Secondary Logo

Journal Logo

Original Article

Comparison of the relative analgesic efficacies of epidural or intramuscular diamorphine following total knee arthroplasty

Green, R. J.*; Chambers, J.; Thomas, P. W.; Monnery, L.; Titley, G.§; Doyle, T.

Author Information
European Journal of Anaesthesiology: November 2007 - Volume 24 - Issue 11 - p 951-957
doi: 10.1017/S0265021507001299



The first documented evidence for the use of epidural opioids was published in the Lancet in 1979 by Behar and colleagues [1], despite the fact that intrathecal opiates had been successfully used nearly a century before [2]. Since then, epidural opioid use has become a routine part of intraoperative and postoperative pain management, analgesia for labour and delivery, and in the management of chronic pain.

Much work has been performed to determine the sites and modes of action of opioids when used in the epidural space. Debate continues as to their true site of action and the influences of concentration, volume and mass on efficacy [3]. The vast majority of this work has focused on the use of epidural opioid delivered as an infusion for postoperative analgesia, or within the obstetric population. Although it has been demonstrated that epidural opioids have increased potency when given in the context of an infusion [4], the evidence supporting a spinal action of epidural opioids in the context of an infusion has been limited. There is, however, evidence to suggest that opioids have a segmental analgesic effect when given as a bolus [5-8], which is supported by physical principles and pharmacological models.

The aim of this study was to compare the analgesic effects of a bolus of diamorphine given by the epidural route with intramuscular (i.m.) administration. Our null hypothesis was that there would be no differences in the analgesic efficacies of diamorphine due to route of administration.


After gaining approval from the West Dorset Local Research Ethics Committee, we recruited patients who were having total knee replacements performed by one orthopaedic surgical team at West Dorset General Hospital. Informed consent was obtained and each patient was randomized to receive either epidural or i.m. diamorphine.

Randomization was performed using a computer-generated sheet, with subject numbers assigned to either treatment modality. On the basis of this, envelopes containing treatment allocation were then assigned to each subject number. The initial computer-generated document was then concealed from those performing the anaesthesia. The patients were blinded as to treatment allocation. A single anaesthetist was responsible for administering the general anaesthetic, epidural injection and subsequent i.m. injection if appropriate to allocation. Follow-up data were collected by recovery nurses and the acute pain service, who were also blinded to the treatment allocation.

Our exclusion criteria included refusal of consent, ASA Grade IV or V, contra-indications to epidural injection, body mass index (BMI) > 35 kg m−2, age > 80 yr, previous opioid use, previous back surgery, non-steroidal anti-inflammatory drugs (NSAID) use on the day of surgery and symptomatic acid reflux. Patients were allowed to choose a regional technique alone, with those opting for this excluded from the trial. Each patient was visited on the night before surgery and all analgesics were stopped. Sedative premedicants were not prescribed. The intravenous (i.v.) access was established in the anaesthetic room, and midazolam (up to 2 mg) was given if required. A lumbar epidural catheter was placed using 10 mL of bupivacaine 0.5% with 1 in 200 000 epinephrine for all patients. Those assigned to the epidural group additionally received 5 mg of diamorphine, which was dissolved in the bupivacaine at this point.

A standardized anaesthetic technique was adopted. Anaesthesia was induced with propofol (2-3 mg kg−1), followed by spontaneous ventilation using oxygen, nitrous oxide and isoflurane through a laryngeal mask airway. It should be noted that opiates were not used intraoperatively unless the anaesthetist considered that epidural anaesthesia had been unsuccessful, based on cardiovascular and respiratory responses to surgical stimulus, in which case, morphine was titrated in 2 mg aliquots until the desired effect had been achieved. These patients were then withdrawn from the trial.

Following induction of anaesthesia, 5 mg of i.m. diamorphine dissolved in 1 in 200 000 epinephrine was administered to those randomized to the i.m. group. All patients had an adhesive plaster placed on their thigh in order to blind recovery and nursing staff to the groups assigned.

Postoperatively, patients were prescribed 10 mg of morphine for rescue analgesia in recovery if required, followed by patient-controlled analgesia (PCA) with a 1 mg of morphine bolus and a 3 min lockout. The PCA system was attached while the patient was asleep. Cyclizine (50 mg i.v.) and ondansetron (4 mg i.v.) were also prescribed for nausea and vomiting as required. Paracetamol and NSAID were withheld for the first 24 h.

Primary outcome measures to examine the analgesic efficacy of each treatment modality included time to first PCA use, total 24 h morphine requirement, i.v. morphine use in recovery, pain scores in recovery and at 24 h. Secondary outcomes assessed the complications of each treatment, specifically pruritis, nausea and vomiting scores in recovery and at 24 h, urinary retention and incontinence.

Pain, nausea and vomiting scores were assessed in recovery when the patient was able to co-operate. A verbal pain scoring system was used where: 0, was no pain at rest or on movement; 1, no pain at rest, mild pain on movement; 2, moderate pain at rest, severe pain on movement; and 3, severe pain at rest. This four-point scoring system was used in preference to the more commonly accepted 10-point visual analogue scale as it was in current use in Dorchester Hospital at the time of the study, and consequently, all staff members were familiar with it.

Nausea and vomiting scores were: 0, no nausea; 1, mild, no treatment required; 2, mild, treatment required; and 3, severe, treatment required. In recovery it was noted whether the patient had been incontinent of urine, and if it had been necessary to administer i.v. morphine. The patient was instructed not to use the PCA until it was required.

Twenty-four hours postoperation, assessment of pain score, nausea score, time from epidural injection to first PCA use and total morphine use in 24 h were noted. Other considerations included the presence of surgical/anaesthetic complications, itching (none/mild/moderate/severe), respiratory depression (minimum respiratory rate and naloxone requirement), urinary complications (retention/incontinence) and mobility (ability to comply with physiotherapy). Follow-up was not performed by the anaesthetist who provided anaesthesia, in order to maintain blinding.

Preliminary data on a small group of patients (n = 10) having epidural diamorphine suggested that mean (SD) for 24-h morphine requirements was 26.2 (19.0) mg. To have 90% power to detect a 50% change in total morphine PCA (i.e. a change of 13.1 mg) in the i.m. diamorphine group relative to the epidural diamorphine group, 44 patients per group were required.

Nominal data were compared between the two groups using Fisher's (or expanded) exact test. Interval data were analysed using t- and U-tests as appropriate. We used P < 0.05 to represent statistical significance. All statistical analyses were performed using Excel for Windows (Microsoft Inc., Redmond, VA, USA) and SPSS for Windows 12.0.1 (SPSS Inc., Chicago, IL, USA).


During the course of the study, bolus epidural techniques had become less commonplace within anaesthetic practice at Dorchester Hospital. When a total of 60 patients had been randomized with 30 being allocated to each group, an interim analysis was performed. Statistical power to detect a 50% (13.1 mg) change was therefore reduced to 64%, and the study had 80% power to detect a difference of 16 mg and 90% power to detect a difference of 18 mg. Of these 60 patients, 48 patients were included in the analysis (see Fig. 1).

Figure 1.
Figure 1.:
Flow chart to show the progress through the phases of the randomized controlled trial.

In total, 11 patients were withdrawn from the trial leaving 23 patients in the epidural diamorphine group and 26 in the i.m. diamorphine group (see Fig. 1). Reasons for withdrawal included two dural punctures, two patients who were given adjuvant analgesia (codeine, paracetamol, NSAID) during the first 24 h, five patients with inadequate follow-up details, one epidural injection that was felt to be inadequate requiring intraoperative morphine, one spinal injection and one patient who received bupivacaine intraarticularly at the end of the procedure. In the i.m. group, recovery data and 24 h data were not collected for two different patients, leaving n = 25 for analysis per group. There was no significant difference for the number of patients withdrawing from either group (Fisher's exact test, P = 0.54). In addition, one patient from the epidural group was withdrawn at 24 h as supplementary analgesia was given. (Fisher's exact test for total withdrawals at 24 h, P = 0.37).

All further analyses are based on patients with outcome data (n = 23 in the epidural group, and n = 26 in the i.m. group). There were 10 males (43%) randomized to the epidural group and 7 males (27%) randomized to the i.m. group (Fisher's exact test, P = 0.25). Mean age in the epidural group was 71 yr (SD 8), with a mean age of 69 (SD 7) in the i.m. group (t-test, P = 0.28).

In the epidural group there was a higher median time to first PCA, lower median total morphine requirement, lower i.v. morphine requirement in recovery and lower pain scores in recovery and at 24 h (Table 1), all of which were statistically significant. There were no significant differences between the groups in secondary outcome measures (Table 2).

Table 1
Table 1:
Results for all primary end-points: (a) =Mann-Whitney U-test, (b) = Fisher's exact test.
Table 2
Table 2:
Results for all secondary end-points.


This study has shown a superior analgesic effect of diamorphine placed in the epidural space compared to diamorphine delivered systemically via i.m. injection. This adds weight to the argument that epidural diamorphine has a direct spinal action. Both groups in the study were equally matched, had the same procedure by the same surgical team and postoperative analgesia provided by PCA alone. The statistical significance of our primary outcomes (time to first PCA use P < 0.001, and total 24 h morphine requirement P < 0.001) leads us to suggest that systemic absorption of diamorphine from the epidural space alone cannot explain the differences between the two groups.

The many barriers opioids face on the passage from the epidural space to laminae II of the dorsal horn grey matter include the contents of the epidural space (fat, venous plexi), spinal meninges, cerebrospinal fluid (CSF) and the lipophilic white matter of the spinal cord [9]. Any drug placed within the epidural space, in accordance with the physical principles, will diffuse down its concentration gradient. The relative lipid/water solubility of the drug will govern the amount of drug available to diffuse (e.g. it is either sequestered in epidural fat or undergoes rapid vascular uptake, reducing drug available to exert a spinal effect [10]). Passing through the vascular meningeal layers also exposes the drug to both systemic absorption and resistance to diffusion. Both strongly hydrophilic and hydrophobic drugs have problems partitioning between the lipid and aqueous states of the arachnoid mater. This leaves drugs of intermediate lipid solubility (e.g. diamorphine) to exhibit the most rapid passage into the CSF [9]. Once in the CSF, rostral spread of opioids is mainly governed by the rate at which the drug is removed into the spinal cord. Finally, the drug must cross into the grey matter of the dorsal horn (hydrophilic tissue) via the more lipophilic white matter (∼80% lipid). The nature of the tissues makes white matter the preferred environment for fentanyl reducing its bioavailability whereas diamorphine and morphine are taken up well by the grey matter increasing its efficacy at a segmental spinal level [11,12].

Diamorphine possesses many of the ideal properties required to maximize delivery to the dorsal horn and provide effective analgesia when administered via the epidural route [3,13,14]. When compared to fentanyl it has less lipid solubility, exhibits lower protein binding and a greater proportion exists in the unionized state at physiological pH due to its lower pKa. All of these properties lead to an increased bioavailability of diamorphine at the segmental opioid receptors on the dorsal horn [9].

There is some debate regarding the more lipid-soluble opioids. A study looking at the fentanyl and sufentanil failed to successfully demonstrate a spinal action when given epidurally [15]. However, there are a number of studies within obstetric patients that have successfully demonstrated that epidural fentanyl may have spinal activity when given both as a bolus and as an infusion [16-18]. Capogna and colleagues [19] looked at the minimum analgesic doses of epidural fentanyl and sufentanil required to produce analgesia in labour. They showed that compared to i.v. administration, the relative potency of sufentanil is diminished, possibly due to its greater lipid solubility and uptake into epidural fat.

Another explanation for the marked difference between the epidural and i.m. groups seen in this study may be the mode of delivery of the epidural opioid. In our study the diamorphine was given as a bolus injection into the epidural space. A study by Ginosar and colleagues [5] looked at the difference between infusion and bolus delivery of epidural fentanyl on the site of action. They demonstrated a segmental analgesic effect following a bolus of epidural fentanyl and a systemic, non-segmental effect during epidural infusion of fentanyl. This hypothesis has been supported by a number of other groups who all demonstrate segmental analgesia following bolus administration of epidural opioids [6-8,18]. This observation has been accounted for through the knowledge that drugs will diffuse down a concentration gradient and the greater the mass of drug within the epidural space, the greater the mass of drug available to reach the dorsal horn. Opioid infusions fail to generate a sufficient concentration gradient and hence fail to demonstrate a spinal action [20-22]. However, Ginosar and colleagues [4] also demonstrated that an infusion of epidural fentanyl was three times more potent than infused i.v. fentanyl, when combined with epidural bupivacaine. They postulated that this could be either a spinal mode of action or a synergistic relationship between the opioid and local anaesthetic [4].

The existence of synergy between epidurally administered local anaesthetics and opioids has recently been questioned in the case of diamorphine where only evidence of addition was found [23].

In this study we compared epidural with i.m. administration. It has been demonstrated that systemic absorption of drug from the epidural space has a similar biphasic blood concentration profile to that of i.m. administration. There is initially a phase of rapid absorption with a peak at 10 min. This is followed by a more sustained slower period of absorption over several hours, presumably relating to distribution into fatty tissue [24]. As a result, one could speculate that the plasma levels in each arm of the trial should have been comparable. However, the epidural diamorphine in this case was administered in a larger volume (10 ml) which may have affected the absorption profile. The pharmacodynamic profile of diamorphine is more complex than presumed. It has been demonstrated that the clinical efficacy of diamorphine when given epidurally is dependent both on the dose and on the concentration of the drug, with smaller volumes being more favourable [3,23], suggesting that dilution, in this case, resulted in the detriment of overall efficacy. Epinephrine 1 in 200 000 was added to both the i.m. and epidural solutions, which, we speculate, may have prolonged the systemic absorption profile, possibly blunting the initial phase of rapid absorption. Previous studies have shown that by adding epinephrine to epidural opioids, it is possible to improve long-term analgesia [25]. This would be consistent with a partly systemic mode of action.

Although not designed to demonstrate a difference, there was no increase in the side-effects associated with systemic absorption of epidural diamorphine. Urinary retention, pruritis or nausea and vomiting scores showed no significant difference between the two groups. This finding is in contrast to studies looking at the use of intrathecal opioids compared with either systemic or epidural diamorphine in which an increase in pruritis was seen [26].

Analysis of those excluded from the trial has shown no significant differences between groups to suggest bias. Despite recruitment to the trial stopping before the sample size target was reached, there were still statistically significant differences in primary outcomes between the two groups. This suggests that the study was not underpowered for the primary outcomes. The possibility that the study was underpowered for detecting differences in the secondary outcomes cannot be excluded. A potential weakness of this study is the fact that plasma drug levels were not measured. However, it has been shown that i.m. administration has similar pharmacodynamic properties to epidural diamorphine, and so one could speculate that the difference seen between the two groups cannot be attributed to either supra-spinal or systemic effects.

We clearly demonstrated a significant difference in the analgesic efficacy of epidural-administered diamorphine compared with i.m. diamorphine. We suggest that the difference seen in this study can be attributed to spinal activity exerted by a bolus dose of diamorphine placed within the epidural space.


We would like to thank Dr Matthew Hough for his help with protocol construction, Dr Penny Alderson and Janice Brown with data collection and Mr Ian Barlow for allowing us to recruit his patients into this trial.


1. Behar M, Magora F, Olshwang D, Davidson JT. Epidural morphine in treatment of pain. Lancet 1979; 10(1): 527-529.
2. Brill S, Gurman GM, Fischer A. A history of neuroaxial administration of local analgesics and opioids. Eur J Anaesth 2004; 21(4): 329-330.
3. McLeod GA, Munishankar B, Columb MO. Is the clinical efficacy of epidural diamorphine concentration-dependent when used as analgesia for labour? Br J Anaesth 2005; 94(2): 229-233.
4. 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: 1439-1445.
5. Ginosar Y, Riley E, Angst M. The site of action of epidural fentanyl in humans: the difference between infusion and bolus administration. Anesth Analg 2003; 97: 1428-1438.
6. Liu SS, Gerancher JC, Bainton BG, Kopacz DJ, Carpenter RL. The effects of electrical stimulation of different frequencies on perception and pain in human volunteers: epidural vs. intravenous administration of fentanyl. Anesth Analg 1996; 82: 98-102.
7. Geller E, Chrubasik J, Graf R, Chrubasik S, Schulte-Monting J. A randomised double-blind comparison of epidural sufentanil vs. intravenous sufentanil or epidural fentanyl analgesia after major abdominal surgery. Anesth Analg 1993; 76: 1243-1250.
8. Cohen S, Pantuck CB, Amar D, Burley E, Pantuck EJ. The primary action of epidural fentanyl after Cesarean delivery is via spinal mechanism. Anesth Analg 2002; 94: 674-679.
9. Bernards C. Understanding the physiology and pharmacology of epidural and intrathecal opioids. Best Pract Res Clin Anaesth 2002; 16: 489-505.
10. Bernards CM, Shen DD, Sterling ES et al. Epidural, cerebrospinal fluid, and plasma pharmacokinetics of epidural opioids (Part 1): differences among opioids. Anesthesiology 2003; 99(2): 455-465.
11. Herz A, Albus K, Metys J, Schubert P, Teschemacher H. On the central sites for the antinociceptive action of morphine and fentanyl. Neuropharmacology 1970; 9: 539-551.
12. Herz A, Teschemacher H. Activities and sites of antinociceptive action of morphine- like analgesics and kinetics of distribution following intravenous, intracerebral and intraventricular application. In: Simmonds A, ed. Advances in Drug Research. London: Academic Press, 1997: 79-117.
13. Kilbride M, Senagore A, Mazier W, Ferguson C, Ufkes T. Epidural analgesia. Surg Gynaecol Obstet 1992; 174: 137-140.
14. Loper KA, Ready LB, Downey M et al. Epidural and intravenous fentanyl infusions are clinically equivalent after knee surgery. Anesth Analg 1990; 70: 72-75.
15. Bernards C, Hill H. Physical and chemical properties of drug molecules governing their diffusion through the spinal meninges. Anesthesiology 1992; 77: 750-756.
16. D'Angelo R, Gerancher JC, Eisenach JC, Raphael BL. Epidural fentanyl produces labour analgesia by spinal mechanism. Anaesthesiology 1998; 88: 1519-1523.
17. Lyons G, Columb M, Hawthorne L, Dresner M. Extradural pain relief in labour: bupivacaine sparing by extradural fentanyl is dose dependent. Br J Anaesth 1997; 78: 493-497.
18. Polley LS, Columb MO, Naughton NN et al. Effect of intravenous vs. epidural fentanyl on the minimum local analgesic concentration of epidural bupivacaine in labor. Anesthesiology 2000; 93(1): 122-128.
19. Capogna G, Camorcia M, Columb MO. Minimum analgesic doses of fentanyl and sufentanil for epidural analgesia in the first stage of labor. Anesth Analg 2003; 96: 1178-1182.
20. van den Nieuwenhuyzen MC, Stienstra R, Burm AG, Vletter AA, van Kleef JW. Alfentanil as an alternative to epidural bupivacaine in the management of postoperative pain after laparotomies: lack of evidence of spinal action. Anaesth Analg 1998; 86: 574-578.
21. Miguel R, Barlow I, Morrell M, Scharf J, Sanusi D, Fu E. A prospective, randomised, double-blind comparison of epidural and intravenous sufentanil infusions. Anesthesiology 1994; 81: 346-352.
22. Menigaux C, Guignard B, Fletcher D, Sessler DI, Levron JC, Chauvin M. More epidural than intravenous sufentanil is required to provide comparable postoperative pain relief. Anesth Analg 2001; 93: 472-476.
23. McLeod GA, Munishankar B, Columb MO. An isobolographic analysis of diamorphine and levobupivacaine for epidural analgesia in early labour. Br J Anaesth 2007; 98(4): 497-502.
24. Mather LE, Cousins MJ. The site of action of epidural fentanyl: what can be learned by studying the difference between infusion and bolus administration? The importance of history, one hopes. Anesth Analg 2003; 97(5): 1211-1213.
25 Keenan GMA, Munishankarappa S, Elphinstone ME et al. Extradural diamorphine with adrenaline in labour: comparison with diamorphine and bupivacaine. Br J Anaesth 1991; 66: 242-246.
26. Bloor K, Thompson M, Chung N. A randomised, double-blind comparison of subarachnoid and epidural diamorphine for elective caesarean section using a combined spinal-epidural technique. Int J Obstet Anaesth 2000; 9(4): 233-237.


© 2007 European Society of Anaesthesiology