Clinical aspects

Description

The general concept of epidural anesthesia is to provide local administration of the analgesic agent into the epidural space. This neuraxial form of anesthesia provides local pain control and reduces the need for systemic analgesics or narcotics. Although indicated for a number of reasons, the general approach to this procedure is relatively consistent (82). Epidural anesthesia is usually administered through a catheter for continuous infusion and allows for prolonged anesthesia and titration of the onset of the anesthetic. Single-shot epidural anesthesia is limited to the duration of action of the injected drug.

The level of insertion of the epidural needle depends on which dermatomes need to be blocked; for example, for a lower abdominal procedure, this will be between TH8–Th12, and for a thoracic procedure, this will be between Th3–Th7.

Two primary techniques exist for reaching the epidural space with the epidural needle: the midline and paramedian approach. In the more common midline approach the performing physician first directs their needle to the space directly between the vertebral spinous processes before advancing. In the less common paramedian approach, the needle is inserted approximately 1 centimeter lateral and caudal to the spinous process and directed in a 35-degree craniomedial direction before advancing through the paraspinal tissues. This technique is more frequently used in patients with narrowed intervertebral spaces, as the location and higher angle avoids more of the bony structures of the vertebrae on insertion.

In both techniques a local anesthetic, typically 1% lidocaine, is injected into the skin and subcutaneous tissue to avoid discomfort prior to needle insertion. Following this administration, a needle, most commonly a 17- or 18-gauge Touhy needle (49), is used to penetrate through the skin and ligamentum flavum into the epidural space. Confirmation of placement in the epidural space can be confirmed in several ways, which are discussed in more detail below. Once placement in the epidural space is confirmed a flexible catheter is then inserted through the needle bore and passed approximately 5 to 10 cm into the epidural space. The needle is withdrawn, and the catheter is pulled back to leave approximately 5 to 6 cm in the epidural space, in addition to the length needed to pass through skin, subcutaneous tissue, and ligamentous structures. It is then immobilized so that continuous administration or multiple injections of medication into the epidural space can be performed (56).

Aspiration of the catheter for blood and CSF is attempted to determine if the catheter tip is not intravascular or in the subarachnoid space. Additionally, test doses (small volumes) of an anesthetic and epinephrine are routinely injected to ensure placement in the epidural space as placement in the subarachnoid space will lead to unexpected spinal block, and intravascular placement will lead to tachycardia or hypertension caused by the epinephrine. Typically, 3 ml of a solution of bupivacaine 2.5 mg/ml or lidocaine 10 mg/ml with epinephrine 1:200.000 is used as a test dose. In the perioperative setting, the catheter is ideally placed well in advance of the surgical procedure in an awake patient. This provides ample time to place the catheter and accurately assess the level of sensory analgesia before surgery begins. This is done by bilaterally assessing cold sensation in the desired dermatomes. Accurate positioning of the catheter is only confirmed by bilateral sensory block. Anything other than an effective bilateral block suggests that the catheter may not be correctly positioned, with pleural puncture as one of the possibilities in a thoracic epidural procedure (92).

A variety of anesthetic or analgesic agents can be injected. Typical anesthetics include lidocaine, ropivacaine, bupivacaine, and levobupivacaine. There are numerous dosing regimens that are used in modern practice. Long-acting agents, such as ropivacaine and bupivacaine, are preferred to prevent the block from wearing off too quickly. Furthermore, it should be noted that the administered volume determines the spread of the local anesthetic in the epidural space and, thereby, the number of dermatomes to be blocked, whereas the concentration of the analgesic determines block density and amount of motor block. In addition to local anesthetics, other medications can be epidurally administered, usually as an adjuvant to the local anesthetic. Common analgesic agents injected include fentanyl, sufentanil, morphine, and clonidine.

Historically, there are two main modes of administering epidural medication, namely continuous epidural infusion (CEI) and patient-controlled epidural analgesia (PCEA), optionally with continuous background infusion. In the latter, bolus doses are administered by the patient with a lockout period ensuring that overdosing is not possible. In epidural labor analgesia, this method seems to lead to less local anesthetic consumption and less motor block (86). Programmed intermitted epidural bolus (PIEB), in which programmed bolus doses are administered at scheduled times, is increasingly used in labor analgesia. The underlying idea of this technique is that repeated bolus doses of local anesthetic will lead to a more widely spread sensory block than continuous administration. A meta-analysis suggests that PIEB combined with PCEA reduces the rate of instrumental delivery, incidence of breakthrough pain, and local anesthetic usage when compared to CEI combined with PCEA (95).

Regarding confirmation of needle placement in the epidural space, the most commonly used techniques are through the loss of resistance and the hanging drop method. In the loss of resistance technique, a needle is advanced through the ligamentum flavum, and resistance to injection of air or saline is continuously or frequently checked. When the tip of the needle is within the ligamentum flavum, air or saline cannot be readily injected. Immediately past the ligamentum flavum, there is a loss of resistance and air or saline can be injected; this indicates that the needle tip has entered the epidural space. In addition to localizing the needle tip within the epidural space, injection of air or saline pushes the dura away from the needle tip, thus, reducing the risk of puncturing or entering the subarachnoid space. The medium for this technique varies and some practitioners use air, some use fluid, and others use a combination of air and fluid to assess loss of resistance. There is no consensus on the relative efficacy of these techniques and in a systematic review with meta-analysis of four older studies, Sanford and colleagues found inconclusive evidence in determining whether a difference in analgesia quality results from the use of air or fluid during the loss of resistance technique (71). It should be noted, however, that inadvertent injection of air in the subarachnoid space can lead to pneumocephalus. In the hanging drop method, a drop of saline is attached to the needle opening as the needle is advanced trough the tissues. The negative pressure inside the epidural space will lead to the suction of the drop into the needle tip, notifying the physician that the epidural space is reached.

Although these techniques remain the most commonly used, research suggests alternatives may prove more effective. In a 2017 meta-analysis, Carvalho and associates found moderate strength evidence that other methods were superior to air or saline injection for the purposes of identifying the epidural space (12). These methods included the use of lidocaine as a loss of resistance medium, the use of the Epidrum (a single use device that gives a visual cue when the epidural space is reached), and acoustic devices (that provide a tone when the space is reached) (02), which were all superior to conventional methods in multiple randomized controlled trials. Electrical stimulation, with paresthesias overlying the target areas generated by peripheral nerve stimulation, have also been proposed as a useful technique for inexperienced providers (46). A small randomized controlled trial showed statistically significant differences in identification of the epidural space and maternal satisfaction. Also of note, the use of imaging-guided technique has become an increasing area of interest. Studies have shown that the use of ultrasound in epidural anesthesia can decrease patient discomfort and time to epidural placement in populations with difficulty anatomy, such as obese or elderly patients (60; 84).

In an attempt to improve the efficacy of labor epidural analgesia, dural puncture epidural has been suggested. In this technique, the dura is intentionally punctured with a spinal needle. Whereas loss of resistance can be subjective, this technique provides a clear endpoint, namely CSF returning trough the spinal needle. Furthermore, the dural puncture may allow transfer of local anesthetic into the subarachnoid space and thereby accelerate analgesic onset. Evidence of the efficacy of this technique is conflicting. However, a randomized controlled trial in obese parturients comparing dural puncture epidural with standard epidural showed no meaningful difference regarding the quality of analgesia (79).

Indications

There are three main indications for epidural anesthesia: perioperative pain management, labor analgesia, and chronic or acute pain. In the perioperative setting, epidural anesthesia can be used for regional anesthesia of the lower extremities, abdomen, and thorax during gynecologic, urologic, cardiothoracic, orthopedic, and general surgical procedures. It is often performed in conjunction with general anesthesia, reducing the need for systemic analgesia intraoperatively and allowing for high-quality postoperative pain control. Advances in other regional anesthetic techniques, such as peripheral nerve blocks of the lower extremity, paravertebral block, erector spinae block, transversus abdominal plane (TAP) block, and rectus sheath catheters, as well as the increase in minimally invasive surgical techniques, have led to a narrowing of the indications for epidural anesthesia. Epidural anesthesia remains an important component of perioperative pain management as it provides excellent analgesia in moderate-to-large thoracotomies and (upper) abdominal laparotomies. Single-injection subcostal TAP block was more effective than intravenous opioid analgesia, whereas continuous thoracic epidural analgesia was more effective than the single-injection subcostal TAP block (94). Other potential benefits of epidural anesthesia are discussed in the Outcomes section of this article. The role of epidural anesthesia in minimally invasive surgery is controversial because its superiority in pain control when compared to intravenous analgesics or other blocks has not been shown in the literature. In general, it is not widely accepted to place an epidural catheter for minimally invasive surgery (54). The decision to place an epidural catheter in the perioperative patient should be made on a case-by-case basis and should include type of surgery and incision as well as patient comorbidity.

Analgesia during labor is the single most common indication for epidural analgesia. It is the gold standard for labor analgesia, and the WHO estimates that epidural analgesia is used in 10% to 64% of births in high-income countries. A lumbar epidural catheter is placed at the L2–3 or L3–4 level in order to achieve a sensory block reaching up to the T10 dermatome while motor function remains intact. This provides relief of visceral pain caused by the contracting uterus and dilating cervix during the first stage of labor. In the second stage of labor, pain is also caused by stretching of the perineum and vagina by the fetal head. Here, the sacral roots are involved, and analgesia may not be as effective as during the first stage of labor.

When compared to systemic opioids, such as remifentanil, epidural analgesia during labor provides superior pain relief and higher maternal satisfaction. On the other hand, it should be noted that there is an increased incidence of hypotension, fever, urinary retention, and longer duration of labor and possibly a higher rate of assisted vaginal delivery (35).

Lumbar epidural analgesia has also been used in patients with severe chronic pain in the lumbosacral segments, such as from cancer or complex regional pain syndrome.

Finally, a thoracic epidural is a good option for acute pain management in patients with multiple rib fractures.

Contraindications

Contraindications to epidural anesthesia are divided between relative and absolute criteria. These are discussed separately below.

Absolute

Given the numerous advances in technique and equipment since the advent of epidural anesthesia, the only current absolute contraindications to epidural anesthesia include patient refusal, elevated intracranial pressure, and associated risk of herniation, or the presence of severe and refractory coagulation abnormalities, ie, disseminated intravascular coagulopathy. All other contraindications to this procedure are relative and the use of this technique in these situations must take into account individual patient characteristics, provider expertise, and risk-benefit analysis.

Relative

Thromboprophylaxis. The risk of catastrophic epidural hematoma is an important consideration in patients requiring epidural anesthesia while on anticoagulant or antithrombotic agents. The American Society of Regional Anesthesia and Pain Medicine routinely reviews and updates consensus guidelines on the use of neuraxial anesthesia in these patients. Current recommendations suggest that aspirin and NSAID use are not contraindications to epidural anesthesia and low-risk chronic pain procedures. For medium- and high-risk chronic pain procedures, these medications should be held for five half-lives prior to procedure when possible and can be resumed 24 hours postprocedure. P2Y12 inhibitors such as clopidogrel must be held for 5 to 7 days prior to procedure and may resume 12 to 24 hours postoperatively. The guidelines for holding and resuming anticoagulants varies based on their mechanism of action. Typical factor Xa inhibitors should be held for 3 days and may be resumed at 24 hours postprocedure. Coumadin is typically held for 5 days or until the INR value has normalized and can be resumed 6 hours postprocedure; however, it is important to know that catheter removal should not occur until the INR has again reached less than 1.4, as removal also imposes a risk of traumatic epidural hematoma. Intravenous heparin can be held 6 hours preprocedure and resumed as soon as 2 hours later. Finally, subcutaneous heparin and low molecular weight heparin formulations in therapeutic dosages can generally be held for 24 hours prior to procedure and resumed 4 to 24 hours postprocedure depending on the risk of the procedure (39).

Sepsis. Historically, the use of epidural anesthesia has been avoided in septic or febrile patients due to concerns regarding hypotension, autonomic instability, the development of coagulopathy, and potential for seeding the epidural space with pathogenic organisms. Multiple animal models have suggested that epidural anesthesia may reduce splanchnic hypoperfusion, end organ damage, and inflammation (20; 27). Unfortunately, this finding has been contraindicated in other studies and a current consensus on the benefit in sepsis does not exist. However, epidural anesthesia may be safely used in patients who have already demonstrated a response to antibiotics (91; 57).

Preexisting neurologic conditions. Preexisting neurologic conditions are typically divided between structural abnormalities and disorders of neural transmission in the consideration of epidural anesthesia. The former, as seen in multiple sclerosis and myasthenia gravis, have been found to be relatively safe for the use of neuraxial anesthesia. These conditions are discussed in further depth in the “special considerations” section. Structural abnormalities, such as spinal stenosis, prior spine surgery, and spina bifida, can result in altered spread of injected agents and difficulty with needle/catheter placement. These conditions require significant risk-benefit assessment based on individual patient factors and physician expertise, but typically, neuraxial block can be safely used in this population (37).

Thrombocytopenia/coagulopathy. The presence of thrombocytopenia or coagulopathy is an important consideration in the use of epidural anesthesia. Currently, no consensus guidelines exist on an appropriate platelet count for the use of epidural anesthesia; however, in clinical practice 70,000 is frequently used. In certain populations such as ITP and gestational thrombocytopenia, functional platelets may exist despite lower counts. A careful consideration of the etiology, platelet trend, and bleeding history should be undertaken prior to epidural anesthesia in this patient population (48). Chi and colleagues concluded that it is possible to offer regional block to women with inherited bleeding disorders provided their coagulation defects have normalized, either spontaneously during pregnancy or following adequate hemostatic cover (15).

Preload dependent state. The use of epidural anesthesia results in reduced systemic vascular resistance, which may produce hypoperfusion and decreased cardiac output in preload-dependent patients. This is commonly seen in patients with aortic stenosis, hypertrophic cardiomyopathy, or hypovolemia. Furthermore, systemic hypotension as a result of a reduced systemic vascular resistance can cause insufficient coronary perfusion in patients with hypertrophic cardiomyopathy, possibly resulting in myocardial ischemia. The severity of illness must be assessed on an individual patient basis.

Outcomes

The length and depth of local anesthetic or analgesic effects can be carefully controlled by the volume, concentration, and frequency of injection as well as potency of the medication. Epidural anesthesia has been demonstrated to provide better postoperative analgesia than parenteral opioids, promote postsurgical recovery, decrease the incidence of postoperative pneumonia, and decrease overall morbidity and mortality when used in conjunction with general anesthesia (08; 64; 65). In abdominal surgery, epidural anesthesia accelerates the return of gastrointestinal transit. There is no evidence for an increased incidence of gastrointestinal anastomotic leak (34). This is reassuring as there are concerns that epidural hypotension and concomitant vasopressor use or fluid administration could compromise anastomotic patency.

In a 2019 meta-analysis, the use of neuraxial blockade in caesarean section was associated with statistically significant improvements in Apgar score and umbilical pH over general anesthesia (45).

In the geriatric population, the use of combined epidural and general anesthesia has previously been shown to be beneficial. A randomized controlled trial of over 1800 participants demonstrated an approximately 33% reduction in the incidence of postoperative delirium in patients receiving this care following major noncardiac surgery (1.8% versus 5%; relative risk 0.351; 95% CI 0.197 to 0.627; P < 0.001) (50). This consideration needs to be weighed against a statistically significant increase in intraoperative hypotension.

Adverse effects

Nonneurologic complications

Nonneurologic complications include inadvertent subarachnoid injection with possible high spinal block and cardiopulmonary arrest, respiratory depression, intravascular injection, hypotension, nausea and vomiting, urinary retention, infection, broken catheter, and pruritus (56; 61).

Frequency and mechanisms of neurologic complications

Frequency. There are many different, but fortunately infrequent, neurologic complications associated with epidural anesthesia. Most studies of the frequency of neurologic complications have been reported by anesthesiologists. The frequency of neurologic deficits reported by anesthesiologists ranges from 0 to 0.10% (85; 42; 73; 80; 19; 05). For example, in the largest series reported, there were 72 patients with complications out of 780,000 procedures, for a complication rate of 0.0092% (1 per 11,000) (85). However, the older studies likely reported only the more severe complications and may have ignored the benign complications of mild lumbar radiculopathy. In a more recent publication, Hosslin and colleagues conducted a single-center retrospective study of complications after thoracic epidural anesthesia and found transient neurologic symptoms to occur in 0.27% of procedures; furthermore, it was found that serious adverse events occurred, with an incidence of 0.01% (90). Two studies by anesthesiologists revealed similar complication frequencies of 1 per 1000 procedures (19) and 1 per 1600 procedures (05). It is also important to note that the majority of neurologic complications appear to be reversible, with 61% to 75% of patients making a complete recovery (17).

Unintended subarachnoid injection. Many neurologic complications result from unintended subarachnoid injection of anesthetic or analgesic agents. Epidural anesthesia differs from spinal anesthesia (intended subarachnoid administration of anesthetics) in that higher concentrations and volumes, and repeated or continuous injections of medications, are administered. When unintended subarachnoid administration of medications occurs during intended epidural anesthesia, the higher doses of medications, the larger needle used, and the insertion of a catheter all increase the risks of neurologic complications. Inadvertent subarachnoid injection may have occurred because of migration of the catheter tip through the dura, migration from the subdural space to the subarachnoid space, or inability to aspirate CSF despite being present in the subarachnoid space (14; 98). The frequency of accidental dural puncture is estimated to be about 0.6% (80). Monsel and colleagues reported a case of accidental epidural administration of distilled water (4 ml) during labor epidural analgesia in a 32-year-old woman, which was associated with severe and instantaneous thoracolumbar pain that was relieved after epidural administration of sodium chloride (NaCl) 0.9% 5 mL, followed by 20 ml of an analgesic mixture (ropivacaine 0.1% + sufentanil 0.25 microg/ml) (55). There was no residual pain or any neurologic deficit until the time of discharge 10 days later. Distilled water had been previously used as an old method to locate the epidural space, though few experimental data suggest that it can be neurotoxic.

Mechanisms of injury. Neurologic complications of lumbar epidural anesthesia and analgesia can be categorized into several mechanisms of action: (1) direct chemical toxicity from the anesthetic or analgesic agent or contaminants; (2) mechanical trauma to neural structures from the needle, catheter, or injectate; (3) bleeding complications; (4) delayed injury to neural structures from processes such as arachnoiditis or Guillain-Barre syndrome; (5) ischemic injury as a result of hypotension or vasospasm; and (6) bacterial infection. Specific examples (Table 1) of these complications are discussed below.

Table 1. Types of Neurologic Complications

In thoracic epidural anesthesia, if the epidural catheter is misplaced, forceful insertion may lead to disastrous complications (01; 31). Sawhney and colleagues described the development of a sterile inflammatory granuloma after perioperative epidural catheter placement, an exceptionally rare occurrence, but a potential complication of epidural catheterization (72).

Radiculopathy

The most common complication is lumbosacral radiculopathy or polyradiculopathy (85; 73; 19; 98). These complications must be differentiated from focal neuropathy or plexopathy due to surgical positioning, childbirth, lithotomy position, or other causes unrelated to epidural anesthesia. Many of the earlier studies of neurologic complications reported by anesthesiologists lacked neurologic detail to confirm localization of the lesion. A series of 12 patients with detailed neurologic examinations was described (98). These 12 patients represented all neurologic consultations at an academic institution over a 6.5-year period who suffered complications of lumbar epidural anesthesia. Of the 12 patients, 11 developed lumbosacral radiculopathy or polyradiculopathy. Patients had motor, sensory, and reflex deficits in typical radicular or polyradicular patterns. MRI of the lumbosacral spine, when performed, did not reveal hematomas, abscesses, or mass lesions. EMG evaluation, when performed, confirmed radiculopathies in three patients. Ten of the 11 patients had epidural anesthesia, whereas one received subarachnoid injection of anesthetic after intended epidural anesthesia.

Nine of the 11 patients with radiculopathies experienced mild to moderately severe neurologic deficits, most often from injury to the L2 root near the lumbar epidural injection (98). All of the patients improved significantly. The mechanism of injury in three patients was likely direct injury of the nerve root by needle puncture, catheter insertion, or interfascicular injection. These mechanisms of injury have been described by others (85; 19). The mechanisms of injury in the other patients were unknown and may have been from direct toxic effects of the anesthetic or analgesic agent on the nerve root, mechanical injury from the needle or catheter, or both.

Anandaswamy and colleagues reported a rare case of transient brachial monoparesis following a standard epidural anesthesia (15 ml of 2% lidocaine with 50 mcg fentanyl given epidurally) to achieve a blockade up to T6 level in a parturient for cesarean section (03). Although the exact mechanism remains unclear, unilateral migration of the catheter to a higher vertebral foramen (97), septate epidural space, and large volumes greater than 40 ml at the lumbar or caudal regions (10) were some known causative factors.

Cauda equina syndrome

Severe polyradiculopathies have been associated with spinal stenosis. In a study, two elderly patients with cauda equina syndrome after epidural anesthesia had severe lumbar spinal stenosis diagnosed on MRI (98). One patient had severe motor axonal loss demonstrated on EMG and nerve conduction studies; she had little recovery from severe, bilateral L2-S2 polyradiculopathies. The other patient had moderately severe, bilateral L2-S5 polyradiculopathies that improved significantly over a few weeks. Other patients with lumbar spinal stenosis have developed cauda equina syndrome after epidural anesthesia (13; 29). Cauda equina syndrome may have developed in these patients because of injection of fluid into a space of limited volume or the development of edema leading to compression of the spinal roots. Alternatively, chronic spinal stenosis may have resulted in increased susceptibility of the nerve roots to the toxic effects of anesthetics because of breakdown of the blood-nerve barrier at that level. Regardless of the cause, spinal stenosis is likely a significant risk factor for the development of cauda equina syndrome. Because spinal stenosis may be asymptomatic, it is difficult to exclude all patients with spinal stenosis from receiving epidural anesthesia.

Inadvertent subarachnoid injection has been reported to cause cauda equina syndrome (14) that probably resulted from toxic effects of the medication rather than mechanical injury (given the multiple roots injured) or structural lesions (MRIs were normal).

Isolated cases of cauda equina syndrome have been reported to result from accidental injection of saline with benzyl alcohol preservative into epidural space, from mass effect of epidural air or the combination of epidural air and lithotripsy, or from unclear mechanisms.

Prognosis for cauda equina syndrome depends on initial severity of neurologic deficits and the degree of axonal loss; EMG evaluation is helpful in quantifying the amount of motor axonal loss. Greater initial severity and greater axonal loss are associated with poor prognosis (98).

Myelopathy

Myelopathy represents one of the most dreaded and severe complications of epidural anesthesia. There are a variety of clinical presentations and etiologies, but all have in common the hallmarks of a thoracic or lumbar myelopathy, namely paraparesis, sensory level, urinary and fecal incontinence, and loss of sexual function. Prognosis varies from near complete recovery to severe residual neurologic deficits depending on the initial severity of the lesion and reversibility of the etiology.

Spinal cord infarction associated with epidural anesthesia is one of the causes of acute myelopathy. Patients with paraparesis, loss of pin and temperature sensation with preservation of vibratory and proprioceptive sensation, and urinary and fecal incontinence have been found to have spinal cord infarct in the anterior spinal artery distribution (21; 85; 51). Other patients may have spinal cord infarction beyond the anterior spinal artery distribution (85). The pathophysiology of spinal artery occlusion or hypoperfusion is uncertain. Contributing factors may include preexisting atherosclerosis or vasculitis, anesthesia-induced hypotension, arterial vasospasm from epinephrine, or arterial compression from large volumes of injectate in the epidural space. Rarely, cord ischemia has occurred in patients with dural arteriovenous fistula (75; 29).

Acute myelopathy can occur following complete spinal block (85; 98). Inadvertent subarachnoid injection of anesthetic agent during intended epidural anesthesia can lead to much larger (up to 10 times) the usual dose of anesthetic used for spinal anesthesia. Subdural injection during intended epidural anesthesia is well documented (68). Patients under general anesthesia are at particular risk for unintended subarachnoid injection because they cannot be monitored for progressive spinal block. After recovering from complete spinal block, usually within a day, patients awaken with a myelopathy. The likely mechanism of myelopathy is direct toxicity from high doses of anesthetic agent within the subarachnoid space.

Spinal cord compression from a hematoma can present either acutely or subacutely. Epidural and subdural hematomas have occurred most commonly in patients who are given warfarin, heparin, or aspirin, or in those with a coagulopathy (85; 19). However, hematomas have also occurred in patients without any obvious bleeding (38). Ehrenfeld and colleagues found an overall incidence (per 10,000 epidural blocks) of epidural hematoma of 1.38 (95% confidence interval, 0-0.002) (25). The mechanism of bleeding is presumably from puncture of venous blood vessels in the epidural space or dura. Arterial bleeding is unlikely given the paucity of and lateral location of arteries within the epidural space (85). Rarely, hemorrhage into spinal tumors after needle puncture has been reported (85). Epidural or subdural spinal hematoma is a medical emergency that requires immediate MRI or CT and appropriate medical (eg, steroids) and surgical intervention. Despite reports of hematomas associated with anticoagulation and antiplatelet therapy, studies suggest that the frequencies of bleeding complications from epidural anesthesia are low in patients receiving low-molecular weight heparin, preoperative antiplatelet therapy, or anticoagulation following placement of the catheter (67).

Epidural abscess constitutes another cause of myelopathy associated with epidural anesthesia (85; 58; 76). Fortunately, with improved aseptic techniques, infectious complications from epidural anesthesia are relatively rare (85). Patients present acutely or subacutely with fever, back pain, local tenderness, and neurologic symptoms and signs of myelopathy or cauda equina syndrome. The mechanism of infection may be from external contamination or hematogenous source. An epidural abscess is a medical emergency, and its suspicion requires immediate MRI, antibiotics, and possibly medical and surgical intervention for cord compression.

Chronically progressive myelopathy can occur from arachnoiditis associated with epidural anesthesia (75; 29). Proposed mechanisms include dural puncture and leaking of epidural anesthetics into the subarachnoid space, contamination of the injectate with detergent, or meningeal reaction from the anesthetic agent or epinephrine (75).

Progressive epidural fibrosis leading to spinal cord compression has been reported in patients receiving long-term epidural morphine (24). Epidural fibrosis may result from a chronic reaction to the catheter or morphine. In two patients, neurologic deficits resolved with removal of the catheter (24). Kalil describes a case of near-complete left hemiparesis that developed following a routine continuous epidural anesthetic for labor resulting from unintended subdural deposit of the local anesthetic (41).

Medication errors are an uncommon but potentially catastrophic cause of morbidity after epidural anesthesia. A summary of case reports of 25 patients receiving inadvertent epidural potassium chloride, a potentially neurotoxic agent, showed that paraplegia was common. The majority of patients recovered spontaneously, but permanent neurologic deficit, and even mortality, were also described. The authors noted that management of inadvertent neuraxial potassium chloride was largely symptomatic. Some authors suggest diluting the agent with saline; there is, however, no evidence for this practice, and one should consider that flushing the epidural space might lead to a further segmental spread of potassium chloride and severity of myelopathy (63). Pysyk and colleagues described the accidental administration of tranexamic acid into the epidural space; it was managed by slowly flushing the epidural space with saline (66). Neurologic evaluations for 48 hours and follow-up at 3 months revealed no motor weakness or seizure-like activity.

Coma

Unintended subarachnoid injection may result in high spinal block, coma, or cardiopulmonary arrest. Patients are frequently undergoing simultaneous epidural and general anesthesia and cannot be monitored for high spinal block. Patients with complete spinal block usually regain consciousness within a few hours, but this may take as long as a day (85; 98).

Headache

A variety of intracranial abnormalities are associated with epidural anesthesia. Postdural puncture headache is an occasional complication (85; 80). In a study, the frequency of headache following inadvertent dural puncture was 0.6% of epidural anesthesia procedures, and, of those, 16% developed headache (80). Patients with postdural puncture headache develop symptoms of a low-pressure postural headache similar to patients with a postlumbar puncture headache. Treatment includes hydration, caffeine, or epidural blood patch. In the study, symptoms in 17 patients subsided in 3 to 19 days after mostly conservative treatment (80).

Rarely, the process of injecting air after entering the epidural space can be complicated by inadvertent injection of air into the subarachnoid space, leading to pneumocephalus and headache (32; 93).

Intracranial hypotension

Besides the more common complication of headache following inadvertent dural puncture, other complications of intracranial hypotension have been reported. Several patients have developed intracranial subdural hematomas from low CSF pressure (69). One patient developed headache and seizures following dural puncture (88). Sixth cranial nerve palsies have been reported to develop following dural puncture, presumably from intracranial hypotension, shifts in the brain, and subsequent stretching of the nerve (23). Prognoses for these disorders are generally good. Yatziv and associates describe a patient with acute comitant esotropia that occurred 1 week after epidural anesthesia for a normal vaginal delivery as a result of unintended dural puncture (96). Magnetic resonance imaging revealed diffuse pachymeningeal enhancement, typically seen after dural puncture. Resolution was spontaneous. Cranial nerve palsy as a complication of epidural anesthesia is rare. When it occurs, it is often due to unintended dural puncture, often unilateral, and because of its long intracranial course, sixth nerve palsy is most common. It can occur in as early as a day after epidural anesthesia or might take as long as 3 weeks. Spontaneous recovery is the rule.

Seizure

Inadvertent intravascular injection of anesthetics during attempted epidural anesthesia is a known complication. Excessive levels of local anesthetic in plasma can lead to a syndrome called local anesthetic systemic toxicity (LAST). This is a life-threatening syndrome characterized by several phases. Mild intoxications may manifest as sensations, such as tingling around the mouth or the patient experiencing a metallic taste. More severe intoxication can lead to seizures and eventual cardiovascular collapse. Seizures should be treated according to protocol, but the syndrome is casually treated with intralipid therapy. Injection of high doses of anesthetic in epidural procedures has been reported and can cause seizures (44; 05). In one case, the seizure responded rapidly to diazepam.

Bacterial meningitis

Bacterial meningitis is a rare complication of epidural anesthesia (85). Sources of infection include external contamination and hematogenous spread. Treatment includes antibiotics and antifungal therapy, and prognosis depends on the severity of the meningitis before treatment.

Horner syndrome

Horner syndrome is a frequent and relatively benign complication (18; 06). The proposed mechanism of injury is from direct neurotoxic effects from rostral spread of the anesthetic within the epidural space. Most patients recover without incident.

Trigeminal neuropathy

Trigeminal neuropathy is an uncommon and relatively benign complication (77). The mechanism of injury may be from rostral spread of the anesthetic within the epidural space and direct neurotoxic effects of the anesthetic. The prognosis for recovery is usually good.

Guillain-Barre syndrome

Guillain-Barre syndrome is a rare complication that may appear in patients 1 to 2 weeks after epidural anesthesia (78). Patients may be at risk for Guillain-Barre syndrome because of interaction between the anesthetic and myelin or nerve trauma from the needle or catheter leading to immunological processes that result in the neuropathy (78). Of the four reported patients, all had complete or near-complete recovery in 1 to 12 months.

Postural hypotension

One of the well-known complications of epidural anesthesia includes postural hypotension, which can delay mobilization in the postoperative period. Gramigni and colleagues revealed that hemodynamic assessment does not predict inability to walk after thoracic and abdominal surgery, and that early mobilization should be tried irrespective of blood pressure or orthostatic changes in postoperative patients with epidural analgesia (33).

Special considerations

Although typically performed by anesthesiologists, there are a number of patient populations who undergo epidural anesthesia and have specific considerations that the practicing neurologist should be aware of. These groups are outlined below.

Pregnant patients. In an academic medical center, approximately one half of the epidural anesthesia procedures were for patients undergoing labor and delivery, and the frequency of neurologic complications was about the same in pregnant women compared to nonpregnant patients (98). This finding was confirmed by other studies (53). Most of the different types of neurologic complications have been reported in pregnant as well as nonpregnant patients. However, the neurologic complications were less severe and long-lasting in the pregnant group. The more severe neurologic complications in the nonpregnant group may be due to a number of factors. First, the nonpregnant group includes many older patients. Elderly patients are more likely to have preexisting neurologic disorders that may make them more susceptible to severe complications. Second, concurrent general anesthesia, which is more common in nonpregnant patients, may mask the early symptoms of neurologic complications, thus, increasing the risk for more severe injuries. Third, many nonpregnant patients have long intraoperative procedures and postoperative epidural analgesia, thus, resulting in much longer durations of epidural anesthesia and analgesia.

Myasthenia gravis. Patients with myasthenia gravis have historically represented a significant challenge to achieving appropriate anesthesia and analgesia during surgical procedures. Due to the lack of fully functional acetylene choline receptors these patients typically have abnormal reactions to neuromuscular blocking agents. They may have increased respiratory complications and prolonged need for mechanical ventilation. The stress of surgery can induce myasthenic crisis (07). Additionally, neuromuscular blockade may be difficult to reverse if the patient is on an acetylcholinesterase inhibitor. However, this concern may be alleviated by the increased use of sugammadex, a reversal agent for the neuromuscular blockade induced by rocuronium and vecuronium, which appears to prevent residual blockade in myasthenic patients (22).

The use of epidural anesthesia or combined anesthesia in this population is widely accepted and has been shown to decrease the need for mechanical ventilation in patients undergoing elective thymectomy (16), decrease the postoperative time to extubation (52), and decrease postoperative opioid consumption. Finally, pregnancy and labor are an important consideration in this population. Although many patients with optimal disease control will have no additional issues in the peripartum period, there is an increased incidence of caesarean section in myasthenic patients. This is thought to be due to maternal fatigue during labor, which can be ameliorated by lumbar epidural anesthesia (83; 26).

Multiple sclerosis. Historically, multiple sclerosis was considered a relative contraindication to epidural anesthesia due to a theoretical risk of local anesthetic induced neurotoxicity and a concern for inducing a multiple sclerosis relapse. Given the prevalence of women of childbearing age in the multiple sclerosis population, this was an area of particular interest. Based on available data, earlier fears appear to have been unfounded. Multiple retrospective and cohort studies have found no correlation between the use of epidural anesthesia and multiple sclerosis relapse (09; 40; 36).

Acute inflammatory demyelinating polyneuropathy. Acute inflammatory demyelinating polyneuropathy, formerly known as Guillain Barre syndrome, is a rare immune-mediated neurologic condition. Despite the relative scarcity of this condition, more than 50 cases of acute inflammatory demyelinating polyneuropathy in pregnancy have been described in the literature and acute inflammatory demyelinating polyneuropathy may be a consideration for other surgical procedures as well (87). The rarity of this condition has impeded the development of randomized controlled trials to assess epidural anesthesia in this population and no consensus guidelines currently exist. Despite this, multiple authors have described the successful use of epidural anesthesia during labor, and it has also been utilized in transurethral prostate resection (99). The use of epidural anesthesia in this population is at the discretion of the practicing provider but appears to be relatively well tolerated based on available literature.

Clinical vignette

A 16-year-old patient with an unremarkable medical history underwent a bilateral derotational osteotomy of the femur under general anesthesia and received a lumbar epidural for postoperative pain management. The patient exhibited no previous neurologic abnormalities. Insertion of the epidural catheter was performed under general anesthesia in the lateral decubitus position. Placement at the L3–4 level was uneventful. The catheter’s correct placement was confirmed by negative aspiration and administration of a test dose of bupivacaine (2.5 mg/ml) with epinephrine (1:200.000), which elicited neither sensory blockade nor tachycardia. A loading dose of 6 ml of bupivacaine 2.5 mg/ml was administered, and, subsequently, continuous infusion of this medication at a rate of 8 ml/hr was initiated. The surgical procedure progressed without complications, and no supplementary analgesics were required during the operation.

Postoperatively, the patient was transferred to the recovery room, where the full anesthesia wore off. The epidural medication was switched to ropivacaine 2 mg/ml. On cooperation, the patient exhibited effective pain control. Motor function of the lower extremities was normal, except for weakness in the right foot flexors. After a regular recovery period, the patient was discharged to the pediatric ward. That same night, about 12 hours later, the patient was readmitted to the recovery room because of increasing pain in the left leg (Numeric Rating Scale > 7) and motor block of the right leg (Bromage 3). Systemic opioids where initiated, and epidural medication was discontinued. Partial resolution of the motor blockade occurred, but the patient still exhibited an inability to flex both knees 8 hours later. There were no signs of cauda equina syndrome; pain was well controlled. A neurologist assessed the patient.

In this case, the differential diagnosis encompassed the possibility of an epidural hematoma, which prompted the performance of an MRI of the lumbar spine and revealed no abnormalities. An expectant approach was pursued, and function of the knee flexors gradually returned over a period of 2 weeks.

Scientific basis

Injection of a local anesthetic into the lumbar epidural space results in diffusion of the drug in the rostral and caudal directions. Direct contact of the drug with the spinal roots as they cross the epidural space and possibly the dorsal root ganglia causes anesthesia at those levels (56). Thus, the level of anesthesia is defined by the extent of rostral and caudal diffusion of the drug, as well as the concentration, volume, and potency of the drug.

The therapeutic effect of local anesthetics is to produce anesthesia by temporary blockade of nerve conduction through reversible inhibition of sodium channels. However, animal studies have demonstrated direct neurotoxic effects of anesthetics at high concentrations, such as are sometimes produced during epidural anesthesia (30; 74). Breakdown of the blood-brain or blood-nerve barrier, such as with intraneural or subarachnoid injection, leads to neurotoxicity at lower concentrations of local anesthetics (30).

The length of the lumbar section of the vertebral column is relatively short, and the dimensions of the lumbar epidural space are fairly constant; this results in only small differences in cranial spread of blockade after injection of local anesthetic at three different lumbar interspaces. In contrast, the thoracic part of the spinal column is longer, and it adjoins many different anatomical structures and spaces. Also, thoracic vertebrae and epidural space varies greatly in shape and size, and the above facts result in varying distribution of neural blockade following epidural injection.

In general, less local anesthetic is required to produce a given level of epidural anesthesia in pregnant patients. Engorgement of epidural veins by increased intraabdominal pressure has often been implied as the mechanism for this phenomenon. During pregnancy, onset of blockade of nerve conduction by local anesthetic is faster and blockade is more intense (11). The recommendation that epidural catheters should be sited at an intervertebral space that represents the middle of the area of surgical incision is no longer tenable when one considers the different patterns of distribution after single injection or continuous infusion of local anesthetic. Also, sympathicolysis, sympathetic epidural blockade in a particular area of the body, may be considered as important as satisfactory analgesia. Naturally, epidural insertion sites for various surgical indications vary to accomplish both goals (89). Epinephrine-augmented hypotensive epidural anesthesia is an effective method to avoid the use of a tourniquet during total knee arthroplasty without the negative effects on perioperative hemoglobin values (47). Parvizi and colleagues retrospectively ascertained by chart review the incidence of epidural hematoma in 11,235 patients who had 12,991 knee arthroplasties and who received oral anticoagulation and epidural anesthesia for their surgery (62). For 1030 patients (1038 knees) whose charts were reviewed in detail, the mean international normalized ratio at the time of removal of the epidural catheter was 1.54 (range 0.93-4.25). Although administration of epidural anesthesia in patients with coagulopathy can be detrimental, they recognized no cases of epidural hematoma causing neurologic symptoms in patients receiving controlled oral anticoagulation after total knee arthroplasty. Kawaguchi and associates reported two cases of epidural anesthesia using ultrasound imaging and concluded that using ultrasound imaging before epidural puncture in obese children is safer and more educational for residents (43).