Both epidural [1-4] and parenteral [5,6] administration of clonidine have been used to treat postoperative pain, but in predetermined doses. As clonidine-induced analgesia is mediated by activation of alpha2-adrenergic receptors on the dorsal horn of the spinal cord , intrathecal or epidural administration of the drug close to its action site seems to be logical. However, clonidine is rapidly and widely absorbed into systemic circulation after epidural administration, reaching many other action sites . As side effects, such as sedation, hypotension, or bradycardia, are at least partly related to systemic absorption , the route that provides the best balance between analgesia and side effects is controversial.
This double-blind randomized study was designed to 1) determine the clonidine dose inducing pain relief after major orthopedic surgery when controlled by the patient, either intravenously or epidurally; and 2) assess whether these two administration routes are clinically equivalent.
This study was approved by our institutional human investigation committee. Written, informed consent was obtained from 24 young ASA physical status I patients scheduled for scoliosis correction. Patients with respiratory insufficiency (scoliosis curvature more than 60 degrees) and those taking opioids, calcium channel blockers, clonidine, and related compounds were excluded. Surgical procedures consisted of distraction and stabilization of the spine by bilateral rods through variable vertebral hooks. No laminectomy was required (meaning that the integrity of the epidural space was preserved). A radial artery cannula, a fiberoptic pulmonary artery catheter, and somatosensory and motor evoked potentials were used to monitor patients during surgery in agreement with our usual practice.
Before surgery, a catheter was introduced into the epidural space via the caudal route. Adequate positioning of the catheter at the lumbar level was checked in all patients by epidurography using iohexol (Omnipaque Registered Trademark; Nycomed, Oslo, Norway), both preoperatively, using radioscopy, and postoperatively, at the time of the radiographic control of the surgical instrumentation. Correct position of the catheter was defined by a rapid and free spread of iohexol in the medullary canal. If the catheter was not correctly positioned, the patient was not studied. Surgery was performed under general anesthesia using propofol (3 mg/kg) for induction, vecuronium (0.1 mg/kg) to facilitate endotracheal intubation, and nitrous oxide 50% in oxygen and isoflurane for maintenance. After surgery, patients were admitted to the recovery room, where the trachea was extubated once the following criteria were met: full consciousness, stable hemodynamic condition, respiratory rate of more than 16 breaths/min, and temperature above 36.5 degrees C.
At the first complaint of pain, patients were assigned, using a random numbers table, to one of two groups receiving clonidine either epidurally or intravenously. An initial dose of 8 micro gram/kg clonidine diluted in 20 mL of isotonic saline was infused during a 30-min period. Then, patients were allowed to self-administer clonidine using a patient-controlled pump (Lifecare 4200 Registered Trademark; Abbott Laboratories, North Chicago, IL), which delivered a bolus dose of 30 micro gram with a 15-min lockout interval. The infusion pump used for the initial dose and then the patient-controlled pump were connected to both epidural and intravenous catheters via a three-way stopcock that was opened to the appropriate route, according to patient randomization. To ensure that the study was double-blind, the stopcock was hidden by an occlusive dressing and was not visible to the patient, nurses, and investigators. Doses of clonidine and infusion rate were selected according to previous reports on postoperative analgesia with clonidine [3,4,9].
Patient demands for clonidine and the administration time of each bolus were recorded. Heart rate, systemic and pulmonary arterial pressures, respiratory rate, arterial oxygen saturation, and mixed venous oxygen saturation were continuously monitored throughout the study. These variables, as well as pain scores measured on a 0-100 visual analog scale, sedation scores measured on a five-point scale (0 = wide awake, 1 = drowsy, 2 = dozing, 3 = mostly sleeping, and 4 = only aroused by tactile stimulation), and cardiac index measured by thermodilution, were obtained before and 15, 30, and 120 min after the loading dose started, and then every 2 h up to 600 min. Arterial blood gases were measured at the 15th, 30th, 240th, and 480th min. At these times, arterial blood samples were also withdrawn, centrifuged, and stored at -70 degrees C for determination of plasma clonidine concentrations by a radioimmunoassay method  (Bioscienta Laboratory, Mainz, Germany), with an assay quantitation limit of 0.1 ng/mL and a detection limit of 0.05 ng/mL.
Side effects and their treatment were defined as follows: 1) inadequate analgesia: the patient required analgesia in addition to the loading dose and self-administered doses of clonidine, in which case ketoprofen 75 mg was given intravenously; 2) hypotension (mean systemic arterial pressure less than 55 mm Hg), which was treated by intravenous fluid administration (if pulmonary artery occluding pressure was lower than 5 mm Hg) or an intravenous 15-mg bolus of ephedrine; 3) bradycardia (heart rate < 45 bpm), which was treated by an intravenous 0.5-mg bolus of atropine; and 4) occurrence within 10 min of more than three episodes of pulse oxygen saturation equal to or less than 90% and lasting more than 20 s, warranting oxygen administration via a face mask and blood gas analysis.
Differences in pain scores and clonidine analgesic use were evaluated using a Mann-Whitney test. The two-factor analysis of variance was used to analyze the difference in hemodynamic parameters, blood gas tensions, pH, arterial oxygen saturation, mixed venous oxygen saturation, and respiratory rate. When P was <0.05 after analysis of variance, Scheffe's multiple comparisons test was used to distinguish differences in measurement variables obtained at data collection intervals. Intergroup categorical data (e.g., requirement for ephedrine, sedation scores) were compared with a contingency table. Demographics and other continuous variables were compared using Student's t-test. Statistical significance was assumed for P < 0.05. Results are expressed as mean +/- SD.
Both groups were comparable for demographics, duration of surgery, and time elapsed between induction of anesthesia and the first postoperative complaint of pain Table 1.
In both groups, pain scores were lower after than before clonidine administration Figure 1. No differences in pain scores were found between groups. One dose of ketoprofen was given to three patients in the intravenous group and two patients in the epidural group (difference no significant [NS]) Figure 1. The initial dose of clonidine was identical by protocol in both groups (442 +/- 66 micro gram in the intravenous group and 417 +/- 80 micro gram in the epidural group [P = 0.41]), but the clonidine dose given via the patient-controlled device to achieve a comparable pain level was lower in the epidural group (P < 0.01) Figure 2. At the end of the study, the self-administered dose was 235 +/- 144 micro gram in the epidural group and 372 +/- 110 micro gram in the intravenous group (P < 0.02). The cumulative clonidine dose, including the initial dose, was 652 +/- 187 micro gram in the epidural group and 814 +/- 144 micro gram in the intravenous group (P < 0.03). Plasma clonidine concentrations were higher in the intravenous group Table 2.
Clonidine produced sedation in both groups Figure 3. Patients receiving clonidine by intravenous route were significantly more sedated than those receiving it by epidural route 15 min after the loading dose started and at 360 and 600 min during the use of the patient-controlled device.
Respiratory rate and blood gases were unchanged throughout the study Table 3. No episodes of arterial oxygen saturation equal to or less than 90%, lasting for 20 s or more, were noted, except for one patient in the intravenous group whose episode occurred about 20 min after the start of the loading dose and lasted for 45 s. The patient was sleeping. Arousal was easily obtained by tactile stimulation, leading to spontaneous normalization of oxygen saturation. No oxygen was given.
Arterial blood pressure decreased similarly in both groups at the end of the initial dose and during the use of the patient-controlled pump. This decrease occurred earlier in the intravenous group Figure 4. Heart rate, cardiac index, and systemic arterial resistances decreased significantly and similarly in both groups Table 4. Ephedrine was required at least once in nine patients in the intravenous group and in five patients in the epidural group (NS). During the 10 h of the study and for the patient population as a whole, the mean ephedrine dose for each patient was 21.2 +/- 16.2 mg in the intravenous group and 15.0 +/- 22.1 mg in the epidural group (NS). The first ephedrine injection was delayed in the epidural group (mean time elapsed before the first ephedrine injection was 288 min [range 240-480 min] in the epidural group and 145 min [range 15-480 min] in the intravenous group; P < 0.05). Cumulative intravenous fluid volume, the volume of blood collected into suction-trap bottles, and diuresis were similar in both groups (1066 +/- 709 mL, 513 +/- 204 mL, and 459 +/- 120 mL, respectively, in the intravenous group, and 974 +/- 513 mL, 367 +/- 189 mL, and 566 +/- 144 mL in the epidural group). No atropine was required.
This study provides an estimation of the mean clonidine dose required to treat pain after major orthopedic surgery. It indicates that comparable levels of analgesia may be achieved when clonidine is given, epidurally or intravenously, as the sole self-administered analgesic. Epidural administration allows for a reduction in clonidine requirement and produces less marked sedation.
As ketoprofen injections were performed late in the course of the study and in a small number of patients, supplemental analgesia apparently had no appreciable influence on the outcome of the study. In these conditions, a mean dose of 650 micro gram of epidural clonidine was sufficient to control postoperative pain during a 10-h period, although wide interpatient variability was noted as may be expected with the use of patient-controlled devices. Doses of epidural clonidine administered to control postoperative pain have progressively increased from initial to more recent studies [1-4,9]. In these studies, doses equal to or greater than 400-600 micro gram as a bolus [1,2] or 800-1000 micro gram over a 12-h period were necessary for pain relief [3,9]. The relatively lower doses of clonidine, used as the sole analgesic in our study, compared with those previously used by others [1-3,9] could be explained by a ceiling effect. Indeed, analgesia induced by clonidine and its opioid sparing effect are related to cerebrospinal fluid concentration . Concentrations higher than 100 ng/mL do not produce an additional analgesic effect. Moreover Meert and De Kock  have recently reported a biphasic effect of clonidine, the largest doses being associated with a reduction in fentanyl potentiation. Finally, a fixed continuous infusion usually overestimates patient requirements when compared to patient-controlled analgesia administration .
Several attempts have been made to compare epidural and systemic administration of clonidine [4,9,13], but no definitive conclusions could be drawn.
In a study using a 150-micro gram bolus injection, we failed to find any significant difference in analgesic efficacy for intramuscular compared to epidural clonidine . However, in view of the small number of patients included in that study, the clonidine dose was probably too low and the resulting effects too limited in duration to produce a significant difference between epidural and intramuscular injections.
In a double-blind study conducted by De Kock et al. , patients received a 4-micro gram/kg initial dose of clonidine followed by a 2-micro gram centered dot kg-1 centered dot h-1 continuous infusion via an epidural or intravenous catheter. Infusions were started intraoperatively and continued postoperatively. Epidural clonidine was more efficient, since self-administered morphine requirements were reduced during the initial postoperative period. However, in this study, residual effects of the different between-group doses of the opioid used to control increased heart rate during surgery may have biased the results, which were collected in the postoperative period. In addition, there is a discrepancy between visual analog scale scores, which were comparable for both groups, and patient analgesia scale scores, which were more favorable for the epidural group.
Recently, Carroll et al.  reported that a 150-micro gram bolus of intravenous clonidine gave better results than the same dose administered epidurally in patients suffering from chronic low-back pain, although differences were observed for only one of the five scales used to assess pain relief. In three patients, unexpected prolonged pain relief lasting up to 56 days was noted after a single low dose of clonidine, which either calls in question the validity of pain assessment or suggests that inclusion in the study protocol gave these patients a psychologic benefit.
In the present study, clonidine was administered by a patient-controlled device to document differences in the clonidine doses required to achieve the same level of analgesia. This end-point was clearly obtained, showing that the epidural and intravenous routes are not equivalent in terms of analgesic efficacy. Reduction of the clonidine requirement by the use of the epidural route provided indirect evidence that the spinal cord is the main site for the analgesic action of clonidine. Thus, for an identical level of analgesia, doses and resulting plasma concentrations of clonidine were lower in the epidural group than in the intravenous group, which might result in some inter-group differences for adverse effects, such as sedation and possibly abnormal ventilatory patterns.
Sedative-hypnotic effects of alpha2-adrenergic agonists are related to the inhibition of neural firing in the locus coeruleus , a brainstem nucleus located in the dorsal part of the medulla. This supraspinal effect is logical after systemic administration of clonidine, but has also been documented after intrathecal injection , probably because of a rostral cerebrospinal fluid spread . After epidural injection, both vascular absorption and rostral cerebrospinal fluid spread might account for this central effect. As plasma clonidine concentrations are comparable after intravenous or epidural administration of a similar dose owing to rapid and extensive absorption of the drug from the epidural space [9,16], reduction of the clonidine dose by the use of the epidural route is logically associated with lower plasma concentration. Use of the epidural rather than the intravenous route resulted in less-sedated patients in our study. This finding can be explained by the difference in the clonidine doses used in our two groups. As the ratio between these doses was low in comparison with the ratio between plasma and cerebrospinal fluid concentrations after epidural administration (clonidine concentrations are 100- to 1,000-fold higher in cerebrospinal fluid , it is likely that the amount of clonidine reaching the locus coeruleus via rostral cerebrospinal fluid movements was very low in our epidural group. Thus, supraspinal effects observed after epidural injection of clonidine could be mainly due to blood absorption of the drug.
As in our previous studies [6,17], analgesia by intravenous clonidine infusion did not result in significant respiratory depression, at least in terms of blood gas analysis and respiratory rate. One episode of oxygen desaturation occurred in the present study near the end of intravenous loading dose administration. It is likely that a major and/or sudden increase in plasma clonidine concentration was necessary to depress respiration. Definitive conclusions could be not validated by this unique observation. However, these data are consistent with results of a recent animal study showing that intravenous, but not intrathecal or intracisternal, administration of alpha2-adrenergic agonist results in substantial respiratory depression .
Clonidine acts at the brainstem level to produce hypotension and/or bradycardia . As supported by the study of Kirmo et al. , supraspinal actions could be responsible for the decrease in sympathetic tone after both intravenous and epidural administration of clonidine. However, there is some evidence that clonidine can also affect arterial pressure by specific actions on sympathetic preganglionic neurons at the spinal cord . In our study, comparable blood pressure was observed in both epidural and intravenous groups, despite lower plasma clonidine concentrations in the epidural group. A possible explanation is that the epidural administration of clonidine was associated with direct actions on sympathetic spinal neurons, leading to hemodynamic effects sufficient to rule out the differences due to lower plasma clonidine concentration.
One criticism of the current study might be that it was difficult to achieve the thoracic level with clonidine administered at the lumbar level, and that efficacy of the epidural clonidine might be underestimated by inadequate pain relief in the upper part of the back. However, because only thoracic intrathecal clonidine injection can decrease arterial blood pressure, whereas similar doses of clonidine do not affect arterial blood pressure when injected at other spinal sites [22,23], it is conceivable that there was sufficient thoracic migration of epidural clonidine in our study to cause sympathetic depression as well as pain alleviation. In this respect, decreases in arterial blood pressure, and thus ephedrine requirements, were delayed in our epidural group, which may indicate that the effective drug concentration at the thoracic spinal sites was obtained after a slight but significant delay due to the kinetics of the drug.
In conclusion, clonidine analgesia may be achieved by both the epidural and intravenous routes in postoperative patients. The overall dose of clonidine is lower by the epidural route, which clearly indicates that the spinal cord is the main site for the analgesic action of clonidine. After epidural administration, part of the hypotension may be due to direct action on spinal sympathetic neurons, whereas sedation seems to be more dependent on vascular absorption. From a clinical point of view, the balance between analgesia and side effects appears to be better after epidural than intravenous clonidine administration, because a reduction in the clonidine dose is associated with a lower level of sedation, which may be a worthwhile advantage in postoperative patients. In both cases, a decrease in arterial blood pressure is easily controlled by low-dose ephedrine.
The authors wish to thank research nurses M-P. Legendre, S. Leveque, B. Marais, C. Morvan, and N. Poirier for technical assistance and Dr. N. Passuti, Professor of Orthopedic Surgery, for his valuable collaboration.
1. Eisenach JC, Lysak SZ, Viscomi CM. Epidural clonidine analgesia following surgery: phase I. Anesthesiology 1989;71:640-6.
2. Mendez R, Eisenach JC, Kashtan K. Epidural clonidine analgesia after cesarean section. Anesthesiology 1990;73:848-52.
3. Huntoon M, Eisenach JC, Boese P. Epidural clonidine after cesarean section. Anesthesiology 1992;76:187-93.
4. Bonnet F, Boico O, Rostaing S, et al. Clonidine-induced analgesia in postoperative patients: epidural versus intramuscular administration. Anesthesiology 1990;72:423-7.
5. Segal IS, Jarvis DJ, Duncan SR, et al. Clinical efficacy of oraltransdermal clonidine combinations during the perioperative period. Anesthesiology 1991;74:220-5.
6. Bernard JM, Hommeril JL, Passuti N, et al. Postoperative analgesia by intravenous clonidine. Anesthesiology 1991;75:577-82.
7. Murata K, Nakagawa I, Kumata Y, et al. Intrathecal clonidine suppresses noxiously evoked activity of spinal wide dynamic range neurons in cat. Anesth Analg 1989;69:185-91.
8. Eisenach JC, Detweiler M, Hood D. Hemodynamic and analgesic actions of epidurally administered clonidine. Anesthesiology 1993;78:277-87.
9. De Kock M, Crochet B, Morimont C, et al. Intravenous or epidural clonidine for intra- and postoperative analgesia. Anesthesiology 1993;79:525-31.
10. Arndts D, Stahle H, Forster HG. Development of a RIA for clonidine and its comparison with the reference methods. J Pharmacol Methods 1981;6:295-307.
11. Meert TF, De Kock M. Potentiation of the analgesic properties of fentanyl-like opioids with alpha2-adrenoceptor agonists in rats. Anesthesiology 1994;81:677-8.
12. Boudreault D, Brasseur L, Samii K, Lemoing JP. Comparison of continuous epidural infusion plus either continuous or patient controlled epidural injection of fentanyl for postoperative analgesia. Anesth Analg 1991;73:132-7.
13. Carroll D, Jadad A, King V, et al. Single dose, randomized, double-blind, double-dummy cross-over comparison of extradural and I.V. clonidine in chronic pain. Br J Anaesth 1993;71:665-9.
14. De Sarro GB, Ascioti C, Froio F, et al. Evidence that locus coeruleus is the site where clonidine and drugs acting at alpha1- and alpha2-adrenoreceptors affect sleep and arousal mechanisms. Br J Pharmacol 1987;90:675-85.
15. Marwaha J, Kehne JH, Commisaris RL, et al. Spinal clonidine inhibits neural firing in locus coeruleus. Brain Res 1983;276:379-82.
16. Post C, Gordh T Jr, Minor BG, et al. Antinociceptive effects and spinal cord tissue concentrations after intrathecal injections of guanfacine or clonidine into rats. Anesth Analg 1987;66:317-24.
17. Bernard J-M, Lagarde D, Souron R. Balanced postoperative analgesia: effect of intravenous clonidine on blood gases and pharmacokinetics of intravenous fentanyl. Anesth Analg 1994;79:1126-32.
18. Sabbe MB, Penning JP, Ozaki GT, et al. Spinal and systemic action of the alpha2 receptor agonist dexmedetomidine in dogs. Anesthesiology 1994;80:1057-72.
19. Kubo T, Misu Y. Pharmacological characterization of the alpha-adrenoreceptor responsible for a decrease of blood pressure in the nucleus tractus solitarii of the rat. Naunyn Schmiedebergs Arch Pharmacol 1981;317:120-5.
20. Kirmo K, Lundin S, Elah M. Epidural clonidine depresses sympathetic nerve activity in humans by a supraspinal mechanism. Anesthesiology 1993;78:1021-7.
21. Guyenet PG, Cabot JB. Inhibition of sympathetic preganglionic neurons by catecholamines and clonidine: mediation by an alpha adrenergic receptor. J Neurosci 1981;1:908-17.
22. Eisenach JC, Tong C. Site of hemodynamic effects of intrathecal alpha2-adrenergic agonists. Anesthesiology 1991;74:766-71.
23. De Kock M. Site of hemodynamic effects of alpha2-adrenergic agonists. Anesthesiology 1991;75:715-6.