An Analysis of Perioperative Surgical Mortality and Morbidity in the Asymptomatic Carotid Atherosclerosis Study
Background and Purpose Our aim was to determine the perioperative morbidity and mortality rates of patients in the surgical arm of the multi-institutional, prospective, randomized Asymptomatic Carotid Atherosclerosis Study (ACAS).
Methods Of 828 patients with carotid stenosis of 60% or more randomized to the surgical arm of ACAS, 721 underwent carotid endarterectomy (CEA). To qualify for participation, surgeons were required to have performed at least 12 CEAs per year with a combined neurological morbidity and mortality rate no greater than 3% for asymptomatic patients and 5% for symptomatic patients. Clinical centers had to demonstrate arteriographic morbidity less than 1% and mortality less than 0.1% per year. Primary events were stroke and death in the period between randomization and 30 days after surgery; secondary events were transient ischemic attack and myocardial infarction occurring in the same period.
Results Of the 721 patients who underwent CEA, 1 died and 10 others had strokes within 30 days (1.5%). Of the 415 who underwent arteriography after randomization but before CEA, 5 (1.2%) suffered transient ischemic attack or stroke caused by arteriography. Thus, a nearly equal risk of stroke was associated with both CEA and carotid arteriography. In addition, 6 transient ischemic attacks and 3 myocardial infarctions could be directly linked to CEA, for a total CEA event rate of 2.6%.
Conclusions Patients with asymptomatic internal carotid artery stenosis exceeding 60% reduction in diameter who are acceptable candidates for elective operation may be considered for CEA if the combined arteriographic and surgical complication rates are 3% or less.
The prospective, randomized ACAS1 documented that asymptomatic patients with stenosis causing reduction of least 60% in the diameter of the internal carotid artery are at a significantly lower risk of subsequent ipsilateral stroke when treated with CEA in addition to risk factor reduction and aspirin than when treated only with risk factor reduction plus aspirin. Patients undergoing CEA had a 53% lower 5-year risk of ipsilateral stroke than patients in the medical arm.
The most important factor in this favorable result was the very low perioperative major morbidity and mortality rate (death, stroke, or MI) of 1.8% in the surgical arm of the trial. This report details the perioperative complications of patients in the surgical arm and analyzes the factors contributing to the low complication rate. The authors also suggest acceptable standards for surgical risks associated with CEA for asymptomatic carotid artery stenosis.
Subjects and Methods
ACAS was a multi-institutional, prospective, randomized clinical study of 1662 patients at 39 participating centers in North America. The study asked the following question: “Will CEA added to aggressive reduction of modifiable risk factors and aspirin 325 mg per day reduce the 5-year risk of ipsilateral cerebral infarction in patients with asymptomatic hemodynamically significant carotid artery stenosis?” All patients were accorded aggressive risk factor reduction and 325 mg aspirin daily. In addition, one half were randomly assigned to CEA as well.
Selection of Surgeons and Clinical Centers
A detailed description of the design and organization of ACAS has been published.2 To qualify for participation, surgeons were required to have performed a minimum of 12 CEAs per year with combined morbidity and mortality rates no greater than 3% for asymptomatic and 5% for symptomatic patients.3 Only 117 (71%) of the 164 surgeons who applied for participation qualified. In addition, clinical centers had to demonstrate the ability to achieve arteriographic morbidity below 1% and mortality below 0.1% per year, the ability to contribute at least 50 randomized patients per year for 2 years, and the ability to follow up with these patients according to the protocol for at least 5 years. Of the 55 centers that applied, 39 (71%) were chosen.1
Evaluation of Study End Points
Each CEA patient was examined within 24 hours after operation by the ACAS surgeon, the ACAS neurologist, and the patient coordinator. Thirty days after the operation, patients were examined once more by an ACAS physician and the ACAS coordinator. All incidents that suggested stroke, TIA, MI, or death during the perioperative period required end point review.1 In addition, a study neurologist (J.P.B.) reviewed the medical records of all patients who had a stroke or TIA associated with CEA or cerebral angiography to determine the timing, cause, and neurological outcome of those patients.
The perioperative events of interest were stroke, TIA, nonfatal MI, and death, analyzed separately, and any combination of these four events. Tables 1 through 4⇓⇓⇓⇓ are based on all 825 patients randomized to the surgical arm, although the number of patients included in specific analyses will vary. Because the events were few, Fisher's exact test was used to assess the significance of the associations between baseline risk factors and event rates. The “total” events column includes stroke, death, TIA, and MI.
Two caveats are relevant. First, no formal adjustment of the probability values for multiple comparisons was performed. Second, because the number of events was small, the power of many comparisons is relatively low. Therefore, a failure to find a significant association must not be interpreted as demonstrating that a meaningful association does not exist.
Baseline Characteristics of Surgical Patients
Baseline characteristics of the 825 patients randomized to the surgical arm are depicted in Table 1⇑. History of TIA or stroke was based on clinical history given by the patient independent of entry CT results. A patient without history of stroke but with CT scan results showing a silent infarct was still considered to have asymptomatic stenosis. Patients were considered to have prior heart failure if they had serious cardiovascular disease that might lessen their life expectancy or make follow-up difficult, including cardiac decompensation, unstable angina, atrial fibrillation, and valvular heart disease. Patients were considered to have had prior MI or prior angina pectoris on the basis of clinical history, the Rose Angina Questionnaire,4 and the judgment of the local investigator.
Table 2⇑ summarizes the timing of the postrandomization events in relation to randomization, arteriography, and CEA. The perioperative period was defined as the time from randomization to 30 days after CEA. For patients randomized to but not undergoing CEA, the perioperative period was defined as 12 days after randomization, plus the day of and the day after arteriography if the arteriogram was performed after that 12-day period. The 12-day period was the median time from randomization to operation, estimated from information available at the time of the first interim analysis. Analysis of the data at the end of the trial showed that the median interval between randomization and operation was 11 days (interquartile range, 7 to 15 days).
Of the 825 patients randomized to CEA, 30 suffered one or more events, including 17 strokes, 9 TIAs, 4 MIs, and 3 deaths, for a total perioperative complication rate of 3.6%. Eight strokes and 4 TIAs were ipsilateral, 6 strokes and 5 TIAs were contralateral, and 3 strokes were vertebrobasilar. However, 2 strokes and 1 death occurred before hospitalization or arteriography (Table 3⇑). In addition, 4 strokes and 3 TIAs were caused by arteriography. One patient had a stroke followed by a fatal MI in the postarteriographic period but before operation. Excluding these events, the perioperative events that can be directly related to CEA are 10 nonfatal strokes, 6 TIAs, 3 MIs, and 1 death (20 events, 19 patients), for a total event rate of 2.6% (Table 2⇑). The perioperative stroke and death rate directly attributable to CEA was 1.5%. Strokes clustered during the first several postoperative days, whereas TIAs and MIs did not cluster in a noticeable way.
Relationship of Baseline and Operative Variables to Perioperative Event Rates
Table 1⇑ shows that there was no relationship between perioperative stroke and sex, race, preoperative hypertension, smoking, heart failure, or previous MI. Diabetes mellitus (P=.043), contralateral siphon stenosis (P=.017), and never drinking (P=.017) were associated with a higher risk of perioperative stroke. Analysis of other events suggested a marginally significant relationship between siphon stenosis and perioperative mortality (P=.053). History of previous stroke (P=.039), contralateral stenosis greater than 60% (P=.036), and never drinking (P=.027) were associated with a higher risk of all perioperative complications combined. Hypertension (P=.002) and contralateral stenosis (P=.043) were associated with a higher risk of TIA.
Table 4⇑ compares the risk of these events across operative variables, methods, and recovery room findings. There was no significant relationship between perioperative stroke and the use of tacking sutures or shunts, ulceration of plaque, plaque hemorrhage, luminal thrombosis, intraoperative EEG changes, or the administration of intraoperative anticoagulants, although the association between intraoperative EEG changes and all events combined was marginally significant (P=.053). Of the operative variables, the length of external carotid artery plaque (P=.031) was significantly associated with subsequent stroke (P=.031) and all-event rates (P=.028). The use of local or regional anesthesia during CEA was associated with a higher risk of TIA (P=.015) and MI (P=.011) (Table 4⇑). Stroke was not significantly associated with cranial nerve deficits, wound complications, or multiple operations (Table 5⇓), although for some of these analyses the small sample size makes meaningful evaluation impossible. There was no significant association between perioperative stroke and the amount of ipsilateral or contralateral cervical carotid stenosis (>80% versus ≤80%) documented by Doppler studies or arteriography (Table 6⇓). Timing, cause, and outcome of strokes and TIAs after CEA are listed in Table 7⇓. Only 5 of the 721 patients (0.7%) who underwent CEA had a postoperative stroke that resulted in significant disability.
Preoperative baseline systolic blood pressure and the maximum systolic blood pressure recorded in the recovery room were significantly related (Table 8⇓). When the baseline systolic blood pressure was normal, the recovery room systolic blood pressure was likely to be normal, and when the baseline systolic blood pressure was elevated, the recovery room systolic blood pressure was also likely to be elevated. There was no association when extreme pressures (>180 mm Hg) were compared, perhaps because the standard practice of administering antihypertensive medications to patients with baseline systolic hypertension may have masked a possible statistical relationship.
Morbidity and Mortality Associated With Cerebral Arteriography
The study required all patients to undergo arteriography before CEA. Thus, 342 patients underwent arteriography before randomization to determine whether they qualified for ACAS. Of those randomized to CEA, 415 were selected on the basis of noninvasive criteria and required cerebral arteriography before CEA. Of these, 1.2% suffered a cerebral infarction, and one death was associated with carotid arteriography. Three patients had an arteriography-associated stroke that resulted in significant disability.
Because significant numbers of our patients underwent arteriography before randomization, we do not have complete data about their angiographic complications. If we assume that complications would have occurred at the same rate as among our postrandomization arteriography patients, 2.7% of the 825 patients in the surgical arm would have suffered a stroke or died.1
Perioperative Event Rate
The very low surgical complication rate of ACAS (1.5% stroke and death rate, 0.7% incidence of perioperative stroke resulting in significant disability and loss of independence) was a key factor in demonstrating the effectiveness of CEA for patients with asymptomatic carotid stenosis of 60% or more. This rate compares favorably with those of other recent trials. In the VA trial of CEA for asymptomatic carotid stenosis,5 6 the permanent stroke and death rate within 30 days after randomization was 4.7% for surgical and arteriographic complications combined. In the CASANOVA Study,7 the perioperative death rate was 1.2%, the stroke rate was 3%, and the TIA rate was 1.2%.
The postoperative complication rates in the VA and CASANOVA studies5 7 are approximately twice those reported by ACAS, perhaps because surgeons did not have to meet specific criteria to participate as in ACAS. Hobson et al5 have recommended that “a low rate of perioperative complications should be confirmed by clinical audit at each institution before a program of operative intervention is begun.” The primary reason for the very low perioperative complication rate in the surgical arm of ACAS was the meticulous selection and monitoring of ACAS surgeons. Although the study populations and entry criteria of each of the three randomized trials differ somewhat (the VA study contained only men, and both the VA and the CASANOVA studies included patients with 50% stenosis), the primary difference appears to be that only ACAS had a surgeon selection protocol.
Predictors of Perioperative Risk
A number of studies have attempted to associate specific surgical methods or characteristics of patients, surgeons, and hospitals with CEA complication risks. A retrospective study of 1160 patients undergoing CEA at 12 major medical centers found no significant relationships between complication rate and the variations in method of performing CEA.8 Others have reported very low perioperative complication rates even for symptomatic elderly patients with a high incidence of previous MIs, perioperative hypertension, and diabetes.9 10 11 Loftus et al9 reported a 0% perioperative mortality rate and a 3.8% morbidity rate in a group of patients older than 70 years (mean age, 73.4 years).
In ACAS, several baseline variables emerged as significant predictors of perioperative complications. Diabetes mellitus, contralateral siphon stenosis, and never drinking were associated with higher risk of perioperative stroke; prior history of stroke, contralateral stenosis greater than 60%, and never drinking were associated with a higher risk of all perioperative complications combined. Of the operative variables, the length of external carotid artery plaque was associated with subsequent stroke and all-event rates; the use of local or regional anesthesia was associated with a higher risk of TIA and MI. The data from ACAS do not explain why these factors are significant, and the findings should be interpreted with caution. Any explanations about the mechanisms of these effects would be purely speculative, and using these data alone to change clinical care would be premature. For example, although the operative variable of regional anesthesia was associated with a higher risk of TIA and MI, there may have been differences in the baseline characteristics of patients for whom regional versus general anesthesia was selected that caused differences in event rates. In addition, although the statistical analysis determined that only these variables were significantly associated with perioperative complications, the number of events that occurred in this study was so small that other variables also associated with complications may not have achieved statistical significance.
Nine (5.8%) of the 155 patients with a maximum recovery room systolic blood pressure higher than 180 mm Hg had a postoperative stroke, TIA, or MI. Those with recovery room systolic blood pressures of 145 mm Hg or lower had significantly fewer complications. As systolic blood pressure increased, the complication rates also increased. Therefore, patients with systolic blood pressures greater than 145 mm Hg in the postoperative period should receive close neurological monitoring and approved control measures.
Towne and Bernard12 also found that patients with postoperative hypertension had a significantly increased incidence of postoperative neurological deficits and mortality. In a study of 330 CEAs, Hans and Glover13 found that postoperative hypertension increased the likelihood of neurological deficit but not of MI. Benzel and Hoppens14 reported that postoperative hypertension in 35 of 100 patients who underwent CEA was significantly associated with preoperative elevation of both systolic and diastolic blood pressure, the use of indwelling shunts, increased age, and race. In contrast to ACAS findings, however, no correlation was found between postoperative hypertension and postoperative complications.
Severe preoperative hypertension, CEA performed in preparation for coronary artery bypass operation, angina, internal carotid artery stenosis near the carotid siphon, age greater than 75 years,8 and diabetes15 have been associated with an increased risk of perioperative complications. In ACAS, preoperative hypertension was not significantly associated with stroke, death, or MI, but it was significantly associated with TIA. The association between perioperative complication risk and both contralateral siphon stenosis and diabetes was confirmed in ACAS. Silent infarcts on preoperative CT did not appear to significantly increase the incidence of perioperative stroke.
Of the five patients with strokes resulting in significant disability, three had strokes within 3 hours after surgery because of technical difficulties such as difficulty with a shunt, plaque dissection, or both; one had a stroke 3 days postoperatively because of a leaking vein patch. The fifth patient suffered intracerebral hemorrhage as the result of a hyperperfusion syndrome on the second postoperative day. All of these strokes occurred during the first 3 days after surgery. However, no operative variable except length of plaque clearly emerged as a risk factor for subsequent stroke, although technical difficulties and postoperative blood pressure were the most important causes of perioperative strokes in ACAS.
Our finding that arteriography was associated with nearly the same risk of stroke as CEA argues the need for stricter standards for performance or for alternatives to arteriography, which has long been held as the “gold standard” for most accurately determining percentage of carotid stenosis. Several recent reports suggest that in selected circumstances performing CEA without arteriography on the basis of duplex ultrasonography, MRA, or both is justified.16 17 18 19 However, considerable controversy remains about whether duplex ultrasound or MRA is accurate enough for a decision regarding CEA.20 A number of recent studies have documented that duplex scanning, when performed in a validated ultrasound laboratory, can achieve accuracy equaling that of arteriography.21 There is also considerable contention about whether it is necessary to demonstrate the intracranial vasculature before making a decision about CEA.20 Chervu and Moore21 found that demonstration of the intracranial vasculature did not change CEA management decisions. However, ACAS data indicate that contralateral siphon stenosis significantly increases the risk of stroke, death, and overall complications. Our trial excluded patients with ipsilateral stenosis; if it had not, the complication rate associated with intracranial disease may have been even higher. This statistically significant association between contralateral siphon stenosis and complications after CEA supports delineation of the intracranial circulation, which can be done noninvasively by MRA or transcranial Doppler ultrasound, as part of the preoperative evaluation.
Recent developments in MRA technology have sufficiently improved visualization of the carotid and intracranial vasculature to correlate highly with conventional arterial arteriography. DeMarco et al22 reported that the results obtained from multiplanar reformation images of the three-dimensional time-of-flight MR angiograms correlated highly with results achieved by arteriography, and that there was no significant difference between the percentages of carotid stenosis determined by the two methods. Blatter and colleagues23 documented that stenosis measurements from high-quality, high-resolution multiple overlapping thin-slab acquisition MRA correlated with measurements obtained by conventional intra-arteriography.
The Stroke Council of the American Heart Association stated that the perioperative death and stroke rate for CEA should be less than 3% if this procedure is to be considered effective for patients with asymptomatic carotid artery stenosis.24 Achieving a complication rate of less than 2% is possible. The surgeons selected for participation in ACAS achieved a perioperative stroke and death rate of 1.5% and a total morbidity and mortality rate (stroke, death, TIA, and MI) of 2.6% when angiographic complications were excluded. We recommend that the standard for the acceptable rate for perioperative stroke and mortality related to CEA be 3% or less. Others have met this standard.25 26 27
In addition, we suggest that hospitals should ensure that their noninvasive imaging laboratories meet the standards suggested by the Intersocietal Commission for the Accreditation of Vascular Laboratories28 and adopt a surgeon credentialing procedure similar to that of ACAS.3 We further suggest that an independent neurologist examine patients postoperatively to help ensure accurate monitoring and the detection of all strokes during the credentialing process.
If the results of ACAS are used to recommend CEA to asymptomatic patients, appropriate actions must be taken to ensure that the surgeon, through technical skill and appropriate patient selection, can achieve the same low perioperative morbidity and mortality rate as that achieved in the trial.
Selected Abbreviations and Acronyms
|ACAS||=||Asymptomatic Carotid Atherosclerosis Study|
|CASANOVA||=||Carotid Artery Stenosis With Asymptomatic Narrowing: Operation Versus Aspirin|
|MRA||=||magnetic resonance angiography|
|TIA||=||transient ischemic attack|
This study was supported by the National Institutes of Health (National Institute of Neurological Disorders and Stroke grant NS22611). The authors gratefully acknowledge the editorial assistance of Flo Witte, ELS, Director of the Publications Office, Department of Surgery, University of Kentucky College of Medicine. We also appreciate the editorial assistance of Debra H. Weiner and Ralph Hicks of the ACAS Operations Center.
Reprint requests to Byron Young, MD, c/o ACAS Editorial Office, Stroke Center, WFU/BGSM Medical Center, Medical Center Blvd, Winston-Salem, NC 27157-1078.
- Received June 3, 1996.
- Revision received August 26, 1996.
- Accepted August 26, 1996.
- Copyright © 1996 by American Heart Association
The Asymptomatic Carotid Atherosclerosis Study Group. Study design for randomized prospective trial of carotid endarterectomy for asymptomatic atherosclerosis. Stroke. 1989;20:844-849.
Moore WS, Vescera CL, Robertson JT, Baker WH, Howard VJ, Toole JF. Selection process for surgeons in the Asymptomatic Carotid Atherosclerosis Study. Stroke. 1991;22:1353-1357.
Rose GA, Blackburn H. Cardiovascular Surgery Methods. Geneva, Switzerland: World Health Organization; 1968:172 (annex 6).
The CASANOVA Study Group. Carotid surgery versus medical therapy in asymptomatic carotid stenosis. Stroke. 1991;22:1229-1235.
McCrory DC, Goldstein LB, Samsa GP, Oddone EZ, Landsman PB, Moore WS, Matchar DB. Predicting complications of carotid endarterectomy. Stroke. 1993;24:1285-1291.
Young GR, Humphrey PRD, Shaw MDM, Nixon TE, Smith ETS. Comparison of magnetic resonance angiography, duplex ultrasound, and digital subtraction angiography in assessment of extracranial internal carotid artery stenosis. J Neurol Neurosurg Psychiatry. 1994;57:1466-1478.
Beebe HG, Clagett GP, DeWeese JA, Moore WS, Robertson JT, Sandok B, Wolf PA. Assessing risk associated with carotid endarterectomy. a statement for health professionals by an Ad Hoc Committee on Carotid Surgery Standards of the Stroke Council, American Heart Association. Circulation. 1989;79:472-473.