More than 20 yr have elapsed since the first reports of epidural opioid administration for the management of pain appeared (1,2). Fentanyl, a lipophilic opioid, is often preferred to morphine, a hydrophilic opioid, because of its more limited rostral spread within the intrathecal space (3,4). Side effects mediated by the brainstem, such as sedation and respiratory depression, may be more pronounced after the epidural administration of a hydrophilic opioid (5). However, there is a debate as to whether the epidural administration of fentanyl (and other lipophilic opioids) is truly justified. This debate is based on the observation that most of the fentanyl injected into the epidural space is taken up into the bloodstream and redistributed to the brain (6) and, thus, may act primarily at supraspinal rather than spinal sites (7–17). If this were true, epidural injection of fentanyl would not confer any benefit over its IV or IM administration but might increase the risk for complications.
The question of whether epidural fentanyl acts at spinal or supraspinal sites has been addressed in multiple clinical and experimental human pain studies, but no clear answer has emerged. Some studies support a spinal mechanism of action by demonstrating a segmental analgesic effect (18), an increased potency for epidural versus IV administration (19–28), and a lack of correlation between the analgesic effect and the plasma concentration after the epidural administration of lipophilic opioid (19,21,22,29). Other studies support a supraspinal mechanism of action by demonstrating a nonsegmental analgesic effect after the epidural administration of drug (7), an equivalent potency when given epidurally and IV (8–17,30,31), and equal plasma drug levels for epidural and IV analgesic regimens of equal potency (7,10–12,14–17,30–32).
Closer inspection of these studies reveals that most investigators who concluded that epidural fentanyl elicits analgesia by a spinal mechanism administered the drug as a bolus, whereas the majority of investigators concluding that epidural fentanyl elicits analgesia by a supraspinal mechanism administered the drug as a continuous infusion (Table 1).
The goal of this study was to test the hypothesis that the analgesic effect observed after the epidural administration of fentanyl is predominantly mediated by a spinal mechanism if the drug is injected as a bolus and by a supraspinal mechanism if the drug is administered as a continuous infusion.
This report summarizes analgesic data obtained with an electrical and thermal experimental pain model in volunteers after lumbar epidural administration of the μ-opioid receptor agonist fentanyl. On different occasions, epidural fentanyl was administered as a bolus injection and as a continuous infusion. Experimental pain testing was performed at a lumbar and a cranial dermatome to determine whether analgesic effects followed a segmental or nonsegmental distribution. Plasma fentanyl concentrations were determined throughout the study.
The IRB of Stanford University approved the study and 10 healthy volunteers gave written informed consent for participation. Seven men and 3 women were studied; the group had a mean age of 27 yr (range, 21–32 yr), mean body weight of 77 kg for men (range, 73–84 kg) and 63 kg for women (range, 57–70 kg), mean height of 181 cm for men (range, 170–185 cm) and 166 cm for women (range, 161–173 cm). All subjects had a normal medical history, and women had a negative pregnancy test. No subject took prescription drugs other than oral contraceptives. Over-the-counter medication was not allowed for 24 h before a study day. Before each study day subjects fasted overnight. The study was conducted in a quiet room at an ambient temperature that was comfortable for subjects. On arrival at the study center, a 20-gauge IV catheter and a 20-gauge radial arterial catheter were placed in one arm. Recording of vital signs (electrocardiogram, invasive blood pressure, hemoglobin oxygen saturation and respiratory rate) was started. Subjects remained semi-recumbent throughout the study.
Using a double-blinded, cross-over design, subjects were randomly assigned to receive epidural fentanyl both as a bolus injection and as a continuous infusion, but on different occasions separated by a minimum of 2 days.
Subjects were placed in the right lateral position and a 17-gauge Tuohy needle (Arrow International Inc., Reading, PA) was inserted into the epidural space at the L2-3 or L3-4 interspace using a midline approach. The epidural space was identified by loss of resistance to saline. An 18-guage epidural catheter (Arrow Flex Tip Plus, Arrow International Inc.) was inserted 4 cm into the epidural space and a negative pressure aspiration test performed. Subjects rested for 30 min after the placement.
At time zero, subjects received 0.03 mg of fentanyl in 0.67 mL of saline injected by hand over a 10-s period (small dose). Subjects received a second bolus of 0.1 mg fentanyl (large dose) in 2 mL of saline 210 min after the first bolus injection.
At time zero an infusion pump (Harvard Pump 22, Harvard Apparatus Inc., South Natick, MA) was started delivering 0.003 mg/mL of fentanyl at a rate of 10 mL/h (small dose). The concentration of the infusion solution was increased to 0.01 mg/mL (large dose) after 210 min. At this concentration, epidural fentanyl was infused for an additional 200 min.
To maintain blinding, each subject received both treatment protocols (i.e., the epidural bolus injections and the epidural infusion) on both study days; one regimen administering epidural fentanyl and the other administering epidural saline placebo. Treatments were switched on the second study day. An investigator otherwise not involved in the study prepared the solutions containing fentanyl and saline placebo.
Four-hundred-ten minutes after starting administration of fentanyl, the μ-opioid receptor antagonist naloxone was injected IV at a dose of 0.4 mg to demonstrate that measured effects were mediated by fentanyl.
At the end of the study and after all analgesic data had been collected, 5 mL of 1.5% lidocaine with 1:200,000 epinephrine was administered via the epidural catheter to exclude intravascular or subarachnoid placement.
Test cycles consisted of experimental pain tests (heat pain and electrical pain), side effects assessment (pruritus, nausea, and sedation), vital signs recording (respiration rate, heart rate, invasive blood pressure), and blood collection (fentanyl plasma concentration and blood gas analysis). Test cycles took 15 min to perform, allowing for a minimum of 15 min rest between test cycles. Test cycles were performed at baseline (3 times) and at 30, 60, 120, 180, 240, 270, 330, 390, and 420 min after starting drug administration. All subjects had been trained with the experimental pain models before study participation.
A thermal sensory analyzer (TSA 2001, Medoc Advanced Medical Systems, Ramat Yishai, Israel) was used to administer nociceptive heat stimuli at the anteromedial right thigh (L2) and at the right cheek (V2). The algorithm to determine the heat pain tolerance (heat stimulus that causes maximum tolerable pain) has previously been described (33). In brief, a hand-held 16 mm × 16 mm thermode was brought into contact with skin. After equilibration between the skin and the thermode temperature at 35°C, the thermode temperature was increased at a rate of 1°C/s. Subjects pushed the button of a hand-held device as soon as the heat pain became intolerable. This procedure was repeated 4 times and the median temperature causing maximum tolerable pain was recorded. The interval between the heat stimuli was 30 s. If a subject was able to tolerate the maximum output temperature of the device (53°C), this was recorded as the maximum-tolerated temperature (44 of 400 measurements).
A constant current device (Neurometer, Neurotron Inc, Baltimore, MD), with a maximum output of 20 mA and delivering 5 Hz sine wave pulses of 3-s duration was used to administer nociceptive electrical stimuli at the anteromedial right thigh (L2) and the pinna of the right ear (C2). The algorithm used to determine the electrical pain tolerance has previously been described (34). In brief, two distinct series of transcutaneous electrical stimuli were administered. A first series using an ascending staircase design was used to estimate the pain threshold (average of the highest electrical current not evoking pain and the lowest electrical current evoking pain) and the pain tolerance (current evoking maximum tolerable pain). The second series was composed of 16 stimuli, evenly spaced, in a range starting 30% below the pain threshold and ending at the pain tolerance. Subjects rated the magnitude of pain evoked by each stimulus on a 100 mm visual analog scale (VAS) (0 = no pain, 100 = maximum tolerable pain). The data (electrical current versus VAS score) were fitted based on a power model and the electrical pain tolerance (current causing a pain corresponding to 100 on the VAS) was derived. The interval between the electrical stimuli was 15 s.
Subjective Side Effects
Subjects rated the magnitude of sedation, pruritus, and nausea on a 100-mm VAS (0 = none, 100 = as much as can be).
Arterial blood was collected into a heparinized tube, which was immediately placed on ice, centrifuged, and stored at −20° Celsius. The plasma fentanyl concentrations were determined at the Analytical Facility, Department of Anesthesiology, University of Washington, Seattle, WA. The assay used gas chromatography (Model 5890 II, Hewlett-Packard, Palo Alto, CA) using J & W columns (30 m × 0.32 mm DB-5 with a 0.25 μm film of phenyl-methyl silicon) with mass spectrometry (Model HP1034C, (Hewlett-Packard). The minimum detectable plasma fentanyl concentration was 0.05 ng/mL. The range of linearity was 0.05–5 ng/mL. The inter-day coefficients of variation were 9.0% and 7.9% for samples containing 0.3 and 3.5 ng/mL fentanyl (accuracy was 106.6% and 104.8%, respectively).
The area under the plasma concentration versus time curve (AUC), the maximum plasma concentration (Cmax), and the time to achieve the maximum plasma concentration (Tmax) were determined. The AUC between any two data points was calculated by using linear interpolation (or the trapezoid rule) as determined by the following formula: AUC = (E1 + E2)/2 × (T2 − T1), where E1 and E2 are effect measures at successive data time points, and where T1 and T2 are the minutes from initial drug administration. The total AUC was determined by the sum of all the component AUCs. Fractionated AUCs for small-dose and large-dose fentanyl were determined by the sum of the component AUCs for the appropriate section of the study.
Pharmacokinetic variables were derived separately for data collected after administration of a small and a large bolus, and during infusion at a slow and at a rapid rate of fentanyl. The paired Student’s t-test or Wilcoxon’s signed-rank test with Bonferroni’s correction was used to detect significant pharmacokinetic differences between bolus administration and infusion of epidural fentanyl.
Analgesic effect variables were expressed as the difference in temperature from baseline for heat pain data (interval scale data), and as the percentage change from baseline for electrical pain data (ratio scale data).
The following pharmacodynamic variables were determined: the area under the analgesic effect variable versus time curve, the maximum analgesic effect observed (Emax), and the time to reach Emax after bolus or starting the infusion of epidural fentanyl (Tmax). The AUC was determined by using linear interpolation. Pharmacodynamic variables were derived separately for data collected after administration of a small and a large bolus, and during infusion at a slow and at a rapid rate of fentanyl. The paired Student’s t-test or Wilcoxon’s signed-rank test with Bonferroni’s correction was used for detecting significant pharmacodynamic differences between measurements obtained at the leg and head (AUC and Emax), and measurements obtained for bolus administration and infusion of epidural fentanyl (Tmax).
A two-way repeated-measures analysis of variance (ANOVA) and Student-Newman-Keuls post hoc test were used to determine whether epidural bolus administration or infusion of fentanyl resulted in a significant analgesic effect and whether this effect was different between measurements obtained at the leg and the head.
Data obtained for epidural bolus administration and infusion of fentanyl were pooled and analyzed separately to determine whether a significant relationship between the plasma concentration of fentanyl and the analgesic effect measure could be identified. Nonlinear regression was used to explore pharmacodynamic models nested in a power model of the general form, MATH where y is the analgesic efficacy variable, x is the plasma concentration, a is the y-intercept, b is the slope of the relationship, and λ is a variable exponent. Contrary to linear regression, nonlinear regression analysis does not reveal an r-squared value. Our group described the statistical approach inherent to nonlinear regression analysis in a recent publication (35). This approach uses the log-likelihood ratio test that is based on χ2 statistics. A decrease of more than 3.84 for the negative product of 2 times the log likelihood indicates that a model parameter (part of the equation) is significant at an α level of 0.05.
Model selection was based on the likelihood ratio test and visual assessment of the goodness of fit. Visual inspection of the data did not support a sigmoidal model and therefore, this model was not explored.
All data are expressed as the mean and the standard deviation or as the median and the interquartile range. As an exception, the 95%-confidence intervals are given for pharmacodynamic model parameters.
All 10 subjects completed the investigation according to the protocol. Oral contraceptive medication taken by one woman was the only continuing drug therapy during the study.
The epidural space was identified in all subjects on the first attempt. After placement of the epidural catheter, all attempts to aspirate blood or cerebrospinal fluid were negative. Correct epidural catheter placement was confirmed in all subjects at the end of each study day by demonstrating varying degrees of bilateral sensory changes and absence of hemodynamic changes after injection of 5 mL 1.5% lidocaine with epinephrine (1:200,000).
The plasma fentanyl concentration versus time is depicted in the bottom graph of Figure 1. All pharmacokinetic variables are listed in Table 2. The plasma concentration was higher, peaked later, and was sustained during infusion as compared with bolus administration of epidural fentanyl. The AUC was 2.3 ± 1.1 times larger during infusion than after bolus administration.
The changes in heat and electrical pain tolerance are depicted versus time in the top and the middle graph of Figure 2. Table 3 lists the absolute values of the heat and electrical pain tolerance as determined at the leg and at the head after bolus administration and during infusion of epidural fentanyl. Table 4 lists the pharmacodynamic variables.
Epidural bolus administration of fentanyl caused significant analgesic effects at 30, 60, 120, 240, 270, and 330 min at the leg (compared with baseline and compared with measurements obtained at the head). No significant analgesic effects were detected at the head. Epidural infusion of fentanyl caused significant analgesic effects at 120, 180, 240, 270, 330, and 390 min at the leg and at 270, 330, and 390 min at the head (compared with baseline). No overall difference was detected between analgesic effects measured at the leg and at the head. Injection of the μ-opioid antagonist naloxone reversed measured analgesic effects.
Epidural bolus administration of fentanyl caused significant analgesic effects at 240, 270, 330, and 390 min (compared with baseline). No significant analgesic effects were detected at the head. Measurements simultaneously obtained at the leg and head were significantly different at 240 and 270 min. Epidural infusion of fentanyl caused significant analgesic effects at 240, 270, 330, and 390 min at the leg and at 270, 330, and 390 min at the head (compared with baseline). No overall difference was detected between analgesic effects measured at the leg and at the head. Injection of the μ-opioid antagonist naloxone reversed measured analgesic effects.
No significant relationship was detected between the analgesic effect measures (heat and electrical pain tolerance) and the plasma concentration after epidural bolus administration of fentanyl (Fig. 2). For epidural infusion of fentanyl a significant linear relationship was identified between the heat pain tolerance and the plasma concentration, MATH and between the electrical pain tolerance and the plasma concentration, MATH where y is the difference in tolerated heat from baseline (baseline = 0°C) or the percentage increase in tolerated current from baseline (baseline = 100%), and x is the plasma concentration. The 95% confidence intervals for the y-intercept and the slope are given in parenthesis. Introducing an exponent different from 1 did not significantly improve the fit.
There was no significant change in heart rate, arterial blood pressure, hemoglobin oxygen saturation, or respiratory rate throughout the study for both treatment arms.
Sedation, pruritus, and nausea were reported by 5, 7, and 1 subjects after bolus administration, and by 5, 5, and 1 subjects during the infusion of epidural fentanyl (Fig. 3). The occurrence of sedation suggests that both bolus administration and infusion of epidural fentanyl resulted in some systemic effects. Nevertheless, none of the data points for sedation, pruritus, or nausea was statistically different from baseline for the sample size studied in either study group.
This study documents a segmental analgesic effect after an epidural bolus injection and a systemic, nonsegmental, analgesic effect during an epidural infusion of the μ-opioid agonist fentanyl in volunteers. A relationship between the fentanyl plasma concentration and the measured analgesic effect was present during the epidural infusion of fentanyl but not after the epidural bolus administration of fentanyl. These findings indicate that epidural fentanyl—at least within the range of doses studied—acts predominantly at spinal sites if administered as a bolus and at supraspinal sites if administered as an infusion.
These findings sit squarely with almost all 26 published clinical and volunteer studies. At first appearance, these studies are divided into approximately equal camps supporting either systemic or spinal mechanisms for epidural fentanyl analgesia. However, when these studies are analyzed according to their modes of drug administration (bolus versus infusion), there is almost unanimous agreement between them, consistent with our hypothesis (Table 1). It should be noted that these published reports include studies using different lipophilic opioid drugs (fentanyl, sufentanil, and alfentanil) and different painful stimuli (labor pain, postoperative pain, intraoperative pain, experimental pain in volunteers) where the pain may not be comparable from a mechanistic point of view and where analgesic efficacy and site of action may be different. In light of these theoretical concerns it is actually remarkable that our finding seems to be echoed by such a broad range of different pain models.
Detailed analysis of the mode of administration versus site of analgesic action has not been published, and this hypothesis has never before been subjected to a hypothesis-driven study. The finding that the predominant site of action of epidural fentanyl is dependent on the mode of its administration may resolve the controversy concerning the site of action of epidural fentanyl.
Discounting studies that administered mixed bolus and infusion regimens, only 3 of the 26 studies summarized in Table 1 reported a site of epidural fentanyl action different from that which may be predicted from the mode of drug administration (bolus or infusion). A study by Guinard et al. (16) compared two groups of patients receiving fentanyl either by epidural or IV bolus administration for intraoperative pain management while undergoing thoracoabdominal esophagectomy. They found that the fentanyl plasma concentration and the need for supplemental analgesia were similar in both groups and concluded that epidural bolus administration did not provide a significant benefit over the IV injection of fentanyl. The average dose of epidural fentanyl studied by Guinard et al. was much larger (1.115 ± 0.430 mg) than the doses studied by other investigators. It is possible that analgesia mediated by a spinal mechanism is obscured at such large doses because of extensive systemic drug uptake from the epidural space with consequent redistribution into the brain or because of a significant rostral spread of fentanyl within the intrathecal space.
In a study by D’Angelo et al. (26), epidural fentanyl was administered as a continuous infusion for the treatment of labor pain. These investigators reported a significant analgesic effect mediated by a spinal mechanism. However, the local anesthetic bupivacaine was coadministered with fentanyl. Local anesthetics act synergistically with opioids (28,36,37). Therefore, an otherwise insignificant analgesic effect by a spinal mechanism may become predominant if epidural fentanyl is infused jointly with a local anesthetic. This view is supported by our parallel study in laboring women suggesting that the required dose of epidural bupivacaine can be reduced by a factor of three when coadministered with an epidural fentanyl infusion as compared with an IV fentanyl infusion (38).
Menigaux et al. (32) studied postoperative pain management with IV and epidural sufentanil administered as small (5 μg) repeated boluses with no background infusion and concluded that epidural sufentanil elicited analgesia by systemic redistribution. Pain scores and plasma sufentanil concentrations were similar in both groups, and significantly larger doses of sufentanil had to be administered in the epidural group to achieve this equivalence of analgesia (32). The authors of that study speculated that the small bolus injections used were unable to generate a sufficient concentration gradient to drive the lipophilic drug through the meninges and into the spinal cord (32).
Our results may explain an apparent contradiction between two experimental pain studies performed by the same research group. A first study reported segmental analgesic effects after bolus administration of a small (0.3 mg) and a large (1 mg) dose of epidural alfentanil (29). In a second study, systemic, nonsegmental analgesic effects were documented after bolus administration of an intermediate dose (0.4 mg) of alfentanil (7). In light of our findings, it is interesting that bolus administration was followed by an infusion of alfentanil (0.4 mg/hr for 2 hours) in the second study but not in the first.
There is consistent evidence that bolus administration of epidural fentanyl causes segmental analgesia by a spinal mechanism of action. This may be surprising in light of pharmacokinetic studies demonstrating a small spinal cord bioavailability after epidural (39,40) or intrathecal (6) administration of lipophilic opioids. These studies demonstrated a small meningeal permeability (39,40), a preferential partition into fat (6), and large systemic drug absorption for epidurally administered lipophilic opioids (6,39,40). A pharmacokinetic study further documented that lipophilic opioids bind much more avidly to myelin and do not penetrate as deeply into the gray matter as hydrophilic opioids (41).
Based on these studies, there was significant concern that a lipophilic opioid administered into the epidural space may not reach the relevant binding sites within the spinal cord and therefore may not exert a significant analgesic action at the level of the spinal cord. However, only pharmacodynamic studies focusing on observed drug effect rather than predictions of drug effect (based on pharmacokinetic studies of drug distribution) can answer whether epidural administration of a lipophilic opioid results in a significant analgesic effect at the level of the spinal cord. Our results and the results of previous studies suggest that epidural bolus administration of a lipophilic opioid results in spinally mediated analgesic effects, despite pharmacokinetic concerns.
We can only speculate about the underlying mechanism responsible for the spinal site of action after epidural bolus administration of fentanyl. A conceivable explanation is that the concentration gradient between the epidural and the intrathecal space is much larger after a bolus administration than during an infusion (the same amount of drug is delivered in a much shorter time interval). The larger the concentration gradient, the more of the total amount of fentanyl administered into the epidural space is likely to be driven into the intrathecal space, and so to reach the spinal cord dorsal horn in sufficient quantities to elicit analgesia by a spinal mechanism, despite its low intrathecal bioavailability. Large concentration gradients are not generated during continuous infusion, so the drug is redistributed to the systemic circulation, where it elicits analgesia by binding to supraspinal sites.
It may be argued that, in our study, the segmental analgesic effects observed after the bolus administration could not be detected during the infusion of epidural fentanyl because the total amount of drug delivered during the observation period of 210 min was different (0.03 and 0.1 mg bolus versus 0.03 and 0.1 mg/h infusion rate). However, direct comparison of the analgesic data collected after the 0.1 mg bolus and during the 0.03 mg/h infusion of epidural fentanyl (approximately the same dose of drug during the observation period) revealed the same result, i.e., that the bolus administration, but not the infusion regimen, produced segmental analgesia.
A potential criticism of our study may concern the lack of study arms for administering IV fentanyl and saline placebo. However, it was not feasible to perform four separate epidurals and study sessions on the same individual subjects. Because data have already been published demonstrating the stability of the pain measures we used over time during saline placebo administration, we decided to demonstrate opioid-specific analgesic effects by reversing these effects at the end of the study with the opioid antagonist naloxone (33,34). Data obtained during an IV infusion of fentanyl would have been useful to determine the degree to which the analgesic profile observed during an epidural infusion of fentanyl corresponded to an analgesic profile obtained with systemic administration of fentanyl and whether there was evidence for some analgesic activity at the spinal site. The data presented in this manuscript do not allow us to make this distinction. However, assuming that systemically administered fentanyl increases the pain tolerance to a similar extent at the leg and at the head, this study does enable us to conclude that 1) a spinal site of action is much more prominent when administrating epidural fentanyl as a bolus rather than as an infusion, and 2) a spinal site of action contributes little to the analgesic effect observed during an infusion of epidural fentanyl.
The volume used to inject epidural fentanyl as a bolus was relatively small in this study, i.e., 0.67 mL (small dose) and 2 mL (large dose). We cannot exclude the possibility that a segmental analgesic effect may have been less pronounced if fentanyl would have been diluted in a larger volume. This may have resulted in a smaller concentration gradient between the epidural space and the intrathecal space, which in turn may have resulted in a reduced uptake of fentanyl into the intrathecal space. However, studies supporting a spinal mechanism for epidural bolus fentanyl administration have injected volumes as large as 6 mL (22,29) and 10 mL (18,23), and yet all demonstrated significant segmental analgesia. This favors the view that the volume used to inject fentanyl into the epidural space is of minor importance for determining whether fentanyl acts at a spinal site.
Figures 1 and 2 show remarkable qualitative similarity between heat and electrical pain data in this study. Nevertheless, the statistical significance of the data points with respect to baseline is different for the heat and electrical pain models used, particularly at the smaller doses. In a previous study published by our group (33) we discussed an analogous finding, namely that heat pain was attenuated more potently than electrical pain. Heat pain (typically burning in character) is mediated by unmyelinated C-fibers (42), whereas electrical pain (typically sharp and pricking in character) is mediated by both C-fibers and myelinated A-fibers (33). Opioids affect pain mediated by C fibers more potently than pain mediated by A-fibers (42,43). Therefore, the observed differential susceptibility of heat and electrical pain to attenuation by opioid analgesia should not be surprising.
For the purposes of this study only the pain tolerance data (maximum tolerable pain), but not the pain threshold data (minimum detectable pain), were collected for two reasons. First, strong pain seems clinically more relevant than pain that is barely detectable. Second, pain tolerance is a more sensitive measure than pain threshold for the assessment of opioid analgesia (33,34).
In summary, this study reports nonsegmental analgesia (suggesting relevant opioid binding in the brain) during the continuous infusion of epidural fentanyl and segmental analgesia (suggesting relevant opioid binding in the spinal cord) after bolus administration. These findings may help resolve the controversy regarding the site of action of epidurally administered lipophilic opioids. Finally, these findings strongly support the value of using lipophilic opioids for administration into the epidural space when delivered by bolus techniques.
1. Behar M, Magora F, Olshwang D, Davidson JT. Epidural morphine in treatment of pain. Lancet 1979; 1: 527–9.
2. Magora F, Olshwang D, Eimerl D, et al. Observations on extradural morphine analgesia in various pain conditions. Br J Anaesth 1980; 52: 247–52.
3. Gourlay GK, Cherry DA, Cousins MJ. Cephalad migration of morphine in CSF following lumbar epidural administration in patients with cancer pain. Pain 1985; 23: 317–26.
4. Gourlay GK, Murphy TM, Plummer JL, et al. Pharmacokinetics of fentanyl in lumbar and cervical CSF following lumbar epidural and intravenous administration. Pain 1989; 38: 253–9.
5. Cousins MJ, Mather LE. Intrathecal and epidural administration of opioids. Anesthesiology 1984; 61: 276–310.
6. Ummenhofer WC, Arends RH, Shen DD, Bernards CM. Comparative spinal distribution and clearance kinetics of intrathecally administered morphine, fentanyl, alfentanil, and sufentanil. Anesthesiology 2000; 92: 739–53.
7. Coda BA, Brown MC, Risler L, et al. Equivalent analgesia and side effects during epidural and pharmacokinetically tailored intravenous infusion with matching plasma alfentanil concentration. Anesthesiology 1999; 90: 98–108.
8. Ellis DJ, Millar WL, Reisner LS. A randomized double-blind comparison of epidural versus intravenous fentanyl infusion for analgesia after cesarean section. Anesthesiology 1990; 72: 981–6.
9. Loper KA, Ready LB, Downey M, et al. Epidural and intravenous fentanyl infusions are clinically equivalent after knee surgery. Anesth Analg 1990; 70: 72–5.
10. Camu F, Debucquoy F. Alfentanil infusion for postoperative pain: a comparison of epidural and intravenous routes. Anesthesiology 1991; 75: 171–8.
11. Glass PS, Estok P, Ginsberg B, et al. Use of patient-controlled analgesia to compare the efficacy of epidural to intravenous fentanyl administration. Anesth Analg 1992; 74: 345–51.
12. Sandler AN, Stringer D, Panos L, et al. A randomized, double-blind comparison of lumbar epidural and intravenous fentanyl infusions for postthoracotomy pain relief. Anesthesiology 1992; 77: 626–34.
13. Guinard JP, Mavrocordatos P, Chiolero R, Carpenter RL. A randomized comparison of intravenous versus lumbar and thoracic epidural fentanyl for analgesia after thoracotomy. Anesthesiology 1992; 77: 1108–15.
14. Chauvin M, Hongnat JM, Mourgeon E, et al. Equivalence of postoperative analgesia with patient-controlled intravenous or epidural alfentanil. Anesth Analg 1993; 76: 1251–8.
15. Miguel R, Barlow I, Morrell M, et al. A prospective, randomized, double-blind comparison of epidural and intravenous sufentanil infusions. Anesthesiology 1994; 81: 346–52.
16. Guinard JP, Carpenter RL, Chassot PG. Epidural and intravenous fentanyl produce equivalent effects during major surgery. Anesthesiology 1995; 82: 377–82.
17. Sinatra RS, Sevarino FB, Paige D. Patient-controlled analgesia with sufentanil: a comparison of two different methods of administration. J Clin Anesth 1996; 8: 123–9.
18. Liu SS, Gerancher JC, Bainton BG, et al. The effects of electrical stimulation at different frequencies on perception and pain in human volunteers: epidural versus intravenous administration of fentanyl. Anesth Analg 1996; 82: 98–102.
19. Justins DM, Francis D, Houlton PG, Reynolds F. A controlled trial of extradural fentanyl in labour. Br J Anaesth 1982; 54: 409–14.
20. Lomessy A, Magnin C, Viale JP, et al. Clinical advantages of fentanyl given epidurally for postoperative analgesia. Anesthesiology 1984; 61: 466–9.
21. Vella LM, Willatts DG, Knott C, et al. Epidural fentanyl in labour: an evaluation of the systemic contribution to analgesia. Anaesthesia 1985; 40: 741–7.
22. Salomaki TE, Laitinen JO, Nuutinen LS. A randomized double-blind comparison of epidural versus intravenous fentanyl infusion for analgesia after thoracotomy. Anesthesiology 1991; 75: 790–5.
23. Inagaki Y, Mashimo T, Yoshiya I. Segmental analgesic effect and reduction of halothane MAC from epidural fentanyl in humans. Anesth Analg 1992; 74: 856–64.
24. Geller E, Chrubasik J, Graf R, et al. A randomized double-blind comparison of epidural sufentanil versus intravenous sufentanil or epidural fentanyl analgesia after major abdominal surgery. Anesth Analg 1993; 76: 1243–50.
25. Harukuni I, Yamaguchi H, Sato S, Naito H. The comparison of epidural fentanyl, epidural lidocaine, and intravenous fentanyl in patients undergoing gastrectomy. Anesth Analg 1995; 81: 1169–74.
26. D’Angelo R, Gerancher JC, Eisenach JC, Raphael BL. Epidural fentanyl produces labor analgesia by a spinal mechanism. Anesthesiology 1998; 88: 1519–23.
27. Polley LS, Columb MO, Naughton NN, et al. Effect of intravenous versus epidural fentanyl on the minimum local analgesic concentration of epidural bupivacaine in labor. Anesthesiology 2000; 93: 122–8.
28. Cohen S, Pantuck CB, Amar D, et al. The primary action of epidural fentanyl after cesarean delivery is via a spinal mechanism. Anesth Analg 2002; 94: 674–9.
29. Coda BA, Brown MC, Schaffer R, et al. Pharmacology of epidural fentanyl, alfentanil, and sufentanil in volunteers. Anesthesiology 1994; 81: 1149–61.
30. van den Nieuwenhuyzen MC, Burm AG, Vletter AA, et al. Epidural vs. intravenous infusion of alfentanil in the management of postoperative pain following laparotomies. Acta Anaesthesiol Scand 1996; 40: 1112–8.
31. van den Nieuwenhuyzen MC, Stienstra R, Burm AG, et al. Alfentanil as an adjuvant to epidural bupivacaine in the management of postoperative pain after laparotomies: lack of evidence of spinal action. Anesth Analg 1998; 86: 574–8.
32. Menigaux C, Guignard B, Fletcher D, et al. More epidural than intravenous sufentanil is required to provide comparable postoperative pain relief. Anesth Analg 2001; 93: 472–6.
33. Angst MS, Ramaswamy B, Riley ET, Stanski DR. Lumbar epidural morphine in humans and supraspinal analgesia to experimental heat pain. Anesthesiology 2000; 92: 312–24.
34. Angst MS, Drover DR, Lotsch J, et al. Pharmacodynamics of orally administered sustained-release hydromorphone in humans. Anesthesiology 2001; 94: 63–73.
35. Lotsch J, Angst M. The m-opioid agonist remifentanil attenuates hyperalgesia evoked by blunt and punctuated stimuli with differential potency: a pharmacological evaluation of the freeze lesion in humans. Pain 2003; 102: 151–61.
36. Akerman B, Arwestrom E, Post C. Local anesthetics potentiate spinal morphine antinociception. Anesth Analg 1988; 67: 943–8.
37. Saito Y, Kaneko M, Kirihara Y, et al. Interaction of intrathecally infused morphine and lidocaine in rats (part 1): synergistic antinociceptive effects. Anesthesiology 1998; 89: 1455–63.
38. Ginosar Y, Columb MO, Cohen SE, et al. The site of action of epidural fentanyl infusions in the presence of local anesthetics: a minimum local analgesic concentration infusion study in nulliparous labor. Anesth Analg 2003; 97: 1439–45.
39. Bernards CM, Hill HF. Morphine and alfentanil permeability through the spinal dura, arachnoid, and pia mater of dogs and monkeys. Anesthesiology 1990; 73: 1214–9.
40. Bernards CM, Hill HF. Physical and chemical properties of drug molecules governing their diffusion through the spinal meninges. Anesthesiology 1992; 77: 750–6.
41. Herz A, Albus K, Metys J, et al. On the central sites for the antinociceptive action of morphine and fentanyl. Neuropharmacology 1970; 9: 539–51.
42. Yeomans DC, Pirec V, Proudfit HK. Nociceptive responses to high and low rates of noxious cutaneous heating are mediated by different nociceptors in the rat: behavioral evidence. Pain 1996; 68: 133–40.
© 2003 International Anesthesia Research Society
43. Le Bars D, Guilbaud G, Jurna I, Besson JM. Differential effects of morphine on responses of dorsal horn lamina V type cells elicited by A and C fibre stimulation in the spinal cat. Brain Res 1976; 115: 518–24.