Volume 56, Issue 1 p. 54-60
Free Access

High-dose intrathecal diamorphine for analgesia after Caesarean section

R. G. W. Stacey

R. G. W. Stacey

Consultant Anaesthetist, Kingston Hospital, Galsworthy Road, Kingston upon Thames KT2 7QB, UK

Search for more papers by this author
R. Jones

R. Jones

Specialist Registrars in Anaesthesia, St George's Hospital, Blackshaw Road, Tooting, London, UK

Search for more papers by this author
G. Kar

G. Kar

Specialist Registrars in Anaesthesia, St George's Hospital, Blackshaw Road, Tooting, London, UK

Search for more papers by this author
A. Poon

A. Poon

Specialist Registrars in Anaesthesia, St George's Hospital, Blackshaw Road, Tooting, London, UK

Search for more papers by this author
First published: 06 August 2008
Citations: 21
Dr R. G. W. Stacey


Forty women undergoing elective Caesarean section under spinal anaesthesia using hyperbaric 0.5% bupivacaine were randomly allocated to receive either 0.5 mg or 1 mg intrathecal diamorphine. All women received diclofenac 100 mg at the end of surgery and morphine via a patient-controlled analgesia system. Oral analgesics were not used. Postoperative analgesia was more prolonged and more reliable in the 1-mg group. Mean time to first analgesia was 10.2 h in the 1-mg group and 6.9 h in the 0.5-mg group, and 45% in the 1-mg group used no morphine, compared with 10% in the 0.5-mg group. Mean morphine consumption over 24 h was 5.2 mg in the 1-mg group and 10.6 mg in the 0.5-mg group. Pain scores all tended to be lower in the 1-mg group but this was only significant at 4 h. There were no serious side-effects. Minor side-effects were common but well tolerated, and the incidence did not differ between the groups. If intrathecal diamorphine is used in combination with rectal diclofenac and without oral analgesia, then 1 mg provides superior analgesia to 0.5 mg without any worsening of the side-effects.

Early published work on the use of intrathecal diamorphine for postoperative analgesia after Caesarean section [1–3] showed the efficacy of the technique, and tended to show more prolonged analgesia and lower morphine consumption with larger intrathecal doses. Subsequent studies have confirmed a dose-related effect [4, 5] although one study failed to show any prolongation of the analgesia [6]. Doses as high as 2.5 mg have been used in an orthopaedic population [7, 8] and our early experience was that larger doses were more effective in preventing the need for postoperative intramuscular analgesia after Caesarean section [9].

Various studies have concluded that the ideal dose of intrathecal diamorphine is 0.3 mg [1, 2, 4, 5] without examining higher doses. At these low doses both the side-effects and quality of analgesia increase with the dose, and the reason for limiting the dose has been to limit the side-effects. While side-effects are common, we have not found those associated with the higher doses used routinely at our institution difficult to deal with. We therefore set out to test whether the analgesia provided by 1 mg of diamorphine was superior to that of 0.5 mg and whether the side-effects were worse.


The investigation was approved by the local research ethics committee. Informed consent was obtained from 40 patients presenting for elective Caesarean section.

Randomisation was performed after recruitment with shuffled sealed envelopes allocating the patient to groups receiving 1 mg or 0.5 mg of diamorphine.

All patients received oral ranitidine on the night before, and the morning of, surgery. In addition, patients received intramuscular metoclopramide 10 mg, 1 h prior to surgery.

After establishing intravenous access with a 14-gauge cannula, a preload of 1500 ml of Hartmann's solution was given with 30 mg ephedrine in the remaining 500 ml. Diamorphine (1 mg or 0.5 mg) in 0.5 ml 0.9% saline was prepared by an independent anaesthetist or the anaesthetic assistant and presented to the anaesthetist conducting the study who was unaware of the group allocation. The patient was placed in the left lateral position with pillows under the shoulder to reproduce the thoracic curve [10]. A 25-gauge pencil-point spinal needle was inserted at either the L2/3 or L3/4 interspace. After obtaining free flow of cerebrospinal fluid, 0.5 ml of the study solution was injected followed by between 2.75 and 3 ml of hyperbaric 0.5% bupivacaine.

A 6-mg bolus of ephedrine was given and the ephedrine infusion started. The patient was turned fully into the right lateral position until the motor block became symmetrical at which point she was turned into the supine position with a wedge under the right hip.

Arterial pressure was recorded at least every minute with an automatic oscillotonometer and the ephedrine infusion adjusted to maintain blood pressure. Additional increments of ephedrine were given as required.

Anaesthesia was considered adequate for surgery when there was total loss of cold sensation to the T4 level using ethyl chloride.

Oxygen was administered via a Hudson mask at 4 l.min−1 until delivery, at which time 10 units of oxytocin was given intravenously.

At the end of surgery, 100 mg diclofenac was administered rectally and a patient-controlled analgesia (PCA) morphine pump (IVAC PCAM model P5000) was connected using a valved Y connector. The system was programmed to deliver a 1-mg patient demand bolus, with a 5-min lockout and no background infusion. Oral analgesia was not administered for the first 24 h.

Patients were admitted to the recovery ward where they stayed for 4–6 h with continuous pulse oximetry and an automatic oscillotonometer. They were seen regularly during this period and the presence of side-effects such as sedation, nausea and itching elicited and treated as necessary. Pain was assessed at 4 and 24 h using a 100-mm visual analogue score (VAS) for pain at rest and also pain on standard movement (reaching across the bed with one arm to touch the opposite side).

After 24 h the PCA pump was disconnected and regular oral analgesia started. The data for morphine consumption and time to first request for analgesia were then obtained from the pump. Patients were asked about any nausea not recorded previously, to estimate the duration and grade the severity of any pruritus as easily tolerated, persistent or irritating, severe or unbearable.

Audit data from 1995 to 1997 showed a 60% need for intramuscular analgesia after Caesarean section with 0.5 mg, and 20% with 1 mg. Based on these figures, we calculated that 40 patients would be required for an 80% power of detecting a reduction of this magnitude, with a significance of 5% [11].

The statistical tests were performed using the stata 5.0 statistical programme (Stata Corporation, Texas).

Confidence intervals for proportions were derived using the method described by Altman [11]. The Chi-squared test was used to analyse the need for postoperative morphine, the Mann–Whitney U-test was used to compare duration of analgesia, morphine consumption, pain scores and the incidence of side-effects. The t-test was used to compare patient characteristics. A p-value of < 0.05 was considered significant.


Forty patients completed the study. The patients in each group were comparable for age, parity, height and weight (Table 1)

Table 1. Patient details. Values are mean (SD)
Diamorphine dose
0.5 mg
(n = 20)
1 mg
(n = 20)
Nulliparous 1 4
Gestation; weeks 38.5 (0.9) 38.6 (1)
Mean age; years 34 (4.3) 35 (3.9)
Height; cm 166 (6.2) 167 (4.2)
Weight; kg 70 (6.6) 72 (5.3)
Mean bupivacaine dose; ml 2.9 (0.1) 2.8 (0.1)

There was shortlived hypotension in six of the 0.5-mg group and three of the 1-mg group. There was shortlived nausea precipitated by hypotension in two subjects in each group, but no intra-operative emetic episodes. There was no intra-operative discomfort or nausea after the start of surgery in any patient.

Women in the 1-mg group were less likely to request any morphine from the PCA device (p = 0.0029), requested analgesia later (p = 0.0049) and used less morphine over the 24-h period (p = 0.0161) than those receiving 0.5 mg (1, 2, Table 2).

Kaplan–Meier plot showing the proportion of women (%) in each group who had not started using PCA morphine (time from spinal injection, h).

Mean (SD) cumulative morphine consumption over 4, 8, 12 and 24 h.

Table 2. Postoperative analgesia requirements during the 24-h study period
Diamorphine dose
0.5 mg
1 mg p
No. not requiring morphine 2 9 0.0029
Proportion [95% CI] 10% [0–23%] 45% [23–67%]
Mean (SD) time to analgesia
in those making analgesic requests; h
6.9 (3.9) 10.2 (5.7) 0.0049
Mean (SD) cumulative morphine
consumption over 24 h; mg
10.6 (9.5) 5.2 (6.8) 0.0161

The majority of patients were extremely comfortable; only two patients (both in the 0.5-mg group) had a VAS score at rest of > 30 mm, whereas six patients in the 0.5-mg group and three in the 1-mg group had a VAS > 30 mm on movement. The pain scores all tended to be lower in the 1-mg group, but this only achieved significance at rest after 4 h (Table 3).

Table 3. Visual analogue scores for pain performed at rest and on movement at 4 and 24 h. Numbers are mean (range)
Diamorphine dose
0.5 mg
1 mg p
4 h at rest 5.9 (0–23) 1.4 (0–9) 0.0324
4 h on movement 13.4 (0–41) 6.25 (0–25) 0.09
24 h at rest 11.5 (0–60) 3.6 (0–15) 0.142
24 h on movement 26.4 (4–60) 19.8 (0–41) 0.30

Postoperative nausea occurred in just under half of the patients in each group, with no differences between the groups. It was treated aggressively with the early use of cyclizine, which was effective on each treatment (Table 4).

Table 4. The incidence of nausea and vomiting
Diamorphine dose
0.5 mg
1 mg
No postoperative nausea 11 11
Mild nausea 5 6
Moderate nausea, < 2 emetic episodes 2 2
Severe nausea, > 2 emetic episodes 2 1
Needing treatment 6 5

The majority of patients had pruritus (92.5%). Most found it easy to tolerate (57.5%), but around one-third (32.5%) complained that, while the pruritis was not severe, its persistence was irritating. Only one patient complained of severe pruritus. Pruritus started intra-operatively in seven patients, two of whom (in the 0.5-mg group) were given naloxone because it was judged to be severe by the anaesthetist. The mean duration of pruritus was difficult to measure because of its tendency to decrease in intensity gradually, and the retrospective nature of the measurement. The duration of pruritus tended to be longer, but no more severe, in the 1-mg group (Table 5)

Table 5. The incidence of pruritis
Diamorphine dose
0.5 mg
1 mg p
No postoperative pruritis 1 2 ns
Easily tolerated pruritis 11 12 ns
Persistent pruritis 7 6 ns
Severe pruritis 1 0 ns
Needing treatment 2 0 ns
Median duration (range); h 3.5 (0–20) 5 (0–20) 0.070

There were no episodes of excessive sedation and no documented episodes of oxygen saturation below 90%, although the oximetry data are unreliable because the probes were removed for large proportions of the time in all women and only replaced when the women were asleep during the first 6 h. We found monitoring oxygen saturation in this population extremely difficult because of the tendency for the awake mothers, or their carers, to take off the finger probe in order to free their hands when caring for the baby.


This study has shown that, when diclofenac 100 mg is given rectally at the end of surgery, and oral analgesia is not started for 24 h after surgery, intrathecal diamorphine 1 mg provides more reliable postoperative analgesia than 0.5 mg and that the side-effects are comparable, albeit with a trend towards a longer duration of pruritis. This differs from the conclusion of Morris et al. [6] who found that increasing doses up to 0.5 mg were not associated with better analgesia, but suggested that there may be an increased incidence of side-effects. This abstract focused on pain scores only, and did not specify what form of postoperative analgesia was used or quote figures for opioid consumption. We recorded pain scores at 4 and 24 h, and found consistent improvements, although these were only statistically significant at 4 h at rest. However, there was a marked reduction in the need for the morphine used by patients to keep themselves comfortable.

Husaini & Russell [12] noted that studies reporting little need for postoperative opioid analgesia tend to use nurse-administered analgesia rather than PCA, and often used regular oral analgesia. They comment that PCA is the only true way of finding patients' true analgesic requirement free of outside interference or restrictions. Postoperative PCA morphine consumption depends on the anaesthetic used and concurrent use of oral analgesics; recent studies with local anaesthetic alone quote a mean 24-h morphine consumption of 42 mg [5], 46 mg [1], 64 mg [4] and 82 mg [13]. Kelly et al. [4] found decreasing morphine consumption with increasing intrathecal diamorphine with 51 mg morphine at 0.125 mg intrathecal diamorphine, 35 mg at 0.25 mg and 28 mg at 0.375 mg. Skilton et al. [5] found lower consumptions ranging from 25 mg morphine at 0.1 mg intrathecal diamorphine, 21 mg at 0.2 mg, decreasing abruptly to 3 mg at 0.3 mg. Our morphine consumption of 10.6 mg (0.5-mg group) and 5.2 mg (1-mg group) compare favourably with these figures.

We found that 45% of the 1-mg group used no morphine from the PCA device (95% CI 23–67%) compared with only 10% in the 0.5 mg group (95% CI 0–23%).

These proportions are similar to those found previously after a dose of 0.3 mg; 30% (95% CI 2–58%) [5] and 35% (95% CI 14–56%) [1].

Previous studies [4, 5] have shown an increase in pruritis and nausea as the dose of intrathecal diamorphine is increased. This has been one reason for limiting the dose used. Our results, which show that these side-effects are no worse with 1 mg than with 0.5 mg, suggest that a plateau in the severity of these side-effects is reached below the doses used in this study.

Almost half of the study population suffered postoperative nausea; treatment was left to the individual anaesthetist and was used in a quarter of the population with no difference between the groups. It is difficult to make meaningful comparisons about the incidence of nausea between studies because of different ways of collecting the data and differences in prophylactic anti-emetic policy. The incidence of nausea quoted in studies has been as high as 65% [12] and 70% [5] using 0.2-mg and 0.3-mg doses, respectively, and as low as 12.5% [4] with a 0.25-mg dose. Our method of detecting nausea may have underestimated the incidence in this study. Vomiting is a much more clearly defined end point and as such it is easier to compare rates. In our study, 17.5% vomited, with no difference between the groups. Rates of vomiting have been as high as 10% [2] with a 0.2-mg dose, and 19% [2] and 35% [4] with 0.3-mg and 0.375-mg doses, respectively. Skilton et al. [5], while not quoting an actual rate for vomiting, found that no subjects needed anti-emetics with doses up to 0.3 mg. Our audit data have shown that intra-operative nausea is usually precipitated by hypotension [14]. We find that postoperative nausea is often precipitated by movement and responds well to cyclizine. We now give prophylactic cyclizine to all patients receiving intrathecal diamorphine.

Pruritus is common even with smaller doses of diamorphine, occurring in 70–82% [2], 90% [4] and 80% [5] of women receiving doses between 0.2 and 0.3 mg. Skilton et al. [5] observed that the pruritis suffered by the majority of patients was mild. The majority of women in our study found that the pruritus was easily tolerated, or lasted long enough to become irritating, but not unbearable. In routine practice we find that, as long as women have been warned about pruritus beforehand and told to expect it for between 2 and 8 h, it is easily tolerated.

The most worrying side-effect is the potential for respiratory depression. Many units observe all patients receiving spinal opioids on a high-dependency unit (HDU). Withholding of spinal opioids because of a lack of HDU facilities makes the assumption that alternative analgesia regimens are safer. This is questionable because no effective opioid-based regimen is without the potentially dangerous side-effect of respiratory depression [15–18]. In a study of oxygen saturation in women after epidural Caesarean section [16], hypoxic episodes occurred in all women whether they had intramuscular or PCA pethidine or epidural morphine. Severe episodes (< 85% saturation for > 30 s) were more common in those receiving epidural morphine (71%) and intramuscular pethidine (63%) than in the PCA group (30%), but the analgesia was superior in the epidural group.

This pronounced episodic desaturation occurs frequently in those receiving opioids, and is associated with the obstructive apnoea that occurs during sleep [15]. Various authors have noted the association of a high degree of sedation, and a normal respiratory rate with severe respiratory depression [19, 20] or obstructive apnoea [21]. This has led to the introduction of sedation scores as part of the routine monitoring in acute pain services [19, 21].

Given that no effective analgesia regimen is absolutely safe, it is not logical to treat those receiving spinal opioids differently to those receiving opioids by other routes. Patients in this study were observed in a routine recovery unit where they stayed for between 4 and 6 h. Attempts at recording continuous oximetry met with failure, because the finger probes were continually removed by the mothers or carers, in order to make handling the babies easier. Kelly et al. [4] had similar difficulty with oximetry monitoring. The policy of allowing the finger probes to be removed if mothers are awake but replacing them if they were asleep did not detect any episodes of < 90% saturation. It is unlikely that there were no hypoxic episodes but we contend that, in this population, shortlived episodes are not important. Importantly, there were no women who were difficult to rouse.

Over a 2-year audit period from April 1998 to March 2000, 869 women received 0.5 mg or more diamorphine intrathecally as part of their anaesthetic for Caesarean section in our unit. In this population there have been no instances of excessive sedation or severe respiratory depression. The concentration of diamorphine in the cerebrospinal fluid after intrathecal administration decreases very rapidly because of its lipophilicity and rapid uptake by the spinal cord [22, 23]. The time of most danger from respiratory depression is likely to be the first few hours after it is given. As long as close observation is possible for the first few hours postoperatively, we believe that high-dependency care for a 24-h period is not necessary.

In summary, this study has shown that if intrathecal diamorphine is used in combination with rectal diclofenac without oral analgesia being given, then 1 mg provides superior analgesia to 0.5 mg without any worsening of the side-effects. The effects of the smaller dose have worn off in 50% of subjects after ≈ 6 h.