not to replace data with plausible values but rather to impute with values that will maintain the covariance structure so that valuable data that are not missing will not be omitted from analysis.
The AORs with associated 95% confidence intervals (CI) for variables of interest were computed. For outcomes where interaction terms were insignificant, the interaction term was dropped and the model was fit again so that more accurate AORs could be calculated. For continuous outcomes (LOS and ventilator hours), general linear models were fit that regressed the outcome as a function of operation type, preoperative NE status, their interaction, and the propensity score. Adjusted mean differences were examined with least-squares means. The data were analyzed with SAS 9.1 software (SAS Institute, Cary, NC). All statistical tests were evaluated at the 
= 0.05 level. All comparisons and model terms were preplanned.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Preoperative and intraoperative data are summarized in Table 1. In general, patients with preoperative NEs had significantly more comorbidities than those without preoperative NEs. Table 2 reports the outcomes according to preoperative NE status, unadjusted for baseline risk. Patients with preoperative NEs were more likely to have postoperative NEs. In addition, in-hospital mortality was significantly higher in patients with preoperative NEs compared with patients without preoperative NEs. Patients with preoperative NEs also had longer hospital stays and a longer duration of postoperative ventilation. Table 3 reports the unadjusted outcomes according to OPCAB vs CCAB. Patients undergoing OPCAB had significantly fewer postoperative NEs (accounted for by fewer postoperative strokes) and a lower unadjusted in-hospital mortality rate.
|
|
The c-index for the model that creates the propensity score was 0.81, indicating that the preoperative risk factors, when considered together, exhibited good discrimination between OPCAB and CCAB cases. Importantly, none of the interaction terms for any of the outcomes was statistically significant, meaning that the effect of operation type was the same in patients with preoperative NEs as it was in patients without preoperative NEs. Thus, the interaction term was dropped and the models were refit.
|
|
|
For the continuous outcomes, LOS and ventilator hours, the general linear model revealed significant advantages for OPCAB patients and patients without preoperative NEs after adjustment for the propensity score. OPCAB and CCAB patients had an adjusted mean LOS difference of 0.6 days (6.3 vs 6.9 days, respectively, p < 0.001) and an adjusted mean ventilator hour difference of 4.4 hours (17.3 vs 21.7, p = 0.008). Patients with and without a preoperative NE had adjusted mean LOS differences of 1.6 days (7.4 vs 5.8, respectively, p < 0.001) and an adjusted mean ventilator hour difference of 7.2 hours (23.1 vs 15.9, p < 0.001).
Finally, preoperative NEs were subcategorized into stroke and TIA to assess the impact of each preoperative event on outcomes. Preoperative stroke was found to maintain statistical significance (comparable with preoperative NEs) on all of the postoperative outcomes (both adjusted and unadjusted), whereas preoperative TIA was not associated with similar statistical outcomes. For the purpose of this study, however, the data are described as preoperative NEs.
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Although the causes of NEs after CABG are multifactorial, embolic phenomena are a major factor [15]. Emboli can arise from intraoperative manipulation of the aorta during aortic cannulation, institution and maintenance of cardiopulmonary bypass, or application and removal of the aortic cross-clamp for cardioplegic arrest or partial clamping for proximal anastomoses. Other sources of emboli can occur from carotid artery disease, particulate matter aspirated from cardiotomy suction devices, or from left atrial thrombi secondary to postoperative atrial fibrillation.
Studies using transcranial Doppler ultrasonography have confirmed the production of aortic emboli associated with cannulation and application of aortic clamps [16–18]. Bowles and colleagues demonstrated large quantities of aortic emboli production during cardiopulmonary bypass without manipulation of the aorta using transcranial Doppler ultrasonography [19]. Furthermore, Kapetanakis and colleagues [1] and Calafiore and associates [20] concluded that aortic manipulation is independently associated with an increased risk of postoperative stroke. Hammon and colleagues [21] reported reduced neuropsychologic deficits in patients undergoing CCAB with a single cross-clamp compared with patients undergoing CCAB with multiple cross-clamping or OPCAB and partial clamping. Thus it would seem intuitive that a reduction in postoperative stroke could be achieved by avoiding or minimizing manipulation of the aorta and avoiding cardiopulmonary bypass.
In addition to known risk factors for postoperative NEs, including age, diabetes, carotid artery disease, perioperative hypotension, and aortic manipulation [1–3], preoperative NEs have been reported to be a risk factor for postoperative NEs [4–7]. In studies examining the effect of preoperative stroke on patients undergoing cardiac procedures using cardiopulmonary bypass, Redmond and colleagues [4] documented a 44% stroke incidence and Rorick and Furlan [7] reported a 13% stroke incidence in patients with a history of stroke. Bucerius and colleagues [22] identified a history of cerebrovascular disease as a significant risk factor (OR, 3.55) for postoperative stroke. Our unadjusted results revealed a postoperative NE rate of 4.6% in patients with a history of preoperative NEs compared with a postoperative NE rate of 1.6% in patients without a history of preoperative NEs. In the multiple logistic regression analysis, preoperative NEs were associated with a threefold increase in postoperative NEs and a twofold increase in in-hospital mortality.
Although OPCAB avoids the use of cardiopulmonary bypass, postoperative NEs have not been significantly reduced in two recent meta-analyses of off- vs on-pump CABG [9, 10]. These analyses included studies with a relatively small number of patients, which individually were underpowered to detect a significant difference in a complication occurring in less than 4% of the study population. There are several factors to consider when evaluating these studies. In patients with a relatively low atherosclerotic burden of the ascending aorta, the risk of aortic emboli is probably low and thus it is unlikely that a difference in postoperative NEs will be appreciated between off- and on-pump CABG. It is plausible that other confounding variables, such as carotid disease, hypoperfusion, or postoperative atrial fibrillation would be more likely to affect postoperative NEs. In addition, partial aortic clamping for construction of proximal anastomoses is still routinely performed in our center and by others for patients undergoing OPCAB. Thus, the benefits of OPCAB with regards to postoperative NEs may be attenuated because of the increased risk associated with partial aortic cross-clamping. Kim and associates [23] reported a lower incidence of postoperative stroke in patients undergoing OPCAB without any manipulation of the aorta compared with patients undergoing OPCAB with partial clamping and patients undergoing CCAB.
The mechanism explaining why patients with a preoperative history of NEs may be at higher risk for postoperative NEs remains elusive. Intuitively, those with significant cerebrovascular disease would be more susceptible to hemodynamic alterations in the intraoperative and perioperative period. In a previous report from our institution, Puskas and colleagues [8] documented that advanced age, previous TIA, and carotid bruits on physical exam were all independent predictors of postoperative stroke. Therefore, we routinely use preoperative screening carotid duplex ultrasonography in all patients aged older than 65 years, those with significant left main stenosis, history of TIA or stroke, or with a carotid bruit on auscultation. Patients with internal carotid artery stenosis of 80% or more are referred to vascular surgery for consideration of carotid endarterectomy.
Study Limitations
Although every effort has been made to control for selection bias, the retrospective nature limits broad inferences in this regard. The lower postoperative NE rates observed in the OPCAB group may suggest a temporal relationship, but because OPCAB became more frequent in the later years of the study, this variable (year of operation) and the surgeon performing the operation were included in the propensity analysis to account for potential confounding.
In addition, we were unable to account for the mechanisms contributing to postoperative NEs in either the CCAB or OPCAB group. Although patients in the OPCAB group had a lower incidence of postoperative atrial fibrillation and a lower incidence of postoperative NEs, a statistically significant interaction was not evident in the multiple logistic regression analysis. Specifically, because of database limitations, we were unable to assess the impact of aortic manipulation, such as partial clamping, epiaortic ultrasound grade, single cross-clamp, and use of anastomotic devices, among others. These variables have recently been included in the data registry, and their impact on postoperative NEs is the subject of current investigation at our institution.
Finally, the impact of conversion from OPCAB to CCAB and its effect on postoperative NEs and in-hospital mortality cannot be ascertained because this variable was not routinely entered in the database until 2004.
Conclusion
Patients with preoperative NEs are at a significantly higher risk for postoperative NEs with its associated increased morbidity and mortality. Patients undergoing OPCAB have a significantly lower incidence of postoperative NEs, although OPCAB does not appear to confer an additional risk reduction when compared with CCAB in patients with preoperative NEs. The mechanisms responsible for the increased risk of postoperative NEs in patients with preoperative NEs, as well as the decreased incidence of postoperative NEs in OPCAB patients should be the focus of future study. Additional strategies to reduce the incidence of this devastating complication and its associated mortality will further improve the results of CABG in the future.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
I do have two questions, though, before we close the book on this issue. As our Secretary stated today, statistics sometimes hide more facts than they bring out, and there are two particular issues that I think need to be discussed. The first is since this is a 10-year data collection period, were there more OPCAB operations done in the more current, more contemporary period than in the earlier period? The second is, as in our institution, technical improvements in the performance of on-pump CABG surgery have accelerated in the recent time as well. And thus, does your institution follow that trend of using coated oxygenators, less traumatic aortic clamps, low velocity aortic cannulas, and emphasizing the use of the cell saver versus cardiotomy suction during on-pump CABG surgery? Thank you again.
DR HALKOS: Thank you, Dr Hammon. I will answer your second question first. Yes, we have over the last 3 to 4 years adopted safer techniques for on-pump patients. So to answer your second question, we have utilized the cell saver device, the different oxygenators and tubing, and low velocity cannulas.
With regards to your first question, I showed that one graph which detailed the longitudinal adoption of off-pump techniques. The crossover point was in 2001, when off-pump surgery was more frequently utilized at Emory Hospitals compared to on-pump cases, and certainly the timing may affect the outcomes that we saw in this group of patients. However, the year of surgery was included in the propensity score analysis. So we tried to accommodate for that variable. Certainly it is a limitation that more off-pump surgery was done in the later years of the study compared to the earlier years.
DR HOOSHANG BOLOOKI (Miami, FL): Thank you very much for a very nice presentation and paper. Certainly Emory has a fantastic database and thanks for bringing that to the attention of this Association.
The question I have is that you have changed from an on-pump service to an off-pump service and showed your curves very nicely crossing each other somewhere about 2001 and 2002. I wonder during these two periods of time if you compared the two groups at the point that you crossed the two lines of 2001 up to now and 2001 before, which was 1998. Were there any differences in the neurological occurrences in those two groups of patients?
DR HALKOS: Thank you for your question. Certainly specific outcomes, especially regarding aortic manipulation, use of proximal anastomotic devices, epiaortic ultrasound, which is now widely used at our institution, were not included in the database for this study, since we went back to 1996. That is currently being investigated. We suspect, and that data should be forthcoming, that these differences may affect the outcome both with on- and off-pump patients.
DR J. W. RANDOLPH BOLTON (Fresno, CA): Just a quick question with regard to the preoperative neurologic events. Did you have any way of evaluating intraoperative TEE [transesophageal echocardiography] to evaluate specifically the left atrial appendage in the on- and off-pump patients during the study, especially in those patients with A-fib [atrial fibrillation]?
DR HALKOS: Intraoperative TEE is routinely used when there are concerns about atrial thrombus in patients with atrial fibrillation. However, this data was not collected in our database.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. A. Miano, M. G. Reichert, T. T. Houle, D. A. MacGregor, E. H. Kincaid, and D. L. Bowton Nosocomial Pneumonia Risk and Stress Ulcer Prophylaxis: A Comparison of Pantoprazole vs Ranitidine in Cardiothoracic Surgery Patients Chest, August 1, 2009; 136(2): 440 - 447. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Chu, F. G. Bakaeen, T. K. Dao, S. A. LeMaire, J. S. Coselli, and J. Huh On-Pump Versus Off-Pump Coronary Artery Bypass Grafting in a Cohort of 63,000 Patients. Ann. Thorac. Surg., June 1, 2009; 87(6): 1820 - 1827. [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 |
