The benefit of epidural versus intravenous fentanyl administration for postoperative analgesia is controversial. In the current study, the intraoperative effects of epidural versus intravenous fentanyl administration were compared during major surgery.


Twenty elective patients scheduled for thoracoabdominal esophagectomy under general anesthesia with propofol infusion were randomly allocated to receive either intravenous or epidural boluses of 50-100 micrograms fentanyl in a double-blind fashion to maintain hemodynamic stability. Plasma cortisol and fentanyl, as well as total urinary catecholamines, were obtained at the end of the operations.


Hemodynamic variations were similar except that patients receiving epidural fentanyl had a lower incidence of heart rate reduction (> 20% reduction from baseline, P < 0.05). There were no differences in mean intraoperative fentanyl (1,115 +/- 430 and 1,010 +/- 377 micrograms, epidural and intravenous, respectively) or propofol (2,281 +/- 645 and 2,452 +/- 1,169 mg) doses, number of boluses of fentanyl (nine in both groups), plasma fentanyl concentration (1.13 +/- 0.4 and 1.02 +/- 0.46 ng/ml), or number of anesthesiologists correctly identifying the site of fentanyl administration. Similarly, there were no differences in plasma glucose (8.9 +/- 1.8 and 9.3 +/- 1.8 mM) and cortisol (696 +/- 446 and 846 +/- 257 mM), or urinary epinephrine (12 +/- 3.7 and 13.1 +/- 9.2, micrograms/sample) and norepinephrine (42.7 +/- 26.7 and 39.1 +/- 27.6, micrograms/sample).


There appears to be no clinical advantage to epidural administration of fentanyl over intravenous administration during anesthesia for major surgery.

Key words: Analgesics, epidural: fentanyl. Analgesics, intravenous: fentanyl. Anesthetic techniques epidural.

EPIDURAL fentanyl is used widely for postoperative analgesia. [1–4] However, numerous studies have been unable to document a therapeutic advantage for epidural fentanyl administration. For example, similar doses of fentanyl result in a similar intensity of analgesia when epidural fentanyl administration was compared to intravenous administration. [5–10] On the other hand, halothane minimum alveolar concentration is reduced to a greater extent when fentanyl is administered by the epidural route than when administered intravenously. [11] Thus, although the therapeutic advantage for epidural administration of lipophilic opioids during the postoperative period is questionable, recent evidence suggests that epidural administration may be advantageous intraoperatively. Therefore, this study was designed to determine whether epidural administration of fentanyl confers any benefit over intravenous administration in terms of (1) magnitude of reduction of intraoperative anesthetic requirements, (2) total intraoperative fentanyl requirements or resulting systemic plasma fentanyl concentrations, (3) intraoperative hemodynamic stability, and (4) ability to blunt metabolic changes associated with surgery.

Materials and Methods

After approval of the investigation committee of the Centre Hospitalier Universitaire Vaudois and informed consent, we undertook this prospective, randomized, double-blind study in 20 consecutive patients. All were scheduled for elective esocoloplasty (Lewis operation) by thoracophrenolaparotomy to treat carcinoma of the superior third of the esophagus. Exclusion criteria included American Society of Anesthesiologists physical status > 3, preoperative opioid consumption, clinical or laboratory contraindication to epidural catheter insertion, and intraoperative major complication, such as blood loss > 2,000 ml or failure to complete the intended operation for surgical reasons.

All patients received 5–20 mg intramuscular midazolam plus 5 mg intramuscular morphine 45 min before anesthetic induction. An antecubital intravenous catheter, a 20-G radial arterial catheter, and an 18-G thoracic (T5-T6) epidural catheter (Perifix, Braun Melsungen, Germany) were inserted under local anesthesia. Intravascular or subarachnoid placement of the epidural catheter was excluded by gentle catheter aspiration followed by injection of 3 ml 1% lidocaine with 15 micro gram epinephrine. After assignment of treatment group according to a randomization table, coded 50-ml syringes containing either fentanyl (10 micro gram/ml) or saline were prepared by a first investigator (P.G.C.) not involved in patient care. One syringe was connected to a side port of a peripheral infusion and the other to the epidural catheter tubing. Baseline blood hemoglobin and plasma glucose were obtained.

General anesthesia was induced in all patients with 2 micro gram/kg fentanyl (blinded to intravenous or epidural route according to randomization) and 2 mg/kg propofol and maintained with a propofol infusion (50–100 micro gram *symbol* kg sup -1 *symbol* min sup -1), pancuronium, oxygen in air, and fentanyl (intravenous or epidural). Fentanyl injections were performed by administering the same volume from each of the coded 50-ml syringes, always injecting with the epidural syringe first to minimize the delay due to epidural tubing resistance. All patients had a left endobronchial double-lumen tube, a five-lumen thermodilution catheter, a thermistor rectal thermometer, and a bladder catheter inserted before incision. All operations were completed by the same surgical team, with similar two-stage technique: first, laparotomy in the supine position to construct a gastric conduit, then thoracotomy in the left lateral decubitus position with one-lung ventilation (left) to replace the carcinomatous esophagus with the gastric conduit. Blood loss was assessed by weighing sponges and measuring aspirate volume within suction bottles. The anesthesiologist to in charge of the patient was instructed to correct signs of inadequate analgesia (increases of > 20% above baseline for heart rate (HR) or mean arterial pressure (MAP) by injecting 50–100-micro gram fentanyl holuses and increasing the rate of propofol infusion (10 micro gram *symbol* kg sup -1 *symbol min sup -1). Hypotension with a > 20% decrease from baseline was treated by increasing the rate of infusion of lactated Ringer's solution (baseline infusion rate 15 ml *symbol* kg sup -1 *symbol* h sup -1), and/or transfusing 1 unit of packed erythrocytes (if estimated blood loss exceeded 1,000 ml), and/or decreasing the propofol infusion rate. The propofol infusions were discontinued before transfer to the intensive care unit, after exchanging a single-lumen orotracheal tube for the double-lumen tube. The incidence of HR and MAP variations of > 20% from baseline lasting more than 5 consecutive minutes, the number of changes in propofol infusion rate, and the number of injections of fentanyl boluses were assessed by a second investigator (J.P.G.), who was not involved in patient care and blinded to syringe content. This investigator also provided new blinded fentanyl syringes when needed. At the end of the anesthetic, anesthesiologists caring for the patient were asked to identify the site of administration of fentanyl (intravenous vs. epidural) based on the patient's intraoperative response.

On arrival in the intensive care unit, blood was drawn from the arterial catheter for determination of hemoglobin, glucose, cortisol, and fentanyl concentrations. Plasma was obtained by immediate centrifugation of blood samples and stored at -30 degrees Celsius until analyzed. Cortisol was measured by radioimmunoassay (Gamma-BCT cortisol RIA kit). [12] All assays were processed in duplicate, with less than 4% and 8% maximal intra- and interassay coefficient of variation, respectively. Fentanyl was measured by radioimmunoassay by the Bioanalytical Laboratory of the Janssen Research Foundation. [13] The lower limit of detection for the fentanyl assay was 0.1 ng/ml, and the coefficient of variation was less than 10% over the range of sample concentration. Urine collected during the operation was refrigerated at 4 degrees Celsius, protected from light, and acidified with 10 ml/l of 5N HCl. Mean urinary epinephrine and norepinephrine concentrations were determined by immunofluorescence, with less than 10% maximal interassay coefficient of variation. [14].

Continuous data are reported as mean plus/minus SD and categorical data as medians and ranges. Differences in demographic and operation data, baseline HR and MAP, and blood and plasma measurements were analyzed with unpaired two-tailed t test. Gender, ASA physical status, number of HR and MAP variations of 20% from baseline, number of variations in propofol infusion rate, number of fentanyl injections, and frequency of correct identification of site of fentanyl administration by the anesthesiologists were compared by construction of contingency tables and subsequent chi-square or Fisher's exact analysis. P < 0.05 was considered significant.


Patient characteristics, duration of operation, mean blood loss, mean volume of lactated Ringer's solution and blood transfused, urine volume, baseline and final hemoglobin levels, and rectal temperatures were similar in the two groups of patients (Table 1).

Table 1. Demographic and Operative Data

Table 1. Demographic and Operative Data
Table 1. Demographic and Operative Data

Baseline HR and MAP were similar (Table 2). During intubation, two patients in the epidural group and one patient in the intravenous group had a significant increase in HR. Three patients in each group had a significant increase in MAP during intubation. Mean number of decreases in HR of > 20% less than baseline during more than 5 consecutive minutes was statistically fewer in patients receiving epidural fentanyl (Table 2). All bradycardia episodes occurred during surgical visceral manipulations, and none required specific treatment. Other hemodynamic variations were similar in the two groups of patients (Table 2).

Table 2. Hemodynamic Data

Table 2. Hemodynamic Data
Table 2. Hemodynamic Data

There were no differences between groups in mean intraoperative fentanyl, propofol, or pancuronium doses, number of boluses of fentanyl administered, number of variations in rate of propofol infusions, mean plasma fentanyl concentration at the end of surgery, or number of anesthesiologists correctly identifying the site of fentanyl administration (Table 3). Anesthesiologists in charge of the patients occasionally elected to inject more than one fentanyl bolus over a short period in response to a single important hemodynamic variation.

Table 3. Anesthetic Drug Data

Table 3. Anesthetic Drug Data
Table 3. Anesthetic Drug Data

Baseline and final plasma glucose, postoperative plasma cortisol, and total intraoperative urinary epinephrine and norepinephrine excretion also were similar in the two groups of patients (Table 4).

Table 4. Metabolic Data

Table 4. Metabolic Data
Table 4. Metabolic Data


Studies in gynecologic and thoracic surgery suggest that epidural opioid administration can provide intraoperative benefits without producing adverse hemodynamic consequences traditionally seen with local anesthetics. [11,15,16],* However, our results show that epidural fentanyl provides no significant benefit over intravenous administration for major surgery lasting 4 h or longer.

Many studies have compared the relative advantages of intravenous versus epidural fentanyl, as well as other lipid-soluble opioids, with contradictory results. Most have found that similar doses of fentanyl were necessary to obtain similar analgesic effects, that these doses produced similar plasma concentrations, and that there were no differences in incidences of side effects. [5–10] In contrast, some studies have demonstrated an advantage for epidural administration of fentanyl. [1–1] The reasons for these contradictory results are not clear. However, to the best of our knowledge, this is the first study to compare epidural and intravenous fentanyl administration during the intraoperative period. Because we were unable to identify a clear advantage for epidural administration of fentanyl during the intraoperative period, our results are consistent with those from previous studies that have found intravenous and epidural administration of lipid-soluble opioids to produce comparable effects.

In our study, dose requirements for supplemental fentanyl did not differ between intravenous and epidural routes. The fact that similar doses resulted in similar plasma fentanyl concentrations for the two groups is consistent with the known rapid systemic absorption of epidurally administered fentanyl. [7,8,10] In addition, hemodynamic variations, propofol infusion rates, and total propofol dose requirements also were similar in the two groups of patients. These results indicate that epidural administration of fentanyl did not provide greater intraoperative analgesia nor did it provide superior hemodynamic protection. In addition, the magnitude of increase in metabolic markers of surgical stress was similar for both groups. These results are consistent with previous studies demonstrating that epidural opioids are no more effective for blunting the stress response than intravenous opioids. [17,18] Finally, anesthesiologists were unable to identify the site of fentanyl administration based on the patient's intraoperative response to the fentanyl boluses. Thus, no differences could be identified between epidural and intravenous fentanyl administration on intraoperative anesthetic management, hemodynamics, or metabolic markers of stress.

Our results are at variance with a recent study demonstrating a greater reduction of halothane MAC by epidural fentanyl when compared with the intravenous route. [11] We speculate that variations in study design may explain these differences. Our study investigated the effects of repeated doses of fentanyl up to a mean dose of 16.9 plus/minus 6.5 micro gram/kg during prolonged thoracoabdominal operations in patients receiving usual anesthetic doses of propofol. By comparison, patients in the previous study received 1–4 micro gram/kg fentanyl as a bolus and were anesthetized with the minimum dose of halothane to tolerate a cutaneous incision (MAC determination). [11] It is possible that the larger dose of fentanyl injected, differences in the magnitude of surgical stimuli, or use of propofol rather than halothane may have offset the potential advantage of epidural fentanyl. However, plasma fentanyl concentrations in both of our groups (1.02–1.13 ng/ml) were not excessive and were within the range previously reported to decrease isoflurane MAC by 40–60%(1–3 ng/ml). [19] Furthermore, plasma fentanyl concentrations were lower than those required to prevent somatic responses in 50% of patients after skin incision (3–4 ng/ml). [20] Similarly, it is unlikely that differences were masked by an excessive dose of propofol, because the mean propofol infusion rate for our patients (6.7 plus/minus 0.9 mg *symbol* kg1*symbol* h1) corresponds to rates that have been recommended in previous studies (6–8 mg *symbol* kg1*symbol* h1). [21] Thus, these discrepancies are unlikely to be due to excessive fentanyl or propofol infusion rates.

The ability of other epidural opioids to provide intraoperative analgesia has produced inconclusive results. Epidural sufentanil was compared with intravenous sufentanil as the major intraoperative analgesic agent in an anesthesia regimen with midazolam and nitrous oxide in thoracotomy patients. [16] Although epidural sufentanil decreased the need for supplementary intravenous analgesia and provided better immediate postoperative analgesia, the same number of sufentanil boluses were necessary in the two groups of patients to maintain a stable intraoperative hemodynamic course. Moreover, interpretation of this study is limited by the absence of plasma opioid levels, metabolic data, or comparison of intraoperative hemodynamics. In another study, epidural sufentanil reduced the incidence of hypotension while providing the same quality of analgesia as thoracic epidural bupivacaine during thoracic surgery. * However, this study did not compare the effect of intravenous versus epidural opioid injection. Morphine has produced similarly inconsistent results. Halothane MAC reduction has been observed after intrathecal morphine injection in an open study in gynecologic surgery. [22] However, no difference was found in a double-blind study during lower abdominal surgery. [23] Furthermore, when compared with intravenous morphine, epidural morphine does not appear to produce a greater reduction in halothane MAC despite significant differences in pain threshold and longer lasting postoperative analgesia. [24] This finding is in direct contrast to a similar investigation by the same authors who attribute the discrepancy to a slower onset of epidural morphine when compared to epidural fentanyl. [11] Thus, the therapeutic advantage for intraoperative epidural opioid administration over intravenous administration remains largely speculative and inconclusive.

Our results should be interpreted with some qualifications. First, the study design allowed adjustments in two drugs (fentanyl and propofol) to control intraoperative hemodynamics. Limitation to only one variable may have permitted more scientifically pure comparisons. Such a limitation, however, would have little relevance to clinical care, in which two or more drugs may be adjusted according to an evolving situation and not in a predetermined way. Moreover, infusion rates of propofol were similar in the two groups of patients and controlled within limits that would have been fixed by a predetermined schedule. [19] Thus, we are confident that our protocol did not mask subtle differences between patients but rather conforms closely to usual clinical practice. Second, our patients had major operations, with thoracoabdominal incisions and significant blood losses, and their lungs were ventilated for 6–12 h after surgery. Our results may not be applicable to patients undergoing minor operations. Moreover, this postoperative regimen precluded evaluation of postoperative analgesia or respiratory parameters. However, previous studies produced conflicting results on the value of postoperative epidural fentanyl analgesia. Thus, major differences in the quality of postoperative analgesia between our two groups of patients would not be expected. [5–10] Third, our study design focused on intraoperative events, and did not allow examination of differences in mortality or hospital stay or cost, which would have required inclusion of a significantly greater number of patients. Fourth, our small number of patients may be viewed as a limitation to a generalization our results. However, a power calculation (alpha 0.05, power 0.8) indicated that at least 240 patients would be required to find a 33% difference in fentanyl dose between groups. Inclusion of such a large number of patients in this type of clinical study is impractical.

In conclusion, we found during major surgery that epidural and intravenous fentanyl produce the same hemodynamic profile, cannot be clinically differentiated, and have similar effects on metabolic responses to surgery. Comparable fentanyl doses were administered by blinded anesthesiologists, resulting in similar plasma fentanyl concentrations. Thus, we were unable to identify a clinical advantage for the epidural route over the intravenous route for fentanyl administration during anesthesia for major surgery.

The authors thank Jean-Francois Cuttat, M. D., who supervised most of surgical procedures, for his help in the performance of the study, as well as Janssen Pharmaceutica, Baar, Switzerland, for processing of the fentanyl assays.

*Haak-van des Lely F, van Kleef JW. Gesink-van der Veer BT, Burm AGL, Bovill JG: Analgesic effect of epidurally administered sufentanil during thoracotomies: A comparison with bupivacaine and saline (abstract). ANESTHESIOLOGY 73:A770, 1990.


Salomaki TE, Laitinen JO, Nuutinen LS: A randomized double-blind comparison of epidural versus intravenous fentanyl infusion for analgesia after thoracotomy. ANESTHESIOLOGY 75:790-795, 1991.
Welchew EA, Thornton JA: Continuous thoracic epidural fentanyl: A comparison of epidural fentanyl with intramuscular papaveretum for postoperative pain. Anaesthesia 37:309-316, 1982.
Vella LM, Willaus DG, Knott C, Lintin DJ, Justins DM, Reynolds F: Epidural fentanyl in labor: An evaluation of the systemic contribution to analgesia. Anaesthesia 40:741-747, 1985.
Grant RP, Dolman JF, Harper DG, White A, Parsons DG, Evans KG, Merrick CP: Patient-controlled lumbar epidural fentanyl compared with patient-controlled intravenous fentanyl for post-thoracotomy pain. Can J Anaesth 39:214-219, 1992.
Guinard JP, Mavrocordatos P, Chiolero R, Carpenter RL. A randomized comparison of intravenous versus lumbar and thoracic epidural fentanyl for analgesia after thoracotomy. ANESTHESIOLOGY 77:1108-1115, 1992.
Loper KA, Ready LB, Downey M, Sandler AN, Nessly M, Rapp S, Badner N: Epidural and intravenous fentanyl infusions are clinically equivalent after knee surgery. Anesth Analg 70:72-75, 1990.
Ellis DJ, Millar WL, Reisner LS: A randomized double-blind comparison of epidural versus intravenous fentanyl infusion for analgesia after cesarean section. ANESTHESIOLOGY 72:981-986, 1990.
Glass PSA, Estok P, Ginsberg B, Goldberg JS, Sladen RN: Use of patient-controlled analgesia to compare the efficacy of epidural to intravenous fentanyl administration. Anesth Analg 74:345-351, 1992.
Sevarino FB, McFarlane C, Sinatra R: Epidural fentanyl does not influence intravenous PCA requirements in the post-Caesarean patient Can J Anaesth 38:450-453, 1991.
Sandler AN, Stringer DG, Panos L, Badner N, Friedlander M, Koren G, Katz J, Klein J: A randomized, double-blind comparison of lumbar epidural and intravenous fentanyl infusions for posthoracotomy pain relief. ANESTHESIOLOGY 77:626-634, 1992.
Inagaki Y, Mashimo T, Yoshiya I: Segmental analgesic effect and reduction of halothane MAC from epidural fentanyl in humans. Anesth Analg 74:856-864, 1992.
Charles G, Ansseau M, Sulon J, Demey-Ponsart E, Meunier JC, Wilmotte J, Legros JJ: Free cortisol and dexamethasone suppression test. Biol Psychiatry 21:549-555, 1986.
Michiels M, Hendricks R, Heykants JJK: A sensitive radioimmunoassay for fentanyl plasma levels in dogs and man. Eur J Clin Pharmacol 12:153-158, 1977.
Crout JR: Catecholamines in urine, Standard Methods of Clinical Chemistry. Volume 3. Edited by Segligson E. Orlando, Academic, 1961, pp 62-80.
McCrae AF, Wildsmith JAW: Prevention and treatment of hypotension during central neural block. Br J Anaesth 70:672-680, 1993.
Rosseel PMJ, van den Broek WGM, Boer EC, Prakash O: Epidural sufentanil for intra- and postoperative analgesia in thoracic surgery: A comparative study with intravenous sufentanil. Acta Anaesthesiol Scand 32:193-198, 1988.
Normandale JP, Schmulian C, Paterson JL, Burrin J, Morgan M, Hall GM: Epidural diamorphine and the metabolic response to upper abdominal surgery Anaesthesia 40:748-753, 1985.
Scott NB, Mogesen T, Bigler D, Lund C, Kehlet H: Continuous thoracic extradural 0.5% bupivacaine with or without morphine: Effect on the quality of blockade, lung function and the surgical stress response. Br J Anaesth 62:253-257, 1989.
McEwan Al, Smith C, Dyar O, Goodman D, Smith LR, Glass PSA: Isoflurane minimum alveolar concentration reduction by fentanyl. ANESTHESIOLOGY 78:864-869, 1993.
Glass PSA, Doherty D, Jacobs JR, Goodman D, Smith LR: Plasma concentration of fentanyl, with 70% nitrous oxide, to prevent movement at skin incision. ANESTHESIOLOGY 78:842-847, 1993.
Roberts FL, Dixon J, Lewis GTR, Tackley RM, Prys-Roberts C: Induction and maintenance of propofol anaesthesia. Anaesthesia 43(S): 14-17, 1988.
Drasner K, Bernards CM, Ozanne GM: Intrathecal morphine reduces minimum alveolar concentration of halothane in humans. ANESTHESIOLOGY 69:310-312, 1988.
Licina MG, Schubert A, Tobin JE, Nicodemus HF, Spitzer L: Intrathecal morphine does not reduce minimum alveolar concentrations of halothane in humans: Results of a double-blind study. ANESTHESIOLOGY 74:660-663, 1991.
Inagaki Y, Mashimo T, Yoshiya I: Time-related differential effects of epidural morphine on the neuraxis. Anesth Analg 76:308-315, 1993.