The influence of anesthetic technique on perioperative complications after carotid endarterectomy

Transcription

1 The influence of anesthetic technique on perioperative complications after carotid endarterectomy Brent T. Allen, MD, Charles B. Anderson, MD, Brian G. Rubin, MD, Robert W. Thompson, MD, M. Wayne FIye, MD, PhD, Patricia Young-Beyer, MD, Peggy Frisella, RN, and Gregorio A. Sicard, MD, St. Louis) Mo. Purpose: This study evaluated the influence of anesthetic techniques on perioperative complications after carotid endarterectomy. Methods: Perioperative complications, the use of a carotid artery shunt, the duration of the operative procedure and postoperative hospital course were retrospectively compared in 584 consecutive patients undergoing 679 carotid endarterectomies with use of either general anesthesia (n = 361) or cervical block regional anesthesia (n = 318). There was no significant difference in the preoperative medical characteristics between the two anesthetic groups. Symptomatic carotid artery disease was the indication for surgery in 247 (68.4%) patients receiving general anesthetics, whereas 180 (56.6%) patients treated with a cervical block anesthetic had a symptomatic carotid artery stenosis (p = 0.02). Results: The perioperative stroke rate and stroke-death rate for the entire series was 2.4% and 3.2%, respectively, and was not significantly different between the anesthetic groups or between patients with symptomatic or asymptomatic disease. A carotid artery shunt was used in 61 (19.2%) patients receiving a cervical block anesthetic and 152 (42.1%) patients treated with a general anesthetic (p < ). Use of cervical block anesthesia was associated with a significantly shorter operative time, fewer perioperative cardiopulmonary complications, and a shorter postoperative hospitalization when compared with general anesthesia. Multivariate risk factor analysis indicated that age greater than 75 years, operative time greater than 3 hours, and the use of a carotid artery shunt were all independent risk factors for perioperative cardiopulmonary complications. When a carotid artery shunt was not analyzed as a multivariate risk factor, then general anesthesia became a significant risk factor for perioperative cardiopulmonary complications (risk ratio 2.08; p = 0.04). Conclusions: We conclude that cervical block anesthesia is safer and results in a more efficient use of hospital resources than general anesthesia in the treatment of patients undergoing carotid endarterectomy. (J VAse SURG 1994;19: ) The North American Symptomatic Carotid Endarterectomy Trial study and the report from the Veterans Affairs Cooperative Study Group have firmly established the benefits of carotid endarterectomy in the treatment of patients with high-grade carotid bifurcation stenoses.1,2 However, in spite of the benefit of surgery when compared with optimal medical From the Section of Vascular Surgery, Department of Surgery, and Department of Anesthesia (Dr. Young-Beyer), Washington University School of Medicine, St. Louis. Presented at the Seventeenth Annual Meeting of the Midwestern Vascular Surgical Society, Chicago, Ill., Sept , Reprint requests: Brent T. Allen, MD, One Barnes Plaza, Suite 5103 QT, St. Louis, MO Copyright 1994 by The Society for Vascular Surgery and International Society for Cardiovascular Surgery, North American Chapter /94/$ /6/ treatment, perioperative morbidity and death, primarily from neurologic and cardiac complications, remain a significant problem in the treatment of cerebrovascular disease. The anesthetic technique used during carotid endarterectomy is an important factor in these perioperative complications. Efforts to minimize neurologic complications during general anesthesia have centered around the intraoperative detection of cerebral ischemia to determine the need for a carotid artery shunt. Electroencephalographic monitoring or carotid artery stump pressure is often used to assess cerebral perfusion during carotid artery occlusion with the patient under general anesthesia but have been shown to be inaccurate by many investigators. 3 - S Clinically obvious or occult heart disease frequently is present in patients with carotid atherosclerosis. 6,7 General anesthesia increases the cardiac

2 JOURNAL OF VASCULAR SURGERY Volume 19, Number 5 Allen et at. 835 risk in such patients and may result in some patients being denied surgery because of severe heart disease. Carotid endarterectomy in the awake patient receiving a local or regional anesthetic has been advocated as a means of avoiding the limitations of general anesthesia. Our group began using cervical block anesthesia in the awake patient as an alternative to general anesthesia in 1990 in an attempt to improve our results with this operation. This change in anesthetic technique has been the only major patient treatment variable in our treatment of carotid atherosclerosis over the past 6 years. This report retrospectively compares our results with carotid endarterectomy with either general anesthesia or cervical block anesthesia during this time period. PATIENTS AND METHODS The vascular registry at Washington University School of Medicine retrospectively reviewed the records of 584 consecutive patients undergoing 679 carotid endarterectomies during the last 6 years (April 1987 to March 1993) on the vascular surgery service at Barnes Hospital in St. Louis, Mo. Patients treated with combined carotid and coronary artery revascularization were excluded. Postoperative follow-up information was obtained through office visit records, letters to referral physicians, and phone calls to patients. Follow-up was available for 617 (90.7%) operations at a mean of 23.0 months after operation. The patients were divided into two groups on the basis of the anesthetic technique used during the operation. The annual frequency of each anesthetic technique over the 6-year study period is shown in Fig. 1. Three hundred sixty-one (53.2%) operations used general anesthesia; 318 (46.8%) were performed with the patient receiving cervical block anesthesia. Cervical block anesthesia used both a deep and superficial block. The block was instituted after the patient was in the operative position and the monitoring equipment (electrocardiograph, arterial line) had been placed. Intravenous sedation with 1 to 2 cc fentanyl citrate and 1 to 2 cc midazolam hydrochloride was administered to all patients before placing the block. The deep cervical block was performed by injecting 5 cc 0.5% bupivacaine hydrochloride approximately 2 mm lateral to each ipsilateral transverse process of C 2, C 3, and C 4 A superficial cervical block was achieved with 10 cc of 0.5% bupivacaine hydrochloride injected subcutaneously along the posterior border of the ipsilateral sternocleidomastoid muscle from origin to insertion. During the course of the operation lidocaine hydrochloride (0.5%) was used as necessary in the surgical field to avoid patient discomfort. The surgical drapes were positioned to construct a "tent" about the patient's face to minimize claustrophobic anxiety. An electric fan was secured to the head of the operating table to keep the patient's face cool and promote air circulation. During the course of the operation the neurologic status of the patient was continuously monitored by the patient's response to frequent questions and by the ability to squeeze a toy squeaker in the contralateral hand. 8 General endotracheal anesthesia was performed with a continuous intravenous infusion of alfentanil hydrochloride and nitrous oxide/oxygen inhalation anesthetic. Intraoperative neurologic evaluation used electroencephalographic monitoring or internal carotid artery stump pressure. Patient anxiety prompted conversion to general anesthesia after placement of the cervical block but before making the skin incision in one (0.1%) patient. The medical characteristics of these two patient groups were not significandy different. (Table I). Bilateral operations were performed in 87 patients. Symptomatic carotid artery disease prompted 247 (68.4%) carotid endarterectomies with general anesthesia, whereas 180 (56.6%) carotid endarterectomies were performed with the patient receiving a cervical block anesthetic for symptomatic carotid artery lesions (p = 0.02). There was no significant difference in the angiographic degree of carotid artery stenosis between the anesthetic groups. Symptomatic carotid artery lesions measured angiographically were 80% stenotic or greater in 290 (67.9%) cases, whereas asymptomatic disease had 80% stenosis or greater in 184 (73.0%) cases. Carotid endarterectomy was performed with standard surgical techniques. A carotid patch angioplasty (autologous vein or prosthetic) was used in 45 (12.5%) operations in the general anesthetic group and 39 (12.3%) operations in the cervical block group. Carotid artery shunts were used selectively in both anesthetic groups. Indications for a carotid artery shunt during general anesthesia included contralateral carotid artery occlusion, recent ipsilateral stroke, electroencephalographic changes after placement of the internal carotid artery clamp and a low carotid artery stump pressure (usually less than 40 mm Hg). A shunt was placed in patients receiving cervical block anesthetic when a deterioration in the clinical neurologic status was noted after ipsilateral internal carotid artery occlusion. A carotid artery shunt was used in 152 (42.1%) of operations performed with the patient receiving general anesthetic and 61 (19.2%) cases performed with the patient receiving cervical block anesthetic (p < ). All patients were monitored in the surgical

3 836 Allen et at. JOURNAL OF VASCULAR SURGERY May 1994 ANESTHESIA FOR CAROTID ENDARTERECTOMY N Vascular Surgery Service Barnes Hospital Washington University School of Medicine General Cervical Block "- I ~ A. ~ -~ ~ L.. """"""" "\ " ~ o ~ PerIod Fig. 1. Frequency of general and cervical block anesthesia during 6-year study period. Table 1. Medical characteristics General anesthesia (n = 361) Age (X ± SEM) (years) 69.2 ± 0.5 Sex Male 213 (59.0%) Female 148 (41.0%) Hypertension 217 (60.1%) Diabetes 90 (24.9%) Smoking (pack years) (X ± SEM) 32.1 ± 1.9 Chronic obstructive pulmonary 71 (19.7%) disease Congestive heart failure 26 (7.2%) Angina 92 (25.5%) Myocardial infarction 53 (14.7%) Coronary artery bypass graft 55 (15.2%) Abnormal electrocardiography 152 (42.1%) result Previous vascular surgery 63 (17.5%) Cervical block anesthesia Total (n = 679) (n = 318) (584 patients) 70.2 ± ± (58.8%) 400 (58.1%) 131 (41.2%) 279 (41.9%) 195 (61.3%) 412 (60.7%) 85 (26.7%) 175 (25.8%) 35.4 ± ± (15.1%) 119 (17.5%) 25 (7.9%) 51 (7.5%) 75 (23.6%) 167 (24.6%) 63 (19.8%) 116 (17.1%) 63 (19.8%) 118 (17.4%) 171 (53.8%) 323 (47.6%) 46 (14.5%) 109 (16.1%) intensive care unit after surgery at least until the first postoperative day regardless of the method of anesthesia. DEFINITIONS AND STATISTICAL ANALYSIS The two anesthetic techniques were compared to identify differences in operative time, hospital stay, and perioperative complications and deaths with use of all the procedures (n = 679) performed in each anesthetic group. Perioperative morbidity was divided into neurologic and nonneurologic compli- cations that occurred within 30 days of surgery. Neurologic complications were subdivided into temporary (transient ischemic attack, amaurosis fujax, seizure) or permanent (stroke, retinal infarction) neurologic events. A stroke was defined as any new postoperative deficit present at hospital discharge or any new radiologic (computed tomography, magnetic resonance imaging) evidence of stroke regardless of symptoms at discharge. Nonneurologic complications were subdivided into cardiopulmonary complications or wound complications. Cardiopulmonary complications were defined as postoperative hyper-

4 JOURNAL OF VASCULAR SURGERY Volume 19, Number 5 Allen et al. 837 tension, hypotension, arrhythmia, congestive heart failure, or angina requiring parenteral pharmacologic treatment. Myocardial infarction and respiratory complications (pneumonia, intubation greater than 48 hours, reintubation) were also included in cardiopulmonary complications. Electrocardiography was performed on all patients on admission to the intensive care unit. The diagnosis of a myocardial infarction was based on elevated myocardial enzymes, which were obtained when myocardial ischemia was suspected (electrocardiographic abnormality, hypotension, arrythmia). Wound complications included hemorrhage, infection, and cranial nerve injuries. Cranial nerve injuries were considered complications only if they were permanent. A univariate and multivariate risk factor analysis was performed to identify risk factors important.in perioperative ( < 30 days) deaths and complications. Each patient was considered only once (n = 584) in this analysis; therefore data from the second operation was not included in patients treated with bilateral carotid endarterectomies (n = 87) or those who underwent operation for recurrent carotid artery disease (n = 6) to avoid excessive statistical weighting by this subset of patients in the risk factor analysis. The risk factors analyzed were age, sex, preoperative medical history (hypertension, diabetes, smoking, lung disease, or heart disease), previous vascular surgery, anesthetic technique, operative time, indication for carotid artery surgery, the use of a carotid artery shunt, and contralateral carotid artery occlusion. General comparisons of operative time, postoperative complications, and deaths for all cases in each anesthetic group were performed with Fisher's twotailed exact test. Comparisons of median postoperative hospitalization used a Wilcoxon rank sum test. Fisher's two-tailed exact test was used for univariate risk factor analysis. Regression analysis was used for multivariate risk factor analysis. All statistical analyses were performed by the Division of Biostatistics and Washington University School of Medicine. RESULTS Perioperative morbidity and mortality Perioperative neurologic events occurred after 35 (5.15%) operations (Table II). Nineteen (2.8%) were temporary neurologic deficits or perioperative seizures, and 16 were strokes for a perioperative stroke rate of 2.4%. Ten (2.8%) of the operations performed with the patient receiving general anesthetic were complicated by stroke, whereas six (1.9%) carotid endarterectomies performed with the Table II. Complications General anesthesia GeYPical block anesthesia Neurologic TIA 12 (3.3%) 7 (2.2%) Stroke 10 (2.8%) ~ (1.9%) TOTAL 22 (6.1%) 13 (4.1%) Cardiopulmonary Blood pressure instability 8 (2.2%) 4 (1.3%) MI 9 (2.5%) 2 (0.6%) Arrhythmia 9 (2.5%) 3 (0.9%) Congestive heart failure, angina o (0%) 2 (0.6%) Respiratory...! (1.1%) ~ (0.6%) TOTAL 30 (8.3%) 13 (4.0%) Wound HematOma/infection 5 (1.4%) 2 (0.6%) Cranial nerve injury...! (0.3%)...! (0.3%) TOTAL ~ (1.7%) 2. (0.9%) Total complications 58 (16.1%) 29 (9.1%) TIA, Transient ischemic artack; MI, myocardial infarction. patients receiving cervical block anesthetic were associated with stroke (p = 0.79). Operations for symptomatic disease were complicated by stroke in 13 (3.0%) patients, whereas perioperative stroke occurred in three (1.2%) patients with an asymptomatic stenosis (p = 0.19). A perioperative neurologic event resulted in prompt reoperation in eight patients after general anesthesia and three patients after cervical block. Reoperation identified a source for the symptoms (thrombotic debris, intimal flap) in the carotid artery in 10 of the 11 (90.9%) patients. Four (50%) patients in the general anesthesia group had persistent deficits (stroke) after reexploration while the deficit resolved after reexploration in all patients in the cervical block group. Cardiopulmonary complications occurred in 30 (8.3%) operations performed with the patientreceiving general anesthetic and in 13 (4.1 %) operations performed with the patient receiving cervical block anesthetic (p = 0.03) (Table II). Myocardial infarction occurred during operation after 11 (1.6%) operations; nine (2.5%) after general anesthesia and two (0.6%) after cervical block (p = 0.07). Arrhythmias complicated nine (2.5%) general anesthetics and three (1.2%) local anesthetics. Blood pressure instability was noted after eight (2.2%) general anesthetics and four (1.3%) cervical blocks. Postoperative hypertension requiring pharmacologic management was necessary in five patients after administration of general anesthetic and three patients after administration of cervical block. Hypotension occurred in three patients receiving general anesthetic and one patient receiving cervical block. Congestive

5 838 Allen et al. JOURNAL OF VASCULAR SURGERY May 1994 Table III. Perioperative deaths Cause Stroke Cardiac Hemorrhage Respiratory Total General anesthesia o (0%) 3 (0.8%) o (0%) 1 (0.3%) 4 (1.1%) Cervical block anesthesia o (0%) 1 (0.3%) 1 (0.3%) o (0%) 2 (0.6%) heart failure required intensive postoperative treatment in two (0.6%) patients receiving a cervical block. Respiratory complications occurred after administration of general anesthetic in four (1.1 % ) patients and after administration of cervical block in two (0.6%) patients. Wound complications occurred after six (1.7%) operations in the general anesthetic group and three (0.9%) operations in the cervical block group. Hematomas accounted for all but three of the wound complications and were surgically treated in four patients. Carotid artery suture line disruption resulted in the two wound hematomas in the cervical block group, each requiring emergency reoperation. This complication led to a severe stroke in one patient and death in the other. One patient had an operative injury to the hypoglossal nerve during general anesthesia, and in one patient persistent hoarseness caused by a vagal nerve injury developed after cervical block anesthesia. A wound infection developed in one patient after general anesthesia. The perioperative ( < 30 days) mortality rate for the entire series was 1.0% (6 patients) and is shown in Table III. Four (1.1%) deaths occurred after general anesthesia and two (0.6%) occurred after cervical block. Four deaths were heart-related in origin, one patient died of respiratory failure, and one patient died after a carotid artery suture line disruption. There were no perioperative deaths from stroke. Risk factor analysis The univariate risk of stroke, stroke and death, myocardial infarction, a cardiopulmonary complication, and all death, stroke, or cardiopulmonary complications between the patients in each anesthetic group are compared in Table IV. Patients with symptoms were not at increased risk for any complication when compared with symptom-free patients. Patients with contralateral occlusions were not at increased risk of stroke when compared with the group as a whole (p = 0.94). There was no significant difference in the number of strokes or strokes and deaths between the anesthetic groups. Patients receiving a general anesthetic were at a significantly higher risk for development of a cardiopulmonary complication than patients receiving local anesthetics (relative risk ratio 2.14). The risk of myocardial infarction was 4.04 times greater during general anesthesia than cervical block, but this difference did not quite reach statistical significance (p = 0.08). The cumulative risk of a stroke, cardiopulmonary complication, or death was significantly greater with general anesthesia by a factor of 1.81 when compared with local anesthesia. Multivariate analysis did not identify any independent risk factors for postoperative stroke or death. Independent risk factors for postoperative cardiopulmonary complications consisted of operative time greater than 3 hours (p = ) and the use of a carotid artery shunt (p = 0.006). Age greater than 75 years approached significance as a risk factor (p = 0.06) (Table V). Because a carotid artery shunt as a risk factor in postoperative cardiopulmonary complications would seem to have no physiologic basis, we recalculated the multivariate risk excluding a carotid artery shunt from the analysis (Table V). In this multivariate analysis, age greater than 75, operative time and anesthetic technique were significant predictors of cardiac complications. General anesthesia was associated with greater than a twofold increased risk of cardiopulmonary complications when compared to cervical block. Operative times and hospitalization Patients requiring staged bilateral carotid endarterectomy and those undergoing surgery for recurrent disease were excluded from this statistical analysis to avoid bias. The average operative time for patients receiving general anesthetic was 2.04 ± 0.04 hours and for those receiving cervical block 1.70 ± 0.02 hours (p = ) (Table VI). The median length of postoperative hospital stay was 4 days (range 1 to 61) for patients in the general anesthetic group and 3 days (range 1 to 66) in patients receiving cervical block anesthetic (p = ). General anesthesia was associated with a hospital stay longer than 3 days in 170 (54.7%) patients, whereas 99 (36.8%) patients remained in the hospital more than 3 days after cervical block anesthesia (p = ). Heart disease was a primary factor accounting for prolonged postoperative hospitalization. Patients with heart disease were nearly two times more likely to remain in the hospital for more than 3 days after carotid endarterectomy

6 JOURNAL OF VASCULAR SURGERY Volume 19, Number 5 Allen et at. 839 Table IV. Death, stroke, or cardiopulmonary complication (Univariate analysis*) General Cerpjcal Relative Confidence Complication anesthesia block risk ratio interval Significance Stroke Strokes and death Cardiopulmonary Myocardial infarction Total (stroke, death, or cardiopulmonary complications) 9 (2.9%) 13 (4.2%) 30 (9.6%) 9 (2.9%) 38 (12.2%) 6 (2.2%) 8 (2.9%) 13 (4.8%) 2 (0.7%) 20 (7.3%) p = 0.79 P = 0.51 P = 0.03 P = 0.08 P = 0.04 *Each patient counted only once. General anesthesia (n = 311), cervical block (n = 273). Table V, A. Cardiopulmonary complications multivariate risk factors Relative risk ratio Confidence interpal Significance Age > 75 years Operative time > 3 hours Carotid artery shunt General anesthesia Table V, B. Cardiopulmonary complications multivariate risk factors (shunt excluded) Age > 75 years Operative time > 3 hours General anesthesia Relative risk ratio Confidence interpal Significance p = 0.04 P = P = 0.04 Table VI. Operative and Hospital Requirements General anesthesia Cerpjcal block anesthesia Significance Operative time (hours) (X ± SEM) Postoperative hospital stay (median days) 2.04 ± (range 1-61) 1.70 ± (range 1-66) p = P = than patients without heart disease (relative risk ratio 1.95; p = ). DISCUSSION The success and safety of carotid endarterectomy is impressive, with several multicenter prospective and retrospective studies reporting operative strokedeath rates of 3% to 6%.1,2,9 The technical aspects of carotid endarterectomy are well established, and it appears unlikely that additional modifications in the existing surgical technique will lead to substantial improvements in operative morbidity and mortality rates. This report shifts the emphasis from surgical to anesthetic technique as a means of improving perioperative morbidity and mortality after carotid artery surgery. This review compared two anesthetic techniques with a statistically valid number of patients from a single institution collected over a 6-year period of time. Although the power of this study is limited by its retrospective nature, comparison of the two patient groups is statistically valid, because during the study period all patients were treated by the same anesthesia and surgical staff in a similar manner except for the method of anesthesia. We found that cervical block anesthesia, when compared with general anesthesia by use of univariate or multivariate analysis, significantly reduced the following variables: operative time, the use of a carotid artery shunt, postoperative cardiopulmonary complications, and the duration of postoperative

7 840 Allen et at. JOURNAL OF VASCULAR SURGERY May 1994 hospitalization. Other authors have made similar observations. Peitzman and coworkers lo noted an incidence of nonneurologic complications in patients receiving general anesthetic of 12.9% (all were cardiopulmonary complications) compared with 2.8% in those receiving regional anesthetic. A nonrandomized study of 399 carotid endarterectomies reported by Corson et al. ll demonstrated that patients receiving general anesthesia had significantly more postoperative blood pressure instability and a longer intensive care unit stay than patients treated with cervical block. Finally, Muskett and associates 12 found that patients receiving cervical block anesthetic during carotid endarterectomy required less intravenous fluid during operation and had a shorter postoperative hospital stay when compared with patients administered general anesthesia. In contrast to these studies Forssell and associates 13 found in a prospective randomized group of 101 patients that local anesthesia was associated with more intraoperative hypertension than general anesthesia. Similarly Gabelman et al. 14 noted significantly greater intraoperative blood pressure instability with local anesthesia versus general anesthesia. However, in spite of a labile intraoperative blood pressure, patients in the study by Gabelman receiving cervical block anesthesia had significantly shorter operative times, intensive care unit stays, and postoperative hospitalizations. Our analysis focused on postoperative rather than intraoperative cardiopulmonary complications. We found that blood pressure instability, arrhythmias, and myocardial infarction were all more common after general anesthesia. The difference in the incidence of myocardial infarction between the two anesthetic techniques approached statistical significance (p = 0.08). Comparison of all cardiopulmonary complications in the aggregate did reach statistical significance in favor of cervical block anesthesia (p = 0.04). The results from all these studies suggest that, although the intraoperative blood pressure may be more labile, overall postoperative cardiopulmonary complications are less with regional anesthesia. Carotid endarterectomy performed with regional anesthesia also seems to consume fewer hospital resources in terms of short operative times and hospital stays. It has been estimated that the cost of 100,000 carotid endarterectomies in 1992 was approximately $2 billion. IS The number of carotid endarterectomies would be expected to rise as the results of the recent prospective randomized studies demonstrating the efficacy of carotid endarterectomy become better known. 1,2 Our findings of decreased operative time and postoperative hospital stay in patients receiving cervical block anesthetic confirm the studies from Gabelman 14 and Muskett 12 and suggest that the use of cervical block anesthesia may be a more cost-effective anesthetic technique for carotid endarterectomy. Indeed, Gabelman et al. 14 noted a 29% savings in total hospital costs when local anesthetic was used for carotid endarterectomy. A carotid artery shunt was used significantly less with cervical block anesthetic in this study. This fact has been well documented in previous investigations. 13,14 Clinical neurologic assessment of the awake patient has been shown to be more accurate in detecting significant cerebral ischemia than electroencephalographic monitoring or the measurement of carotid stump pressure. In a study of 125 awake patients treated with carotid endarterectomy, Kwaan et al. 4 found that 25 (20%) patients demonstrated no neurologic deficits after internal carotid artery occlusion, even though the carotid artery stump pressure was < 50 mm Hg. In 24 of their patients in whom a neurologic deficit developed after cross-clamping, the stump pressure ranged from 15 to 85 mm Hg and was greater than 50 mm Hg in eight of the 24 (33%) patients. Optimal cerebral monitoring was prospectively investigated by Evans et al. 3 in 134 carotid endarterectomies performed in the awake patient monitored with electroencephalographic and carotid stump pressures. They concluded that neither electroencephalographic monitoring nor carotid artery stump measurements identified all patients who required a carotid artery shunt, and carotid artery clamping in the awake patient appeared to be the most reliable indicator of the need for a carotid artery shunt. Similar conclusions were made in a large (1200 carotid endarterectomies) prospective report of carotid endarterectomy with use of regional anesthetic by Hafner and Evans. S They found that 343 of 401 (86%) patients with a stump pressure less than 50 mm Hg would have been shunted urmecessarily on the basis of neurologic examination results during the operation. Neither stump pressure nor electroencephalographic changes accurately predicted postoperative stroke or the need for shunting. The use of a shunt is not without risk. Prioleau and associates 16 noted that the use of a carotid artery shunt was more commonly associated with embolic stroke during carotid endarterectomy. Fode et al. 17 in a multicenter review of more than 3000 cases of carotid endarterectomy noted that nonmonitored patients who underwent operation without a shunt did significantly better after operation than nonmonitored patients who underwent shunting. We did not observe an increased risk for neurologic complica-

8 JOURNAL OF VASCULAR SURGERY Volume 19, Number 5 Allen et at. 841 tions with the use of a shunt, but did find it to be a prominent risk factor for cardiopulmonary complications. The physiologic basis for a relationship between a' carotid artery shunt and perioperative cardiopulmonary complications is not obvious. We suspect that because a shunt was used more than twice as often in patients during general anesthesia than during cervical block, the use of a shunt and the method of anesthesia may be acting as statistical proxies for one another. The fact that general anesthesia significantly increased the risk of a cardiopulmonary complication on univariate analysis and that general anesthesia became a significant independent risk factor for cardiopulmonary complications when the use of a carotid artery shunt was deleted from the multivariate analysis lends support to this hypotheses. These observations suggest that general anesthesia is the more important of these two variables in the development of a perioperative cardiopulmonary complication. However, even if the use of a shunt during carotid endarterectomy does predispose to a cardiopulmonary complication, cervical block anesthesia would be associated with less risk than general anesthesia because of the marked difference in the use of a shunt between these anesthetic techniques. We conclude from this review that carotid endarterectomy performed with the patient receiving cervical block anesthesia is safer and uses less hospital resources when compared with use of a general anesthetic. A crucial element required for the success of regional anesthesia during carotid artery surgery is cooperation between the patient, surgeon, and anesthesiologist. A thorough preoperative informative discussion will minimize the patient's anxiety about the procedure. The patient should be made to feel part of the operative team and reassured that any discomfort or anxiety experienced during the procedure will be rapidly relieved. The anesthesiologist and surgeon need to work together to maintain a stable airway, provide adequate analgesia, and assess the neurologic status of the patient. The success of this approach in our practice is evidenced by Fig. 1 showing the frequency of each anesthetic technique over the 6-year study period. As we became more familiar with cervical block anesthesia, its use increased rapidly. It has been well received by patients, who refused this method of anesthesia very rarely (less than 10 times over the study period). Perhaps the best measure of acceptance of cervical block anesthesia is the number of times the original (cervical block or general) anesthetic technique was changed in the 87 patients who underwent a second carotid endarterectomy for contralateral disease. Three (6%) patients who had a general anesthetic for their first carotid artery surgery switched to cervical block for the second operation. Only one (2.6%) patient who underwent a cervical block for the first carotid endarterectomy received a general anesthetic. for the second operation. The change in anesthetic technique in this patient was requested by the surgeon because of a very distal and calcific carotid artery lesion. We conclude from our experience that cervical block anesthesia offers significant advantages over general anesthesia in overcoming the limitations in intraoperative cerebral monitoring and perioperative cardiac risk. REFERENCES 1. North American Symptomatic Carotid Endarterectomy Trial Collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high grade carotid stenosis. N Engl J Med 1991;325: Hobson R W II, Weiss DG, Fields WS, et al. Efficacy of carotid endarterectomy for asymptomatic carotid stenosis. N Engl J Med 1993;328: Evans WE, Hayes JP, Waltke EA, Vermilion BD. Optimal cerebral monitoring during carotid endarterectomy: neurologic response under local anesthesia. J VASC SURG 1985;2: Kwaan JHM, Peterson GJ, Connolly JE. Stump pressure. Arch Surg 1980;115: Hafuer CD, Evans WE. Carotid endarterectomy with local anesthesia: results and advantages. J VASC SURG 1988;7: Hertzer NR, Young JR, Beven EG, et al. Coronary angiography in 506 patients with extracranial cerebrovascular disease. Arch Intern Med 1985;145: Hertzer NR. Basic data concerning associated coronary disease in peripheral vascular patients. Ann Vasc Surg 1987;1: Spielberger L, Turnderf H, Culliford A, Imparato A. Handheld toy squeaker during carotid endarterectomy in the awake patient. Arch Surg 1979;114: Rubin JR, Pitluk HC, King T A, Hurton M, Kieger EF, Plecha FR, Hertzer NR. Carotid endarterectomy in a metropolitan community: the early results after 8535 operations. J V ASC SURG 1988;7: Peitzman AB, Webster MW, Loubeau JM, Grundy BL, Bahnson HT. Carotid endarterectomy under regional (conductive) anesthesia. Ann Surg 1982;196: Corson JD, Chang BB, Shah DM, Leather RP, DeLeo BM, Karmody AM. The influence of anesthetic choice on carotid endarterectomy outcome. Arch Surg 1987;122: Muskett A, McGreevy J, Miller M. Detailed comparison of regional and general anesthesia for carotid endarterectomy. Amer J Surg 1986;152: Forssell C, Takolander R, Bergqvist D, Johansson A, Persson NH. Local versus general anaesthesia in carotid surgery: a prospective, randornised study. Eur J Vasc Surg 1989;3:

9 842 Allen et al. JOURNAL OF VASCULAR SURGERY May Gabelman CG, Gann DS, Ashworth CJ Jr, Carney WI Jr. One hundred consecutive carotid reconstructions: local versus general anesthesia. Amer J Surg 1983;145: Dyken ML. Carotid endarterecromy studies: a glimmering of science. Editorial. Stroke 1986;17: Prioleau WH Jr, Aiken AF, Hairston P. Carotid endarterectomy: neurologic complications as related to surgical techniques. Ann Surg 1977;185: Fode NC, Sundt TM Jr, Robertson JT, Peerless SJ, Shields CB. Multicenter retrospective review of results and complications of carotid endarterectomy in Stroke 1986;17: Submitted Sept. 16, 1993; accepted Dec. 28, DISCUSSION Dr. Jeffrey R. Rubin (Youngstown, Ohio). Anesthetic and operative techniques are equally important when looking at operative outcome. This study very carefully examines the influence of anesthetic technique on perioperative complications after carotid endarterectomy. Based on an in-depth data analysis, it was demonstrated that age greater than 75 years, operative time greater than 3 hours, and general anesthesia were all independent risk factors for perioperative cardiopulmonary complications. In addition, they showed cervical block anesthesia was associated with shorter operative times, fewer perioperative cardiopulmonary complications, and a shorter postoperative hospitalization when compared to general anesthesia. They concluded therefore that cervical block anesthesia was safer and more cost effective than general anesthesia for the treatment of patients undergoing endarterectomy. These conclusions were based on a retrospective analysis of 584 consecutive patients undergoing 679 carotid endarterectomies between 1987 and There was an almost equal distribution of individuals undergoing operations with cervical blocks and general anesthetics. Were the patients randomly chosen to receive general versus regional anesthetic, or did the surgeons or anesthesiologists selectively use one or the other techniques routinely? In addition, I would like the authors to comment on the timing of anesthesia techniques, specifically, were there more general anesthetics used early in the series as opposed to cervical blocks or were both anesthesia techniques used equally throughout the course of the study? Cardiopulmonary complications occurred in almost twice as many patients who received general anesthetic versus a cervical block in this study. Cumulatively, these complications reached statistical significance, although myocardial infarction, arrhythmias, and blood pressure instability were not significantly different individually. It would appear that if the numbers of patients were greater, these numbers may have reached statistical significance independent of each other. On reviewing the anesthesia records, did you note any difference in 'the treatment of these patients with regard to the administration of antiarrhythmics, blood pressure control medications, or other drugs such as nitroglycerine? In addition, was there any difference in anesthesia staffing for the patients in the cervical block group versus the patients in the general anesthesia group? Lastly, I found it interesting that the time of the operation and the postoperative hospital stay were both significantly longer for patients who received general anesthetic as opposed to cervical block. Can the authors explain these findings? In summary, Dr. Allen and his associates have attempted to determine whether regional anesthesia is better than general anesthesia for patients undergoing carotid endarterectomy. Although they have demonstrated a tendency for patients to have increased cardiopulmonary complications after general anesthesia, I think there are enough additional questions that remain unanswered that begs for further study. Dr. Brent T. Allen. The first question was the distribution of patients and whether it was equal over the duration of the 6-year study period. The distribution of patients wasn't equal. We began using cervical block anesthesia in Before that we used general anesthesia almost exclusively. The reason for the change from general anesthesia to cervical block anesthesia was really the arrival of an anesthesiologist who was skilled in the administration of a cervical block anesthetic. He rapidly taught the rest of the anesthesia team to perform that procedure, and we subsequently adopted it as our experience with cervical block anesthesia increased and we became more aware of its advantages. We chose that study period, however, from 1987 to 1993, as the 3 years before we started using cervical block anesthesia and the 3 years after we started using cervical block anesthesia. During that time, the method of anesthesia was the only significant patient treatment variable in our treatment of patients with carotid artery disease: The next question was whether general anesthetics were used early on. We predominantly used general anesthetics early and cervical block in later years. Was the intraoperative treatment of these patients different? Well, I think clearly it was. The patients who had general anesthetic had more arrhythmias and more evidence of myocardial ischemia than the patients who received cervical block anesthetic. There was no difference in the way that these hemodynamic parameters were managed from the outset, but when evidence of myocardial ischemia or hemodynamic perturbations was observed by the anesthesiologists, they were treated appropriately.

10 JOURNAL OF VASCULAR SURGERY Volume 19, Number 5 Allen et al. 843 The anesthesia staffing was essentially the same throughout the entire study period. As I mentioned, one anesthesiologist arrived who was well skilled in the use of cervical block anesthesia, but he was not a vascular anesthesiologist, so he subsequently taught our vascular anesthesiologists the technique, and they used it for us. The difference in operating room time and the hospital stay is reflective of the reduced amount of stress that this type of anesthesia places on patients. I think it is well described in previous investigations by other authors that operating room time is less simply because the patients wake up quicker, and their neurologic examination is more rapidly evaluated. They are subsequently transferred out of the operating room more rapidly. Similarly, if there are fewer cardiopulmonary complications after an operation, the hospital stay will be less. Dr. William R. Flinn (Baltimore, Md.). Concern always comes up about the development of seizures in these patients receiving regional anesthetic. Have you had any? Concern also arises about the compromise of the training of residents or fellows to perform carotid endarterectomy with the patient receiving regional anesthetic. Was there any difference in the two groups of general anesthesia versus regional anesthesia in terms of percentage of cases done by residents or fellows? That could impact on operating time observed. Since you observed that increased perioperative risk was associated with increased operating time, did you stratify your results in any way regarding which carotid arteries were patched versus the ones that weren't patched? Dr. Allen. We patch selectively. The number of patches that were used in this series numbered 84. The number of carotid artery patches used in each group was similar. We have a vascular fellowship, and we are very interested in training both vascular fellows and general surgeons in carotid artery surgery. We have been impressed that the level of anxiety during a carotid endarterectomy performed with the patient receiving cervical block anesthetic is so much less than when the patient receives general anesthetic that you are more willing to take the time to instruct your resident or fellow in the technique than if you are uncertain about the neurologic examination by basing it on a stump pressure or an electroencephalograph monitor. We've all been impressed with how well tolerated these operations are by patients that undergo cervical block anesthesia. We typically use midazolam maleate and fentanyl sedation but still have the patient awake enough to respond easily to verbal commands and questions and to move their contralateral extremity. It makes for a very comfortable operative environment and one that is conducive to teaching. Finally, what about the incidence of seizures or major cardiac events that may complicate the operation in an awake patient? During this series of 679 patients, two patients had intraoperative seizures. Both of those patients had a mask placed and were ventilated through a mask until a carotid artery shunt could be placed. Once the shunt was placed, their mental status came back and the operation was continued with use of cervical block anesthetics. When it came time for the shunt to be taken out, that same procedure was repeated again.