A sixty-four-year-old man with a medical history that included hypertension and non-insulin-dependent diabetes mellitus was managed with a right total knee arthroplasty for osteoarthrosis on October 18, 1993. The procedure was performed with the patient under spinal anesthesia. Preoperative examination of the knee had revealed a range of motion of 15 to 120 degrees and a varus deformity of 5 degrees.
Constant passive motion and physical therapy for range-of-motion exercises and walking were initiated on the first day after the operation. Venous duplex examination of the common femoral, superficial femoral, popliteal, and tibial veins was performed bilaterally on the fourth day after the operation because of bilateral pain in the calf; it showed no evidence of deep venous thrombosis. The pain in the calves resolved twenty-four hours after it had begun; at the time of discharge, the patient only had pain in the knee that was attributable to the operation, and motion of the knee was 15 to 80 degrees.
During his hospital stay, the patient received prophylaxis for thromboembolism, with adjusted-dose warfarin used to maintain the international normalized ratio between 1.8 and 2.4. This kept the prothrombin time between 14.6 and 18.2 seconds (normal, 11.0 to 12.5 seconds). The first dose was given on the day of the operation, and subsequent doses were given daily thereafter. The patient was discharged on the eleventh postoperative day and completed a twenty-one-day course of warfarin therapy.
A follow-up examination was performed on the twenty-second postoperative day. The range of motion was 0 to 50 degrees, even though outpatient physical therapy had been continued biweekly. The patient had no pain or tenderness in the calf. Manipulation of the knee and intensive inpatient physical therapy were scheduled for the following day.
An anesthetic was administered epidurally on a continuous basis to relieve pain in the knee and to allow manipulation. The knee was flexed to 115 degrees with some palpable tearing of adhesions. Approximately one minute after the initiation of the manipulation, the patient abruptly became bradycardic, hypotensive, hypoxic, and unresponsive. Resuscitation was attempted for sixty minutes without success.
Postmortem examination of the thorax and right lower extremity revealed a large embolus in the pulmonary trunk and extending into both main pulmonary arteries. The embolus was a thrombus with some areas showing valve markings from deep veins. Histologically, the thrombus appeared to have been formed within a few days before the patient's death. Examination of the right popliteal space revealed a thrombus filling the right popliteal vein. Veins distal to the popliteal vein and contralateral limb were not examined.
Surgeons vary considerably regarding the use of prophylactic measures to prevent the development of deep venous thrombosis after total knee arthroplasty, indications for diagnostic screening tests, and treatment of thrombosis of the calf. We try to minimize thromboembolic complications with the routine use of warfarin for three weeks after the operation. We do not perform specific screening tests to detect asymptomatic thrombi. The purpose of this report is to heighten awareness that pulmonary embolism can occur during manipulation after total knee arthroplasty.
Khaw et al. reported symptomatic pulmonary embolism in seven (1 per cent) of 499 patients who had total knee arthroplasty without anticoagulation therapy. One additional patient (0.2 per cent) died of fatal pulmonary embolism, and the authors concluded that routine anticoagulation is of doubtful value for these patients because of the low prevalence of this complication. Although the thromboembolism in our patient represents a single event, it indicates that this complication is an important risk of total knee arthroplasty.
Our patient received a three-week course of warfarin therapy and had a negative duplex examination of the involved limb on the fourth day after the operation. Duplex examination is an appealing alternative to venography for the diagnosis of a deep venous thrombosis because it is non-invasive; however, it is dependent on the experience of the technician21. Our vascular laboratory is staffed with one full-time and one part-time technician who have been performing duplex examinations since 1987. Eight hundred and eighty-seven scans were performed in 1993, and each scan was interpreted by an attending vascular surgeon. Although no formal study for comparison of duplex scanning with venography has been conducted at our institution, we have abandoned the use of venography for detection of clinically suspected deep venous thrombosis because of our confidence in the vascular laboratory. The negative duplex scan on the fourth postoperative day and the postmortem documentation of a recently formed thrombus that had embolized to the lung on the twenty-third postoperative day indicate that the thrombus most likely developed relatively late in the three-week period after the operation. It is also clear that the thrombus was dislodged as a result of the manipulation of the knee.
The indications and results of manipulation after total knee arthroplasty are variable. Shoji et al. recommended manipulation for knees that did not have 90 degrees of flexion by the tenth postoperative day. Twenty-seven (54 per cent) of fifty knees were manipulated, and the range of motion improved an average of 30 degrees. Daluga et al. reported a rate of manipulation of 12 per cent and compared the results for ninety-four knees (sixty patients) that were manipulated with those for forty-one knees (twenty-eight patients) that were not manipulated. The anteroposterior dimension of the knee was measured on a lateral radiograph as the combined anteroposterior dimension of the distal part of the femur and the patella. This measurement before the operation was compared with the same measurement after it. An increase in the anteroposterior dimension of the knee as well as quadriceps adhesions were associated with the performance of a manipulation.
Fox and Poss reported on a consecutive series of patients who had been managed with 343 total knee arthroplasties. Eighty-one of the knees did not have 90 degrees of comfortable active flexion by two weeks and were manipulated with the patient under general anesthesia. These knees had a mean immediate gain of 37 degrees of flexion, but they had retained only 17 degrees after one week. After one year, the ranges of motion of manipulated and non-manipulated knees were similar. The authors concluded that the ultimate flexion of the knee did not depend on manipulation, and they recommended the procedure, to facilitate participation in physical therapy, for patients who stop gaining comfortable active motion early in the postoperative period. A fatal pulmonary embolism developed in one patient in their series, within twenty-four hours after the manipulation, from deep venous thrombosis in the calf of the manipulated limb. The authors did not discuss the role of diagnostic screening for deep venous thrombosis before manipulation.
The charge at our institution for a venous duplex scan is $150, which includes the fees of both the technician and the vascular surgeon. As described, in the series of eighty-one manipulations of the knee reported by Fox and Poss, fatal pulmonary embolism developed in one patient. For 343 patients, the cost of venous duplex screening, which could potentially prevent one fatal pulmonary embolism, would be $12,150. If, for every five patients who are managed with a total knee replacement, one needs a manipulation, and if this patient has a duplex scan at a cost of $150, the cost for this scan per patient having a total knee replacement distributed over the group of five patients would be $30 per patient. This is less than the cost of determining two prothrombin times or of administering one dose of cefazolin at our institution.
Pulmonary embolism is a recognized complication of total knee arthroplasty and accounts for approximately 68 per cent of deaths of patients managed with a total joint replacement20. Various methods of prophylactic anticoagulation with warfarin, dextran, or heparin have been advocated, as has external compression of the limb and rapid mobilization of the patient. However, it is apparent that in our patient a deep venous thrombosis developed and was dislodged as a result of manipulation two days after the completion of a twenty-one-day course of warfarin therapy. This resulted in a fatal pulmonary embolism. We now recommend that an assessment of the deep venous system be considered before manipulation of the knee is undertaken in the immediate postoperative period after an arthroplasty.