Stroke during the perioperative period is a devastating complication associated with a high rate of morbidity and mortality1-3. The prevalence of this complication varies according to the type and complexity of the surgery, the timing of the procedure, and the patient's age3,4. Although the rate of stroke following general surgery procedures has been reported to be as low as 0.2%, the rate of this complication following cardiac, vascular, and neurologic procedures is higher, rising to 9.7% after double or triple heart valve surgery1,5-9. While the issue of stroke following cardiac and vascular surgery has received considerable attention in the literature, there is a relative lack of studies examining this issue following total joint (hip and knee) arthroplasty.
Although the exact prevalence of perioperative stroke following joint replacement is not known, these operations are not considered to be intrinsically high-risk procedures for stroke3. However, joint arthroplasty is being performed more frequently and is being offered to older patients, and the proportion of complex procedures such as revisions and bilateral procedures is growing. Because older patients who undergo more complex procedures tend to be at greater risk for stroke, even if the nominal risk of stroke is low following total joint arthroplasty, the number of high-risk patients who are undergoing those procedures implies that the overall risk for stroke following total joint arthroplasty is also increasing10.
Perioperative stroke can result in mortality rates of as high as 26%, and this complication is associated with increased rates of disability, prolonged hospital stays, and discharge to rehabilitation facilities8,11. Several case reports have mentioned stroke after total joint arthroplasty; however, to our knowledge, there is a relative lack of recognition and appreciation of the risk factors of stroke following these procedures.
The purpose of the present study was to review the prevalence of perioperative stroke in patients undergoing total joint arthroplasty and to identify possible risk factors and predictors. We also sought to determine how this complication affected the outcome of joint replacement surgery.
In this case-controlled study, we evaluated all patients undergoing joint arthroplasty at our institution from January 1, 2000, to December 31, 2007. We have a prospective electronic database at our institution that records any complications that occur following joint arthroplasty during hospitalization and during the postoperative period. Patients who were diagnosed with cerebrovascular accident or stroke during their hospital admission were identified on the basis of discharge abstracts or examination of daily complications forms. These forms, which outline the nature of each in-hospital complication and the treatment rendered, are completed by orthopaedic surgery residents and are uploaded into our database. The same form is then administered to each patient during each follow-up visit. In addition, we have another prospective database (based on billing records) to identify any patients who may be evaluated in the emergency department or readmitted to the hospital. Patients who had a stroke after hospital discharge were identified by means of an examination of electronic records of follow-up clinic visits and reoperation records. All patients who were deemed to have had a stroke within thirty days after joint arthroplasty were identified. A total of 18,745 patients underwent primary or revision total knee or total hip arthroplasty during the period of study. We identified thirty-six patients who had a stroke within the same hospital admission or within thirty days after total joint arthroplasty. We defined stroke as any new, rapidly developing episode of focal or global loss of cerebral function in which symptoms lasted for more than twenty-four hours or that had a fatal outcome. We then matched the patients who had had a stroke, in a 3:1 ratio and on the basis of the date of surgery and surgeon, with 108 patients who underwent total joint arthroplasty during the same time period. In some instances, matching revealed numerous patients for a given day of surgery and surgeon. In that circumstance, we only selected three control patients for that particular day to allow for the inclusion of control patients from various time points.
A detailed retrospective review of the medical records of these patients was conducted to extract all pertinent information. Numerous patient and surgery-related factors were then evaluated to identify risk factors for developing this complication and/or affecting the outcome. We collected the medical history for each patient, which recorded information on the history of hypertension, diabetes, hyperlipidemia, smoking, myocardial infarction, coronary artery bypass graft, congestive heart failure, left ventricular dysfunction, low ejection fraction, arrhythmia, atrial fibrillation, valvular heart disease, stroke, transient ischemic attack, systemic embolism, carotid artery disease, peripheral vascular disease, and associated conditions including chronic obstructive pulmonary disease, chronic bronchitis, asthma, hypothyroidism, gastrointestinal disease, genitourinary disease, abnormal renal function, and malignancy. We also calculated the Charlson index12 on the basis of patient comorbidities for all patients and controls. In terms of physical examination data, we recorded each patient's height and weight and calculated each patient's body-mass index and body surface area. We also obtained data on preoperative medication use, including use of antithrombotic medications. The patients in the cohort were categorized into two groups: urgent and elective. The patients in the urgent group were those who needed surgery as a result of admission through the emergency room because of a hip fracture, infection, dislocation, or periprosthetic fracture. The patients in the elective group were those for whom surgical intervention was scheduled at the time of an office visit.
We recorded the American Society of Anesthesiologists (ASA) score for each patient13. Details of the surgical procedure in terms of the type of surgery, the mode of fixation of components, operative time, the type of anesthesia, blood pressure, estimated blood loss, and the amount of blood transfused intraoperatively and postoperatively were also collected. We also obtained the results of computerized tomographic scans (completed for all patients) as well as the results of magnetic resonance imaging (performed for thirty-one patients). Finally, we identified several different in-hospital complications in twenty-one patients.
Statistical Analysis
Baseline and intraoperative covariates were tested between cases and controls with use of the t test for continuous variables and the Fisher exact test for categorical variables. All reported p values are two-sided. We combined potential preoperative predictors into composite categories in order to create a more stable model for estimating risk. Patients with a history of myocardial infarction or coronary artery bypass graft were categorized as having a history of coronary artery disease; patients with a history of stroke or transient ischemic attack were categorized as having a history of cerebrovascular disease. A category for "any atherosclerosis history" included patients who had a history of any of following: history of coronary artery disease, cerebrovascular disease, peripheral vascular disease, and carotid artery disease. We also had a category for patients with a history of "noncoronary cardiac disease," which included patients with history of atrial fibrillation, arrhythmia, valvular cardiac disease, congestive heart failure, or left ventricular dysfunction. All covariates were tested with use of an unadjusted logistic regression. We report the corresponding odds ratios, 95% confidence intervals, and p values. Any covariates with significant findings in the unadjusted analysis (p < 0.05) were then tested in a multivariate logistic regression model that was adjusted for age, sex, history of stroke, and history of coronary artery disease.
Source of Funding
We did not receive any outside funding in support of this study.
Cohort Description
Thirty-six cases of perioperative stroke were diagnosed during the study period. The cohort included seventeen men and nineteen women with a mean age of 68.2 years (range, forty-five to eighty-seven years). Five patients developed a stroke on the day of surgery; in one of these patients, the stroke was diagnosed in the recovery room. Stroke developed on the first postoperative day in sixteen patients, on the second postoperative day in four, on the third postoperative day in two, and on the fourth postoperative day in two. Three additional patients had development of a stroke on the fifth, sixth, and seventh postoperative days. The remaining four patients had detection of a stroke during the second week after surgery while in a rehabilitation facility. All but two strokes were ischemic; the remaining two were hemorrhagic. Moreover, both of the hemorrhagic strokes occurred on the first postoperative day. One of these strokes was due to hemorrhage in a cortical neoplastic lesion, and the other was due to a subarachnoid hemorrhage that presumably was due to a vascular malformation. Four patients (including the two who had a hemorrhagic stroke) died during the hospital stay; one died on the first postoperative day, two died on the second postoperative day, and one died on the nineteenth postoperative day. Twenty-seven patients were discharged to rehabilitation facilities, and five patients were discharged to their homes or to nursing homes. Of the thirty-two patients who were ultimately discharged, eight died before the latest follow-up. Eleven patients had residual problems that continued to affect their daily activities. The other thirteen patients had a complete recovery without any residual symptoms and were functionally independent in terms of activities of daily living at the time of the latest follow-up. The right side of the brain was involved in twenty-one patients, and the left side was affected in thirteen patients. In the two patients who had a hemorrhagic stroke, both sides of the brain were affected.
Three patients underwent thrombolytic therapy; in one of these patients, intra-arterial infusion of the middle cerebral artery resulted in a hemorrhagic conversion of the stroke and, on the following day, death. The other two patients responded well to thrombolysis without any major complication; one of them was alive without any residual problem related to the stroke at the time of the latest follow-up, and the other died one year after surgery. The results of post-stroke echocardiography were available for twenty-four of the thirty-six patients, and seventeen (71%) of these studies showed abnormal findings in the heart. A patent foramen ovale was found in three patients. Finally, we identified several different in-hospital complications in twenty-one patients.
We successfully matched all thirty-six patients who had a stroke (the case group) with 108 controls on the basis of both surgeon and year of surgery. Eight different surgeons, including two of the authors (J.P. and R.H.R.), operated on the patients in our case and control groups, with the number of patients per surgeon ranging from four to forty-four. The mean age of the total group (including the case group and the control group) was 63.1 ± 13.4 years. Eighty-seven percent of the procedures (in 125 patients) were primary arthroplasties, and 13% (in nineteen patients) were revisions. Eighteen percent of the procedures (in twenty-six patients) were simultaneous bilateral procedures, and 81% (in 117 patients) were unilateral procedures. One patient underwent simultaneous hip and knee replacement. Overall, ninety patients (63%) underwent hip replacement and fifty-four (38%) underwent knee replacement.
Results of Univariate Analysis
The variables in the case and control groups were categorized as patient-related and surgery-related factors (Tables I and II). There was no difference between the groups in terms of the risk of cerebrovascular complications due to sex, body-mass index, body surface area, smoking, hypertension, diabetes, hyperlipidemia, chronic obstructive pulmonary disease, obstructive sleep apnea, congestive heart failure, carotid artery disease, peripheral vascular disease, history of systemic embolism, renal disease, hypothyroidism, gastrointestinal disease, genitourinary problems, malignancy, or history of anticoagulation medications before surgery. Factors associated with complications included advanced age, history of a myocardial infarction, history of a coronary artery bypass graft, history of a cerebrovascular accident, history of a transient ischemic attack, having atrial fibrillation or rhythm abnormality on a preoperative electrocardiogram, left ventricular dysfunction, cardiac valvular disease, and history of cardiac valvular disease. All consolidated variables, including history of coronary artery disease, history of cerebrovascular disease, history of atherosclerotic disease, and history of noncoronary cardiac disease, showed significant association with stroke on univariate analysis (Table I).
With regard to surgery-related factors, the maximum, minimum, and range of mean arterial blood pressure and heart rate during surgery; the site of surgery (hip or knee); the type of procedure (revision or primary); whether the surgery was bilateral or unilateral; the duration of surgery; the use of cement; and the estimated blood loss showed no association with stroke. On the other hand, higher ASA class, mean alterations in heart rate at the time of the operation, rhythm disturbances during surgery, and both blood transfusion and the quantity of blood transfused were associated with cerebrovascular complications (Table II).
Results of Multivariate Analysis
For variables that were significant on univariate analysis, we ran a multivariate analysis that was adjusted for age, sex, history of cerebrovascular disease, and history of coronary artery disease. This analysis demonstrated that a history of noncoronary cardiac disease (odds ratio, 4.13; 95% confidence interval, 1.66 to 10.25), the priority of the surgery (urgent versus elective) (odds ratio, 5.89; 95% confidence interval, 1.29 to 26.82), general anesthesia (odds ratio, 3.54; 95% confidence interval, 1.01 to 12.39), and intraoperative arrhythmia (atrial fibrillation) or changes in mean heart rate during the surgery (odds ratio, 1.06; 95% confidence interval, 1.02 to 1.11) were independent predictors of postoperative stroke following total joint arthroplasty.
Outcome
The mean duration of follow-up for all patients and controls in the present study was sixty-two months (range, zero to 124.9 months). There were fourteen deaths (39%) among the thirty-six patients in the stroke group and fourteen deaths (13%) among the 108 patients in the control group (p = 0.001). The first-year mortality among stroke patients was 25% (nine of thirty-six). There were four in-hospital deaths in the stroke group, whereas all deaths in the control group occurred after hospital discharge. In addition, and not surprisingly, the duration of hospital stay was significantly longer for patients with a stroke than for controls (mean, 12.75 compared with 3.63 days; p < 0.001).
The issue of stroke following total joint (hip and knee) arthroplasty has received minimal attention in the literature. The present study, from a very large institutional joint arthroplasty database, identified a rate of 0.2% for perioperative stroke following total joint arthroplasty. The prevalence increased twofold, to 0.4%, among patients with an age of seventy-five years or more. Our results are similar to those in a recently published study by Bateman et al.14, which was based on discharge reports extracted from the Nationwide Inpatient Sample. That group also reported a 0.2% prevalence of stroke following total hip arthroplasty.
The present study demonstrates several important points, some of which also have been demonstrated in patients who have not undergone arthroplasty. First and foremost, our study confirmed that perioperative stroke has a profoundly deleterious effect on the outcome of joint arthroplasty. Perioperative stroke greatly increases the odds of in-hospital mortality or discharge to a medical or chronic-care facility (as opposed to discharge to home), and it also increases the duration of hospital stay. In addition, the sequelae of stroke (that is, hemiplegia or hemiparesis) lead to a deterioration of patient function and to more morbid conditions. Finally, the overall death rate in this group of patients was significantly higher than it was among patients without this complication. These results are supported by previous reports in the literature; stroke during the perioperative period has been shown to significantly increase the mortality and morbidity of patients undergoing surgery1-3.
Second, the results of the present study confirm that several traditional risk factors for perioperative stroke also apply to the population of patients undergoing total joint arthroplasty. Advanced age, a history of cerebrovascular disease, a history of coronary artery disease, a history of atherosclerotic disease, a history of cardiac valvular disease, and atrial fibrillation were all shown to be associated with perioperative stroke. In addition, our study demonstrated that a history of noncoronary cardiac disease is a significant independent risk factor for the development of perioperative stroke following total joint arthroplasty. Among different noncoronary cardiac problems, having a history of atrial fibrillation is of particular importance as it can increase the absolute risk of stroke by as much as twentyfold, and at the same time it is potentially modifiable15.
Third, our study revealed that urgent surgery is an independent predictor of perioperative stroke following total joint arthroplasty. It has been shown that emergency cardiac surgery for the treatment of unstable angina or infective endocarditis is an independent risk factor for stroke5,16; our results confirm that urgent surgery is also an independent risk factor for perioperative stroke even in patients undergoing noncardiac surgery. This result is probably due to less-than-optimal treatment of associated medical comorbidities in urgent cases. Patients undergoing urgent surgery who have major medical comorbidities are managed before surgery to bring their laboratory values to within normal ranges for surgery. The present study suggests that additional factors (such as arrhythmias and noncoronary cardiac disease) should be considered when preparing patients for urgent surgery because these factors are independent predictors of stroke. In cases in which appropriate treatment of medical comorbidities is not possible, patients undergoing urgent surgery should be informed of the increased potential for stroke as a complication of surgery.
Finally, our study identified several important surgery-related factors that are independent risk factors for stroke. We found that the use of general anesthesia (as opposed to regional anesthesia) was a risk factor for stroke. It has been shown that patients who receive regional anesthesia are less likely to develop perioperative stroke as compared with those who receive general anesthesia17. This finding illustrates another benefit of regional anesthesia in patients undergoing total joint arthroplasty. There were two other potentially modifiable intraoperative findings that independently increased the odds of perioperative stroke in the present study. The presence of intraoperative arrhythmia, primarily atrial fibrillation, has been shown to be a predictor for stroke in patients undergoing surgery18, and our findings confirmed that it is a risk factor of stroke following total joint arthroplasty. In addition, we found that greater changes in mean heart rate during surgery also increase the risk of stroke. On the basis of these findings, we recommend that patients at high risk for stroke be managed aggressively for arrhythmias and that the heart rates for such patients should be closely monitored during surgery.
We acknowledge that the present study had some limitations. First, the risk factors that we included in our analysis were limited to those that we were able to obtain from the medical records. It is possible that some risk factors may not have been captured or recorded. Second, because of the rarity of this complication in patients undergoing joint replacement, the number of cases was small, and therefore it was difficult to draw definitive conclusions. Another potential limitation is that the present study was from a single center with a high volume of joint arthroplasty procedures where there is access to emergency neurologic and neurosurgical care. Thus, the findings of the present study may not be applicable to centers in which a smaller number of arthroplasties are performed. Finally, it is possible (although unlikely) that some patients who had development of a stroke following discharge from the hospital were managed elsewhere and may have escaped follow-up and gone undetected. While the latter limitation does not affect our findings regarding risk factors for stroke, we acknowledge that it may underestimate the prevalence of this complication.
We do not believe that these shortcomings undermine the important findings of this study, namely, that perioperative stroke following total joint arthroplasty is an important source of postoperative mortality and morbidity, particularly in elderly patients. Additionally, we emphasize the fact that the risk of stroke following total joint arthroplasty appears to be lower than the risk of stroke following other operations such as cardiovascular operations. Nevertheless, the present study did identify several risk factors that can lead to increased cerebrovascular complications after total joint arthroplasty surgery. Identification of these risk factors may lead to strategies that can minimize or prevent this complication.