Dr R. J. McInnes, Division of Nursing and Midwifery, University of Glasgow, G12 8LW, UK.
Objectives To compare the efficacy of diamorphine administered by a patient-controlled pump (patient-controlled analgesia) with intramuscular administration for pain relief in labour.
Design Randomised controlled trial.
Setting The South Glasgow University Hospitals NHS Trust.
Sample Primigravidae and multigravidae in labour at term (37–42 weeks).
Methods Women were randomised in labour to the study (patient-controlled analgesia) or control group (intramuscular). Randomisation was achieved through a random permuted block design stratified by parity. Study group women were given a loading dose of 1.2 mg diamorphine intravenously and then attached to the pump. Control group women received intramuscular diamorphine as per hospital protocol. Participants were also given 3 mg of buccal Stemetil. Data were collected throughout labour and at six postnatal weeks.
Main outcome measures Analgesia requirements during labour and women's satisfaction with the method of pain relief.
Results Women in the study group (patient-controlled analgesia) used significantly less diamorphine than women in the control group (intramuscular) but were significantly more likely to state that they were very dissatisfied with their use of diamorphine and were significantly more likely to opt out of the trial before the birth of the baby. The majority of women in both groups used other analgesia concurrent with diamorphine such as Entonox, aromatherapy or TENS.
Conclusions Patient-controlled analgesia administration of diamorphine for the relief of pain in labour offers no significant advantages over intramuscular administration. The results also suggest that diamorphine is a poor analgesic for labour pain irrespective of the mode of administration.
In recent years, policy changes within the maternity services have recommended that care should be woman-centred offering more choice and control to women.1 With regard to the provision of pain relief in labour, a number of recent studies2–4 have shown that woman are often faced with limited choices controlled ultimately by those who provide them. The CRAG/SCOTMEG3 report on pain relief in labour highlighted that links exist between pain, anxiety and distress and how much control women have over their own pain relief. The major determinants of the use of pain relief in labour are the method planned by the woman and the duration of labour. These factors in turn are modulated by the mode of labour onset and the parity of the woman.3 The National Birthday Trust survey4 conducted in 1990 found that most women used some form of pain relief in labour, and of those 38% used intramuscular opioids, mainly pethidine.
In Scotland, diamorphine has overtaken the use of pethidine as the most popular intramuscular analgesic in labour.3 Diamorphine has been reported to provide effective analgesia associated with euphoria and less maternal vomiting than traditionally associated with pethidine use.5 In clinical use, diamorphine may cause nausea, vomiting, feelings of loss of control in the mother and/or neonatal sedation.
Satisfaction in childbirth is not necessarily dependent on the complete absence of pain. Many women accept that some pain is inevitable, but they do not want it to overwhelm them. The need for a perceived sense of control during labour has been demonstrated in a number of studies.6–8 Loss of control during labour is associated with lower maternal satisfaction and postnatal wellbeing scores. The literature suggests that there are a number of advantages of self-administration of opioid analgesia. The psychological influence of self-controlling analgesic administration appears to have a beneficial effect on the overall pain experience and may lead to a greater tolerance of pain.9 Perhaps even most importantly, it may enhance the woman's feeling of control in labour leading to increased maternal satisfaction.
The purpose of the study was to compare the efficacy of diamorphine administered by a patient-controlled pump with intramuscular administration by means of a randomised controlled trial. The specific objectives of the study were to compare the two modalities of administration with regard to the following outcomes:
Primary outcome measures:
•analgesia requirements during labour
•women's satisfaction with the method of pain relief
Secondary outcome measures
•women's perceptions of pain in labour
•presence of side effects (e.g. nausea, vomiting, drowsiness, disorientation)
•clinical outcomes for the mother and baby
The study was conducted in the South Glasgow University Hospitals NHS Trust, which serves the population of the south side of Glasgow and conducts approximately 3000 deliveries per year. Ethical approval was obtained before commencement of the study from the Ethics Committee of the hospital. Women were recruited from hospital and community-based antenatal clinics and both primigravidae and multigravidae were included.
The sample consists of pregnant women attending antenatal clinics at the South Glasgow University Hospitals NHS Trust, Glasgow between August 2000 and March 2002, who intended to give birth in the hospital and who fulfilled the criteria for eligibility. Eligible women were those who were in labour at term, met the criteria for midwife care (standard protocol for SGH), requested diamorphine as their analgesia of choice and gave written consent to participate. All women who potentially met the criteria for entry were given an information leaflet about the study at around 30 weeks of gestation. Contact details for the research midwife were included so that women could obtain further information about the study if desired. This allowed women to consider participation in the study prior to the onset of labour. When women were seen at the hospital antenatal clinic at 36 weeks of gestation, they were asked if they were willing to participate in the study. If they agreed, a consent form and an antenatal questionnaire were completed and the case notes tagged. When study participants were admitted to hospital in labour, eligibility and confirmation of continued willingness to participate (informed consent) in the study were ascertained by the research midwife or the co-ordinating labour ward midwife.
The sample size calculation was based on the percentage of women who would require 7.5 mg or less in the intramuscular and patient-controlled analgesia groups. In the pilot study,10 these were 70% and 90%, respectively. If the true difference was 10% rather than the 20% observed in the pilot (e.g. 75%vs 85%), a study with 400 women per group (800 in total) would have 95% power to detect a difference of this size at the 5% level of significance. The pilot study also found that 40% of women overall required rescue analgesia. A sample of this size will have 82% power to detect a difference of 10% between the percentages of women requiring rescue analgesia (35%vs 45%).
Randomisation to the study (patient-controlled analgesia) or control (intramuscular analgesia) groups was achieved although a random permuted block design stratified by parity. The randomisation lists were prepared in advance by the research team and group allocation details were sealed in opaque envelopes by the research team secretary. Envelopes were numbered to enable assessment of the integrity of the randomisation procedure. Randomisation to either the study group or the control group took place when women who had consented to take part were in labour. At this time, an envelope containing group allocation, study instructions and the intrapartum data collection tool was opened.
Women allocated to the patient-controlled analgesia group were given a loading dose of 1.2 mg (4 mL) of diamorphine intravenously then attached to a Vygon disposable pump (manufactured in the EC), containing 15 mg of diamorphine in 50 mL of normal saline. The pump was set to deliver 0.5 mL a dose with a 5 minute cut out, thus the maximum dosage available in an hour would be 1.8 mg. The pump would last 8 hours if delivering the maximum dose per hour. The Vygon pump was selected on the basis of ease of setup and portability which allowed women to ambulate in labour if they wished. Women were also given 3 mg of buccal stemetil at this time.
Women allocated to the intramuscular group received either 5 mg diamorphine (multigravidae) or 7.5 mg diamorphine (primigravidae) by intramuscular injection and 3 mg buccal stemetil. This was the normal diamorphine prescription available to all women in labour in the SGH.
A training programme was developed to enable core labour ward staff to set up and administer patient-controlled analgesia diamorphine. Refresher training was provided throughout the study period to maintain staff skills and confidence. During the study, the patient-controlled analgesia was set up by a trained member of staff who was available on the shift.
Data were collected from women at three time points:
1An antenatal questionnaire completed at around 36 weeks of gestation provided information on some of the variables known to influence perception of pain (ethnicity, anxiety) as well as previous experience of labour and birth. This form was designed to be self-completed by eligible women when they attended the antenatal clinic and after informed consent had been obtained.
2An intrapartum data collection tool was completed by the attending midwife. This included clinical data for the mother (mode of labour onset, cervical dilatation on admission to the study, length of labour, augmentation, analgesia requirements (including the need for rescue analgesia, mode of delivery) and the baby (Apgar scores, need for resuscitation, cord pH, feeding and birthweight). The woman's perception of pain in labour was measured by a verbal descriptor and a visual analogue scale. The verbal descriptor described pain at four levels (no pain, some pain, a great deal but bearable and a great deal unbearable). The visual analogue scale indicated pain level on a 10 cm coloured slide rule. Pain scores were repeated every hour between contractions. Measurements of pain scores, effectiveness of analgesia and monitoring of side effects continued as long as the participant used diamorphine.
3A postnatal questionnaire was mailed to participants six weeks after the birth. Mothers were asked to recall their labour and comment on the effectiveness of the pain relief and their satisfaction with their pain relief.
Data were entered into a database created in Access by the project secretary. Random cross-checking of data to ensure accuracy was conducted by the research midwife. Any missing data were retrieved from the case notes where possible. Data analysis was performed by the statistician using Minitab. Analysis was conducted on the basis of intention to treat, except for the comparison of satisfaction scores which were based on those who responded to the postnatal questionnaire (79% response rate). Because of the differences in response between primigravidae and multigravidae, separate analyses were carried out for these two subgroups. Percentages were compared between the patient-controlled analgesia and the intramuscular treatment groups using χ2 tests or Fisher's exact test if numbers were small. Results are presented as the ratio of percentages, or risk ratio (patient-controlled analgesia/intramuscular). Two-sample t tests or Mann–Whitney U tests were used to compare the means (or medians) of numerical variables between the patient-controlled analgesia and intramuscular treatment groups. It was originally planned to make adjustments for the effects of confounding variables using multiple linear regression and multiple logistic regression, but these analyses were not carried out in view of the results obtained from the basic analysis. No formal adjustment has been made for multiple comparisons, but the fact that multiple comparisons have been made for secondary outcome variables should be borne in mind when interpreting confidence interval for differences in means which just fail to include 0 and confidence interval for ratios of percentages which just fail to include 1.
A total of 1161 women (628 primigravidae and 533 multigravidae) completed the first antenatal form and consented to participate should they choose to use diamorphine as labour analgesia. Of these, 356 were actually randomised to the study and control groups (228 primigravidae and 128 multigravidae). Five primigravidae and eight multigravidae were given intramuscular instead of patient-controlled diamorphine, however, intention-to-treat analysis is used throughout. Intrapartum data collection forms were completed for all women randomised. Postnatal questionnaires were sent to all 356 women at six weeks and returned by 282 (79.2%) (Fig. 1).
There were no significant differences in baseline variables in either primigravidae or multigravidae (Table 1).
Table 1. Baseline variables. Values are expressed as n/n (%), mean [SD] or mean (range). Response numbers for each variable differ slightly due to missing data or incomplete forms.
Primigravidae (N= 228)
Multigravidae (N= 128)
Patient-controlled analgesia (n= 113)
Intramuscular (n= 115)
Patient-controlled analgesia (n= 66)
Intramuscular (n= 62)
Ethnicity: proportion who were white
Intended to breastfeed
Mean weight at end of pregnancy (kg)
Support person in labour
Analgesia used prior to diamorphine
Spontaneous onset of labour
Mean gestation (weeks)
Mean cervical dilatation prior to diamorphine (cm)
Mean visual analogue scale pain score prior to diamorphine (cm)
Verbal descriptor for pain prior to diamorphine (% stating ‘a great deal of pain’)
Analgesia requirements in labour
Analgesia requirements, including the need for rescue analgesia, were collected on the intrapartum data collection form.
In primigravidae, the patient-controlled analgesia group used significantly less diamorphine than the intramuscular group even after adjusting for the fact that duration of diamorphine in the intramuscular group was longer (intramuscular mean 3.2 mg per hour, patient-controlled mean 1.7 mg per hour; difference 1.5 mg per hour: 95% CI 1.1 to 1.9 mg per hour, P < 0.001). However, there was a greater need for ‘rescue’ analgesia among the patient-controlled analgesia group with a greater number opting for an epidural, although these differences were not statistically significant (patient-controlled 60%, intramuscular 52%; RR 1.15: 95% CI 0.92 to 1.45). Fewer patient-controlled analgesia users remained in the trial for the duration of their labour and birth compared with intramuscular users (Table 2). The majority of women in both groups used some other form of analgesia concurrent with diamorphine use such as Entonox, aromatherapy or TENS.
Table 2. Analgesia requirements in labour. Values are given as mean [SD] or n (%). Confidence intervals are for differences in means or for risk ratios. Response numbers for each variable differ slightly due to missing data or incomplete forms.
The results for multigravidae were similar in that the patient-controlled analgesia group used significantly less diamorphine compared with the intramuscular group. The intramuscular group used diamorphine for slightly longer but the difference was less marked than it had been in the primigravidae. Even after adjusting for this difference, patient-controlled analgesia users still used significantly less diamorphine per hour (intramuscular mean 3.1 mg per hour, patient-controlled mean 1.5 mg per hour; difference 1.6 mg per hour: 95% CI 1.1 to 2.0 mg per hour, P < 0.001; Table 2).
Significantly fewer multigravidae in the patient-controlled group (61%) completed their labour using diamorphine compared with intramuscular users (79%, RR 0.77; 95% CI 0.61 to 0.97), but the need for an epidural was similar between the two groups and much lower than it had been in the primigravid group. Primigravidae were also significantly less likely to complete the trial than multigravid participants (37%vs 70%; RR 0.53; 95% CI 0.43 to 0.65). The majority of both groups used some other form of analgesia concurrent with diamorphine use such as Entonox, aromatherapy or TENS.
Women's satisfaction with method of pain relief
Satisfaction with the analgesia used was measured at six weeks after delivery. Women were asked to recall their experience of childbirth and to answer a series of questions related to both the first and second stages of labour.
Primigravidae allocated to the patient-controlled group were significantly more likely to state that they were very dissatisfied with their use of diamorphine compared with those in the intramuscular group (patient-controlled 35%, intramuscular 7%; RR 5.08; 95% CI 2.22 to 11.61) and, conversely those allocated to the intramuscular group were more likely to state that they were very satisfied. Only 34% of primigravidae in the patient-controlled analgesia group stated that they would use diamorphine again compared with 61% of intramuscular users (RR 0.56; 95% CI 0.40 to 0.79; Table 3). Fewer primigravidae in the patient-controlled analgesia group said they felt in control during labour and enjoyed the birth of their baby and more remembered their labour as being painful, although these results did not reach statistical significance. The majority of primigravidae agreed that they had received pain relief at the right time (71% intramuscular, 61% patient-controlled analgesia).
Table 3. Satisfaction with analgesia. Values are given as n (%). Confidence intervals are for risk ratios. Response numbers for each variable differ slightly due to missing data or incomplete forms.
Felt in control of labour: completely or quite (%)
0.87 (0.70 to 1.08)
0.88 (0.69 to 1.11)
Felt in control of birth: completely or quite (%)
1.18 (0.88 to 1.58)
0.99 (0.70 to 1.39)
Pain in labour was unbearable (%)
1.48 (0.89 to 2.47)
1.56 (0.91 to 2.65)
Very satisfied with pain relief in labour (%)
0.65 (0.38 to 1.13)
0.39 (0.15 to 1.01)
Received labour analgesia too late (%)
1.61 (0.94 to 2.73)
2.32 (1.20 to 4.49)
Very satisfied with diamorphine in labour (%)
0.27 (0.12 to 0.63)
0.07 (0.01 to 0.53)
Very dissatisfied with diamorphine in labour (%)
5.08 (2.22 to 11.61)
4.29 (1.33 to 13.80)
Would use diamorphine again (%)
0.56 (0.40 to 0.79)
0.59 (0.42 to 0.84)
The results for multigravidae were similar. Women allocated to the patient-controlled analgesia group were significantly less likely to state that they were satisfied with their pain relief in labour (and with the use of diamorphine in particular, patient-controlled analgesia 2% satisfied, intramuscular 29% satisfied; RR 0.07; 95% CI 0.01 to 0.53). Forty-four percent of those in the patient-controlled analgesia group stated that they received their pain relief too late compared with 19% of intramuscular users (RR 2.32; 95% CI 1.20 to 4.49). Significantly fewer women in the patient-controlled analgesia group said they would use diamorphine again (intramuscular 75%, patient-controlled analgesia 44%; RR 0.59; 95% CI 0.42 to 0.84; Table 3).
Women's perceptions of pain in labour
The woman's perception of pain in labour was measured by a verbal descriptor and a visual analogue scale. The verbal descriptor described pain at four levels (no pain, some pain, a great deal but bearable and a great deal unbearable). The visual analogue scale indicated pain level on a 10 cm coloured slide rule. Pain scores were repeated every hour between contractions. Measurements of pain scores, effectiveness of analgesia and monitoring of side effects continued as long as the participant used diamorphine (Table 4).
Table 4. Women's perceptions of pain in labour. Values are mean [SD] or n (%). Confidence intervals are for differences in means or for risk ratios. Response numbers for each variable differ slightly due to missing data or incomplete forms.
Verbal descriptor minimum (% stating pain was unbearable)
4.71 (2.01 to 11.01)
1.70 (0.89 to 3.23)
Verbal descriptor maximum (% stating pain was unbearable)
1.15 (0.91 to 1.45)
1.05 (0.75 to 1.46)
The minimum and maximum recorded pain scores tended to be lower in the intramuscular group for both primigravidae and multigravidae and some of these differences were statistically significant (Table 4). Women allocated to the patient-controlled analgesia group were less likely to suffer nausea and vomiting than those in the intramuscular group, but these differences were not significant.
There were no statistically significant differences in the clinical outcomes for either primigravidae or multigravidae except that the mean length of labour was shorter for primigravidae in the patient-controlled analgesia group (Table 5).
Table 5. Clinical outcomes in women. Values are mean [SD] or n (%). Confidence intervals are for differences in means or for risk ratios. Response numbers for each variable differ slightly due to missing data or incomplete forms.
The mean 5 minute Apgar score was higher for babies born to both primigravidae and multigravidae in the patient-controlled analgesia group, although this is of no real clinical significance. The need for resuscitation, oxygen and intermittent positive pressure ventilation was lower in the patient-controlled analgesia group but this was not statistically significant (Table 6).
Table 6. Clinical outcomes in infants. Values are given as mean [SD] or n (%). Confidence intervals are for differences in means or for risk ratios. Response numbers for each variable differ slightly due to missing data or incomplete forms.
The purpose of the study was to compare the efficacy of diamorphine administered by a patient-controlled pump with intramuscular administration by means of a randomised controlled trial. A total of 1161 women agreed to participate in the study and of these, 356 were actually randomised. This was a pragmatic trial therefore recruitment and follow up should also be considered within the constraints of a busy labour ward. Several factors may have affected the use of diamorphine during the study period: media coverage; the participating hospital was working towards achieving Baby Friendly Accreditation and opiates were cited as being a possible hindrance to breastfeeding; as well as the increasing availability and use of aromatherapy and other alternative therapies and the promotion of a more active birthing culture. Ultimately, the choice of analgesia was made by the woman in labour regardless of any antenatal consent to participate in a trial and many chose other options which suited them better at the time, some requiring no analgesia. Sample size calculations had been based on diamorphine usage in an earlier study, however, given the changes which now influenced diamorphine usage, the recruitment rate was slower than anticipated and the sample was not attainable with the time and funding available. Even with a sample smaller than originally anticipated, the study still had acceptable power to detect differences between groups. For example, the study had 90% power to detect a difference of 13% between the percentages of women requiring 7.5 mg or less in the intramuscular and patient-controlled analgesia groups (e.g.75%vs 88%).
The primary outcome measures used in the study were women's analgesia requirements during labour and women's satisfaction with the method of pain relief. Both primigravidae and multigravidae allocated to the patient-controlled group used significantly less diamorphine than women allocated to the intramuscular group. However, women in the patient-controlled analgesia group were also significantly less likely to complete labour on the allocated treatment. The proportion of primigravidae (60.2% patient-controlled analgesia vs 52.2% intramuscular) and multigravidae (15.2% patient-controlled analgesia vs 14.5% intramuscular) that required rescue analgesia (epidural) was not significantly different between the groups.
Women allocated to the patient-controlled analgesia group had access to the same amount of diamorphine as intramuscular users but most used only a small amount of diamorphine and then moved quickly onto epidural or other types of analgesia. This seems to suggest that they were not confident that the patient-controlled analgesia would provide them with effective analgesia. It is possible that the beliefs and opinions of the attending staff may have affected the perceptions of the women and her ability to use the patient-controlled analgesia pump effectively. It was not possible to measure accurately the length of time it took from the woman's request for analgesia until the actual administration of the diamorphine, however, anecdotal evidence suggests that it took longer for women allocated to the patient-controlled analgesia group to receive their medication. This was due to the length of time it took to set up the intravenous cannula and the patient-controlled analgesia device. Attending midwives voiced concerns about delays in setting up the patient-controlled analgesia system in a very busy labour ward where only the core staff had been trained in its use. The majority of staff appeared more comfortable with the use of intramuscular diamorphine, which is familiar and quicker to deliver. The psychological effects of taking part in trials and the expectations of both those administering and receiving treatment are well documented.11 Some women stated a dislike of having an intravenous cannula in situ and this again may have affected their perceptions of the effectiveness of patient-controlled analgesia.
Previous studies have related satisfaction with childbirth to feelings of being in control or coping with labour while avoiding large amounts of analgesia.6,7,12,13 In this trial, there was no significant difference in the proportion of patient-controlled analgesia or intramuscular users who said they felt in control during labour, however, multigravidae allocated to the patient-controlled analgesia group were significantly less likely to say they enjoyed the birth (58.8%vs 77.6%) and both primigravidae and multigravidae in the patient-controlled analgesia group were significantly more likely to say they were very dissatisfied with the use of diamorphine in labour and less likely to say they would use diamorphine again. It was hypothesised that patient-controlled analgesia users would feel more in control of their analgesia and therefore more satisfied with their birth experience. However, the initiation of patient-controlled analgesia was still ultimately controlled by the midwife and although pain and its relief may not be important in satisfaction with childbirth, they may become more important where expectations about pain and/or pain relief are not met.13 There is some evidence that midwives are poor judges of pain in women in labour tending both to over-estimate and under-estimate need12,14; this was supported in the present study where significantly more primigravidae in the patient-controlled analgesia group felt that they received the diamorphine ‘too late’.
Epidural analgesia clearly provides effective pain relief for labour,15 however there continues to be variation in epidural provision across Scotland16 and, in any case, not all women will wish epidural analgesia as their first choice of pain relief in labour. At present, large numbers of women continue to receive systemic opioids in labour, however, there is increasing evidence that opioids do not provide good analgesia for labour pain.17,18
There has been comparatively little research into the use of diamorphine as an analgesia during childbirth, although in one study it was shown to be more effective and with fewer side effects than pethidine.5 However, in the same trial, around half of the women who received either diamorphine or pethidine rated that the drug they had been given offered poor pain relief and around 40% of women required further analgesia. Labour pain has been shown to be insensitive to morphine and pethidine, both of which cause heavy sedation17 and a Cochrane review19 stated that there was not enough evidence to compare the efficacy and safety of the various intramuscular opioids used in labour (although this review did not include the use of diamorphine).
The results from the current study suggest that diamorphine is also a poor analgesic for labour pain, irrespective of the mode of administration. So there is an urgent need to consider alternatives. Preliminary investigations of the ultra-short-acting opioid remifentanil and the agent sevoflurane may offer promise for intermittent administration during labour. This would require drug administration to be co-ordinated with the onset of a uterine contraction, in the same way as inhalational analgesia, such as Entonox, is administered.20
There is therefore a need for well-designed and suitably sized trials of optimal pain relief in labour. Such studies should consider not only drugs and their modes of administration, but also the use of alternatives such as non-pharmacological methods, which are largely inadequately researched or incompletely understood.
The results from this study suggest that patient-controlled analgesia administration of diamorphine for the relief of pain in labour offers no significant advantages over intramuscular administration. However, the results do suggest that diamorphine is a poor analgesic for labour pain irrespective of the mode of administration, and that there is an urgent need to consider alternatives.
The authors would like to thank the following people who supported and assisted in this study: Karen Kane (project secretary); the staff of the Southern General Hospital including midwives, record staff, auxiliaries and medical staff; the members of the steering group (L. Wojciechowska, Dr R. O'Connor, Dr M. Carty, I.Woods, M.Gillies) and all the women and mothers who participated in the setting up of the study and those who took part.