Total knee replacement can be performed with either general anesthesia or regional anesthesia. Epidural anesthesia has a number of potential advantages over general anesthesia. Epidural anesthesia avoids central nervous system depression; has a different spectrum of effects on the cardiopulmonary system; may modify the stress response to surgery; facilitates delivery of postoperative analgesia; and has been associated with a lower risk of deep vein thrombosis after total hip replacement. General anesthesia, however, is a technique with which all anesthesiologists are familiar, and general anesthesia can be induced rapidly.
To compare the effects of epidural anesthesia and general anesthesia on the major clinical outcomes after primary unilateral total knee replacement, a randomized clinical trial of epidural anesthesia versus general anesthesia was performed. Williams-Russo et al26 have previously reported the effects on long term cognitive function, postoperative delirium, and cardiac complications. This paper compares the impact of epidural anesthesia and general anesthesia on the rate of attainment of rehabilitation goals during hospitalization which indicate readiness for discharge. The incidence of thromboembolism was also investigated in the study subset of 178 patients who were prescribed a common thromboembolic prophylaxis and detection protocol. Thromboembolic complications are the most frequent potentially serious medical complications of lower extremity joint replacement.
MATERIALS AND METHODS
All patients undergoing elective primary unilateral total knee replacement with participating orthopaedic surgeons at the Hospital for Special Surgery between October 1989 and October 1992 were screened for eligibility. Exclusion criteria included age less than or equal to 40 years, a history of surgery performed with either a regional or general anesthetic in the 3 months preceding total knee replacement, or the presence of a contraindication to either epidural anesthesia or general anesthesia.26 The protocol was reviewed and approved by the Institutional Human Rights Committee. After patients gave written informed consent, they were randomly assigned to receive either epidural anesthesia or general anesthesia by a blocked schedule based on a random numbers table.
Anesthesia Protocol- Epidural Anesthesia
All patients received oxygen supplementation via a nasal cannula throughout the case. After sterile preparation, a Number 17 gauge Tuohy needle (Braun Medical, Bethlehem, PA) was inserted with a paramedian approach and a catheter placed 3 to 4 cm into the epidural space between the third and fourth or between the second and third lumbar vertebrae. Patients received either lidocaine 2% or bupivacaine 0.75%. Adjunctive medications for sedation included midazolam and fentanyl.
Anesthesia Protocol- General Anesthesia
Induction was accomplished with thiopental sodium, fentanyl, and vecuronium. Intubation was by endotracheal tube. Maintenance of anesthesia was accomplished with fentanyl, inhaled nitrous oxide (70%), and isoflurane. Neuromuscular blockade was reversed with neostigmine and atropine or glycopyrrolate.
Surgery was performed with a thigh tourniquet inflated to 350 mm Hg after exsanguination of the limb with an Esmarch bandage. After the fixation of components with methylmethacrylate, the tourniquet was deflated for several minutes to achieve hemostasis; the operative leg was reexsanguinated with an Esmarch bandage and the tourniquet reinflated. At the end of surgery, the tourniquet was again deflated after the dressings were applied. All patients received bicondylar cemented components.
Postoperative analgesia was initiated as soon as patients complained of pain in the postanesthesia care unit. The route of administration and dosing of postoperative analgesics were left to the discretion of the treating anesthesiologist. During the first year of the study, the patients in the epidural anesthesia group received epidural analgesia for 12 to 24 hours postoperatively, whereas the patients in the general anesthesia group received systemic narcotics during the same period. When they returned to their hospital rooms, both groups of patients received narcotics on an as needed basis (intramuscularly, subcutaneously, or orally) for postoperative analgesia. Early in the second year of the study, patient controlled analgesia was introduced and made available in the postanesthesia care unit and in the patients' rooms. Thereafter, patients in the epidural anesthesia group had epidural analgesia for 48 to 72 hours postoperatively, whereas patients in the general anesthesia group had intravenous patient controlled analgesia for the same period.
All patients received the same perioperative physical therapy protocol. Continuous passive motion was applied within the first 24 hours postoperatively. Transfers in and out of bed, ambulation, weightbearing as tolerated with a walker, and therapeutic exercise was initiated on postoperative Day 2. Physical therapy was temporarily discontinued if the prothrombin time was greater than 20 seconds or if there was excessive wound drainage as determined by the surgeon.
The course of rehabilitation progress was documented on a previously validated instrument, the Hospital for Special Surgery total knee replacement functional milestone from.10 Each functional milestone is scaled into 2 levels of achievement, performance assisted and performance unassisted. Assisted is defined as requiring aid from another person to perform the activity, that is, physical assistance, contact guarding, verbal cuing or supervision. Unassisted is defined as the ability to perform each milestone without the assistance of another person, and with or without an ambulatory aid (walker, cane, crutches). The need for ambulatory aids was not scaled because every patient discharged from the hospital left with an assistive device.
During the study period, readiness for discharge required attainment of the following functional milestones: unassisted ability to transfer in and out of bed, unassisted walker or cane ambulation, unassisted negotiation of stairs, and 90 ° active knee flexion.
Perioperative Thromboembolic Prophylaxis and Surveillance Protocol
Five of the 7 participating surgeons routinely used a common thromboembolic prophylaxis and surveillance protocol. The prophylaxis protocol included oral acetylsalicylic acid 325 mg twice a day starting on postoperative Day 1, the use of graded elastic stockings on the nonoperative limb, and early mobilization as described in the rehabilitation protocol. Patients prescribed the thromboembolic protocol did not receive perioperative heparin or sequential pneumatic compression boots. Usual contraindications to inclusion for this protocol included chronic anticoagulation with warfarin or heparin for medical comorbid conditions such as cardiac or neurologic disease, and contraindications to aspirin use including moderate or severe renal insufficiency or recent active peptic ulcer disease or gastritis. Inclusion of patients with a history of deep vein thrombosis or pulmonary embolism was at the discretion of the orthopaedist and consulting internist. All patients were instructed to discontinue aspirin use at least 1 week and nonsteroidal antiinflammatory drug use at least 48 hours before surgery.
The routine deep vein thrombosis surveillance protocol of the participating orthopaedic surgeons included an ascending contrast operative limb venogram performed on postoperative Days 4 or 5 according to the technique of Rabinov and Paulin.20 Patients with a negative venogram continued to receive aspirin 325 mg twice a day orally for 6 weeks after surgery. Patients with venographic evidence of deep vein thrombosis (calf or proximal clot) had aspirin discontinued and were started on oral warfarin, which was continued for 3 months after surgery.
The routine pulmonary embolism surveillance protocol of the participating surgeons included a perfusion lung scan performed 7 to 10 days preoperatively using injections of 4 mCi of technetium 99 (99 mTc) macroaggregated albumin. A series of 8 images were taken. Perfusion lung scans were repeated on postoperative Days 4 or 5. Results were classified as low, medium, or high probability of pulmonary embolism based on the radiologist's comparison of the preoperative and postoperative scans. The radiologists reading the venograms and the lung scans were blind to anesthesia status.
All patients were interviewed and examined by study physicians at least once daily through postoperative Day 7 for any symptoms or signs suggestive of deep vein thrombosis or pulmonary embolism. In addition, all patients were interviewed 6 months postoperatively, and any interim adverse medical outcomes were documented.
Comparison between the 2 anesthesia groups of the mean length of time to achieve rehabilitation milestones was assessed using Student's test for 2 samples, using Student's t test procedure, and by nonparametric analysis of the postoperative day on which each milestone was reached using the nonparametric analysis of variance on ranks, both available in the SAS Version 5 (SAS Institute, Cary, NC).21 Chi square and when appropriate, Fisher's exact test, was used to compare the proportions of patients reaching a rehabilitation milestone before discharge, and to compare the incidence of deep vein thrombosis and pulmonary embolism between the 2 anesthesia groups. Chi square and Fisher's exact test were also used for comparison of incidence rates in different subgroups based on reported risk factors for thromboembolic complications. All p values were 2-tailed.
Patient Enrollment and Compliance With Anesthesia Intervention
As previously described,26 the target sample size of 262 patients was calculated based on the primary outcome of long term cognitive complications. To achieve enrollment of 262 patients, 452 eligible patients were approached, and 58% agreed to participate. One hundred thirty-four patients were randomized to receive epidural anesthesia, and 128 patients to receive general anesthesia. There were 6 deviations from the assigned anesthesia regimen. One patient randomized to epidural anesthesia had a history of lumbar laminectomy, had a failed block, and received general anesthesia. A second patient in the epidural anesthesia group with a history of lumbar spinal fusion had an adequate block for the first hour of the case and then, because of inadequate muscle relaxation required light general anesthesia for an additional 30 minutes. A third patient in the epidural anesthesia group had an inadvertent spinal anesthetic with a high thoracic block and required intubation for respiratory support with light general anesthesia. Three patients randomized to general anesthesia instead received epidural anesthesia.
Baseline Comparability of the 2 Anesthesia Groups
As previously reported, there were no significant differences between the 2 anesthesia groups at baseline in the distribution of characteristics including age, gender, comorbid medical conditions, or preoperative laboratory values.26 Of the 262 patients, 9 patients were not included in the analysis of the rehabilitation data: 3 of these patients had postoperative complications leading to transfer to an intensive care unit with cessation of physical therapy for at least the first week postoperatively; 2 patients received combined epidural anesthesia and general anesthesia as described above; and 4 patients' functional milestone data forms were incomplete. Thus, 253 patients had complete data for inclusion in the analysis. Again, there were no differences in the baseline distribution of patient characteristics between the 2 anesthesia groups for these patients.
Thromboembolic Protocol Subset and Baseline Comparability
Of the total of 262 patients in the main trial, 188 patients were prescribed the common thromboembolic prophylaxis and surveillance protocol. Of the 74 patients not prescribed the thromboembolic protocol, 35 were patients of the 2 surgeons who followed a different thromboembolic prophylaxis routine protocol perioperatively (warfarin prophylaxis and no venogram); 21 had a history of deep vein thrombosis or pulmonary embolism and their surgeons or internists elected to follow a different prophylaxis protocol (warfarin); 10 were patients with a contraindication to aspirin use; and 8 were on chronic warfarin anticoagulation for cardiac or neurologic disease. The 74 patients not on thromboembolic protocol included 36 patients in the epidural anesthesia group and 38 patients in the general anesthesia group.
Of the 188 patients prescribed the common thromboembolic prophylaxis protocol, 10 patients did not have an adequate venogram performed. This included 5 patients in whom the radiologists were technically unable to cannulate a vein despite repeated attempts; 2 patients who refused; 1 patient with a history of dye allergy; and 2 patients in whom new postoperative increases in creatinine developed. Complete data on deep vein thrombosis incidence are thus available for 178 patients, including 97 randomized to epidural anesthesia and 81 to randomized to general anesthesia.
Table 1 shows the baseline comparability of the 2 anesthesia groups for the 178 patients prescribed the thromboembolic prophylaxis and surveillance protocol. The mean age was 68 in both groups. The proportions of women in each group were similar at 74% and 65%. Both groups were similar regarding the comorbid conditions of obesity, (33% in the epidural anesthesia and 38% in the general anesthesia groups as defined by a Quetelet Index2 ≥ 30), hypertension (38% and 39%, respectively), and the Charlson Comorbidity Index.3 However, slightly fewer patients in the epidural anesthesia group, 3%, had a history of deep vein thrombosis or pulmonary embolism compared with 10% of the patients in the general anesthesia group (p =.1).
Comparison of the proportions of patients in each anesthesia group who successfully achieved each milestone before discharge revealed a consistent trend favoring epidural anesthesia, but this reached statistical significance only for stairs assisted (p <.001, Fisher's Exact test). As shown in Table 2, all functional milestones were achieved at a slightly faster rate in the epidural anesthesia group than in the general anesthesia group, but again, this difference reached statistical significance only for the length of time to achieve assisted stair climbing, which was reached on average 1½ days earlier in the epidural anesthesia group (p <.002, Student's t test). On nonparametric analysis of the data, there was again a consistent trend favoring epidural anesthesia, although only the postoperative day that stairs assisted was achieved reached statistical significance (p <.009). The postoperative day on which patients achieved the goal of 90 ° active knee flexion was also significantly earlier for patients in the epidural anesthesia group than patients in the general anesthesia group for those patients who reached this goal (p <.03, Student's t test).
The mean length of stay was not significantly different in the 2 groups, 12.1 days for epidural anesthesia versus 12.7 days for general anesthesia patients. Discharge from the hospital lagged an average of 1 to 2 days after rehabilitation goals had been met.
Subgroup analysis showed that age and body mass index were significant predictors of the rate of attainment of rehabilitation goals. Patients aged 70 years and older reached the milestones of walker unassisted, cane unassisted, and stairs unassisted more slowly than did younger patients (p <.005 for all). Patients with a Quetelet Index of 30 or greater reached the milestones of cane assisted, cane unassisted, and stairs assisted more slowly than did nonobese patients (p <.01 for all). As noted previously, there was no difference in the distributions of age and obesity between the 2 anesthetic groups.
The overall incidence of deep vein thrombosis was 44% (78/178). All clots were limited to the deep calf veins, with no patient having evidence of clot proximal to the calf. There were no complications of venography. The incidence of deep vein thrombosis in the epidural anesthesia group was 40% compared with 48% in the general anesthesia group (p =.3), a difference in incidence rates of 8%. The power of this trial to have detected a 10% difference in incidence rates is 61%, and its power to have detected a 20% difference in incidence is 95%.
Regarding other potential risk factors for deep vein thrombosis, no significant associations between deep vein thrombosis and age, gender, obesity, or history of thromboembolic disease were observed. Of the 11 patients with a history of thromboembolism, a calf clot developed in 4.
Lung Perfusion Scan Results
The overall incidence of symptomatic pulmonary emboli was 1.7% (3/178). Of the 178 patients who had a venogram, 153 patients had a postoperative lung perfusion scan performed. Twenty-five patients did not have a lung scan. Because their venograms showed a new calf clot, warfarin anticoagulation was initiated, and their treating physicians thought there would be no change in management based on the lung scan results.
The overall incidence of new perfusion defects postoperatively compared with preoperative lung perfusion scans was 10.5% (161/153). Five of the 16 patients with new perfusion defects had multiple bilateral subsegmental defects and were deemed to have a high likelihood of postoperative pulmonary emboli and did not go on to have a combined ventilation perfusion scan. The 3 symptomatic patients belonged to this category. Eight of the 16 patients with new perfusion defects had confirmatory ventilation perfusion scans performed showing unmatched defects read as highly probable for pulmonary emboli. Three of the 16 patients with new perfusion defects had intermediate probability scans.
The incidence of new perfusion defects was similar in both groups: 12% (10/86) in patients in the epidural anesthesia group and 9% (6/67) in patients in the general anesthesia group (p =.6).
The duration of surgery was not significantly different by type of anesthesia, with a mean and standard deviation of 85 ± 33 minutes in the epidural anesthesia group versus 88 ± 32 minutes in the general anesthesia group. The length of stay was also not significantly different between the 2 anesthesia groups (Table 1). There were no inhospital deaths. There were 2 deaths within the first 2 months after discharge. A patient in the epidural anesthesia group had a cerebral vascular accident 48 hours after surgery, and after discharge to a rehabilitation facility, died secondary to a second cerebral vascular accident. A patient in the general anesthesia group died unexpectedly at home 8 weeks after surgery. No autopsy was performed, and the specific cause of death was not determined. The patient's early recovery had been complicated only by a calf deep vein thrombosis treated with warfarin.
This is the largest randomized clinical trial of epidural anesthesia versus general anesthesia for total knee replacement reported to date. The detailed methods and the primary outcomes of the trial, have been previously reported26; no significant differences between epidural anesthesia and general anesthesia were observed in the incidence of long term cognitive complications, postoperative delirium, or perioperative cardiac complications. This paper reports on other clinical outcomes of major importance to orthopaedists, namely, progression of rehabilitation and thromboembolic complications. A consistent trend toward a faster rate of inhospital rehabilitation was noted with epidural anesthesia. This reached statistical significance for the ability to climb stairs without the assistance of another person. Patients who received epidural anesthesia plus postoperative epidural analgesia climbed stairs unassisted 1 ½ days earlier on average than the patients who received general anesthesia. Ninety-eight percent of the patients receiving epidural anesthesia, but only 88% receiving general anesthesia, were able to climb stairs before discharge. The ability to climb stairs requires motivation, adequate range of motion, strength in the operative limb, and a relatively pain free state. By contrast, other milestones such as the ability to stand or to achieve 90 ° knee flexion are less complex and depend primarily on specific factors such as level of consciousness, surgical technique, or adequacy of pain control.
It has been proposed that regional anesthesia and epidural analgesia with local anesthesia may obtund the stress response to surgery, minimizing the catabolic effect.9 In the early postoperative period, epidural anesthesia plus epidural analgesia also minimizes pain with much less central nervous system depression, and thus facilitates earlier effective exercise therapy.4 More than 95% of the patients who had epidural anesthesia had continuous postoperative epidural analgesia with bupivacaine and fentanyl for 12 to 72 hours after surgery. As the trial progressed, the technique of postoperative epidural analgesia changed by using more narcotic, less local anesthetic, a patient controlled mode, and the infusions were maintained for 72 hours in most cases. It is not clear what role epidural anesthesia compared with postoperative epidural analgesia played in the improved rehabilitation.5,14 Whatever the mechanism, it seems that patients who received epidural anesthesia had a more rapid early recovery after total knee replacement which has implications for earlier discharge. An effect of anesthesia type on length of stay was not seen in this study; however, discharge usually lagged 1 to 3 days after rehabilitation readiness for discharge. This delay was due to a variety of patient specific factors, such as availability of home services or transportation, and to medical issues such as reaching therapeutic anticoagulation levels in patients with thromboembolic complications.
This is also the largest randomized trial reported to date investigating the effect of epidural versus general anesthesia on thromboembolic complications after total knee replacement. This was not 1 of the primary outcomes of the study, and thus enrollment into the main trial did not require surgeon or patient participation in this aspect of the trial. The authors were able to take advantage of the fact that the majority of the participating surgeons followed the same standardized thromboembolic prophylaxis and surveillance protocol. This allowed the patients to be observed prospectively with daily physical examinations, to determine the reasons for adherence or nonadherence to the usual protocol, and to blind the outcome assessors to which type of anesthesia the patients had received.
The rate of venographically documented deep vein thrombosis in the operative leg was 40% in the epidural anesthesia and 48% in the general anesthesia groups. This difference in rate does not reach statistical significance; however, the power to detect a true difference in rates of 10% was only 61%, or stated differently, the probability of falsely concluding that there is not at least a 10% difference between the rates is 39%. As noted in the methods, the study sample size was calculated before the trial on the basis of the primary outcome and long term cognitive complications, and not on the basis of deep vein thrombosis rates.
A previous randomized study of total knee replacement showed a reduction of deep vein thrombosis rate with epidural anesthesia, but only 48 patients were studied.8 In another randomized trial of total knee replacement, no significant difference in venographically documented deep vein thrombosis rate was noted, although proximal thrombi were less frequent with epidural anesthesia.12 A large retrospective historical cohort study at the authors' institution found that after the introduction of epidural anesthesia, a significantly lower rate of deep vein thrombosis was observed after unilateral total knee replacement compared with the rate during an earlier period when only general anesthesia was available (42 versus 56%, respectively).22 The authors therefore sought to investigate this issue in this concurrent, prospective randomized trial.
In contrast, significantly lower rates of deep vein thrombosis have been documented using regional anesthesia compared with general anesthesia after hip surgery.14,19,25 At the authors' institution, the rate of deep vein thrombosis after total hip replacement has fallen to less than 15% in patients receiving hypotensive epidural anesthesia.11,24 The difference between the impact of anesthesia on deep vein thrombosis for total hip replacement versus total knee replacement may relate to the use of a thigh tourniquet during total knee replacement. During total hip replacement, preservation or augmentation of lower extremity blood flow occurs during surgery performed under epidural anesthesia.1,6,13 By contrast, during total knee arthroplasty, inflation of the tourniquet prevents this increase in flow in the operative leg. The tourniquet leads to ischemia of the limb, and a prothrombotic state is established at the endothelial surface.16,18 This may be the physiologic mechanism of thrombogenesis leading to the high frequency of deep vein thrombosis occurring with total knee arthroplasty.23
Despite the high incidence of postoperative deep vein thrombosis, no patient in this study had evidence of a clot proximal to the popliteal veins. Nonetheless, there was convincing evidence of postoperative pulmonary embolization in patients without proximal clots in the operative leg. Three of the 16 patients with new perfusion defects had negative venograms with no evidence of calf or proximal clots. All 3 patients were in the ventilation perfusion scan confirmed group. Bilateral lower extremity venography was not done, and thus it cannot be ruled out that patients with pulmonary emboli did not have a proximal clot in the nonoperative leg. Nonetheless, this observation reinforces the importance of not considering a calf deep vein thrombosis as clinically inconsequential.7,15,17
This randomized clinical trial of epidural versus general anesthesia for primary unilateral total knee replacement showed an enhanced rate of early rehabilitation in patients who received epidural anesthesia. In contrast to studies of hip surgery, the authors did not find a significant reduction in postoperative deep vein thrombosis rate with regional anesthesia. This may reflect the use of a thigh tourniquet during total knee arthroplasty which interferes with the salutary mechanisms of epidural anesthesia during hip surgery.
1. Bading B, Blank S, Sculco TP, Pickering TG, Sharrock NE: Augmentation of calf blood flow by epinephrine infusion during lumbar epidural anesthesia. Anesth Analg 78:1119-1124, 1994.
2. Bray GA: Obesity: Definitions, diagnoses and disadvantages. Med J Aust 142:52-58, 1985.
3. Charlson M, Pompei P, Ales K, MacKenzie C: A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chron Dis 40:373-383, 1987.
4. Dahl JB, Daugaard JJ, Rasmussen B, et al: Immediate and prolonged effects of pre- versus postoperative epidural analgesia with bupivacaine and morphine on pain at rest and during mobilization after total knee arthroplasty. Acta Anaesthesiol Scand 38:557-561, 1994.
5. Dalldorf PG, Perkins FM, Totterman S, Pellegrini Jr VD: Deep venous thrombosis following total hip arthroplasty. Effects of prolonged postoperative epidural anesthesia. J Arthroplasty 9:611-616, 1994.
6. Davis FM, Laurenson VG, Gillespie WJ, Foate J, Seager AD: Leg blood flow during total hip replacement under spinal or general anaesthesia. Anaesth Intensive Care 17:136-142, 1989.
7. Giachino A: Relationship between deep-vein thrombosis in the calf and fatal pulmonary embolism. Can J Surg 31:129-130, 1988.
8. Jørgensen LN, Lind B, Hauch O, et al: Thrombin-antithrombin III-complex & fibrin degradation products in plasma: Surgery and postoperative deep venous thrombosis. Thromb Res 59:69-76, 1990.
9. Kehlet H: Surgical stress: The role of pain and analgesia. Br J Anaesth 63:189-195, 1989.
10. Kroll M, Ganz S, Backus S, et al: A tool for measuring functional outcomes after total hip arthroplasty. Arthritis Care Res 7:78-84, 1994.
11. Lieberman JR, Huo MM, Hanway J, et al: The prevalence of deep venous thrombosis after total hip arthroplasty with hypotensive epidural anesthesia. J Bone Joint Surg 76A:341-348, 1994.
12. Mitchell D, Friedman RJ, Baker III JD, et al: Prevention of thromboembolic disease following total knee arthroplasty. Clin Orthop 269:109-112, 1991.
13. Modig J: Influence of regional anesthesia, local anesthetics and sympathicomimetics on the pathophysiology of deep vein thrombosis. Acta Chir Scand (Suppl) 550:119-127, 1988.
14. Modig J, Maripuu E, Sahlstedt B: Thromboembolism following total hip replacement: A prospective investigation of 94 patients with emphasis on the efficacy of lumbar epidural anesthesia in prophylaxis. Reg Anesth 11:72-79, 1986.
15. Monreal M, Ruiz J, Loazabal A, Arias A, Roca J: Deep venous thrombosis and the risk of pulmonary embolism. Chest 102:677-81, 1992.
16. Ogawa S, Gerlach H, Esposito C, et al: Hypoxia modulates the barrier and coagulant function of cultured bovine endothelium. J Clin Invest 85:1090-1098, 1990.
17. Philbrick JT, Becker DM: Calf deep venous thrombosis: A wolf in sheep's clothing? Arch Intern Med 148:2131-2138, 1988.
18. Pinsky D, Oz M, Liao H, et al: Restoration of the cAMP second messenger pathway enhances cardiac preservation for transplantation in a heterotopic rat model. J Clin Invest 92:2994-3002, 1993.
19. Prins MH, Hirsh J: A comparison of general anesthesia and regional anesthesia as a risk factor for deep vein thrombosis following total hip surgery: A critical review. Thromb Haemost 64:497-500, 1990.
20. Rabinov K, Paulin S: Roentgen diagnosis of venous thrombosis in the leg. Arch Surg 104:134-144, 1972.
21. SAS Institute Inc: SAS User's Guide: Statistics. Cary, NC, SAS Institute Inc 1985.
22. Sharrock NE, Haas SB, Hargett MJ, et al: Effects of epidural anesthesia on the incidence of deep-vein thrombosis after total knee arthroplasty. J Bone Joint Surg 73A:502-506, 1991.
23. Sharrock NE, Hargett MJ, Urquhart B, et al: Factors affecting deep vein thrombosis rate following total knee arthroplasty under epidural anesthesia. J Arthroplasty 8:133-139, 1993.
24. Sharrock NE, Ranawat CS, Urquhart B, Peterson M: Factors influencing deep vein thrombosis following total hip arthroplasty. Anesth Analg 76:765-771, 1993.
25. Sorensen RM, Pace NL: Anesthetic techniques during surgical repair of femoral neck fractures: A meta analysis. Anesthesiology 77:1095-1104, 1992.
26. Williams-Russo P, Sharrock NE, Mattis S, Szatrowski TP, Charlson ME: Cognitive effects after epidural vs general anesthesia in older adults. A randomized trial. JAMA 274:44-50, 1995.
SYMPOSIUM© Lippincott-Raven Publishers.