Predictors and Outcomes of Seizures After Cardiac Surgery: A Multivariable Analysis of 2,578 Patients

Transcription

1 Predictors and Outcomes of Seizures After Cardiac Surgery: A Multivariable Analysis of 2,578 Patients Andrew B. Goldstone, BA, David J. Bronster, MD, Anelechi C. Anyanwu, MD, Martin A. Goldstein, MD, Farzan Filsoufi, MD, David H. Adams, MD, and Joanna Chikwe, MD Departments of Cardiothoracic Surgery and Neurology, Mount Sinai School of Medicine, New York, New York Background. Limited data exist regarding the epidemiology and prognosis of seizures in adults after cardiac surgery. The aim of this study was to define preoperative predictors of seizures and impact on postoperative outcomes. Methods. A prospective database of 2,578 consecutive adults who underwent cardiac surgery at a single institution between April 2007 and December 2009 was retrospectively analyzed to determine risk factors for postoperative seizures and prognostic impact. No patient received tranexamic acid or aprotinin. Results. Seizures occurred in 31 patients (1%) at a median of 2 days postoperatively (1 patient had a preoperative diagnosis of seizure disorder). Seizures were classified as generalized tonic-clonic (71%), simple/complex partial (26%), or status epilepticus (3%). Incidence differed according to procedure (coronary bypass 0.1%, isolated valve 1%, valve/coronary bypass 3%, aorta 5%; p <0.001). Patients who experienced seizures had nearly a fivefold higher operative mortality than patients who did not (29% versus 6%, p <0.001). Head computed tomography was performed in 26 patients (84%), of whom ischemic strokes were identified in 14 cases (53%). These were embolic in 9 patients (34%), and watershed in 3 (12%). Hemorrhagic strokes were observed in 2 patients (8%). The occurrence of a new postoperative stroke in seizure patients did not significantly increase the likelihood of in-hospital death. Thirteen patients (41%) were discharged without neurologic deficit. Multivariable logistic regression analysis identified three risk factors for seizure: deep hypothermic circulatory arrest, aortic calcification or atheroma, and critical preoperative state. Conclusions. Seizures in adults after cardiac surgery are strong independent predictors of permanent neurologic deficit and increased operative mortality. Early head computed tomography may be indicated to identify treatable pathology. (Ann Thorac Surg 2011;91:514 9) 2011 by The Society of Thoracic Surgeons Seizures occurring after cardiac surgery are thought to be the result of focal or global cerebral ischemia from hypoperfusion, particulate or air emboli, metabolic derangements [1], or drug reactions including withdrawal. In infants and children, early postoperative seizures after cardiac surgery are markers of central nervous system injury and have been associated with adverse neurologic outcomes [2, 3]. While postoperative stroke and cognitive dysfunction in adult cardiac surgical cohorts are well characterized, there are few data on the incidence, epidemiology, and outcome of postoperative seizures. This study was, therefore, designed to analyze the epidemiology and outcome of seizures in a contemporary adult cardiac surgery population, with the aim of identifying predictors of seizures and determining their impact on postoperative morbidity, mortality, and 1-year survival. Accepted for publication Oct 4, Address correspondence to Dr Chikwe, Department of Cardiothoracic Surgery, 1190 Fifth Ave, Box 1028, New York, NY 10029; joanna.chikwe@mountsinai.org. Patients and Methods Patient Population and Data This study represents a retrospective analysis of 2,578 consecutive adult patients undergoing cardiac procedures at our institution between April 2007 and December The protocol was approved by the local Institutional Review Board with a waiver of informed consent. All patients at our institution are entered prospectively into a New York State mandated cardiac surgical database at the time of surgery. Complications occurring during the hospitalization (including stroke but excluding seizure) are recorded upon hospital discharge. Patient records were retrospectively reviewed for postoperative seizures and added to the prospective database. Patient characteristics and operations performed are presented in Table 1. A description of data collection methods and variable definitions is obtainable from the New York State Department of Health. Critical preoperative state was defined as the presence of ventricular tachycardia or fibrillation, preoperative cardiac massage, ventilation, inotropes, or intraaortic balloon pump. A seizure was defined as a sudden episode of transient neurologic symptoms featuring involuntary motor movements, or by electroencephalography in patients with nonconvulsive 2011 by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc doi: /j.athoracsur

2 Ann Thorac Surg GOLDSTONE ET AL 2011;91:514 9 SEIZURES AFTER SURGERY 515 Table 1. Patient Characteristics Variable No Seizures 2,547 (98.8%) n (%) Seizures 31 (1.2%) n (%) p Value Age, years, median (IQR) 64 (54 74) 73 (58 79) Female 943 (37.0%) 10 (32.3%) Diabetes mellitus 626 (24.6%) 2 (6.5%) Congestive heart failure 556 (21.8%) 9 (29.0%) Peripheral vascular disease 335 (13.2%) 13 (41.9%) Cerebrovascular disease 270 (10.6%) 5 (16.1%) Extensive aortic calcification or atherosclerosis 129 (5.1%) 6 (19.4%) Critical preoperative state 163 (6.4%) 4 (12.9%) Left ventricular function Poor 328 (12.9%) 5 (16.1%) Moderate 1,064 (41.8%) 12 (38.7%) Chronic obstructive pulmonary disease 154 (6.0%) 3 (9.7%) Prior coronary graft performed 122 (4.8%) 4 (12.9%) Reoperation 412 (16.2%) 10 (32.3%) Renal failure 89 (3.5%) 2 (6.5%) Urgent or emergent operation 876 (34.4%) 11 (35.5%) Prior myocardial infarction 483 (19.0%) 6 (19.4%) Endocarditis 91 (3.6%) 0 (0%) Hepatic failure 16 (0.6%) 0 (0%) Operation a CABG 764 (30.0%) 1 (3.2%) Isolated valve b 1,025 (40.2%) 10 (32.3%) Valve with CABG b 234 (9.2%) 6 (19.4%) Aorta 277 (10.9%) 13 (41.9%) Transplant or ventricular assist device 127 (5.0%) 1 (3.2%) Other 120 (4.7%) 0 (0%) Deep hypothermic circulatory arrest 204 (8.0%) 17 (54.8%) 0.01 a Includes reoperative procedures. Aorta includes ascending, arch, and descending thoracic aortic repair or replacement. procedures. CABG coronary artery bypass graft; IQR interquartile range. b Includes multivalve seizures. Status epilepticus was defined as continuous seizures or intermittent seizures without return of consciousness, lasting longer than 30 minutes. Seizures were classified by an attending neurologist based on clinical history and examination, reversing sedation and paralysis in all patients, with electroencephalography used to identify nonconvulsive seizure states. Neurology consultation was obtained in all cases of seizure. Computed tomography (CT) scans are currently performed for all seizures, but this protocol evolved over the study period: early practice was to not to perform scans for isolated seizures that resolved spontaneously with no residual neurologic defect after 24 hours. There was, however, no systematic evaluation of patients for postoperative seizures; recording of a seizure was based on clinical diagnosis in the course of standard patient management, and electrical encephalograms in patients suspected of seizure activity. Magnetic resonance imaging (MRI) scans were performed in patients with persistent neurologic deficits without positive findings on CT scan, if the patients were not inotrope dependent. Operative mortality was defined as death within the index hospitalization or within 30 days of operation, regardless of cause. Surgical Techniques Anesthesia, cardiopulmonary bypass, and surgical techniques varied according to physician preference. However, general patterns were noted as a result of institutional preferred practices and have been previously reported [4]. Briefly, all patients had intraoperative transesophageal echocardiographic evaluation of the thoracic aorta for atheroma and routine epiaortic ultrasonography to identify optimal cannulation and clamping sites. The ascending aorta was the preferred site for arterial cannulation. In patients undergoing aortic surgery, patients with aortic atheroma, and selected reoperations, the axillary artery was favored for arterial inflow. All patients except those with known prothrombotic states received aminocaproic acid at a dose of 150 mg/kg over 30 minutes, then 25 mg kg 1 h 1 for the duration of the case. Aprotinin and tranexamic acid were not used. Systemic cooling varied between 25 C and 34 C according to physician preference. Profound hypothermia (16 C to

3 516 GOLDSTONE ET AL Ann Thorac Surg SEIZURES AFTER SURGERY 2011;91: C) was consistently employed for procedures requiring circulatory arrest. On-pump coronary artery bypass grafting (CABG) was typically performed with a singleclamp technique. For off-pump CABG, proximal aortic anastomoses were executed with side-biting clamps or an anastomotic device. Deairing and venting procedures differed among surgeons, but carbon dioxide insufflation was routinely used in valve procedures and transesophageal echocardiography universally employed to visually assist deairing. Vents were discontinued while closing the atriotomies or aortotomies, often with warm blood being given through a coronary sinus catheter allowing the heart chambers to fill with blood. Before removal of the cross clamp, patients were placed in steep Trendelenberg position, caval snares removed, venous blood allowed to return to the heart, and air allowed to vent from the right atrium. The anesthetist performed recruitment maneuvers, the aortic root vent was opened to suction, and with the heart still full, the surgeon compressed the left ventricle to expel air. Venous return was restored to the bypass circuit, and the cross clamp was then removed during a period of low flow. A split 22G catheter was usually placed as an additional aortic vent. After removal of the cross clamp, those maneuvers were repeated with the heart beating under transesophageal echocardiographic guidance. Bypass was not discontinued until all gross air was either aspirated or ejected from the heart. In procedures performed under deep hypothermic circulatory arrest, particular care was applied to deairing the aorta, carotid circulation, and grafts before reclamping and restoration of flow. Database Validation Our prospective database is maintained in accordance with the New York State Department of Health for the auditing of cardiac surgical results in New York State. All hospitals within the state of New York divulge information to the Department of Health on every patient that underwent cardiac surgery. Approximately 45 risk factor variables and 8 outcome variables are gathered on each patient. The New York State Department of Health periodically validates our database by reviewing unusual reporting frequencies, cross-referencing our database with other data sources, and occasionally visiting our institution to review the medical records of a selected sample of cases. Fig 1. Association between seizures and postoperative morbidity; p less than for stroke, respiratory failure, renal failure, and sepsis; p less than 0.05 for bleeding and sternal infection. Statistical Analysis Comparisons between groups were assessed with the 2 test or Fisher s exact test for categorical data, the independent samples Student s t test for normally distributed continuous data, and the Mann-Whitney U test for nonnormally distributed data. Survival was evaluated with the Kaplan-Meier method, and groups were compared with log-rank testing. Univariate logistic regression was utilized to ascertain associations between preoperative characteristics and the occurrence of seizures. Variables with p less than 0.15 were included in a multivariable logistic regression analysis. A multivariable logistic regression model with stepwise selection was performed to identify independent predictors of seizures with p less than 0.10 required for inclusion. All tests were two-tailed, and a p value less than 0.05 was considered statistically significant. The statistical analysis was performed using SPSS for Macintosh, version 18.0 (SPSS, Chicago, IL). Results Incidence of Seizures During the study period, postoperative seizures were observed in 31 patients (incidence 1.2%). Of these, only 1 patient had a preoperative diagnosis of seizure disorder. Seizures occurred at a median of 2 days postoperatively, with 48% occurring within 24 hours of the index surgical procedure. Seizures were generalized tonic-clonic 71% (22 of 31), simple/complex partial 26% (8 of 31), or status epilepticus 3% (1 of 31). The incidence of seizures differed significantly according to procedure: coronary bypass 0.1% (1 of 765), isolated valve 1.0% (10 of 1,035), valve with coronary bypass 2.5% (6 of 240), aorta 4.5% (13 of 290), and ventricular assist device or transplant 0.8% (1 of 128; p 0.001). Burden of Seizures Patients who experienced seizures had nearly a fivefold higher hospital mortality than patients who did not (29% [n 9] versus 6% [n 153], p 0.001). Additionally, a significantly higher incidence of all major postoperative complications was noted in patients who had seizures (Fig 1). The difference in outcomes between seizure type did not reach statistical significance. Head CT was performed in 26 patients (84%), in whom pathology was identified in 14 cases (53%). These included embolic infarcts in 9 patients (34%) and watershed infarcts in 3 (12%). Intracranial hemorrhages were observed in 2 patients (8%). Although half of patients with seizures suffered a stroke, the occurrence of a new postoperative stroke in seizure patients did not significantly further increase the likelihood of in-hospital death. Of the 22 patients who survived to discharge, 13 (60%) were discharged without neurologic deficit. The burden of seizures

4 Ann Thorac Surg GOLDSTONE ET AL 2011;91:514 9 SEIZURES AFTER SURGERY 517 Fig 2. One-year survival after cardiac surgery stratified by occurrence of postoperative seizures. persisted after hospital discharge, with a lower 1-year survival rate for patients having postoperative seizures compared with patients not having seizures (53% 10% versus 84% 1%, respectively; p 0.001; Fig 2). Predictors of Seizures Eight of the 25 factors examined on univariate analysis were eligible for inclusion into the multivariable analysis. Multivariable analysis revealed three independent preoperative risk factors for seizures: deep hypothermic circulatory arrest, aortic calcification or atheroma, and critical preoperative state (Table 2). The area under the receiver operating characteristic curve for the multivariable model was 0.8. Temperature and duration of circulatory arrest were not significantly different between seizure and non-seizure groups. Comment Incidence and Epidemiology The incidence of postoperative seizures in this contemporary series of adult patients undergoing cardiac surgery was 1.2%. In the three studies we were able to identify that reported seizures after cardiac surgery (all of which were designed to evaluate the impact of antifibrinolytics on postoperative outcomes), the incidence ranged from 0.4% among patients undergoing coronary bypass surgery [5], to 1.3% in a series of patients undergoing valve, coronary, and complex aortic surgery [6], and 3.8% in adult cardiac surgery patients [7]. In all of these retrospective studies, which rely on clinical observation and documentation rather than systematic prospective screening, the incidence of seizures may be underestimated, particularly in the immediate postoperative period when patients are paralyzed or sedated. The main risk factors for seizures identified by the latter two studies included valve and open aortic surgery. Tranexamic acid, which was identified as a risk factor for postoperative seizures in all three studies, was not used in our patient series: all patients except those with known prothrombotic states received aminocaproic acid. The only study to report results of cross-sectional brain imaging in patients with postoperative seizures excluded patients with significant new lesions identified on brain CT from further analysis [6]. In the remaining 24 patients, small acute infarcts were identified in 3 cases, old infarcts in 4, and a subdural in 1 patient [6]. In our cohort, just over half of patients with seizures had intracranial pathology identified on CT, with embolic or watershed ischemic infarcts accounting for 12 of 14 of these cases. Although clearly it is impossible to state with certainty that these positive CT findings represent the main etiology of seizures in these patients, these data suggest that seizures are a marker of both focal and global cerebral injury, most commonly from ischemic events. The increased risk we observed from open heart as opposed to isolated coronary surgery suggests a causative role for air and particulate embolism. Outcome The increase in morbidity and mortality associated with seizures may be explained in several ways. Firstly, patients with seizures were at significantly higher risk than patients who did not have seizures, with a greater prevalence of risk factors associated with postoperative morbidity and decreased survival (Table 1), and that may only be partly accounted for by multivariable analysis. Secondly, seizures were probably more likely to occur in patients with increased risk of poor postoperative outcomes due to risk factors not included in our predictive model, such as inadequate intraoperative cerebral protection, drug withdrawal including alcohol abuse, and critical postoperative states. Finally, seizures may be a marker of significant neurologic injury even in the absence of acute positive findings on CT, or clinically detectable permanent neurologic deficit. Table 2. Multivariable Model for Predictors of Postoperative Seizures a Variable Deep hypothermic circulatory arrest Extensive aortic calcification/atheroma Critical preoperative state Constant Coefficient SEM OR p Value a c statistic 0.763; Hosmer-Lemeshow p OR odds ratio; SEM standard error of the mean.

5 518 GOLDSTONE ET AL Ann Thorac Surg SEIZURES AFTER SURGERY 2011;91:514 9 Implications for Clinical Practice We have found working closely with a neurology team with a specific interest in this area to be particularly helpful. Early CT may have a role in detecting potentially reversible causes of neurologic injury such as large emboli, cerebral edema, or intracranial bleeding. Such treatable pathology, however, represents a small proportion of the neurologic events in our series, underlining the importance of preventative strategies. Adjuncts to minimize cerebral injury in our practice include routine use of epiaortic scanning, carbon dioxide insufflation, intraoperative cerebral oxygen saturation monitoring, retrograde cerebral perfusion with hypothermia or deep hypothermia during circulatory arrest, and echocardiography-guided deairing in all patients. These adjuncts may have been associated with a decrease in the risk of postoperative stroke in our practice [8, 9], although this study was not designed to provide data on whether they have reduced the risk of seizures. Based on these data, it is difficult to advocate a conservative approach to the use of circulatory arrest in cases where this approach is optional rather than mandated by surgical anatomy, as the adverse sequels of seizures and neurologic deficit may be offset by the advantages of circulatory arrest. Is there a role for more extensive use of MRI in these patients to better define the etiology of neurologic injury? Magnetic resonance imaging was performed in patients with persistent neurologic deficits without clear positive findings on CT scan if the patients were not inotrope dependent. We did not routinely perform MRI scans, however, for several reasons. The first issue with MRI in this context is that it is not a particularly specific test: approximately 5% of patients have a diffusion defect visible on MRI before cardiac surgery [10], as many as 45% of patients have new lesions on MRI after cardiac surgery [11 13], and no studies have managed to correlate acute findings with neurocognitive deficit. Secondly, in a patient group where air embolism is a particular concern, not only does MRI not differentiate between air and particulate embolism, but also CT scan is a much more sensitive diagnostic test for large air embolism. Finally, our institutional protocols preclude MRI imaging in patients with epicardial pacing wires, which form the majority of cases in the early perioperative period. Strengths and Limitations Our study provides new data for the incidence, epidemiology, and outcome of postoperative seizures in adult cardiac surgery patients. A more detailed data analysis is limited by the small number of patients who presented with seizures, and as outlined above, it is impossible to account for important confounding variables such as inadequate intraoperative cerebral protection. The observational nature of this study precludes drawing firm conclusions about the role of preventative strategies in reducing the incidence of seizures in this patient group. The lack of prospective, systematic screening means that the incidence of seizures, particularly isolated focal seizures during the immediate postoperative period, is likely underestimated. In conclusion, seizures occur in approximately 1% of adults after cardiac surgery and are strong predictors of permanent neurologic deficit and increased operative mortality. Early head CT may be indicated to identify treatable pathology, but preventive strategies aimed at optimizing cerebral protection intraoperatively are a key part of improving outcomes for at-risk patients. This work was supported by a grant from the Doris Duke Charitable Foundation to the Mount Sinai School of Medicine to fund Clinical Research Fellow Andrew B. Goldstone. References 1. Bronster DJ. Neurologic complications of cardiac surgery: current concepts and recent advances. Curr Cardiol Rep 2006;8: Bellinger DC, Jonas RA, Rappaport LA, et al. Developmental and neurologic status of children after heart surgery with hypothermic circulatory arrest or low-flow cardiopulmonary bypass. N Engl J Med 1995;332: Bellinger DC, Wypij D, Kuban KC, et al. Developmental and neurological status of children at 4 years of age after heart surgery with hypothermic circulatory arrest or low-flow cardiopulmonary bypass. Circulation 1999;100: Anyanwu AC, Filsoufi F, Salzberg SP, Bronster DJ, Adams DH. Epidemiology of stroke after cardiac surgery in the current era. J Thorac Cardiovasc Surg 2007;134: Roach GW, Kanchuger M, Mangano CM, et al. Adverse cerebral outcomes after coronary bypass surgery. Multicenter Study of Perioperative Ischemia Research Group and the Ischemia Research and Education Foundation Investigators. N Engl J Med 1996;335: Murkin JM, Falter F, Granton J, Young B, Burt C, Chu M. High-dose tranexamic acid is associated with nonischemic clinical seizures in cardiac surgical patients. Anesth Analg 2010;110: Martin K, Wiesner G, Breuer T, Lange R, Tassani P. The risks of aprotinin and tranexamic acid in cardiac surgery: a one-year follow-up of 1188 consecutive patients. Anesth Analg 2008;107: Filsoufi F, Rahmanian PB, Castillo JG, Bronster D, Adams DH. Incidence, imaging analysis, and early and late outcomes of stroke after cardiac valve operation. Am J Cardiol 2008;101: Filsoufi F, Rahmanian PB, Castillo JG, Bronster D, Adams DH. Incidence, topography, predictors and long-term survival after stroke in patients undergoing coronary artery bypass grafting. Ann Thorac Surg 2008;85: Maekawa K, Goto T, Baba T, Yoshitake A, Morishita S, Koshiji T. Abnormalities in the brain before elective cardiac surgery detected by diffusion-weighted magnetic resonance imaging. Ann Thorac Surg 2008;86: Gerriets T, Schwarz N, Sammer G, et al. Protecting the brain from gaseous and solid micro-emboli during coronary artery bypass grafting: a randomized controlled trial. Eur Heart J 2010;31: Knipp SC, Matatko N, Wilhelm H, et al. Evaluation of brain injury after coronary artery bypass grafting. A prospective study using neuropsychological assessment and diffusionweighted magnetic resonance imaging. Eur J Cardiothorac Surg 2004;25: Restrepo L, Wityk RJ, Grega MA, et al. Diffusion- and perfusion-weighted magnetic resonance imaging of the brain before and after coronary artery bypass grafting surgery. Stroke 2002;33: