PERIPHERAL nerve block has emerged as an attractive anesthetic option when postoperative anticoagulation is planned and concern regarding spinal hematoma might preclude continuous epidural anesthesia. 1When the patient undergoes anticoagulation at the time of block placement, neuraxial block is contraindicated, and investigators have also recommended avoidance of deep blocks (such as lumbar plexus block) in anticoagulated patients out of concern for incompressible arterial bleeding. 2In our practice, we have routinely employed lumbar plexus blockade (LPB) instead of neuraxial blockade when patients with normal coagulation have postoperative anticoagulation planned. We now report two cases of major, delayed retroperitoneal hemorrhage following LPB in patients with normal coagulation at the time of needle placement, who received anticoagulants postoperatively for thromboprophylaxis.
An active 85-yr-old woman (weight, 66 kg) presented for unicompartmental right knee arthroplasty. Her history included osteoarthritis, depression, and hypothyroidism, and her medications were mirtazapine, levothyroxine, omeprazole, zolpidem, and calcium. Laboratory data included hemoglobin level 14.4 g/dl (normal, 12–16), platelet count 223 × 103(normal, 160–360 × 103), prothrombin time (PT) of 12 sec (normal, 10.8–13.9 sec), International Normalized Ratio (INR) of 0.94, and activated partial thromboplastin time (aPTT) of 24.5 sec (normal, <31 sec).
Sciatic and continuous LPB via the posterior approach were performed with a peripheral nerve stimulator (PNS) in the preoperative holding area. A 90-mm, pencil-point, 21-gauge needle/catheter assembly (Polymedic®, Temena SARL, Bondy, France) was used for the LPB. After contact with the transverse process (presumed L4) at the level of the iliac crests on the first pass, the needle was redirected caudad and a quadriceps twitch was elicited at 0.4 mA (0.1 ms pulse width). After negative aspiration, 25 ml of 0.5% ropivacaine with 1/300,000 epinephrine and clonidine 50 μg was incrementally injected. The 18-gauge over-the-needle catheter was left in place, and a 20-gauge stimulating catheter threaded easily through it. A quadriceps twitch was again elicited at 2 mA and was maintained as the catheter was threaded 9–10 cm. A single-shot sciatic block was then performed via the Labat approach, and 25 ml of the same local anesthetic injected incrementally after eliciting a sciatic motor response. The lumbar plexus catheter was then fitted with a Tuohy-Borst adaptor, at which point blood could be steadily aspirated. The catheter was withdrawn 2–3 cm, flushed with saline, and after negative aspiration a test dose of 3 ml of 1% lidocaine with 1/200,000 epinephrine was negative. The patient was unable to lift her leg or feel pinprick on the thigh and calf after 45 min, but she still had sensation in the anteromedial leg and some dorsiflexion of her foot. A supplemental sciatic block at the midpoint of the line between ischial tuberosity and greater trochanter, and femoral block at the groin were performed with PNS and 15 ml and 7 ml, respectively of ropivacaine 0.5% with 1/300,000 epinephrine. The 1.5 h surgery then proceeded uneventfully with propofol 1%, ketamine 0.1% infusion for sedation. The hemoglobin level in the postanesthesia care unit (PACU) was 14.1. The lumbar plexus infusion was initiated in the PACU with 8 ml/h of 0.2% ropivacaine/0.0002% clonidine. The patient had excellent postoperative analgesia and easily participated in physical therapy.
The lumbar plexus infusion was stopped the evening of postoperative day (POD) 1 in anticipation of discharge on POD2. The next morning (POD 2), the hemoglobin level was 12.4, platelet count 157 × 103, and the patient was comfortable. She was given her first dose of enoxaparin 30 mg subcutaneously at 0900, and the lumbar plexus catheter was removed intact at 1040. Four hours later, physical therapy was prevented by new, significant back pain (9/10 on VAS scale). Morphine was given, and enoxaparin was continued at 30 mg subcutaneously every 12 h. On POD 3 she was again evaluated by the acute pain management service (APS) that noted stable vital signs and right paravertebral pain, no knee pain, and no neurologic deficit. The former catheter site was unremarkable and the hemoglobin level was 9.1. The APS attending was concerned with the possibility of paravertebral bleeding and recommended computerized tomography (CT) scan if her pain persisted. Ten hours later, her pain had decreased by 50%, vital signs were stable, and hemoglobin level was 9.4 with no neurologic deficit. The next day, her pain had diminished, she was walking without assistance, and discharge was considered until the hemoglobin level was measured at 7.1, (repeat was 6.5). Her PT, aPTT, platelet count, and fibrinogen were normal, and vital signs and physical exam were unchanged. The enoxaparin was stopped, she was transfused with two units of packed red blood cells (PRBCs), and a CT scan demonstrated an extensive retroperitoneal hematoma that extended from the retrohepatic space to the pelvis and displaced the kidney anteriorly (fig. 1). She later developed hypotension, oliguria and her hemoglobin level was 8 the following morning. Her vital signs normalized and her hemoglobin level increased to 8.9 with two more units of PRBCs.
Her postoperative course was protracted with transient elevation of creatinine to 3.3, ileus requiring NG drainage, pulmonary edema, and atrial fibrillation for 24 h. She had no laboratory evidence for myocardial infarction. She never developed a neurologic deficit and was able to ambulate and participate in physical therapy, but complained of continued pain in the flank and hip and eventually developed extensive ecchymosis by POD 5 (fig. 2). She was discharged on POD 20.
A 65-yr-old man (weight, 82 kg) presented for arthroscopy of the left knee because of chronic pain from a meniscal tear and presumed anterior cruciate ligament injury. His medical history included a coronary artery bypass graft with mechanical aortic valve replacement 5 yr prior to admission (PTA) requiring chronic anticoagulation, angioplasty of two coronary grafts 1 yr PTA, and a negative dobutamine stress echocardiographic study 4 months PTA. His history also included a gunshot wound to the thoracic spine 30 yr prior with Brown-Sequard lesion: left leg numbness and right leg weakness. He had a cerebrovascular accident 15 yr PTA without residual deficit, and an innominate artery aneurysm. Medications included coumadin 5 mg/day, aspirin 81 mg twice a day, metoprolol, furosemide, ranitidine, and amitriptyline. The physical exam was notable for a systolic click/murmur, and right leg weakness. Admission laboratory data revealed PT 29.1 sec (INR 5.19), aPTT 37.5 sec, hemoglobin level of 12.8 g/dl, platelet count 204 × 103.
The cardiologist admitted the patient, coumadin was stopped, and heparin started in preparation for surgery. When his INR was measured at 9.27 twelve hours later, he was given 1 unit of FFP and 10 mg Vitamin K. One day before surgery his PT had corrected to 12.9 (INR 1.08), and the morning of surgery the PT was 11.9 sec (INR 0.92). His heparin was discontinued at 0100 on the day of surgery, and the aPTT was 23.9 sec at 0700. We felt that peripheral nerve block would provide prolonged postoperative analgesia with minimal residual sedation in a patient at risk for stroke. Single-injection LPB (posterior approach) and sciatic blocks were performed in the preoperative holding area with a PNS and a 21 g, 90-mm insulated needle (Polymedic®, Temena SARL, Bondy, France). A quadriceps twitch was readily elicited at 0.4 mA, and no blood or paresthesia was noted. Incremental injection of 25 ml of 0.3% ropivacaine/0.8% mepivacaine with 1:300,000 epinephrine was used for the LPB, and 25 ml for the sciatic block after stimulation at 0.35 mA. The patient received an additional 1 mg midazolam, 50 mcg fentanyl, and propofol at 30 mcg/kg/min for sedation during the 45 min surgery.
His immediate postoperative course was complicated and bizarre. In the PACU, he complained of long-term memory loss and chest pain radiating to his left arm. He was evaluated by the cardiologist and had emergency echocardiography that demonstrated no new wall-motion abnormalities. Because reheparinization was planned and he exhibited mental status changes, he had an emergent CT scan that showed no intracranial bleeding. He was admitted to the ICU for observation and myocardial enzymes were normal. His PACU hemoglobin level was 13.7.
His heparin infusion was reinitiated at 1200 U/h 8 h after the LPB; his aPTT on POD 1 was 77.7 and hemoglobin level was 13.4. Coumadin was restarted the evening of POD 1. Consultants from Internal Medicine and Neurology continued to assess the patient for his long-term amnesia (short-term memory was preserved), headache, chest, and arm pain, and left leg pain and numbness. His only new objective finding was left arm weakness; results from the carotid doppler and repeat contrast head CT were negative. Psychiatric evaluation suggested his symptoms were consistent with a conversion reaction.
On POD 3 the hemoglobin level was 12.6, aPTT > 100, and INR 1.4. For the first time, he complained of back pain at the site of the LPB. The heparin infusion was adjusted, the hemoglobin level was 10.5 the following morning (POD 4), and the aPTT was 60.2 sec. His vital signs remained stable. Because of the continued flank pain and the decrease in hemoglobin level, an abdominal CT was performed which showed a “moderate-sized retroperitoneal hematoma that appears to originate in the left psoas muscle” (fig. 3). The hemoglobin level decreased further to 8.8 on POD 5, all anticoagulation was discontinued, and the patient received vitamin K 1 mg subcutaneously, and 2 units of PRBCs. The patient continued to complain of chest and arm pain daily without evidence of ischemia, but his other pain complaints resolved. He was discharged on POD 10 with a plan to restart anticoagulation 2 weeks after discharge. Because of the bleeding complication, the cardiologist felt a delay in restarting anticoagulation was necessary in spite of the embolic risk.
Major bleeding complications reported following regional anesthesia and anticoagulation have almost exclusively focused on spinal and epidural anesthesia, since even modest bleeding in the noncompliant spinal canal may cause neurologic compromise. This is particularly apparent following the 1993 U.S. Food and Drug Administration approval of enoxaparin, which has been associated with 80 cases of epidural hematoma reported to the FDA. Because of the association of bleeding and neurologic complications with neuraxial anesthesia and enoxaparin, peripheral nerve blocks such as LPB have emerged as alternatives to central blocks for procedures, such as knee replacement which require intensive analgesia as well as perioperative anticoagulation. 3The American Society of Regional Anesthesia has published Consensus Statements with respect to neuraxial anesthesia and anticoagulation, 4but no such guidelines exist for peripheral nerve or plexus catheters and perioperative anticoagulation. The paucity of data in this area makes it difficult to develop recommendations. In addition, since peripheral nerve blocks are not typically performed in noncompliant tissue spaces, substantial bleeding may occur with less dramatic or no neurologic symptoms compared to epidural bleeding.
A single case report of retroperitoneal hematoma following attempted LPB was published in 1997. 2Unlike our cases, this patient was on enoxaparin at the time of needle placement, and the lumbar plexus could not be located “despite several attempts,” suggesting more needle passes than in our two cases. This previously reported patient developed unilateral leg paralysis, which led to a CT scan and the diagnosis. Significant retroperitoneal hematoma complicating lumbar sympathetic block was recently reported in two patients. 5Again in contrast to our cases, these patients were taking irreversible platelet inhibitors with residual effect at the time of needle placement, and both developed numbness in the thigh as a symptom of the bleeding.
In case 1 of our report, vascular trauma was evident from the initial blood aspiration through the lumbar plexus catheter, but it was felt the catheter had been successfully withdrawn from the vessel. Subsequent plexus stimulation and successful analgesia via the catheter indicated proper location of the catheter adjacent to the plexus. The patient was on no antiplatelet agents, and the enoxaparin was not given until 40 h after the block, but the timing of catheter removal occurred near the probable peak of anticoagulant activity of her first dose of enoxaparin. 6Although she had some decrease in hemoglobin level before catheter removal, the larger hemoglobin level decrease and her back pain suggest the major bleeding occurred after removal. In an earlier review of cases of epidural hematoma following neuraxial anesthesia, 7approximately half of the cases were diagnosed after epidural catheter removal, indicating the possibility of vascular trauma with removal as well as placement of epidural catheters. In truth, it is difficult to time the onset of bleeding with continuous techniques; it is possible that slow, continued bleeding occurred after initial placement of the lumbar plexus catheter, and was only diagnosed after catheter removal. In spite of substantial retroperitoneal bleeding, this patient showed no apparent neurologic deficit.
In case 2 of our study, coagulation was normal at the time of block and there was no apparent vascular trauma. Full anticoagulation was initiated soon after the block, but the hemoglobin level remained relatively stable until POD 4. Given the precise localization of the bleeding on CT scan to the site of the LPB and the time course of pain complaints and the drop in hemoglobin level, it seems most likely that unrecognized vascular trauma occurred from successful needle placement, but that bleeding was exacerbated by excessive anticoagulation on the POD 3. The patient's numerous somatic complaints and preop neurologic abnormalities may have obscured a new deficit from detection, but the neurology consultants diagnosed no new lower extremity deficits. Some consider neurologic disease a relative contraindication to regional anesthesia since postop deficits may be attributed to the anesthetic. 8In addition, a block may obscure or delay the diagnosis of a complication identified by neurologic symptoms. 9In this case as in our first case, however, it was back pain and anemia rather than a neurologic complaint that led to the CT scan and diagnosis.
Spontaneous retroperitoneal hematoma is a known, serious complication in patients on anticoagulants, and the risk increases with increasing intensity of anticoagulation, age greater than 70, and associated antiplatelet therapy. 10The first spontaneous retroperitoneal hematoma with therapeutic (1 mg/kg every 12 h) enoxaparin therapy for deep vein thrombosis was reported in 1999 and was fatal. 11The patient presented with hypotension and anemia.
Retroperitoneal hematoma is also a well known, potentially fatal complication following femoral arterial cannulation for angiography with a reported incidence of 0.12%. 12In a 13-month study of patients undergoing cardiac catheterization, 11 patients developed retroperitoneal hematoma requiring surgical intervention. The most common signs were falling hematocrit and hypotension, and 2 patients died. Femoral nerve palsy was evident in 6 patients. The authors recommended cessation or minimization of anticoagulation, serial CT scanning, and surgical exploration for continued bleeding or significant nerve dysfunction. 13The thromboembolic risk of discontinuing anticoagulation in a patient such as our case 2 is unknown. However, one series of 28 patients with prosthetic heart valves on coumadin who required reversal of anticoagulation when major hemorrhage occurred has been described. Anticoagulation was withheld for a mean of 15 ± 4 days, and no thromboembolic complication occurred. 14
Although isolated complications such as total spinal anesthesia 15and systemic toxicity 16have been reported, LPB has been shown to provide safe and effective surgical anesthesia and analgesia for knee arthroplasty. 17Continuous femoral sheath block is an alternative to posterior LPB but is less successful in achieving surgical anesthesia of the obturator and lateral femoral cutaneous nerve unless the catheter can be advanced to the sacroiliac level. 18
These two cases demonstrate the risk of significant, concealed bleeding from needle placement in an area that cannot be observed when anticoagulation is initiated after nerve block. In addition, the signs of substantial, occult bleeding from LPB in these cases were anemia and back pain without apparent new neurologic deficit. Both patients required blood transfusion and prolonged hospitalization. Since the occurrence of these two cases at our institution, we have changed our practice. We now manage patients having LPB in much the same way as we manage those having neuraxial block when thromboprophylaxis is ordered. If a patient is on prophylactic enoxaparin, we avoid LPB unless the last dose was given at least 12 h before. We do not place lumbar plexus catheters in patients who will be anticoagulated unless the catheter is removed before initiating anticoagulation, and we do not start anticoagulation until 2 h after catheter removal. We also follow patients with heightened vigilance for signs of bleeding if they have deep blocks and are anticoagulated postoperatively.