There is limited information about the frequency of perioperative complications after elective primary orthopedic total hip and knee arthroplasty in contemporary practice. The purpose of this study was to determine the frequency of clinically relevant myocardial infarction, pulmonary embolism, deep venous thrombosis, and death within 30 days after elective primary hip or knee arthroplasty treated according to contemporary perioperative management.
The authors examined the medical records of consecutive patients undergoing hip or knee arthroplasty at their institution in a 10-yr period. Prospectively collected databases were used to identify patients with the diagnosis of myocardial infarction, pulmonary embolism, deep venous thrombosis, or death using strict validation criteria and diagnostic-certainty categories.
A total of 10,244 patients underwent primary total hip or knee arthroplasty in the period of study. Of these, 224 patients had one or more adverse events (overall event rate: 2.2%; myocardial infarction: 0.4%; pulmonary embolism: 0.7%; deep venous thrombosis: 1.5%; death: 0.5%). Most adverse events (myocardial infarction, pulmonary embolism, and death) increased in frequency with older age, particularly for patients aged 70 yr or older. Myocardial infarction occurred more frequently in male patients. There were no differences in the overall event frequency between types of procedure. However, pulmonary embolism was highest in patients undergoing bilateral knee operations.
The overall frequency of serious complications within 30 days after primary total hip or knee arthroplasty with contemporary practice was 2.2%. Accurate knowledge of the perioperative risks associated with widely performed elective operations can be used to implement management strategies that may further improve patient outcomes and decrease cost.
Click on the links below to access all the ArticlePlus for this article.
Please note that ArticlePlus files may launch a viewer application outside of your web browser.
INCREASING numbers of operative procedures are being performed in older patients. 1–3Because the prevalence of hip and knee osteoarthritis increases with age, major orthopedic surgery is now commonplace in ever-older individuals—a population in which some degree of cardiac, pulmonary, cerebral, renal, and hepatic disease is often present. 1Importantly, these procedures are commonly performed in low-volume surgical settings. For example, in the United Kingdom, although nearly 35,000 total knee replacements are implanted annually, only 9% of orthopedic surgeons perform more than 90 per year. 4Although somewhat controversial, there are no definitive data to support the assumption that improved perioperative outcomes are achieved in hospitals with larger surgical volumes. 5,6Therefore, a thorough understanding of the potential complications in this operative group is paramount to the delivery of the highest-quality medical care.
Previous studies 3,7–15have examined the frequency of myocardial infarction, deep venous thrombosis, pulmonary embolism, or death after lower extremity arthroplasty. Most of these studies were performed in small patient populations in selected hospitals or predate current perioperative care of the arthroplasty patient. 1,2,16–25Recent studies, including large numbers of patients and using current medical treatments for patients undergoing major orthopedic operations, are still lacking.
The goal of this study was to determine the frequency of clinically significant complications in the perioperative period for patients undergoing primary total hip or knee joint replacement at a single institution with a large, broadly based patient population. Accurate knowledge of the actual contemporary risks of perioperative complications is valuable to the individual's decision-making process when considering an elective operation. The identification of factors that predispose patients to perioperative complications offers the potential to optimize the outcome of operative procedures and to focus on strategies that may favorably impact the cost of medical care. 12,14,26–30
Patients and Methods
After institutional review board approval (Mayo Clinic, Rochester, Minnesota), two prospectively collected databases were used to identify all patients at our institution who underwent elective primary total hip or knee arthroplasty between January 1, 1986 to December 31, 1995 and who experienced an adverse cardiac or pulmonary complication in the perioperative period. In accordance with Minnesota state law, patients who had denied access to their medical records were excluded from the study.
The Total Joint Registry is a prospectively collected, computerized database of all joint replacements performed at the Mayo Clinic since 1969 and has been described previously. 31Detailed information about operative approach and prosthesis data are abstracted and entered in a computerized database by joint registry personnel. Data about operative indication, preoperative diagnoses, and postoperative medical and surgical complications are collected in standardized data collection forms at hospital discharge. Patients are routinely seen at 2 or 3 months after their procedure. The Total Joint Registry is highly integrated with other institutional systems (e.g. , registration, billing, appointment scheduling), which provide up-to-date contact information and facilitate patient follow-up. A standardized form and formal chart review are performed at 1, 2, and 5 yr and every 5 yr thereafter. The overall rate of clinical follow-up in the Total Joint Registry exceeds 95%. 31The Total Joint Registry was screened to identify all patients who received one or more elective primary total hip or knee replacements in the study period. Patients treated with nonelective hip arthroplasty for hip fracture were excluded. For patients who underwent more than one operative procedure during the study period, only the first procedure was considered; thus, only one operative episode was examined for each patient. Using the Total Joint Registry, all patients known to have experienced arteriovascular or thromboembolic complications or death within the first year after the date of operation were identified. The medical records of those patients were retrospectively reviewed.
The institutional unified medical record system contains information about all patients who have ever been examined at our institution. 32The Medical Index is a set of databases that serve as an index to the patient's medical record. This index also was used to identify all study patients who had a diagnosis of coronary artery disease, myocardial ischemia, myocardial infarction, pulmonary embolism, deep venous thrombosis, or death at any time during the previous year, the year of the surgical event, or the year after the surgical event. The medical records of these patients also were retrospectively reviewed.
This broad initial screening methodology, which identified all patients experiencing potential adverse events occurring within 1 yr after the operation, was used to maximize case ascertainment. Strict validation criteria were subsequently used for the definition of an adverse perioperative event (myocardial infarction, pulmonary embolism, deep venous thrombosis, or death). Each episode of myocardial infarction, pulmonary embolism, or deep venous thrombosis occurring during the initial 30 days after the date of operation was categorized into the highest level of diagnostic certainty—definite, prob-able, or no event—according to previously defined criteria. 27,33–35
Definite perioperative myocardial infarction was defined as any episode occurring within 30 days after the date of operation in which either routine or clinically relevant cardiac evaluation (new onset of symptoms consistent with cardiac morbidity) demonstrated any two of the following three criteria: (1) creatine kinase MB fraction greater than 3 times the upper limit of normal (or a creatine kinase > 3 times normal if no creatine kinase MB fraction was obtained; (2) electrocardiographic evidence of new Q waves greater or equal to 0.04 s; or (3) new wall motion abnormality (by echocardiography or scintillation scanning). Probable perioperative myocardial infarction included episodes occurring within 30 days after the date of operation in which (1) one of the three criteria above was met and (2) the medical record indicated that a physician had made the diagnosis of myocardial infarction. Perioperative pulmonary embolism was categorized as definite when it occurred within 30 days after the date of operation and either (1) pulmonary angiography, computed tomographic scan, magnetic resonance imaging, echocardiographic visualization, or pathologic examination of thrombus removed at surgery or autopsy confirmed pulmonary embolism; or (2) the onset of new symptoms highly suggestive of acute pulmonary embolism were associated with a perfusion or ventilation–perfusion lung scan interpreted as high probability for pulmonary embolism. Perioperative pulmonary embolism was categorized as probable if confirmatory tests were either not performed or were indeterminate and (1) the medical record indicated that a physician made a diagnosis of pulmonary embolism; (2) signs and symptoms consistent with pulmonary embolism were present; and (3) the patient received a course of anticoagulation therapy with heparin, warfarin, or a similar agent, or placement of an inferior vena cava filter. Interim anticoagulation therapy while awaiting diagnostic evaluation for suspected deep vein thrombosis or pulmonary embolism was not included as evidence of an event. Perioperative deep venous thrombosis was categorized as definite when (1) it occurred within 30 days after the date of operation when confirmed by venography, computed tomographic scan, magnetic resonance imaging, or pathologic examination of thrombus removed at surgery or autopsy; or (2) the onset of local or systemic symptoms suggestive of deep venous thrombosis was associated with a documented positive noninvasive diagnostic test: impedance plethysmography, continuous wave Doppler ultrasound, compression duplex ultrasonography, radionuclide venography, or radiolabeled fibrinogen leg scan. Perioperative deep venous thrombosis was categorized as probable if testing for the definite level of diagnostic certainty was either not performed or was indeterminate and (1) the medical record indicated that a physician made a diagnosis of deep vein thrombosis; (2) signs and symptoms consistent with deep vein thrombosis were present; and (3) the patient received a therapeutic course of anticoagulation with heparin, warfarin, or a similar agent or a surgical procedure for deep vein thrombosis. The end-point of death was obtained from documented events in the institutional databases and the social security index.
In patients with more than one adverse event (myocardial infarction, pulmonary embolism, deep venous thrombosis, or death), each episode was categorized into the highest level of diagnostic certainty present for each manifestation. Patients experiencing a perioperative pulmonary embolism were considered to have an associated probable deep venous thrombosis when no lower extremity diagnostic testing was performed. All patient records were abstracted by using a standardized data collection form. Demographic data, surgical information and date, and timing of adverse event and detailed diagnostic criteria were recorded.
The diagnostic certainty categories of definite and probable were combined for analysis to maximize the clinical relevance of these events. A preliminary analysis was performed to confirm that the ages and sexes of the definite and probable event groups did not differ significantly. The frequencies of myocardial infarction, pulmonary embolism, deep venous thrombosis, and death were calculated overall and also separately for patients undergoing unilateral knee, bilateral knee, and total hip arthroplasty. Age- and sex-specific frequencies were also calculated. For each event type, multiple logistic regression analyses were performed to determine whether the frequency of the event differed significantly across calendar time, age, sex, or type of procedure. For these analyses, calendar time and age were treated as continuous variables, whereas sex and type of procedure were treated as categorical variables. In all cases, two-tailed P values less than or equal to 0.05 were considered statistically significant.
A total of 10,244 patients (4,775 male, 5,469 female) who underwent elective primary total hip or knee joint replacement surgery in the 10-yr period of study are included in this article (table 1). There were an additional 158 patients who underwent one of these procedures during the study period who were excluded from the study because they had denied access to their medical records. Most patients (80.6%) were 60 yr of age or older at the time of surgery. The total number of patients undergoing primary hip or knee arthroplasty, for diagnoses other than fracture, remained relatively consistent during the 10-yr interval from 1986 through 1995. No differences in sex or age distributions were found over the study period. Overall, 5,233 patients underwent primary hip replacement, 3,601 underwent unilateral knee replacement, and 1,410 underwent bilateral knee replacement surgery (table 1). Nearly all (> 95%) of the surgeries included in this study were performed by a group of 20 orthopedic surgeons, with each surgeon performing a median of 453 surgeries (range, 132–1,273).
After the initial database screen, 423 patients were identified as potentially having one or more adverse events (myocardial infarction, pulmonary embolism, deep venous thrombosis, or death) within 1 yr after the hip or knee replacement surgery. The medical records of these 423 patients were reviewed, and perioperative adverse events were assigned diagnostic certainty categories as described in the Methods section. Two hundred twenty-four patients were found to have experienced one or more definite or probable adverse events in the 30 days after surgery, corresponding to an overall event rate of 2.2% (myocardial infarction: 0.4%; pulmonary embolism: 0.7%; deep venous thrombosis: 1.5%; death: 0.5%). The overall frequency of adverse events and the frequency of individual adverse events did not change significantly over the 10-yr period of study, with the exception of myocardial infarction, for which there was a slight increase over time (0.2% in 1986 and 1987, and 0.6% in 1994 and 1995;P = 0.004).
The frequencies of myocardial infarction, pulmonary embolism, deep venous thrombosis, and death are presented in table 2and figure 1, according to age and type of procedure. The age- and sex-specific frequencies of these events are presented in figure 2. The frequency of myocardial infarction increased markedly with older age (P < 0.001) and was higher for male patients compared with female patients (P = 0.016, fig. 2). For patients aged 60–69 yr, 70–79 yr, and 80 yr or more, the corresponding frequencies of myocardial infarction were 0.1, 0.4, and 1.3% for female patients and 0.4, 0.7, and 2.2% for male patients. The median time to myocardial infarction was 1 day; 38 of 46 (83%) events occurred within 3 days; and 43 (93%) occurred within 14 days after the operation. The frequency of pulmonary embolism was similar for male and female patients but was increased in older patients (P = 0.002) and was found to be significantly different (P < 0.001) across procedure types. For patients aged 60–69 yr, 70–79 yr, and 80 yr or more, the corresponding frequencies of pulmonary embolism were 0.2, 0.8, and 0.9% for unilateral knee procedures; 1.0, 1.8, and 2.7% for bilateral knee procedures; and 0.4, 0.4, and 1.3% for hip procedures. In general, patients undergoing bilateral knee procedures were observed to have a frequency of pulmonary embolism nearly double that of patients undergoing unilateral procedures (fig. 1). The median time to pulmonary embolism was 4 days; 45 of 68 (66%) events occurred within 7 days; and 53 (78%) occurred within 14 days after the operation. The overall frequency of clinically detected deep venous thrombosis was 1.5%. The frequency of deep venous thrombosis was not found to differ significantly with age, sex, or type of procedure. The median time to deep venous thrombosis was 7 days; 110 of 153 (72%) events occurred within 14 days after the operation. Perioperative mortality was similar for male and female patients and was consistent across procedure types but increased significantly (P < 0.001) with older age. The frequency of death was less than 0.2% for patients aged less than 70 yr, 0.4% for patients aged between 70 and 79 yr, and 2.1% for patients aged 80 yr or more. The median time to death was 9 days, with 30 of 47 (64%) events occurring within 14 days after the operation. Thirteen (28%) of 46 patients who experienced probable or definite myocardial infarction and 10 (14%) of 69 patients who experienced probable or definite pulmonary embolism died within 30 days after surgery. Of the 155 patients experiencing deep venous thrombosis, there were 10 deaths, all of which occurred in patients who also experienced pulmonary embolism.
Overall, in these patients undergoing primary hip or knee arthroplasty, the percentage who experienced one or more of these adverse events (myocardial infarction, pulmonary embolism, deep venous thrombosis, or death) increased with older age. Of 5,433 patients aged less than 70 yr, 1.7% experienced one or more of these events. This percentage increased to 2.2% for 3,692 patients aged 70–79 yr and to 4.6% for 1,119 patients aged 80 yr or more.
This review provides physicians involved in the perioperative care of arthroplasty patients with information about the rate of clinically relevant perioperative morbidity and mortality after major elective orthopedic surgery of the lower extremities performed with contemporary operative and perioperative care. The frequencies of myocardial infarction, pulmonary embolism, and death were higher in older patients, particularly those aged 70 yr or older. Given the increasing age of the population, lower extremity joint replacement operations are now considered and performed in many older patients with serious coexisting diseases. Therefore, an accurate description of the risks associated with these surgeries should help clinicians and their patients in the decision regarding treatment options.
The risk of myocardial infarction, pulmonary embolism, deep venous thrombosis, or death found in the current study is in accordance with the results of several recent studies. 3,8,14,15,36Previous studies have consisted mostly of small series or single surgeon case reports without prospective case ascertainment. Consequently, comparison with historical control studies proves inadequate for estimation of the event rate over time.
Perioperative myocardial infarction is a significant cause of morbidity and mortality in the noncardiac operative population because of the high prevalence of risk factors for coronary artery disease. 12,29,37Major orthopedic operations are commonly performed in older patients, possibly with significant coronary artery disease, hypertension, and atherosclerosis. 1–3The American College of Cardiology/American Heart Association Task Force suggests that the risk of overall cardiac morbidity in the orthopedic population is intermediate, in the order of 1–5%. 12There are little data available in the literature regarding the current frequency of perioperative myocardial infarction after lower extremity arthroplasty operations. Marsch et al. 20prospectively examined myocardial ischemia using perioperative Holter monitoring and reported a frequency of myocardial infarction of 1.9% in 52 patients after total hip arthroplasty operations. In the current study, we found a 0.4% frequency of clinically relevant myocardial infarction within 30 days after elective primary total hip or knee arthroplasty operations. A clear age- and sex-specific effect on frequency was evident. However, potential differences in underlying comorbid diseases may contribute to the sex effect on the rate of myocardial infarction and cannot be excluded.
The Fifth American College of Chest Physicians Consensus Conference on antithrombotic therapy 26evaluated available literature and reported total prevalence rates of deep venous thrombosis after hip replacement surgery (including asymptomatic cases detected by venography) between 45 and 57%, with proximal deep venous thrombosis rates between 23 and 36%. Similarly, total prevalence rates of deep venous thrombosis after total knee replacement were reported between 40 and 84%, with proximal deep venous thrombosis rates of 9–20%. Total prevalence rates of pulmonary embolism were 0.7–30% after hip replacement and 1.8–7% after knee replacement surgeries. 26In this same document, prevalence rates for fatal pulmonary embolism were reported between 0.34 and 6% after hip arthroplasty and between 0.2 and 0.7% after knee arthroplasty. In the current study, we found an overall frequency of clinically significant and symptomatic deep venous thrombosis of 1.5% in patients treated with elective primary lower extremity arthroplasty. This rate is certainly lower than that reported when prospective evaluation of deep venous thrombosis using venography is undertaken 16,18,19,21,24,25but is comparable to previous reports based on clinical symptoms. 1,7,9,10,13–15We found a 0.7% frequency of clinically relevant pulmonary embolism within 30 days after elective primary total hip or knee arthroplasty operations, an incidence slightly lower than that of earlier investigations. Importantly, despite the successful reduction of asymptomatic thromboembolic disease with the use of routine anticoagulation, an actual reduction of clinically relevant adverse events has been difficult to demonstrate in everyday medical practice. 36Although there is a large amount of clinical data documenting the efficacy and cost effectiveness of routine thromboprophylaxis in patients undergoing elective total hip or knee replacement surgery, 26it is likely that selected patient populations might receive the most benefit from specific prophylactic measures. 36For example, in our study, patients undergoing bilateral knee replacement procedures had a much higher frequency of pulmonary embolism when compared with unilateral knee or even hip replacement patients. However, the frequency of death in patients treated with elective primary bilateral knee arthroplasty was not different from that of patients undergoing unilateral knee or hip replacement operative procedures.
Our report is the largest, detailed, single-institution study of perioperative morbidity and mortality after primary lower extremity arthroplasty. Comparison of the rate of clinically relevant adverse events from study to study and over time is complicated by the differing end points and criteria used for diagnosis of adverse events. A review of available studies does not show an obvious reduction in event rates with time. However, it is important to note that the population of patients considered as candidates for elective arthroplasty operations has probably changed over time, and more patients who are older, sicker, and at greater risk are offered elective operations now than in the past. In addition, adverse events are more likely to occur during the patient's hospital stay and subsequent rehabilitation. In the current study, the median times to myocardial infarction, pulmonary embolism, and death were 1, 4, and 9 days, respectively. Identification of patients at risk for serious complications and accurate knowledge of the timing of adverse events may allow early diagnosis and intervention through heightened surveillance.
There are limitations inherent to the design of this study. The quality of the data, especially regarding events, is dependent on the accuracy and completeness of the clinical record. In fact, it is possible for some adverse perioperative events, such as a non–Q-wave myocardial infarction or small pulmonary embolus without hemodynamic changes, to go unrecognized by the physician in charge of the patient's care. However, all the data were collected prospectively, and the extent of follow-up of patients undergoing arthroplasty surgery at our institution is such that the likelihood of missing clinically relevant adverse events is very small. In the current study, a single person collected all data using standardized data collection forms and strict criteria for each adverse event. Distinct levels of diagnostic certainty were used according to the evidence present in each diagnosis. We assume that all clinically relevant events came to clinical attention and were noted on the record and thus were available for analysis.
The applicability of our results, which represent a large-volume academic center, to other clinical settings is unknown. Although a relation between surgical volume and outcome has been suggested, the results are conflicting. Taylor et al. 6reported a decreased 30-day mortality rate after total hip arthroplasty in hospitals with higher surgical volumes, but no attempt was made to adjust for the presence of comorbidities or degree of surgical difficulty. Conversely, in a review of the Veteran's Health Administration multihospital healthcare system, Khuri et al. 5found that specialty and procedure volume did not affect the risk-adjusted 30-day mortality for eight commonly performed procedures of intermediate complexity, including total hip replacement. Therefore, there are no definitive data to suggest that the surgeon's experience and surgical volume affect mortality. However, additional investigations are needed to determine the role of surgical volume and patient comorbidities in perioperative morbidity, including myocardial infarction, pulmonary embolism, and death.
In conclusion, this study represents one of the largest investigations evaluating the frequency of adverse perioperative morbidity and mortality in patients undergoing lower extremity arthroplasty using current medical treatment. The observed frequency of adverse events in this patient population provides information that may be helpful when counseling patients about the risks associated with elective primary lower extremity joint replacement operations.