Skip to main content


Archived Comments for: Intrapartum epidural analgesia and breastfeeding: a prospective cohort study

Back to article

  1. Epidural analgesia and breastfeeding

    William Camann, Brigham and Women's Hospital

    8 January 2007

    To the editor;

    I read with interest the recent article by Torvaldsen (Intrapartum epidural analgesia and breastfeeding: a prospective cohort study. Int Breastfeeding Journal, December 2006).

    Cursory analysis of the methodology leads one to the inevitable conclusion that the findings of this study are not supported by the actual data. Most notably, the authors focused on the putative effect of epidural fentanyl as the factor resulting in adverse breastfeeding consequences. It is well recognized that other investigators have studied epidural techniques in other settings, including analysis of the possible effects of fentanyl, and as appropriately noted and cited by both Torvaldsen,, and Jordan in the accompanying editorial (Infant feeding and analgesia in labour: the evidence is accumulating), the data is conflicting. My commentary in this letter is to highlight only the methods and conclusions of the Torvaldsen study.

    The most important problem is that this study, a secondary analysis of data collected for another investigative purpose, using patients who delivered in 1997, reveals that all patients in this cohort who used epidural analgesia for labor also used parenteral pethidine. Thus, it is mathematically and statistically impossible in this cohort to separate any effect of pethidine from any effect of the epidural component. There is no “labor epidural only” group in this study. In fact, if anything, the data actually supports a likely negative effect of parenteral pethidine, rather than that of fentanyl. Figure 3 clearly indicates that the decay rate towards breastfeeding cessation for a group receiving only parenteral pethidine (and no epidural) is identical to the group receiving epidural + pethidine. This is quite suggestive evidence that the pethidine component, not the epidural component, is the factor that needs additional attention.

    The authors do attempt to sort out the pethidine dilemma, by analyzing a group that had “epidural only” vs. “epidural plus pethidine”. However, as noted above, there is no “epidural-only” group due to the universal use of pethidine in the laboring cohort. Thus, for purposes of analysis, the so-called “epidural-only” group was comprised of those women having elective cesarean delivery under epidural anesthesia. The inclusion and analysis of patients having elective cesarean delivery is not appropriate for comparison in a cohort that is ostensibly designed to study the effect of a labour analgesic technique. Proper analysis should have limited the study to include only subjects in labour. This non-randomised study, which includes patients delivering at home, in birthing centers, as well as hospitals, has more than enough obvious confounding variables without adding the additional confounder of the differences between those who underwent labour and those who delivered by elective cesarean. Moreover, an important additional factor here is that the epidural anesthetic technique for cesarean is not the same regional anesthetic technique as used for labour, further rendering comparisons inappropriate.

    Pethidine is a long-acting opioid, with long-acting active metabolites, and is a drug that has for many years been observed to have adverse neonatal effects. Most USA hospitals no longer use pethidine (or meperidine, as it is called in USA) for labour analgesia, due to both fetal and maternal side-effects. In fact, the general trend in labour analgesia today is to avoid any parenteral opioids to the greatest extent possible, owing to the recognition of neonatal effects. Regional analgesia, as provided by modern-day epidural techniques, is the least depressant of the pharmacological methods used for pain relief in labour. In essence, the study by Torvaldsen is quite supportive of this trend. Their data suggests that parenteral pethidine is a likely factor associated with breastfeeding cessation. Thus, rather than implicating epidural fentanyl, something which, as noted above, is statistically impossible based on the methods of Torvaldsen, I believe that that additional attention needs to be paid to the clear adverse effects of parenteral opioids in labour.

    Finally, I agree with the authors that adequate breastfeeding assistance and support is crucial for promoting this important aspect of parenting. The evidence is quite clear that breastfeeding is the optimal form of nutrition for babies. The evidence is also quite clear that very successful breastfeeding outcomes are entirely compatible with a high utilization of epidural analgesia. Patient comfort and satisfaction in labor and optimal infant nutrition are not mutually exclusive goals. I believe that the lactation community and the obstetric anesthesia community need to work in closer collaboration to allow for the best possible maternal and fetal outcomes.

    William Camann, MD

    Director, Obstetric Anesthesia

    Brigham and Women’s Hospital

    Associate Professor of Anesthesia

    Harvard Medical School

    Boston MA USA

    Immediate past-president, Society for Obstetric Anesthesia and Perinatology (SOAP)

    Competing interests


  2. Response to A/Prof William Camann's comment

    Siranda Torvaldsen, NSW Centre for Overweight and Obesity

    29 January 2007

    We appreciate A/Prof William Camann’s interest and thoughtful comments on our paper. We focused more on epidurals than on pethidine because the aim of our study was to investigate whether or not there was any association between epidural analgesia and breastfeeding. The reason behind this aim was that the lactation consultants at our local hospital told us that, since the addition of fentanyl to epidurals, they noticed that women who had had epidurals for the birth of their baby were experiencing more difficulty establishing breastfeeding than women who had not had epidurals. They noticed this in women who had had caesarean sections as well as those women who delivered vaginally. Hence our aim was subtly different from the one suggested by Camann (to study the effect of a labour analgesic technique) and therefore we felt it appropriate to include women who had a caesarean section (37% of whom laboured). We would also like to point out that although delivery type was associated with partial breastfeeding and breastfeeding difficulty in the first week postpartum, it was NOT significant in the survival analyses (P=0.36) and therefore cannot be confounding the relationship between analgesia and breastfeeding cessation. As we state in the paper, we did undertake additional analyses which were restricted to women who gave birth vaginally and the results were essentially unchanged. For readers’ information, the hazard ratios, adjusted for maternal age and education, among women who delivered vaginally were 2.01 (95% CI 1.49–2.72) for the epidural group and 1.66 (95% CI 1.23–2.43) for the pethidine only group.

    We agree with Camann that it is impossible to separate the effects of pethidine and epidural analgesia among women who delivered vaginally because, as we state in the RESULTS section, all women who delivered vaginally with epidurals also used pethidine. We were surprised by this result and it was the reason we did not have an epidural only group in the primary analyses. The reason that so many women had pethidine is not clear. As stated in the paper, it is based on self-report and we did not check any individual patient records. The instruction in the questionnaire asked women to circle ALL answers which apply to the question ‘During labour, what did you use to help relieve the pain’. The third option of the answer was ‘Injection of pethidine (or similar drug)’ and the fourth option was ‘Epidural’.

    It is also very difficult to separate the effects of delivery type and analgesia on breastfeeding in the first few days after birth, as we point out in the DISCUSSION. However this type of analysis does allow for the identification of women who are more likely to be experiencing difficulty and may require additional breastfeeding support. The other reason for focussing more on epidurals than on pethidine was that only epidurals were associated with both breastfeeding difficulties in the first week AND breastfeeding cessation in the first 24 weeks. The pethidine only group was not associated with breastfeeding difficulty or partial breastfeeding in the first week after birth. For readers’ information, the odds ratio for pethidine (adjusted for parity) among women reporting breastfeeding difficulty in the first week was 1.30 (95% CI 0.85–2.00).

    We agree with Camann that the survival analyses show that women who did not have an epidural but received intrapartum pethidine were more likely to stop breastfeeding in the first 24 weeks than women who received no pharmacological analgesia, and there is no significant difference in hazard ratios between women who received pethidine and women who had epidurals (+/- pethidine). In the RESULTS section we state that women who used pethidine were more likely to stop breastfeeding in the first 24 weeks than women who did not use any pharmacological methods of pain relief and readers are directed to the hazard ratios in Table 3. It did occur to us that the effect in the epidural group could have been due to pethidine and, as Camman says, we did attempt to sort this out by re-analysing the data using a pethidine only group. As we state in the paper, the results were not significantly different, and the hazard ratio for the epidural only group was higher (although not significantly so) than the pethidine plus epidural group, which, if the effect was mostly due to pethidine, we would not expect. We note that none of the epidural only group gave birth vaginally so the results must be interpreted with caution. However, as mentioned previously, delivery type was not associated with breastfeeding cessation in the first 24 weeks. Camman makes the valid point that we focus more on fentanyl than on pethidine in the DISCUSSION. This wasn’t our intention and this section is longer than in earlier draft versions of the paper because two important studies investigating the effects of fentanyl on breastfeeding were published after we had finished our analyses (Jordan, 2005 and Beilin, 2005). We felt our discussion would have been incomplete without briefly mentioning these studies.

    We clearly state in the DISCUSSION that some or all of the association could be due to confounding and we do not agree that our conclusions are not supported by the data. Our conclusion merely states that an association between epidural analgesia and breastfeeding exists and that women at higher risk of breastfeeding problems should be provided with adequate assistance and support. We even preface it with ‘Whatever the underlying mechanism…’ We realise that our study raises as many if not more questions as it answers but we believe it provides a useful background for future research in this important area.

    Since this paper was published we have had a few enquiries about the anaesthetic technique used for women undergoing elective caesarean sections. We apologise for not having provided this in our original paper. It is our understanding that large numbers of anaesthetists in the ACT started using spinal anaesthesia exclusively for caesarean section in 1998 (the year after our study), following a debate in October that year on the advantages of spinal anaesthesia over epidural anaesthesia for caesarean section. However some women in our study who did not labour may have received a spinal rather than epidural anaesthetic and it is not possible to identify them from the data collected for this study. It should be noted that the proportion of women who did not labour, some of whom may have had a spinal, was not large (139/1280=11%) and as already mentioned, excluding women who had caesareans from the survival analyses did not alter the results. The recommended technique in 2007 is a combined spinal-epidural technique. Also, the epidural solution mentioned in our study methods was that used for women who delivered vaginally. Women who had a caesarean section would have received higher concentrations of both local anaesthetic and opioid components. Lignocaine 2% ± adrenalin 5ug/ml and /or bupivicaine 0.5% with fentanyl 5ug/ml were commonly used for caesarean sections. We are very grateful to Dr Frank Lah, VMO Anaesthetist at the Canberra Hospital for providing us with this information. Dr Lah also makes the point that epidural anaesthesia and analgesia are extremely safe and effective means of providing a stress free environment for mother and baby during the delivery process.

    We agree wholeheartedly that women’s comfort and satisfaction in labour and optimal infant nutrition are not mutually exclusive goals and it is unfortunate that some of the media coverage has suggested otherwise. Whenever we have had the chance to prevent this portrayal we have done so. For example the Reuters report, which many news outlets used as their source, states:

    “I think the most important message for pregnant women is to get good advice and help with breastfeeding,” Torvaldsen said. Lactation consultants, she noted, can help women learn how to best support breastfeeding and overcome any difficulties they may encounter.

    “For many women, the benefits of epidural analgesia will outweigh the risks and it is important that women feel supported whatever decision they make,” Torvaldsen said.

    Competing interests