Although severe or disabling neurologic complications are rare, recent epidemiologic series suggest the frequency of some serious complications, including spinal hematoma and central nervous system (CNS) infections, is increasing. A prospective survey in France recently evaluated the incidence and characteristics of serious complications related to regional anesthesia (2). Participating anesthesiologists kept a log of all cases and detailed information of serious complications occurring during or after regional anesthetics. All patients with a neurologic deficit lasting more than 2 days were examined by a neurologist; patients with cauda equina syndrome were evaluated with a computed tomography (CT) scan to rule out compressive etiology. A total of 103,730 regional anesthetics, including 40,640 spinal and 30,413 epidural anesthetics, were performed over a 5–month period. The incidence of cardiac arrest and neurologic complications was significantly higher after spinal anesthesia than other types of regional procedures (Table 12-1). Neurologic recovery was complete within 3 months in 29 of 34 patients with deficits. In 12 of 19 cases of radiculopathy after spinal anesthesia, and in all cases of radiculopathy after epidural or peripheral block, needle placement was associated with either paresthesia during needle insertion or pain with injection. In all cases, the radiculopathy had the same topography as the associated paresthesia. The authors concluded that needle trauma and local anesthetic neurotoxicity were the etiologies of most neurologic complications. In a follow-up investigation performed with similar methodology 5 years later, the investigators reported a slight decrease of neurologic complications related to regional anesthetic technique (10).

An epidemiologic study evaluating severe neurologic complications after neuraxial block conducted in Sweden between 1990 and 1999 reported some disturbing trends (8). During the 10–year study period, approximately 1,260,000 spinal and 450,000 epidural (including 200,000 epidural blocks for labor analgesia) were performed. A total of 127 serious complications were noted, including spinal hematoma (33), cauda equina (32), meningitis (29), and epidural abscess (13) (Table 12-2). The nerve damage was permanent in 85 patients. Complications occurred more often after epidural than spinal blockade, and were different in character: cauda equina syndrome, spinal hematoma, and epidural abscess were more likely to occur after epidural block, whereas meningitis was more often associated with a spinal technique. Undiagnosed spinal stenosis (detected during evaluation of the new neurologic deficits) was a risk factor for cauda equina syndrome and paraparesis with both techniques. In the 18 cases of cauda equina syndrome following spinal anesthesia, 5% hyperbaric lidocaine was administered in eight cases, whereas bupivacaine (hyperbaric or isobaric) was the local anesthetic in 11 cases. This large series suggests that the incidence of severe anesthesia-related complications is not as low as previously reported. Moreover, since serious complications were noted to occur even in the presence of experienced anesthesiologists, continued vigilance in patients undergoing neuraxial anesthesia is warranted.

For example, Cheney and colleagues (11) examined the American Society of Anesthesiologists (ASA) Closed Claims
database to determine the role of nerve damage following regional/pain block or general anesthesia in malpractice claims filed against anesthesia care providers. Of the 4,183 claims reviewed, 670 (16%) were for anesthesia-related nerve injury, including 189 claims involving the lumbosacral roots (105 claims) or spinal cord (84 claims); spinal cord injuries were the leading cause of claims for nerve injury that occurred in the 1990s, whereas previously, injuries to the ulnar nerve or brachial plexus were more common. In addition, lumbosacral nerve root injuries having identifiable etiology were associated predominantly with a regional (compared to general) anesthetic technique (92%), and were related to paresthesias during needle or catheter placement or pain during injection of local anesthetic. Major factors associated with spinal cord injury were blocks for chronic pain management and systemic anticoagulation in the presence of neuraxial block (see also Chapter 50). A more recent Closed Claims analysis of the 1,005 cases of regional anesthesia claims from 1980–1999, reported that the majority of neuraxial complications associated with regional anesthesia claims resulted in permanent neurologic deficits (12). Hematoma was the most common cause of neuraxial injuries, and the majority of these cases were associated with either an intrinsic or iatrogenic coagulopathy; 89% of patients had a permanent deficit. Conversely, complications caused by meningitis or abscess were more likely to be temporary. In a subset comparison of obstetric versus nonobstetric neuraxial anesthesia claims, obstetrics had a higher proportion of claims with low-severity and temporary injuries.

Direct needle- or catheter-induced trauma rarely results in permanent or severe neurologic injury. A recent retrospective study of 4,767 spinal anesthetics noted the presence of a paresthesia during needle placement in 298 (6.3%) of patients. Importantly, four of the six patients with a persistent paresthesia postoperatively complained of a paresthesia during needle placement, identifying elicitation of a paresthesia as a risk factor for a persistent paresthesia (13). In the series by Auroy and colleagues (2), two-thirds of the patients with neurologic complications experienced pain during needle placement or injection of local anesthetic. In all cases, the neurologic deficit had the same distribution as the elicited paresthesia. In addition, the neurologic injury occurred even though the investigators did not continue to inject in the presence of pain. It is unknown whether clinicians should abandon the procedure if a paresthesia is elicited (rather than replacing the needle) in an effort to decrease the risk of nerve injury. This decision is complicated by the series of conus medullaris injuries following spinal (three cases)
or combined spinal-epidural (four cases) anesthesia using a pencil-point needle reported by Reynolds (14). All seven cases complained of pain on needle insertion (only one noted pain on injection) and suffered damage to more than a single nerve root. In all patients, the anesthesiologist believed needle placement to have occurred at or below L2–L3. A syrinx was noted on magnetic resonance imaging (MRI) in six cases, suggesting intracord injection was the etiology of the deficits. Cases of cord damage from needle insertion were also reported in the series by Auroy and colleagues (10) and Moen and colleagues (8). Importantly, in all cases, the proceduralist had presumed the level of insertion to be below L1. These cases support the recommendation to insert needles below L3 to reduce the risk of direct needle trauma (14,15).

The passage and presence of an indwelling catheter into the subarachnoid or epidural space presents an additional source of direct trauma. However, a lower frequency of persistent paresthesia/radiculopathy occurs following epidural techniques, which are typically associated with catheter placement, compared to single-injection spinal anesthesia (2,10). Although the incidence of neurologic complications associated with thoracic epidural techniques has historically been judged to be higher than that of lumbar placement, Giebler and colleagues (16) noted only a 0.2% incidence of postoperative radicular pain in 4,185 patients undergoing thoracic epidural catheterization; all cases were responsive to catheter removal. Placement of a subarachnoid catheter most likely further increases the risk of neurologic dysfunction. In one study, the incidence of paresthesias was 13% with a single-dose and 30% with a continuous catheter spinal anesthetic (CSA) technique (17). The incidence of postoperative neurologic deficits was also significantly increased following CSA (0.66%) compared to single-dose techniques (0.13%). Laboratory studies have demonstrated demyelination and inflammation adjacent to the catheter tract in both the spinal root and cord of rats following placement of indwelling subarachnoid catheters (18). The use of a catheter may indirectly contribute to neurologic injury. Poor mixing resulting from very slow injection rates through spinal microcatheters may increase the risk of developing high concentrations of hyperbaric local anesthetics in dependent areas of the spinal canal. This is the presumed mechanism of cauda equina syndrome following continuous spinal anesthesia (19,20,21,22) (see also the discussion in the section Local Anesthetic Toxicity). In a series of 603 consecutive CSAs, including 127 delivered through a 28-gauge microcatheter, three patients reported pain (persistent paresthesia) postoperatively. In two patients, the symptoms resolved in 4 days; the other patient was discharged 8 days postoperatively with residual foot pain. One patient with aseptic meningitis and one patient with a sensory cauda equina syndrome (still present after 15 months) were also reported (23). A recent multicenter study compared the efficacy and safety of continuous spinal (329 patients) with continuous epidural (100 patients) anesthesia for labor analgesia (24). In the CSA group, sufentanil was continuously infused and bupivacaine administered as a bolus (0.25 mg isobaric bolus, with 1 hour maximum = 7.5 mg). No neurologic complications occurred; however, in 15% of cases, the intrathecal catheters were moderately difficult/difficult to remove and one catheter broke during extraction, leaving a 4-cm remnant in the patient. The catheter was left in situ with no apparent complication. Overall, the initial analgesia was superior in the CSA group, but there were also more technical difficulties and catheter failures, compared to continuous epidural analgesia. Although this study is the largest study to date evaluating spinal microcatheter techniques, the use of an infusion (which potentiates maldistribution) and administration of small (labor analgesia) doses of local anesthetic does not allow for extrapolation to other patient populations. In addition, the lack of a clear superiority with CSA also suggests that application of this technique may be limited.

Neurologic complications after neuraxial anesthesia may be a direct result of local anesthetic toxicity. Both laboratory and clinical evidence suggests that local anesthetic solutions are potentially neurotoxic and that the neurotoxicity varies among local anesthetic solutions (5,18,20,25). Neurotoxicity is dependent on pKa, lipid solubility, protein binding, and potency. In histopathologic, electrophysiologic, and neuronal cell models, lidocaine and tetracaine appear to have a greater potential for neurotoxicity than bupivacaine at clinically relevant concentrations (26). Additives such as epinephrine and bicarbonate may also affect neurotoxicity. The presence of a preexisting neurologic condition may predispose the nerve to the neurotoxic effects of local anesthetics (25).

Although most local anesthetics administered in clinical concentrations and doses do not cause nerve damage (27), prolonged exposure, high dose, and/or high concentrations of local anesthetic solutions at the spinal roots may result in permanent neurologic deficits (22). For example, cauda equina syndrome has been reported after single-dose and continuous spinal anesthesia, intrathecal injection during intended epidural anesthesia, and repeated intrathecal injection after failed spinal block with lidocaine (2,20,28). In the late 1980s, three manufacturers introduced 27- to 32-gauge “microcatheters” capable of passage through standard 22- to 26-gauge needles. Experience with these devices was just gaining popularity when the occurrence of eleven cases of cauda equina syndrome led to their withdrawal from the U.S. market by the U.S. Food and Drug Administration (FDA) in 1992 (29). Presumably, injection (and/or reinjection) results in high concentrations of local anesthetic within a restricted area of the intrathecal space and causes neurotoxic injury. In the study by Auroy and co-workers, (2) 75% of the neurologic complications after uneventful (atraumatic) spinal anesthesia occurred in patients who received hyperbaric lidocaine, including one patient who received 350 mg over 5 hours with a 5% lidocaine infusion. Drasner (30) has recommended a maximum dose of 60 mg of lidocaine and the avoidance of epinephrine to prolong lidocaine spinal anesthesia. In addition, many clinicians recommend the use of isobaric solutions to reduce the risk of nonuniform distribution within the intrathecal space. Attention to patient positioning, total local anesthetic dose, and careful neurologic examination (evaluating for preferential sacral block) will assist in the decision to inject additional local anesthetic in the face of a patchy or failed block (31) (Tables 12-3 and 12-4).

2-Chloroprocaine was introduced nearly 50 years ago as a local anesthetic for epidural administration. However, concern for neurotoxicity emerged two decades ago, with a series of eight cases of neurologic injury associated with the use of Nesacaine-CE, a chloroprocaine solution containing the antioxidant sodium bisulfite. In all cases, the injury occurred after a large volume of anesthetic solution intended for the epidural space was accidentally administered intrathecally. Subsequent laboratory investigations evaluating the toxic contributions of 2-chloroprocaine, bisulfite, epinephrine, and pH reported that the commercial solution of 3% chloroprocaine (containing 0.2% sodium bisulfite, pH 3) produced irreversible block, but exposure to the same solution buffered to pH 7.3 resulted in complete recovery (32). It was assumed that bisulfite was the
source of neurotoxicity and that solutions that were bisulfite-free were safe for intrathecal use. More recently, these experiments were repeated with a more appropriate animal model and yielded different results: nerve injury scores were greater after administration of plain chloroprocaine compared to those of chloroprocaine containing bisulfite. These findings suggest clinical deficits associated with unintentional intrathecal injection of chloroprocaine likely resulted from a direct effect of the anesthetic, not the preservative. In addition, the data suggest that bisulfite can actually reduce neurotoxic damage induced by intrathecal local anesthetic (33). Although recent clinical and volunteer studies (34) have not reported neurologic symptoms following spinal anesthesia with low-dose 2-chloroprocaine (30–40 mg), the laboratory evidence for toxicity warrants a cautious approach until additional toxicity data are available.

Transient neurologic symptoms (TNS) were first formally described in 1993. Schneider and colleagues (3) reported four cases of severe radicular back pain occurring after resolution of hyperbaric lidocaine spinal anesthesia. All four patients had undergone surgery in the lithotomy position. No sensory or motor deficits were detected on examination, and the symptoms resolved spontaneously within several days. Multiple laboratory and clinical studies have been performed in an attempt to define the etiology, clinical significance, and risk factors associated with TNS. However, our understanding remains incomplete.

The incidence of TNS has ranged between 0% and 37% (35,36), and is dependent on anesthetic, surgical, and probably undefined patient factors. A prospective randomized study reported a 16% incidence of TNS in patients receiving either hyperbaric 5% lidocaine with epinephrine or 2% isobaric lidocaine. However, no patient receiving 0.75% hyperbaric bupivacaine developed TNS (35). In addition, the incidence was higher among patients positioned with knees or hips flexed (genitourinary, arthroscopy) than in patients positioned supine (herniorrhaphy), presumably because the flexion results in additional stretch on the nerve roots. A subsequent study comparing the incidence of TNS in knee arthroscopy patients undergoing spinal anesthesia with 50 mg of lidocaine in 2%, 1%, and 0.5% concentrations also failed to note a concentration effect; the incidence was similar in all groups (37). The lack of concentration (37) or dose effect (38) suggests that neurotoxicity is not the etiology of TNS, but does not rule out an alternative intrathecal source. Neurophysiologic evaluation in volunteers during TNS did not reveal abnormalities in somatosensory evoked potentials, electromyography, or nerve conduction studies (39).

A large multicenter epidemiologic study involving 1,863 patients was recently performed to identify potential risk factors for TNS (38). The incidence of TNS with lidocaine (11.9%) was significantly higher than that with tetracaine (1.6%) or bupivacaine (1.3%). The pain was described as severe in 30% of patients and resolved within a week in over 90% of cases. Outpatient status, obesity, and lithotomy position also increase the risk of TNS for patients who receive lidocaine. This suggests that the risk of TNS is high among outpatients in the lithotomy position (24.3%) and low for inpatients having surgery in positions other than lithotomy (3.1%). However, these variables were not risk factors with tetracaine or bupivacaine. The authors also reported that neither gender, age, history of back pain or neurologic disorder, lidocaine dose/concentration, spinal needle/size, aperture direction, nor addition of epinephrine increased the risk of TNS (Table 12-5). A previous study has identified the addition of phenylephrine as a risk factor for TNS with tetracaine spinal anesthesia (40). These findings were confirmed in a systematic review of TNS. The analysis included 14 trials reporting 1,347 patients (117 of whom developed TNS). The relative risk for developing TNS after spinal anesthesia with lidocaine was higher (4.35 [95% confidence interval (CI), 1.98–9.54]) than with other local anesthetics (bupivacaine, prilocaine, procaine, and mepivacaine) (41). There was no evidence of neurologic deficits; in all patients, the symptoms disappeared spontaneously by the 10th postoperative day.

The high frequency of TNS with lidocaine spinal anesthesia has resulted in a search for a safe and effective alternative. The intrathecal administration of 2-chloroprocaine is under reconsideration due to the concern regarding toxicity, as previously mentioned. Mepivacaine may be a suitable substitute; in a series of 1,273 patients undergoing spinal or combined spinal-epidural anesthesia, TNS occurred in only 78 (6.4%; 95% CI, 5.1%–8%) (42).

The etiology and clinical significance of TNS are unknown. Recent studies suggest a local anesthetic toxicity, although the mechanism may not be identical to that of cauda equina syndrome (43). Although many anesthesiologists believe that the reversible radicular pain is on one side of a continuum leading to irreversible cauda equina syndrome, no data support this concept. It is important to distinguish between factors associated with serious neurologic complications, such as cauda equina syndrome, and transient symptoms when making recommendations for the clinical management of patients. For example, increasing the concentration/dose of lidocaine and adding epinephrine increases the risk of irreversible neurotoxicity, but has little effect on the risk of TNS. Therefore, the clinician must determine the appropriate intrathecal solution, including adjuvants, given the surgical duration and intraoperative position for each individual patient.

The blood supply to the spinal cord is precarious due to the relatively large distances between the radicular vessels. Systemic hypotension or localized vascular insufficiency with or without a spinal anesthetic may produce spinal cord ischemia resulting in flaccid paralysis of the lower extremities (including sphincter dysfunction) or anterior spinal artery syndrome (44). Classically, proprioception and sensation are spared or preserved, relative to motor loss. Characteristics of anterior spinal artery syndrome, spinal abscess, and spinal hematoma are reported in (45) (Table 12-6). Local anesthetic solutions have a varied effect on spinal cord blood flow. For example, lidocaine and tetracaine either maintain or increase blood flow, whereas bupivacaine and levobupivacaine result in a decrease (46,47,48). The addition of epinephrine or phenylephrine results in a further decrease. However, in laboratory investigations, the alterations in blood flow are not accompanied by changes in histology or behavior. Likewise, large clinical studies have failed to identify the use of vasoconstrictors as a risk factor for temporary or permanent deficits. Most presumed cases of vasoconstrictor-induced neurologic deficits have been reported as single case reports, often with several other risk factors present (2,49).

Finally, the addition of vasoconstrictors may potentiate the neurotoxic effects of local anesthetics. In a laboratory model, it was determined that the neurotoxicity of intrathecally administered lidocaine was increased by the addition of epinephrine (50). A recent investigation by Sakura and colleagues (40) noted that the addition of phenylephrine increased the risk of TNS in patients undergoing tetracaine spinal anesthesia (although no patient had sensory or motor deficits). The actual risk of significant neurologic ischemia causing neurologic compromise in patients administered local anesthetic solutions
containing vasoconstrictors appears to be very low. Clinicians should be aware of other surgical and patient factors predisposing to spinal cord ischemia, including major aortic vascular or spinal column procedures, arthrosclerosis, sustained hypotension, and anemia. The decision to perform a neuraxial block in these patients is based on risk–benefit evaluation and the ability to diagnosis/intervene should a reversible etiology of ischemia occur.

The actual incidence of neurologic dysfunction resulting from hemorrhagic complications associated with neuraxial blockade is unknown; however, the incidence cited in the literature is estimated to be less than one in 150,000 epidural and less than one in 220,000 spinal anesthetics (51). In a review of the literature between 1906 and 1994, Vandermeulen and colleagues (52) reported 61 cases of spinal hematoma associated with epidural or spinal anesthesia. In 87% of patients, a hemostatic abnormality or traumatic/difficult needle placement was present. More than one risk factor was present in 20 of 61 cases. Importantly, although only 38% of patients had partial or good neurologic recovery, spinal cord ischemia tended to be reversible in patients who underwent laminectomy within 8 hours of onset of neurologic dysfunction.

The need for prompt diagnosis and intervention in the event of a spinal hematoma was also demonstrated in a review of the ASA Closed Claims project, which noted that spinal cord injuries were the leading cause of claims in the 1990s (11). Spinal hematomas accounted for nearly half of the spinal cord injuries. Risk factors for spinal hematoma included epidural anesthesia in the presence of intravenous (IV) heparin during a vascular surgical or diagnostic procedure. Importantly, the presence of postoperative numbness or weakness was typically attributed to local anesthetic effect rather than spinal cord ischemia, which delayed the diagnosis. Patient care was rarely judged to have met standards (1 of 13 cases), and the median payment was very high.

It is impossible to conclusively determine risk factors for the development of spinal hematoma in patients undergoing neuraxial blockade solely through review of the case series, which represent only patients with the complication and do not define those who underwent uneventful neuraxial analgesia. However, large inclusive surveys that evaluate the frequencies of complications (including spinal hematoma), as well as identify subgroups of patients with higher or lower risk, enhance risk stratification. In the series by Moen and co-workers (8) involving nearly 2 million neuraxial blocks, 33 spinal hematomas occurred. The methodology allowed for the calculation of frequency of spinal hematoma among patient populations. For example, the risk associated with epidural analgesia in women undergoing childbirth was significantly less (one in 200,000) than that in elderly women undergoing knee arthroplasty (one in 3,600, p <0.0001). Likewise, women undergoing hip fracture surgery under spinal anesthesia had an increased risk of spinal hematoma (one in 22,000) compared to all patients undergoing spinal anesthesia (one in 480,000).

Overall, these series suggest that the risk of clinically significant bleeding varies with age (and associated abnormalities of the spinal cord or vertebral column), the presence of an underlying coagulopathy, difficulty during needle placement, and an indwelling neuraxial catheter during sustained anticoagulation (particularly with standard or low-molecular-weight heparin [LMWH]). They also consistently demonstrate the need for prompt diagnosis and intervention.

Prevention of venous thromboembolism remains a crucial component of patient care following major surgery. Although neuraxial anesthesia and analgesia reduce the risk of venous thrombosis, a significant risk remains, even in the presence of a continuous epidural infusion containing a local anesthetic (53). As a result, pharmacologic (and/or mechanical) prophylaxis is warranted. Thromboprophylaxis is based upon identification of risk factors. The risk factors for thromboembolism include trauma, immobility/paresis, malignancy, previous thromboembolism, increasing age (over 40 years), pregnancy, estrogen therapy, obesity, smoking history, varicose veins, and inherited or congenital thrombophilia. Not surprisingly, only the healthiest patients undergoing minor surgery are not considered candidates for thromboprophylaxis postoperatively.

Guidelines for antithrombotic therapy including appropriate pharmacologic agent, degree of anticoagulation desired, and duration of therapy continue to evolve. There is a trend toward initiating thromboprophylaxis in close proximity to surgery. However, early postoperative (and intraoperative) dosing of LMWH was associated with an increased risk of neuraxial bleeding. Likewise, the duration of prophylaxis has been extended to a minimum of 10 days following total joint replacement or hip fracture surgery, whereas the recommended duration for hip procedures is 28 to 35 days. It has been demonstrated that the risk of bleeding complications is increased with the duration of anticoagulant therapy. The interaction of prolonged thromboprophylaxis and previous neuraxial instrumentation, including difficult or traumatic needle insertion, is unknown.

Recommendations from the Seventh American College of Chest Physicians (ACCP) in 2004 are based upon prospective randomized studies that assess the efficacy of therapy using contrast venography or fibrinogen leg scanning to diagnose asymptomatic thrombi (54) (Table 12-7). Clinical outcomes, such as fatal pulmonary embolism and symptomatic deep venous thrombosis are not primary endpoints. Despite the successful reduction of asymptomatic thromboembolic events with routine use of antithrombotic therapy, an actual reduction of clinically relevant events has been more difficult to demonstrate.