Epidural Anesthesia and Analgesia: Their Role in Postoperative Outcome

Approximately one in eight patients who undergo surgery in the United States has risk factors for or has known coronary artery disease. [21]Some patient populations are at especially high risk for coronary artery disease. For example, asymptomatic patients with peripheral vascular disease have a 40-70% incidence of significant coronary stenoses (documented with coronary angiography). [22,23]It is no surprise therefore that cardiac morbidity is the primary cause of death after anesthesia and surgery with reported incidences in high-risk populations ranging from 2% to 15%. [24]Thus, anesthetic and analgesic techniques that reduce postoperative cardiac morbidity should significantly improve patient outcome.

Experimental Effects. Previous studies in nonsurgical populations have demonstrated that activation of the sympathetic nervous system may result in myocardial ischemia and infarction. [25-27]Similarly, interventions that inhibit the sympathetic response can reduce cardiac morbidity. [28-33]Because sympathetic activation occurs during the perioperative period, inhibition of sympathetic activity during the perioperative period may reduce postoperative myocardial morbidity (Figure 1). The mechanisms whereby sympathetic activation causes cardiac morbidity can be mediated both by increases in myocardial oxygen demand or reductions in myocardial oxygen supply. For example, pain from surgical stimuli can activate sympathetic efferent nerves and increase heart rate, inotropy, and blood pressure (Figure 1). Although activation of the sympathetic nervous system can increase indexes of myocardial oxygen demand and result in ischemia, most episodes of myocardial ischemia occur in the absence of major hemodynamic changes. [34-37]In fact, most episodes of "silent ischemia" are not preceded by increases in myocardial oxygen demand with the exception of small increases in heart rate. [38,39]Thus, reductions in myocardial oxygen supply may be the primary cause of silent ischemia.

Reduction in myocardial oxygen supply may result from coronary vasoconstriction or thrombosis in coronary arteries and may be exacerbated by episodes of perioperative hypoxemia. [40,41]Activation of the sympathetic nervous system may trigger both of these mechanisms. For example, activation of cardiac sympathetic nerves is correlated with coronary vasoconstriction, [42,43]paradoxical coronary vasoconstrictor response to intrinsic vasodilators, [44]and poststenotic coronary vasoconstriction (Figure 2) [45]with resultant signs of myocardial ischemia such as ST-segment changes, angina, dysrhythmias, and increases in myocardial infarct size. [46-50]Sympathetic activation has also been proposed to cause postoperative hypercoagulable states [51]and thus may be a factor in thrombotic causes of myocardial ischemia. Once ischemia develops, secondary increases in heart rate and blood pressure can increase myocardial oxygen demand and further aggravate ischemia (Figure 3). Thus, sympathetically mediated decreases in myocardial oxygen supply may be a major factor in postoperative cardiac morbidity.

Selective blockade of cardiac sympathetic innervation (T1-T5) can most easily be achieved by administrating local anesthetics through an epidural catheter placed at an upper thoracic level, a technique commonly known as thoracic epidural anesthesia (TEA). In patients with coronary artery disease, TEA produces small reductions in cardiac output, heart rate, and blood pressure and may thus decrease myocardial oxygen demand. [46,52]Perhaps more important, determinants of oxygen supply may also improve after TEA. Although total coronary blood flow remains unchanged, [42,53]blood flow to ischemic regions of myocardium increases, [42,53,54]regional distribution of myocardial blood flow improves by increasing the endocardial to epicardial blood flow ratio, [42,47,54]and sympathetically mediated coronary constriction distal to the coronary stenosis is inhibited. [45,53]Thus, use of TEA in patients with ischemic myocardium may improve myocardial oxygen supply, as long as blood pressure is maintained in a relatively normotensive range.

These potential improvements in myocardial oxygen supply after TEA with local anesthetics may be attributable to blockade of both alpha- and beta-adrenergic sympathetic stimulatory effects. Activation of cardiac sympathetic nerves can cause vasoconstriction of large coronary epicardial arteries through stimulation of alpha-adrenergic receptors. [55,56]Because approximately 75% of atherosclerotic coronary stenosis are dynamic and not fixed, [57]the majority of stenosis are still able to constrict in response to sympathetic stimulation. [58]This vasoconstriction could further reduce coronary blood flow and increase the severity of myocardial ischemia (Figure 2). Indeed, high TEA (T1-T6) with bupivacaine during coronary angiography in patients with severe coronary artery disease, increased the luminal diameter of 16 (62%) of 26 stenotic epicardial coronary arteries without changing the diameter of non stenotic segments. [53]Thus, alpha-adrenergic blockade may increase blood flow through stenotic coronary arteries.

Sympathetic stimulation of alpha-adrenergic receptors in coronary arterioles also plays an important role in distribution of blood flow within the myocardium. Several studies suggest that vasoconstriction of coronary arterioles in an area distal to a critical coronary stenosis (within the area of ischemic myocardium) may actually shunt blood flow to the area of myocardium at greatest risk of infarction (typically from epicardium to endocardium) (Figure 2). [59,60]Local anesthetic blockade of cardiac sympathetic innervation can decrease the magnitude of poststenotic coronary vasoconstriction and actually decrease myocardial ischemia. [60]Although effects of adrenergic coronary vasoconstriction in the ischemic myocardium remain controversial, [55,59]studies assessing regional myocardial blood flow distribution in intact animals indicate that TEA with local anesthetics results in a beneficial effect by increasing endocardial-to-epicardial blood flow ratios. [47,54,61]Thus, TEA may increase myocardial blood flow both by dilating coronary arteries, improving blood flow distribution through coronary arterioles, and by reducing the stimulus (pain) for activation of cardiac sympathetic nerves. These effects on myocardial oxygen supply in combination with reduction in indexes of myocardial oxygen demand suggests that TEA with local anesthetics provides a beneficial effect on the oxygen supply-demand ratio in ischemic myocardium.

Studies in intact animals have demonstrated that use of TEA after acute coronary artery occlusion decreases the magnitude of ST-segment changes, [62,63]reduces myocardial infarct size (Figure 4), reduces the incidence of ventricular dysrhythmias, [54,64]and lessens ischemia-induced decreases in myocardial pH. [62]Of note, beneficial effects of TEA do not appear to be the result of systemic absorption of the local anesthetic, [54]and effects of epidural opioids on ischemic myocardium are unknown. These studies in animals suggest that TEA with local anesthetics may be effective in the treatment, and perhaps prevention, of myocardial ischemia.

Correspondingly, TEA has been found to be an effective treatment for humans with myocardial ischemia refractory to conventional medical therapy. [46,49,65,66]TEA has been used as a long-term therapy (> 3 yr) for patients not only refractory to medical therapy but also judged to be too high risk for surgical coronary bypass. [67]Initiation of TEA in patients with refractory myocardial ischemia (refractory to therapy with beta-adrenergic blocking agents, calcium antagonists, nitrates, low-dose heparin, or nitroglycerin infusion for > 24 h) resulted in rapid relief of chest pain and reduced ischemic ST-segment changes. [46,49]Relief of ischemia probably occurred both through increases in myocardial oxygen delivery by coronary artery vasodilation and also through further decreases in indexes of myocardial oxygen demand (Figure 5). [68]Hemodynamic changes after induction of TEA were mild, and in no patient did significant hypotension or bradycardia develop. The lack of significant decreases in systemic blood pressure and heart rate may be explained by the production by TEA of minimal cardiac depression in the presence of preexisting beta-adrenergic blockade (eight of nine patients were already being treated with beta-adrenergic blocking agents). [69]These results suggest that TEA not only improves the oxygen supply-demand ratio in ischemic regions of the myocardium in humans but may also be superior to conventional antianginal therapy.

Although TEA may be a useful therapy for angina, epidural anesthesia is more frequently used for major surgical procedures. The combination of TEA and general anesthesia has been shown to produce beneficial effects such as reduced hemodynamic correlates of myocardial oxygen consumption, [70-73]improved intraoperative hemodynamic stability, [70,74]and reduced stress response. [52,70]Obviously, combination of two anesthetic techniques also has the potential to produce detrimental hemodynamic effects and reduced coronary perfusion if anesthesia is not carefully administered. [75,76]Nonetheless, TEA with local anesthetics, with or without general anesthesia, offers potential benefits such as improved balance of myocardial oxygen supply-demand and greater intraoperative hemodynamic stability for patients with coronary artery disease undergoing surgery.

Although TEA has not been directly compared with lumbar epidural anesthesia (LEA), LEA does not appear to offer the same degree of benefits as TEA. In contrast to TEA, LEA is less commonly extended to attain block levels sufficient to produce the beneficial effects of thoracic sympathectomy (at or higher than T1). Thus, if hypotension occurs during LEA and reduces myocardial oxygen supply, this reduction in supply may not be counterbalanced by a simultaneous reduction in demand. [77]In fact, coronary blood flow may be further decreased if compensatory sympathetic activation occurs in the unblocked upper thoracic dermatomes. [78]It should then come as no surprise that the effects of LEA on myocardial ischemia are less encouraging. [71,79]Indeed, reductions in mean arterial pressure of approximately 20% during LEA with an upper level of sensory analgesia between T6-T12 may increase myocardial ischemia in some patients with coronary artery disease. [79]In contrast, similar reductions in blood pressure during TEA were not associated with myocardial ischemia during TEA. [77]Thus, the ability to selectively block cardiac sympathetic fibers (T1-T5) suggest an advantage of TEA over LEA during hypotension in patients at risk for cardiac morbidity.

In summary, animal and human data indicate that TEA with or without general anesthesia can favorably alter the balance of oxygen supply and demand in ischemic myocardium by selective thoracic sympathetic blockade. In fact, TEA may have efficacy equal to or greater than conventional antianginal therapy during acute myocardial ischemia. LEA may reduce myocardial morbidity for operations on the lower extremity or lower abdomen through a reduction in pain and the surgical stress response. However, the ability to directly inhibit cardiac sympathetic innervation may confer a physiologic advantage to TEA over LEA.

Clinical Effects. Initial efforts by anesthesiologists to reduce perioperative myocardial morbidity and mortality focused on variations in intraoperative anesthetic technique and found little ability to alter outcome. For example, neither general anesthetic technique nor routine use of intensive monitoring (pulmonary artery catheters) appears to affect postoperative cardiac morbidity significantly. [34,80,81]Similarly, intraoperative use of TEA can produce beneficial intraoperative hemodynamic effects between induction and tracheal extubation, yet it may have little effect on the incidence of postoperative myocardial ischemia. [76]A probable cause for this lack of effect is that the incidence and severity of intraoperative ischemia appears to be no greater than the incidence in the immediate preoperative period. [35]Furthermore, episodes of intraoperative myocardial ischemia most commonly occur without obvious hemodynamic causes: only 10-30% of episodes of myocardial ischemia are associated with intraoperative hemodynamic perturbations. [34-37]These results suggest that the intraoperative period is usually no more arduous than daily life and that ischemia that occurs intraoperatively may serve only to identify high-risk patient populations. Thus, although intraoperative ischemia can be correlated with postoperative cardiac morbidity, [82]it may not be a direct cause. Therefore, the disparity between beneficial intraoperative hemodynamic effects of TEA and lack of reduction in postoperative myocardial ischemia in the majority of studies is not surprising. [34-37].

On the other hand, ischemia that occurs postoperatively is more frequent, more severe, and more prolonged than ischemia occurring during pre- and intraoperative periods. [35]Furthermore, early postoperative ischemia appears to be an important correlate of cardiac morbidity in high-risk patients undergoing non cardiac surgery. [21]These observations caused Mangano et al. to conclude that "the postoperative period may warrant special attention including therapeutic trials to determine whether extended monitoring and aggressive therapy for control of pain and heart rate are warranted during the 1st 2 days after surgery." [21]This conclusion was supported by subsequent studies demonstrating that intensive perioperative analgesia with large doses of systemic opioids (high enough to require mechanical ventilation) can reduce myocardial morbidity and mortality. [4,5]However, myocardial infarction commonly occurs on the 3rd or 4th day after surgery, [83,84]suggesting that aggressive therapy may need to be extended for at least 4 days. If so, the need for prolonged ventilatory support would limit the usefulness of high-dose postoperative opioid analgesia. Thus, the greatest potential to influence cardiac outcome may result from application of analgesic techniques in the postoperative period. If epidural analgesia is expected to be beneficial, it makes sense to focus evaluations on the postoperative period.

Although administration of epidural local anesthetics or opioids in the postoperative period provides better analgesia, suppresses the stress response to surgery, and reduces incidences of myocardial ischemia and dysrhythmias when compared with systemic opioids, [6,70,85]effects of epidural analgesia on cardiac morbidity are controversial. [42,73,76,86-91]However, most studies do not have the power to detect clinically significant differences in cardiac outcomes because of the low incidence of these events and small numbers of patients studied, [89,91]and no single study to date provides a definitive answer. Consequently, conclusions at this time must be based on interpretation of trends from randomized, prospective, controlled studies that suggest that postoperative epidural analgesia is associated with reduced postoperative cardiac morbidity and mortality in high-risk populations (Table 1).

The importance of studying high-risk patient populations (i.e., patients that typically experience a high incidence of complications) cannot be overemphasized. It is in these patient groups that the greatest effects of epidural anesthesia and analgesia have been demonstrated. For example, Yeager et al. studied only patients scheduled for "intrathoracic, intraabdominal, or major (noncerebral) vascular surgery" that were also "scheduled preoperatively by the surgical staff to receive postoperative care in the intensive care unit due either to the severity of preexisting disease(s), the magnitude of the anticipated surgical procedure, or both." [73]Of note, patients in this study had an average American Society of Anesthesiologists (ASA) physical status classification of 3. Similarly, Tuman et al., studied only patients undergoing major peripheral vascular surgery (approximately 45% of operations involved the abdominal aorta). [88]Patients in these studies had high incidences of postoperative myocardial morbidity in the general anesthesia groups and statistically significant reductions in morbidity were identified in the epidural analgesia groups.

However, not all studies investigating high-risk populations have observed a benefit from epidural anesthesia or analgesia. Perhaps the best-designed study that did not find a benefit for epidural anesthesia in a high-risk patient population is a study by Baron et al. [76]No difference in the incidence of cardiac complications was observed in aortic reconstruction patients randomized to general anesthesia versus combined TEA and general anesthesia. Although many aspects of study design were excellent, postoperative analgesia techniques were not controlled or randomized. For example, only 37% of the patients in the epidural anesthesia group received epidural local anesthetics for postoperative analgesia, whereas 59% of the patients in the general anesthesia group received postoperative analgesia with epidural local anesthetic or opioid. Thus, this study assessed only the effect of intraoperative anesthetic technique on postoperative outcome. As previously discussed, there is minimal basis for believing that intraoperative TEA alone will affect postoperative cardiac morbidity. Of note, absolute morbidity rates reported by Baron et al. [76]were comparable to those reported by Yeager et al. [73]for the epidural group (Table 2). One possible conclusion is that the general anesthetic technique used by Yeager et al. was associated with increased postoperative morbidity. However, direct comparison with absolute rates of morbidity reported by Tuman et al. [88]suggests that postoperative epidural analgesia with a combination of local anesthetic and opioid may be associated with a reduction in morbidity. Although this speculation is stimulating, it seems problematic to directly compare absolute outcome data from different institutions, after different operations, performed years apart, in different countries, and with different patient populations. Nevertheless, the crux of the as-yet unanswered modern question is whether epidural analgesia produces better outcomes than an optimal general anesthetic technique followed by optimal systemic analgesia.

In contrast to these studies investigating high-risk populations, Hjortso et al. [91]studied lower-risk patient populations (> 90% ASA physical status classification 1 or 2), and Christopherson et al. [89]studied a high-risk population having a less invasive surgical procedure (infrainguinal revascularizations) with a lesser magnitude of surgical insult. [92]Both studies had low incidences of myocardial morbidity in both control and epidural groups, and neither was able to identify significant effects on cardiac morbidity. [89,91]Thus, use of postoperative epidural analgesia after intraoperative epidural anesthesia has been associated with reductions in postoperative cardiac morbidity only in high-risk patients undergoing major operations.

In summary, no current study includes a sufficient number of patients to provide definitive conclusions. However, differences appear to be largest when high-risk populations undergoing major operations are studied. Thoracic epidural catheter placement and use of local anesthetics have theoretical advantages that have not been evaluated in clinical trials. If these potential benefits are proven significant in follow up studies with larger numbers of patients, epidural analgesia has the potential to dramatically reduce perioperative myocardial morbidity after major operations in high-risk patient populations.

Major surgery is associated with a hypercoagulable state that persists well into the postoperative period. [93-95]Increases in perioperative coagulation are associated with vasoocclusive and thromboembolic events that may result in postoperative morbidity and mortality. [88,89,96,97]Although the etiology of this postoperative increase in coagulability is uncertain, the stress response appears to be an important initiator. [51,98,99]Postoperative changes occur in all arms of the coagulation system and include increased concentrations of coagulation factors, [94]decreased concentration of coagulation inhibitors, [100]enhanced platelet activity, [101]and impaired fibrinolysis. [93,102]General anesthesia with parenteral opioid analgesia has little effect on postoperative increases in coagulability. [51,88,89,103,104]However, epidural anesthesia and analgesia may reduce increases in postoperative coagulability and thus may improve clinical outcome.

Experimental Effects. Epidural administration of local anesthetic has multiple effects on coagulation that may be beneficial. [105,106]For example, general anesthesia and surgery produce a sustained reduction of blood flow in deep veins that may predispose to vascular graft occlusion and formation of deep venous thrombosis. Epidural anesthesia with local anesthetics improves lower extremity blood flow by increasing arterial inflow and venous emptying rate. [105]Addition of epinephrine to the local anesthetic increases the magnitude of these beneficial effects. [105]Second, epidural anesthesia enhances fibrinolytic activity [51,95,107-109]by prevention of postoperative increase in plasminogen activator inhibitor 1, [51]more rapid return of antithrombin III from increased to normal values, [95]and lessening of postoperative increases in platelet aggregation. [88]In addition, systemic absorption of local anesthetic during epidural anesthesia produces plasma concentrations sufficient to directly impair platelet aggregation [103,110,111]and reduce plasma and whole blood viscosity. [112]Although epidural opioids can also reduce the surgical stress response, no studies have specifically examined effects on coagulation. Finally, beneficial effects of epidural local anesthetics may be attenuated by addition of general anesthesia. [95]Thus, epidural administration of local anesthetics modifies the perioperative hypercoagulable state through several mechanisms, including blockade of sympathetic efferent nerves, attenuation of excessive coagulability, and anticoagulant properties of systemically absorbed local anesthetics.

Clinical Effects. Epidural anesthesia and analgesia appears to significantly reduce the incidence of thromboses of vascular grafts in patients undergoing lower extremity revascularization. Clinical markers for graft failure such as incidence of reoperation for graft occlusion or amputation are decreased with use of epidural anesthesia and analgesia. For example, Tuman et al. [88]randomized patients to receive either general anesthesia with parenteral opioids or general-epidural anesthesia followed by postoperative bupivacaine-fentanyl epidural infusion. Use of epidural analgesia was associated with a ninefold decrease in incidence of vascular graft occlusion. Similarly, Christopherson et al. [89]compared the effects of epidural and general anesthesia in peripheral vascular surgery patients and found that the epidural group required fivefold fewer reoperations for graft failure within 1 month. To date, impressive differences associated with epidural anesthesia and analgesia have been seen only in patients inherently at high risk for vasoocclusive events. [113]For example, patients with atherosclerotic disease are known to be hypercoagulable [88,114]and to have abnormal responses to intrinsic vasodilators. [44]Whether or not other patient groups benefit from epidural anesthesia and analgesia remains to be seen. Nonetheless, these well-conducted studies in high-risk groups demonstrate a reduction in vasoocclusive events associated with use of epidural anesthesia and analgesia.

Deep venous thrombosis with resultant pulmonary embolism is another manifestation of increased perioperative coagulability. Patients undergoing total hip replacement appear to be at especially high risk for thromboembolic complications because of immobility, decreased deep vein blood flow, [97,108]surgical manipulation of the femoral vein, decreased cardiac output from anesthetic-related effects, and prolonged surgical time. [115]Modig et al. [97,107,108]reported a series of studies on the effects of LEA followed by 24 h of analgesia with bupivacaine and epinephrine on deep venous thrombosis in patients undergoing hip replacement surgery. When compared with patients who received general anesthesia and intramuscular opioid analgesia, the epidural group had a 2.5-5-fold reduction in deep thigh venous thrombosis, a 1.5-fold reduction in calf venous thrombosis, and a 3-fold reduction in pulmonary embolism. Reduction of deep thigh venous thromboses is particularly significant, because these thromboses are almost exclusively associated with emboli to the pulmonary circulation. Similar reductions in thromboembolic complications have been reported with epidural anesthesia and analgesia (with local anesthetic) by other authors [116]and after other operations such as open prostatectomy [117]and knee arthroplasty. [118].

Thus, reduced incidences of vascular and thromboembolic complications with epidural anesthesia are well documented; however, vasoocclusive and thromboembolic events are postoperative occurrences that probably begin intraoperatively. [115]Is it the intraoperative use of epidural anesthesia or the postoperative use of epidural analgesia that diminishes hypercoagulability and reduces related complications? Christopherson et al.'s [89]finding that most reoperations occurred soon after surgery suggests that use of intraoperative epidural local anesthetic was critical. Correspondingly, their use of epidural fentanyl alone for postoperative analgesia suggests that either epidural opioids maintain a reduction in postoperative hypercoagulability or that intraoperative effects of epidural local anesthetic overshadow effects from postoperative analgesia. Further evidence for the importance of the intraoperative effect is suggested by a metaanalysis of anesthesia for hip repair (11 spinal anesthesia and 2 epidural anesthesia studies) that documented a 31% reduction in deep venous thrombosis as compared with general anesthesia. [119]Thus, the relative importance of postoperative epidural analgesia for reduction in vasoocclusive and thromboembolic complications is unclear.

In summary, reduced incidences of vascular graft occlusion and thromboembolic complications are associated with the use of epidural anesthesia and analgesia. Furthermore, there appears to a strong association between the use of intraoperative epidural anesthesia and reduction in morbidity, whereas the role of postoperative epidural analgesia remains uncertain.

Postoperative pulmonary dysfunction occurs as a result of surgery and anesthesia-related physiologic perturbations and remains a major cause of postoperative morbidity. Thus, techniques that reduce postoperative pulmonary dysfunction may result in improved clinical outcome.

Experimental Effects. Upper abdominal and thoracic incisions significantly reduce post operative pulmonary function, [120-122]whereas laparoscopic [123,124]and peripheral operations have little effect. Pulmonary dysfunction after upper abdominal surgery occurs because of pain, [121]abnormal diaphragmatic function, [125]and increased abdominal and lower intercostal muscle tone during exhalation. [126]Pulmonary dysfunction begins with incision and remains diminished for 7-14 days postoperatively. [127]The most important alteration of respiratory function is decreased functional residual capacity, which begins about 16 h postoperatively, reaches a nadir at 24-48 h, and usually resolves within 1 week. [120,121,127,128]Decreased functional residual capacity may result in atelectasis and ventilation-perfusion abnormalities leading to hypoxemia, pneumonia, and postoperative pulmonary complications (Figure 6). [120,121,129,130]Patients especially at risk for reduction of functional residual capacity and resultant pulmonary complications include those with preexisting pulmonary disease, [122,131,132]upper abdominal and thoracic incisions, [124]advanced age, [133]obesity, [134]and those in severe pain. [135].

Choice of anesthetic technique affects the degree of postoperative pulmonary dysfunction. Use of general anesthesia may briefly exacerbate surgery-induced pulmonary dysfunction. [136]Mechanical ventilation, paralysis, inhaled anesthetics, and opioids all contribute to reduce pulmonary function. [137]In contrast to the exacerbation of postoperative pulmonary dysfunction seen with general anesthesia, TEA has minimal effect on pulmonary function [138-140]and may offset detrimental changes in pulmonary function induced by general anesthesia. [132]Observation of differential effects on pulmonary function between general and epidural anesthesia appears to depend on the extent of surgery. For example, laparoscopic cholecystectomy results in little postoperative pulmonary dysfunction, and thus epidural anesthesia provides little benefit after this operation. [124]Finally, most of the general anesthesia-related pulmonary dysfunction reverts to baseline early in the postoperative period and may have minor effect on postoperative pulmonary morbidity. [137]Thus, intraoperative epidural anesthesia appears to offer relatively minor advantages in preservation of early postoperative pulmonary function.

On the other hand, previous studies suggest that postoperative use of epidural analgesia has the potential to reduce pulmonary morbidity by providing better analgesia, improved diaphragmatic function, and reduced frequency and severity of postoperative hypoxemia. For example, epidural analgesia using local anesthetic, [124,135]opioid, [12,141]and especially both [142-144]provides better analgesia than systemic opioid, including delivery by intravenous PCA. [12,141]Because relief of pain improves pulmonary function after abdominal or thoracic surgery, [121,145]it is not surprising that use of epidural analgesia is associated with improved pulmonary function when compared with intramuscular and intravenous opioid. [132,134-136,146,147]That not all studies demonstrating superior analgesia in the epidural group also demonstrate improvements in pulmonary function [12,148,149]suggests that factors other than pain are also important etiologic factors for postoperative pulmonary impairment.

For example, diaphragmatic function is impaired after abdominal or thoracic surgery and may also contribute to pulmonary dysfunction. [150]Diaphragmatic dysfunction appears to result from reflex inhibition of phrenic nerve activity [125,145,150]and is not significantly altered by relief of pain. [151-153]Thus, analgesia provided with parenteral or epidural administration of opioids alone does not produce appreciable improvement in diaphragmatic function. [151]On the other hand, thoracic epidural blockade with local anesthetic may improve postoperative diaphragmatic function. This improvement in function probably results from neural blockade of the inhibitory reflex [125,145]and perhaps through changes in chest wall compliance. [152,153]Thus, epidural administration of local anesthetics for postoperative analgesia can result in improved pulmonary function in part through relief of pain but also by improving postoperative diaphragm function.

Pulmonary dysfunction may also produce morbidity through hypoxemia. Episodic hypoxemia is common in the postoperative period, particularly during sleep. [40,154-156]Postoperative hypoxemia serves as a marker for patient's with postoperative pulmonary morbidity [154]and is associated with myocardial ischemia. [40]Analgesia with either parenteral or epidural opioid is associated with equally high incidences of episodic postoperative hypoxemia. [155,157]However, use of epidural analgesia with local anesthetics is associated with a reduced incidence and severity of hypoxemia in the early postoperative period [6]perhaps because of improved pulmonary function or reduced sedation. [136].

In summary, postoperative pulmonary dysfunction can be attenuated by the intraoperative use of epidural local anesthetics. More important, continuation of epidural analgesia with local anesthetic into the postoperative period may maintain improvements in postoperative pulmonary function. These effects may be attributable to to the benefits in addition to analgesia provided by epidural administration of local anesthetics, such as limitation of the degree of postoperative diaphragmatic dysfunction, improvement of abdominal or chest wall compliance, and limitation of episodes of postoperative hypoxemia.

Clinical Effects. Although previous studies have demonstrated reduced incidence of radiologic markers of pulmonary morbidity in patients receiving epidural anesthesia and analgesia as compared with general anesthesia followed by intramuscular opioid, [158-160]effects of epidural analgesia on serious pulmonary complications remain unclear (Table 3). Several studies noted improved clinical outcome in patients receiving epidural analgesia in terms of decreased incidences of postoperative pneumonia and respiratory failure (Table 3) [73,88,134,135]Studies that observed benefit from epidural anesthesia and analgesia examined high-risk patients (e.g., obese patients undergoing upper abdominal surgery), used intraoperative epidural anesthesia (with or without general anesthesia), and continued epidural analgesia (local anesthetic or opioid) into the postoperative period.

On the other hand, several studies have failed to show an advantage of epidural analgesia over parenteral analgesia in incidence of pulmonary morbidity (Table 3). [16,76,91,148,149]Studies not observing differences between epidural anesthesia and analgesia versus general anesthesia and systemic opioids either used healthy patients, did not study high-risk operations, or did not control postoperative analgesia. Thus, it appears that epidural anesthesia or analgesia may be beneficial only in patients at high risk for postoperative pulmonary complications and also may be observed only when adequate epidural analgesia is maintained into the postoperative period.

Postoperative ileus is a temporary impairment of gastrointestinal motility that occurs after surgery. Although most common and severe after major abdominal procedures, ileus also occurs after peripheral operations, general trauma, or other stressful situations. [161,162]Ilieus delays resumption of an enteral diet, and this delay may contribute to postoperative morbidity. For example, early enteral feeding has been shown to reduce the surgical stress response, reduce postoperative septic complications, and improve wound healing. [163-166]The financial costs of postoperative ileus are substantial and have been estimated at $750,000,000 annually. [161]Ileus is thus a major surgical morbidity. Anesthetic or analgesic techniques that speed recovery from or prevent postoperative ileus have the potential to substantially reduce postoperative morbidity, duration of hospitalization, and costs of surgical care.

Experimental Effects. Postoperative ileus affects all segments of the gastrointestinal tract with variable duration and intensity. [167,168]Motility in the stomach and small intestines generally recovers within 24 h after abdominal surgery, [169,170]whereas colonic motility is inhibited for 48-72 h. [169,171]Although the pathophysiologic features of postoperative ileus remains ill defined, [161,167]the most commonly accepted theory is that abdominal pain activates a spinal reflex arc that inhibits intestinal motility. [172]In addition, surgical stress induces sympathetic hyperactivity, and excessive sympathetic stimulation of the bowel inhibits organized propulsive activity. [161,173]Thus, both nociceptive afferent and sympathetic efferent nerves are believed to be key initiators of ileus. Although stimulation of sympathetic nerves inhibits contractile activity, [174,175]therapeutic interventions with adrenergic blocking drugs do not speed resolution of ileus. [161,176]In contrast to the lack of efficacy of adrenergic-blocking drugs, section of sympathetic innervation pathways to the gastrointestinal tract can block ileus after abdominal surgery. [177]This observation provides a theoretical framework suggesting that blockade of abdominal nociceptive afferent or sympathetic efferent pathways might abolish inhibition of gastrointestinal motility induced by abdominal surgery.

Epidural local anesthetics can theoretically improve bowel motility through both of these mechanisms. Blockade of nociceptive afferent pathways may disrupt the afferent limb of the spinal reflex arc that has been postulated to mediate postoperative ileus. [172]Furthermore, epidural local anesthetics can block the efferent limb of the reflex by blocking thoracolumbar sympathetic efferent nerves. [170,177,178]Of note, activation of gastrointestinal parasympathetic nerves stimulates propulsive contractions, and parasympathetic innervation of the colon is provided by the vagus nerve and pelvic nerves originating from the sacral region of the spinal cord (Figure 7). This anatomic arrangement should make TEA with local anesthetics particularly beneficial. Segmental neural blockade of thoracic dermatomes with local anesthetics will selectively block nociceptive afferent and sympathetic efferent nerves while leaving parasympathetic innervation (by the vagus and pelvic nerves) intact. The resulting shift in autonomic balance toward a relative increase in parasympathetic tone during TEA should increase propulsive activity in the colon.

Another means whereby epidural local anesthetics may increase gastrointestinal function is through systemic absorption of the epidurally administered local anesthetic. For example, intraperitoneal or intravenous administration of local anesthetic speeds return of propulsive motility in the colon and shortens the duration of postoperative ileus. [179,180]Purported mechanisms for systemic effects of local anesthetics on the gastrointestinal system include a direct excitatory effect on intestinal smooth muscle, reduction of opioid dose requirement, and blockade of the inhibitory spinal reflex arc. [180]Thus, in addition to the potential beneficial effects resulting from blockade of nociceptive afferent and sympathetic efferent fibers, it appears that systemic absorption of local anesthetic can also contribute to reducing the duration of postoperative ileus.

A final potential benefit of the local anesthetic induced sympathectomy is the associated increase in gastrointestinal blood flow. [181,182]Because blood flow to the bowel is a critical factor for gastrointestinal motility [183]and for healing of gastrointestinal anastomoses, [184]some authors have suggested that increased colonic blood flow resulting from the sympathetic blockade should result in reduction of ileus [183]and improved healing of surgical anastomoses. [185]In contrast, other authors have proposed that epidural analgesia with local anesthetics can lead to excessive propulsive activity causing dehiscence of colonic anastomoses. [186,187]However, this concern is based on a few case reports, is inconsistent with animal studies that demonstrate excellent healing of anastomoses, [188]and is refuted by larger clinical studies that actually demonstrate a modest trend toward a lower incidence of dehiscence. [185,189]Thus, epidural analgesia increases intestinal blood flow that should theoretically enhance postoperative recovery of bowel motility and promote healing of gastrointestinal anastomoses.

In contrast to local anesthetics, epidural opioids may not block transmission in somatic or sympathetic nerves [190]and may thus be intrinsically less effective in reduction of ileus than local anesthetics. Furthermore, epidural opioids may directly inhibit gastrointestinal motility. For example, it is well known that systemic administration of morphine delays gastric emptying and impairs colonic transport. [176,191,192]Similarly, animal and volunteer studies suggest that actions of opioids in the spinal cord inhibit motility in both the proximal and distal colon [193,194]and prolong gastric and intestinal transit time to a similar degree as parenteral opioids. [170]On the other hand, epidural opioids may be advantageous during recovery from ileus. During the 1st 2 days after abdominal operations, opioids have no detectable effect on colonic myoelectric activity. [195]However, beginning on the 3rd day after surgery (when colonic function begins to recover) systemic administration of opioids inhibits propulsive myoelectric activity. During the same time, epidural administration of morphine does not inhibit bowel motility. [195]Thus, it appears that systemic opioid administration may contribute to and prolong postoperative ileus, whereas the effects of epidural opioid on colonic motility remains unclear.

In summary, experimental evidence suggests that epidural local anesthetics can reduce postoperative ileus through relief of pain, systemic absorption of the local anesthetic, blockade of sympathetic innervation of the bowel, and reduction of requirements for systemic opioids.

Clinical Effects. Numerous clinical studies have demonstrated more rapid recovery of bowel function when postoperative analgesia is provided with epidural local anesthetics rather than systemic opioids (Table 4). For example, postoperative TEA with bupivacaine resulted in lower pain scores and earlier passage of flatus, and earlier bowel movements than was observed with systemic opioid analgesia. [149,187,196,197]Admittedly, time to passage of flatus or first bowel movement may not be the most sensitive markers of gastrointestinal function. [198]However, more sensitive markers of gastrointestinal propulsive activity such as transit of barium contrast [199]and radiopaque markers, [200]and propulsive gastrointestinal myoelectric activity [201]also recover more rapidly when postoperative analgesia with epidural local anesthetic was compared with systemic opioid administration. Thus, studies using a variety of assessment techniques have consistently observed more rapid recovery from postoperative ileus when postoperative analgesia is provided with epidural local anesthetics.

In contrast, one major study found that epidural bupivacaine did not improve recovery of bowel function when compared with systemic opioids (Table 4). However, this study administered epidural local anesthetic for only 24 h and then provided analgesia with parenteral or epidural opioids during the time that gastrointestinal function was beginning to recover (2-4 days after surgery). [91,198]Thus, it seems likely that early discontinuation of epidural local anesthetic administration may account for these negative findings. If so, these results suggest that epidural local anesthetics should be administered a minimum of 2 to 3 days after surgery to provide a beneficial effect on recovery from postoperative ileus.

Finally, few studies have directly compared epidural administration of opioids with epidural local anesthetics [197,202,203](Table 5). However, these studies indicate that epidural local anesthetics may provide greater benefit than epidural opioids. For example, postoperative analgesia provided with epidural bupivacaine was associated with more rapid return of gastric emptying [204]and shorter duration until passage of flatus and feces than when analgesia was provided with epidural morphine. [197,203]In contrast, one study did not demonstrate a benefit for epidural local anesthetics. Of note, patients in this study had a particularly prolonged postoperative ileus (Table 5), which outlasted the duration of epidural analgesia by an average of more than 2 days. Thus, the vast majority of patients in this study were receiving systemic opioids during the time that bowel function was returning to normal, and this may explain the lack of benefit observed in this study. On balance, these results suggest that epidural analgesia with local anesthetics may speed resolution of postoperative ileus when compared with epidural opioids.

In summary, clinical studies consistently demonstrate that postoperative administration of epidural local anesthetics can reduce the duration of postoperative ileus. Although use of epidural analgesia with local anesthetics may attenuate the decrease in bowel motility that typically occurs during postoperative ileus, bowel motility does not return to the presurgical baseline after peripheral [162]or abdominal surgery. [205]These results suggest that factors other than pain and sympathetic hyperactivity also contribute to postoperative ileus. Epidural opioids may also improve recovery of gastrointestinal function when compared with parenteral opioids but more data are needed to make definitive conclusions. The benefit of local anesthetics may result from the ability not only to block pain pathways but also to block sympathetic innervation of the gastrointestinal tract. To produce a benefit reliably, it appears that epidural analgesia (with local anesthetics) should be continued until postoperative ileus is resolved (typically 48-72 h after surgery). Despite earlier recovery of bowel function, epidural analgesia has led to a shorter duration of hospitalization in only one study. [134]The lack of improvement in patient outcome may have occurred because convalescence protocols were not adjusted to take advantage of the earlier recovery of bowel motility. For example, none of the studies mention a controlled effort to remove nasogastric tubes or feed patients any earlier. A recent study supports this contention and suggests that postoperative analgesia provided with a combination epidural local anesthetic and opioid can speed recovery from colon surgery when used in conjunction with an aggressive convalescence program. [206]These preliminary results suggest that epidural analgesia may contribute not only to an earlier recovery from postoperative ileus but also to a shorter duration of hospitalization.

Surgical stress consistently elicits a metabolic response by activation of the sympathetic and somatic nervous system and through local trauma. [207]Responses to surgical stress include release of neuroendocrine hormones [208]and local release of cytokines (Figure 8). [207]Effects of this stress response may be detrimental: infusions of neuroendocrine hormones and cytokines provoke tachycardia, fever, shock, and increased minute ventilation. [209-211]These effects are dose related, and serum concentrations of these factors correlate with severity of injury and with outcome after injury. [92,209,212]Thus, the concept has arisen that inhibiting the stress response may improve surgical outcome.

Experimental Effects. The stress response can not be prevented, and may be exacerbated, by administration of general anesthetics with the exception of large doses of opioids (e.g., 50-100 micro gram/kg fentanyl) [213-216]before incision. [217]In contrast, epidural administration of either local anesthetics or opioids can blunt the surgical stress response. [3,98,218-220]Indeed, intraoperative use of epidural anesthesia with local anesthetics completely suppresses the stress response resulting from procedures below the umbilicus. [220,221]However, complete suppression of the stress response requires complete sympathetic and somatic blockade of the surgical site such as can be provided with extensive epidural anesthesia (T4-S5 to pinprick). [222]Although less extensive blockade provides anesthesia, the stress response is incompletely blocked because of incomplete suppression of sympathetic and somatic signals. [98,222,223]Thus, pain relief and prevention of the stress response do not appear to be directly coupled, and analgesia alone may not guarantee that activation of the stress response will be prevented. Further support for this concept is found when the effects of epidural opioid analgesia are considered. Virtually all opioids commonly used for epidural analgesia can attenuate the stress response. [207]However, regimens of epidural opioids that produce equivalent postoperative analgesia to epidural local anesthetics produce a less effective block of the stress response. [218,224]The relative lack of efficacy of opioids is consistent with the concept that nociceptive pathways are only partially responsible for activation of the stress response. [214,218,225]In other words, the greater efficacy of epidural local anesthetics for inhibition of the stress response may be attributable to the ability to block nociceptive and nonnociceptive pathways, such as the sympathetic innervation to the adrenal glands, whereas epidural opioids may modulate only nociceptive pathways (Figure 8). [190]Thus, previous studies indicate that epidural administration of local anesthetics can completely suppress the stress response after procedures below the umbilicus, whereas epidural opioids only lessen the stress response.

In comparison with procedures below the umbilicus, epidural anesthesia has less effect on the stress response resulting from procedures above the umbilicus. Epidural local anesthetics reduce but do not completely prevent the stress response from these operations and do not have greater efficacy than epidural opioids. [214,224,226-228]There are three likely factors for the incomplete effects of epidural anesthesia on the stress response from procedures above the umbilicus. First, doses of local anesthetic used in previous studies were unlikely to produce complete neural blockade of the surgical site. [229]Because the degree of neural blockade can be correlated with magnitude of the surgical stress response, [229,230]it is possible that greater doses of local anesthetic may be needed to more effectively inhibit the stress response from procedures above the umbilicus. Second, the type of local anesthetic used for epidural anesthesia can also affect the efficacy of attenuation of the stress response. [231]Third, some components of the stress response (cytokines) are directly released into the bloodstream by local trauma and injury (Figure 8). [232]These mediators directly activate the stress response and are released in greater quantities after surgery above the umbilicus. [223,233,234]Thus, dose-related differences in extent and density of neural blockade, choice of local anesthetic, and local release of stress response mediators may explain incomplete effects of epidural local anesthetics on the stress response for surgical procedures above the umbilicus.

Either intra- or postoperative administration of epidural local anesthetics and opioids can lessen the stress response. [98,218,224,235,236]However, only preincisional establishment of epidural anesthesia with local anesthetics can prevent the stress response and maintain mediator concentrations at preoperative values. [235]Once the stress response is initiated, postincisional administration of epidural anesthesia can only attenuate the response. Furthermore, epidural analgesia with local anesthetics or opioids should be continued into the postoperative period to maximally reduce the stress response. [220]Use of epidural analgesia in the postoperative period may be critical, as maximal increases in stress response occur immediately after surgery [237]and may continue for as long as 5 days postoperatively. [92,238]Although the necessary duration of epidural analgesia for maximal benefit is unknown, use of epidural analgesia with local anesthetic for only 24 h can suppress the catabolic component of the stress response (nitrogen balance, muscle amino acid composition) for as long as 5 days. [239-241]In contrast, use of intravenous PCA for postoperative analgesia relieves pain, but the stress response is unaltered. [242]Thus, use of epidural anesthesia followed by epidural analgesia with local anesthetic results in the greatest observed reduction of the perioperative stress response.

In summary, epidural local anesthetics or opioids effectively inhibit the stress response from surgical procedures, especially below the umbilicus. However, the greatest suppression of the stress response has been observed after epidural anesthesia with local anesthetics followed by postoperative epidural analgesia with local anesthetics. Thus, epidural local anesthetics appear to have greater efficacy than opioids in reduction of the stress response, perhaps because opioids and local anesthetics have different mechanisms of action. Opioids produce analgesia by modulating nociceptive pathways within the central nervous system, whereas local anesthetics can nonspecifically block both nociceptive and nonnociceptive pathways. The superior efficacy of local anesthetics for blunting the stress response supports the concept that relief of pain is only partially responsible for effects of epidural analgesia.

Clinical Effects. The stress response initiates potentially detrimental physiologic responses (Figure 8), peaks during the postoperative period, and is temporally associated with peak episodes of postoperative morbidity. [89,243]Thus, a popular theory is that increases in the stress response are an important marker for patients at increased risk for adverse outcome. Although there is little evidence that the stress response per se results in morbidity, several potentially detrimental physiologic effects are modulated through the stress response. For example, cardiac morbidity may be increased through release of neuroendocrine hormones with resultant reductions in myocardial oxygen supply or increases in demand. [43]In addition, the stress response enhances development of a hypercoagulable state that predisposes to vascular thrombosis both through enhancement of coagulation [99,244-246]and inhibition of fibrinolysis. [247]Furthermore, stress response mediators are potent inhibitors of the immune system [248]and may contribute to postoperative immunosuppression and infection. This concept is addressed in the following section. Thus, modulation of the surgical stress response may theoretically affect rates of morbidity in multiple organ systems.

Only a few studies have examined both stress response and postoperative morbidity in well-conducted comparisons of general anesthesia versus epidural anesthesia and analgesia. Interestingly, these studies have observed parallel movements in these measurements. [73,89,91,98]For example, Yeager et al. [73]observed marked decreases in cardiac morbidity in patients receiving epidural anesthesia and analgesia (Table 1). Urinary cortisol excretion was measured as a marker of the stress response and was significantly less in the epidural group during the first 24 h after surgery. Christopherson et al. [89,98]observed a marked increase in patency of vascular grafts in patients randomized to epidural anesthesia followed by epidural fentanyl. Plasma catecholamines were measured as markers for the stress response and were significantly decreased in the epidural group during the postoperative period. In contrast, Hjortso et al. [91]were unable to identify differences in morbidity (Table 3) between patients randomized to general anesthesia followed by on demand "systemic" morphine or general anesthesia combined with epidural anesthesia followed by epidural bupivacaine for 24 h and epidural morphine for 72 h. Serum concentrations of albumin and transferrin were measured as markers of the stress response and also did not differ between groups. Although these observations are provocative and a theoretical framework linking reduction of the stress response with reduction of cardiac and vascular morbidity exists, data are currently insufficient to establish a clear relation. More studies are needed to determine if a relation exists between the stress response and postoperative morbidity before the clinical importance of reduction of the stress response with epidural anesthesia and analgesia can be determined.

Both cellular and humoral immune function are suppressed after surgery. [248,249]Although the etiology of postoperative immunosuppression is unclear, many known mediators of the stress response are potent immunosuppressants. [250]Postoperative immunosuppression typically lasts for several days, may last longer in inherently immunosuppressed patients (i.e., those with malignant disease or acquired immunodeficiency syndrome), [251-253]and may predispose to the development of postoperative infections [254,255]or facilitate postoperative tumor growth and metastases. [256-259]General anesthetics (with the exception of large doses of opioids) can not suppress the stress response and may exacerbate postoperative immunosuppression by depression of cellular and humoral immune function. [260-264]For example, immunosuppression resulting from general anesthetics occurs within 15 min after induction [265]and may persist for as long as 3-11 days. [261,266]However, the clinical importance of general anesthesia induced or stress response mediated immunosuppression remains uncertain.

Experimental Effects. Although high concentrations of local anesthetics directly inhibit leukocyte activity, [267,268]epidural anesthesia followed by epidural analgesia with local anesthetics has been associated with a modest preservation of cellular and humoral immune function, especially for procedures below the umbilicus. [265,266,269,270]Preservation of immune function continues beyond the duration of epidural analgesia and is thought to result from attenuation of the stress response. [266]Addition of general anesthesia to epidural anesthesia does not appear to antagonize preservation of immune function, provided epidural analgesia is maintained into the postoperative period. [269]Finally, low plasma concentrations of amide local anesthetic, such as are seen after epidural administration, have antiinflammatory properties, [271]prevent release of tissue damaging superoxide anions and lysosomal enzymes, [272-274]and may therefore reduce wound infection through enhanced healing of tissues. [271,274,275]Thus, use of epidural anesthesia followed by epidural analgesia with local anesthetics may aid in preservation of perioperative immune function that persist beyond the duration of epidural analgesia.

Postoperative immunosuppression can promote tumor growth [276]and therefore may affect outcome after oncologic surgery. Surgery and the resultant stress response leads to suppression of natural killer (NK) cell function, [277-279]which is associated with increased tumor growth and metastases. [280,281]Even transiently immunosuppressive interventions such as blood transfusions may adversely affect outcome after oncologic surgery. [282]Thus, anesthetic and analgesic choice may also affect outcome after oncologic surgery. General anesthetics directly suppress NK cell function, [261,283]and administration of general anesthetics in animal models leads to depression of NK cell activity and increased growth and spread of inoculated tumor cells. [256,276,284]On the other hand, epidural anesthesia followed by postoperative epidural analgesia with local anesthetics attenuates stress response induced depression of NK cell function, does not directly suppress NK cell function, [266]and thus represents a theoretical means to reduce perioperative tumor growth and improve patient outcome.

Clinical Effects. The immune system is complex and difficult to assess in the perioperative period. Individual tests of cellular and humoral immune function may have little reflection on the immune system's total ability to eradicate infections and neoplasms. [264]Nevertheless, clinical studies suggest that use of epidural anesthesia followed by epidural analgesia with local anesthetics may be associated with decreased incidences of postoperative infectious complications (sepsis, wound infections, and pneumonia). [73,88,91,135,149]Studies observing a beneficial effect have consistently examined high-risk patients undergoing major surgery and have used epidural anesthesia followed by adequate postoperative epidural analgesia with local anesthetics. Currently, there are no clinical studies evaluating the influence of choice of anesthesia and analgesia on outcome after oncologic surgery. However, anesthetic and analgesic techniques may decrease the magnitude of postoperative immunosuppression and may thus affect outcome after oncologic surgery.

Transient postoperative reduction in cognitive function is common in all patients and is poorly understood. [285]Reduction in cognitive function may be especially common (10-50% incidence) and severe in the elderly. [286,287]The nadir of cognitive function typically occurs on the 2nd postoperative day with full recovery occurring within 1 week. [288]However, elderly patients may require months to return to full preoperative mental capacity. [289]Furthermore, impaired cognitive function in the elderly is associated with increased incidence of postoperative complications such as depression, decubitus ulcers, stroke, urologic complications, and delayed return to independence, all of which may lead to longer hospitalization. [286]Purported causes of postoperative cognitive dysfunction include episodes of hypoxemia, use of psychoactive medication, and preoperative affective depression. [286,290,291]However, correlation with these suggested etiologies has been inconsistent, [292]and there are no definite links between perioperative physiologic perturbations and reduction in cognition. Thus, attempts to reduce the severity of postoperative cognitive dysfunction have been hampered by lack of a definite etiology.

Initial studies comparing general and epidural anesthesia demonstrated dramatic improvement in postoperative mental status with use of epidural anesthesia in patients undergoing total hip arthroplasty. [290]However, observers were not blinded, and objective tests of psychological function were not used. In fact, subsequent studies have failed to demonstrate differences in adverse cognitive sequelae. Studies randomizing patients to epidural anesthesia, general anesthesia, and combined epidural-general anesthesia have assessed postoperative cognitive function over a 3-month period by using tests such as psychological scales, organic brain syndrome scales, and batteries of memory, psychomotor, and skilled performance tests. [286,288,293,294]These well-designed studies consistently demonstrated similar severity of postoperative cognitive dysfunction in patients regardless of anesthetic selection.

However, none of these studies addressed the effects of postoperative analgesia on cognitive function. Because the nadir in cognitive function typically occurs on the 2nd postoperative day, it seems likely that the choice of postoperative analgesic technique rather than intraoperative anesthetic technique may affect cognition. Although there are no well-conducted studies examining effects of postoperative analgesia on cognition, previous studies have demonstrated less sedation in patients receiving epidural analgesia (local anesthetic or opioid) than those receiving parenteral opioid. [136,236]No objective psychomotor tests were performed, but these data suggest potential for reduction in postoperative cognitive dysfunction from use of epidural analgesia.

In summary, patients may experience impaired cognitive function and inability to perform activities of daily living after surgery. Although the etiology remains uncertain, reduction in cognitive function may last longer and be especially severe in the elderly, intraoperative anesthetic technique does not appear to affect incidence or severity of postoperative cognitive dysfunction. On the other hand, cognitive function reaches its nadir in the postoperative period, and effects of postoperative use of epidural analgesia have not been evaluated.

Perioperative hypothermia occurs frequently and is associated with both adverse and beneficial physiologic effects. [295,296]Hypothermia may be caused by several factors including low operating room ambient temperature, [297]heat loss during skin preparation and surgical exposure, [298]and use of anesthetics. [297]Both general and epidural anesthesia are associated with impaired thermoregulation and increased heat loss. [299,300]Epidural anesthesia with local anesthetics induces several thermoregulatory changes that lead to hypothermia such as increased cutaneous blood flow, decreased input from cutaneous thermal receptors, and redistribution of central heat. [301]In addition, augmentation of general anesthesia with epidural anesthesia appears to lead to greater degrees of hypothermia than use of general anesthesia alone. [302]Surprisingly, epidural analgesia has also been associated with otherwise unexplained hyperthermia in women during labor. [303]Thus, effects of epidural analgesia on human thermoregulation are complex and incompletely understood.

Several clinical studies have attempted to determine whether use of epidural or general anesthesia resulted in greater incidences of perioperative hypothermia. The results from these studies are conflicting, as greater incidences have been found with epidural anesthesia, [304,305]with general anesthesia, [297,306]or no difference has been found. [307]Although definite conclusions can not be drawn from these results, perioperative hypothermia can be effectively prevented (e.g., by passive insulators and active warming systems). [308-310]Therefore, careful perioperative maintenance of patient temperature can effectively prevent hypothermia and may make thermoregulatory preservation a minor consideration in selection of type of anesthesia.

Potential beneficial physiologic effects of epidural anesthesia and analgesia are accompanied by risks for potential complications. Complications from epidural analgesia may be separated into complications from placement of epidural catheters and complications from use of local anesthetics and opioids.

The most common complication from placement of epidural catheters is accidental dural puncture with resultant post-dural puncture headache. Recent large surveys encompassing 51,000 epidural catheter placements performed during the past 3 decades suggest that the incidence of dural puncture is 0.16-1.3%. [311,312]Subsequent development of post-dural puncture headache depends on several factors [313]and ranges from 16% to 86%. Less frequent complications are paresthesias and neurologic injury. The incidence of these complications is likely 0.01-0.001%. [311,314]Most of the neurologic injuries were self-limited and did not require treatment. [314].

An extremely rare but potentially disastrous complication of epidural analgesia is paraplegia. Paraplegia associated with epidural analgesia appears to result most commonly from formation of epidural hematoma. [315]Prompt recognition and surgical decompression of symptomatic epidural hematoma can prevent permanent neurologic injury. [316]Puncture of epidural vessels during placement of epidural catheters occurs in approximately 3-12% of cases. [317]Although subsequent formation of asymptomatic hematomas may be common, [318]formation of symptomatic epidural hematoma is a rare event. The incidence of symptomatic epidural hematoma associated with epidural analgesia is difficult to estimate, but combined case series of more than 100,000 epidural anesthetics have been reported without a single epidural hematoma. [319]Currently, the risk of formation of epidural hematoma after epidural analgesia appears to be increased if patients are receiving anticoagulant agents or have a coagulation disorder. [317,319,320]The majority of case reports of paraplegia resulting from epidural hematoma (with or without epidural catheters) were associated with these factors. [314-316,319,321]However, use of anticoagulants may not be an absolute contraindication to epidural analgesia. In fact, several large series have documented safe use of epidural analgesia in patients receiving antiplatelet therapy, [320,322,323]warfarin sodium, [324]and high- and low-dose heparin. [317,320,325-327]In fact, current clinical trials are examining the role of TEA in patients undergoing coronary bypass surgery. [328]These patients have epidural catheters placed preoperatively and undergo complete anticoagulation with heparin during surgery. Thus, with the possible exception of thrombolytic agents, [329]the risk of formation of symptomatic epidural hematoma in patients receiving anticoagulants may be small. Currently, it remains unclear what degree of laboratory abnormality identifies patients at increased risk [330-332]and when to remove epidural catheters if a serious coagulation disorder develops. [323,329,333]Other causes of paraplegia associated with epidural analgesia are probably even more rare and include epidural abscess, [334,335]epidural air from loss of resistance technique, [336]and preexisting medical conditions. [337].

Finally, recent surveys of more than 18,000 epidural anesthetics performed at university teaching hospitals suggest that placement of epidural catheters at thoracic vertebral levels does not confer higher risk than placement at lumbar levels. [311,338]In fact, incidences of complications resulting from placement of epidural catheters at thoracic levels were comparable to or lower than complications at the lumbar levels. Thus, risks of complications from placement of epidural catheters are small and are not higher for thoracic levels as compared with lumbar levels.

Administration of epidural local anesthetics for surgical anesthesia frequently results in multiple hemodynamic changes including decreased chronotropy, inotropy, dromotropy, systemic vascular resistance, cardiac output, and myocardial oxygen consumption. [339,340]Hemodynamic effects from epidural administration of local anesthetics depend on the extent of neural blockade, intravascular volume status of the patient, and use of epinephrine containing local anesthetic solutions. [340,341].

Accidental intravascular or intrathecal injection of local anesthetics may cause complications including dysrhythmias, cardiovascular collapse, or high spinal anesthesia. [342,343]Use of a test dose of local anesthetic with epinephrine [344]and incremental rather than bolus injection of local anesthetics [345]may help decrease the incidence of these complications but does not eliminate the risk of complications. [346].

Complications from low concentrations of local anesthetics used for postoperative analgesia are also common. For example, postoperative use of concentrations of bupivacaine greater than 0.1% for LEA has been associated with high incidences (37-80%) of motor and autonomic blockade. [159,347]Complications resulting from motor and autonomic blockade include hypotension, urinary retention, impaired ambulation, and pressure necrosis. [347-349]The use of lower concentrations of bupivacaine such as 0.015-0.08% has been reported to result in lower incidences of complications (0-50%). [350,351]Furthermore, preliminary studies suggest that vertebral level of epidural catheter placement may significantly affect rate of local anesthetic complications (difficulty ambulating and orthostatic hypotension). [352,353],* These studies demonstrated that TEA is associated with less sympathetic block of the lower extremities than LEA. [352]Furthermore, use of thoracic epidural catheters for postoperative analgesia was associated with lower rates of complications than lumbar epidural catheters (2.5% vs. 20%) * and with equivalent rates of complications as parenteral morphine. [353]These findings are consistent with the segmental nature of epidural administration of local anesthetics. [352]TEA has relatively little effect on lumbar and sacral areas, [352]and these areas can initiate compensatory vasoconstriction. [78,354]Thus, preliminary results suggest that complications resulting from postoperative epidural analgesia with local anesthetics may be reduced when epidural catheters are inserted at a thoracic level.

The most feared complication from epidural administration of opioids is delayed respiratory depression. The risk of respiratory depression after central neuraxial administration of opioids appears to be dose-dependent. [355,356]Although virtually all opioids commonly used for epidural analgesia have been reported to cause respiratory depression, [356]morphine is associated with the highest risk of delayed respiratory depression. Nonetheless, the incidence of delayed respiratory depression after epidural administration is small. Combined case surveys including more than 25,000 patients suggest that the incidence of delayed respiratory depression from epidural morphine is less than 1%. [356,357]This low incidence compares favorably with reported incidences of 0.9% for severe respiratory depression associated with oral and parenteral morphine. [358]Furthermore, several surveys have reported that use of epidural morphine in patient wards is safe, provided that adequate nursing monitoring is available. [357,359]Thus, risk of delayed respiratory depression after epidural opioids is small, and it appears that epidural opioids can be safely used on hospital wards.

Other complications from administration of epidural opioid include pruritus (reported incidences range from 28% to 100%), nausea (30-100%), and urinary retention (15-90%). [356,358,360-363]Incidences vary depending on specific opioid used, but in general, morphine appears to have the highest incidences of these side effects, whereas fentanyl has the lowest. It remains unclear whether these side effects are dose related. [364,365].

Finally, although analgesic synergism may exist between epidural local anesthetics and opioids, [14,15]it remains unclear whether combinations of local anesthetics and opioids reduce the incidences of respiratory depression, pruritus, nausea, and urinary retention. For example, addition of epidural bupivacaine in concentrations as low as 0.01% can reduce epidural fentanyl usage by 50% in patients receiving postthoracotomy epidural analgesia. [366]Use of small doses of local anesthetic and opioid to provide combination analgesia may lead to lower incidences of complications. [367]However, only one study has documented a reduction in side effect (pruritus) by combining local anesthetic and opioid. [368]Thus, further studies are necessary to identify optimal combinations of local anesthetics and opioids for provision of optimal analgesia, minimal complications, and reduced incidences of morbidity.

In summary, use of epidural anesthesia and analgesia may provide potential benefits but is also associated with potential detrimental effects. Currently, we have inadequate data to determine risk-benefit ratios; however, serious risks associated with epidural anesthesia and analgesia appear to be low. As in all of our practice, we must carefully weigh potential risks and benefits for each patient before initiating epidural anesthesia and analgesia.

Health care consumes more than 12% of our gross national product, [369]and it appears that the United States is moving toward some form of health care reform. Competition for health care resources will increase, [370]and medical practices will undergo greater scrutiny regarding cost effectiveness by both internal and external regulators. [371]Thus, our specialty must take the initiative in examining cost effectiveness of our practice, [369,372-374]especially in new areas such as acute pain management.

Currently, fewer than 2% of published anesthesia literature include useful cost information. [372]Furthermore, cost effectiveness analyses are rarely performed. For example, Yeager et al. demonstrated a lower average hospital cost ($11,218 vs. $20,380) for high-risk patients receiving epidural anesthesia and analgesia as compared with general anesthesia. [73]However, actual cost savings may not be as large, as several direct and indirect costs are unknown. For example, direct costs of postoperative analgesia are unknown (equipment and services) and differ markedly among institutions. [359,375]Furthermore, indirect costs such as out-of-hospital costs (including home nursing, home medications, and days away from work) or expense of treating complications resulting from provision of epidural analgesia are also unknown. Finally, distribution of costs may be especially important. Yeager et al. [73]studied only high-risk patients and demonstrated impressive decreases in morbidity and cost. However, cost savings may be substantially diluted by provision of epidural analgesia to low-risk groups where there is less potential to decrease morbidity. Thus, direct costs, indirect costs, and distribution of costs can all affect cost effectiveness.

Correspondingly, long-term benefits may also magnify cost effectiveness of an intervention. For example, it is likely that the reduction of morbidity and mortality seen in Yeager's study added years of life and improved quality of life in the epidurai group. These long-lasting benefits gained through use of epidural anesthesia and analgesia may outweigh the total cost of providing epidural analgesia. However, these data are difficult to collect, [376]and there is neither a standard tool for measurement nor a consensus regarding relative weighting of outcome variables. [377].

In summary, although cost effectiveness of epidural analgesia remains unknown, potential long-term benefits from reduction in morbidity associated with epidural analgesia may outweigh total costs. Further studies are needed to identify the full extent of potential benefits from epidural analgesia, to identify optimal techniques, and especially to identify populations that receive the greatest benefit. Based on this information, our specialty can make rational decisions regarding cost-effective postoperative analgesic therapy.

Many studies have observed better pain relief with epidural analgesia than with systemic opioids. [12,91,378]Improved analgesia is particularly evident when local anesthetics and opioids are combined and analgesia is assessed during activity. [15]However, the ability of epidural analgesia to significantly alter other clinical outcomes is less easily defined. Our review indicates that use of epidurai anesthesia and analgesia may be associated with reductions in incidence and severity of many perioperative physiologic perturbations. Whether these effects will translate into improved patient outcome remains to be established. Preliminary evidence suggests that the perioperative coagulability can be reduced with epidural analgesia and this may significantly reduce the incidence of venous and arterial thromboses. [88,89,97]Similarly, significant improvements in gastrointestinal motility documented by indirect measures (e.g., passage of flatus) require further study to determine whether these intermediate benefits will result in a more rapid functional recovery from surgery. Evidence for clinically significant improvements in perioperative morbidity involving other organ systems is currently insufficient to allow conclusions.

It is clear that "epidural analgesia" is not a generic term. Its effects on outcome may differ depending on whether epidural injections consist of opioid, local anesthetic, or both. Local anesthetics may be more effective than opioids for attenuating many aspects of postoperative morbidity. However, theoretical benefits of local anesthetics on perioperative morbidity have not been confirmed in clinical studies, and the relative risk-benefit ratio for use of local anesthetics has not been defined. In addition, the insertion site for the epidural catheter (lumbar, low thoracic, or high thoracic) will significantly alter physiologic effects when local anesthetics are used. Consequently, site of catheter insertion and drugs used for analgesia must be considered when evaluating data from clinical trials.

Finally, we would like to promote the concept that improved analgesia may not be the only mechanism for improved outcome--at least not through a direct cause-and-effect relation. Epidurally administered analgesics may also produce beneficial effects through other mechanisms such as inhibition of efferent pathways (particularly sympathetic efferent pathways) or inhibition of neural reflex arcs.

Our ability to define conclusively the role of epidural anesthesia and analgesia in clinical practice is currently limited by a paucity of definitive data. Numerous questions remain unanswered. What is the optimal concentration, or dose, of local anesthetic? What is the necessary duration of epidural analgesia after surgery? Does the combination of local anesthetic and opioid provide benefits over the use of either drug alone? Does the site of catheter insertion make a difference? Most important, does epidural analgesia provide benefits over an optimal general anesthetic technique followed by analgesia with intravenous PCA opioid and additional systemic drugs that blunt detrimental physiologic responses to surgery (e.g., beta-adrenergic blocking agents or anticoagulants)?

Although epidural anesthesia and analgesia may improve outcome in high-risk patient populations, potential benefits in low-risk populations remains poorly defined because of the low rate of morbidity in these patient groups. In essence, our ability to detect differences is limited because of the relative safety of anesthetic care. Each intervention that successfully reduces postoperative morbidity requires even greater numbers of patients to detect beneficial effects in subsequent studies. To date, significant beneficial effects of epidural analgesia have been observed only in high-risk patient groups and often only for intermediate outcomes (e.g., myocardial ischemia). Perhaps future studies should attempt to determine the effect of analgesic therapy on ultimate patient outcome (e.g., death, myocardial infarction, or return to work). This effort will likely require formation of multicenter study groups and necessitate enrollment of thousands of patients. Although this approach is relatively uncommon in anesthesia research, it is a common practice in other medical specialties. Beneficial effects of numerous medical therapies have been conclusively documented only through multicenter trials (e.g., effect of beta-adrenergic blockade on myocardial morbidity [28]). It appears that multicenter studies will soon become necessary in anesthesia outcome research.

When designing outcome studies, we must strike a balance between two opposing goals: first, the scientific need to investigate effects of a single intervention by keeping all other aspects of care constant, and second, the clinical reality that postoperative recovery is multifaceted and that a solitary intervention is unlikely to alter outcome. For example, better pain management may not benefit ultimate recovery unless other aspects of convalescence are altered to take advantage of the improved analgesia. When designing studies, we must ask the following questions. Why does the patient need to be in the hospital today? What factors make it impossible for this patient to be discharged today (e.g., pain, ileus, or lack of mobility)? It is only after these factors have been identified that comprehensive convalescence programs can be developed. We must take a leadership role in defining which cases will benefit from aggressive analgesic management.

Finally, we must cooperate in developing postoperative analgesic regimens that are not only scientifically sound but also cost effective. Future studies should include measures of outcome and cost [373]and also adjust outcomes for quality (years of life added or improvement in quality of life). [379]Studies should include cost per increment of benefit gained (e.g., cost to save 1 day of hospitalization) or ratio of total costs to total benefits. Although many of these outcomes are relatively subjective and difficult to quantify, [377]we must expand our horizon beyond the intraoperative or even intrahospital period to see whether our interventions influence ultimate patient recovery (e.g., return to work). The changing economic environment means that it is no longer sufficient for anesthesiologists merely to determine the best method for postoperative analgesia or merely to believe that patients do better. In the future, we must also demonstrate cost effectiveness for our analgesic techniques. The challenge is ours.

The authors thank Sture G. Blomberg, M.D., Ph.D, Oscar A. De Leon-Casasola, M.D, and Christopher M. Bernards, M.D., for critical commentary.

*Wild LR: Does continuous infusion of epidural bupivacaine for postoperative pain impair functional mobility? (abstract). 7th World Congress on Pain, August 22-27, 1993, Paris, France.