To determine the use of pharmacologic analgesia during childbirth when antenatal hypnosis is added to standard care.
To determine the use of pharmacologic analgesia during childbirth when antenatal hypnosis is added to standard care.
Randomised controlled clinical trial, conducted from December 2005 to December 2010.
The largest tertiary referral centre for maternity care in South Australia.
A cohort of 448 women at >34 weeks of gestation, with a singleton pregnancy and cephalic presentation, planning a vaginal birth. Exclusions were: the need for an interpreter; pre-existing pain; psychiatric illness; younger than 18 years; and previous experience of hypnosis for childbirth.
All participants received usual care. The group of women termed Hypnosis + CD (hypnotherapist guided) were offered three antenatal live hypnosis sessions plus each session's corresponding audio CD for further practise, as well as a final fourth CD to listen to during labour. The group of women termed CD only (nurse administered) were played the same antenatal hypnosis CDs as group 1, but did not receive live hypnosis training. The control group participants were given no additional intervention or CDs.
Use of pharmacological analgesia during labour and childbirth.
No difference in the use of pharmacological analgesia during labour and childbirth was found comparing hypnosis + CD with control (81.2 versus 76.2%; relative risk, RR 1.07; 95% confidence interval, 95% CI 0.95–1.20), or comparing CD only with control (76.9 versus 76.2%, RR 1.01, 95% CI 0.89–1.15).
Antenatal group hypnosis using the Hypnosis Antenatal Training for Childbirth (HATCh) intervention in late pregnancy does not reduce the use of pharmacological analgesia during labour and childbirth.
Hypnosis is difficult to define, but appears to be a daydream-like, conscious state of focused attention that can result in a reduced awareness of external stimuli, and an increased response to suggestion.[2, 3] Suggestions are verbal or non-verbal communications that lead to apparent non-volitional changes in patient perception or behaviour, and are pivotal in eliciting hypnotic responses.[4, 5]
Pain relief during childbirth has been a primary target of clinical hypnosis for more than a century. Anaesthesia and clinical hypnosis have had a long association. The introduction of chemical analgesia in the 19th century and negative myths associated with hypnosis have led to a decline in its use. However, since the 1950s there has been renewed interest after hypnosis was formally recognised by the British Medical Association (BMA) Working Party as an effective clinical tool. More recently, positron emission tomography or functional magnetic resonance imaging have shown the consistent finding of anterior cingulate modulation when hypnosis is used to induce analgesia.[10-12]
Perceptions of the increased medicalisation of childbirth have led many women to look for alternative means of relieving pain during labour.[13, 14] Hypnosis, despite being used for some remarkable cases of painless surgery,[15, 16] has not been considered to be a reliable method of labour analgesia. Systematic reviews of randomised controlled trials (RCTs) investigating the effects of hypnosis during childbirth have suggested that hypnosis decreases analgesia requirements during labour, decreases the use of oxytocic labour augmentation, and increases the incidence of vaginal birth.[14, 18, 19] This evidence has come from small or poorly designed trials,[20-24] which were inadequate to confirm these apparent benefits.[25, 26] Antenatal hypnosis training of women in the first and second trimester of pregnancy was found to reduce analgesia use during childbirth, and further study in late pregnancy was advocated. We investigated whether standardised group hypnosis training in late pregnancy reduced analgesia requirements during childbirth.
The Hypnosis Antenatal Training for Childbirth (HATCh) trial was an RCT using a three-arm parallel group design, with a 1:1:1 ratio. It was conducted in the largest tertiary referral centre for maternity care in South Australia, between December 2005 and December 2010. Women between 34+0 and 39+0 weeks of gestation, who were planning a vaginal birth, were not in active labour, and were carrying a singleton, viable fetus of vertex presentation were eligible for inclusion. Exclusion criteria were: previous hypnosis preparation for childbirth; poor understanding of English, requiring a translator; current enrolment in another pregnancy trial in which analgesia requirements were an outcome measure; active psychological or psychiatric problems; and severe intellectual disability. Women with previous hypnosis childbirth training or with pre-existing pain were also excluded.
Following trial registration (ACTRN012605000018617, 18 July 2005; NCT00282204, 24 January 2006), and approval from the Women's and Children's Hospital Human Research Ethics Committee in 2004 and subsequently extended by the Child Youth and Women's Health Service (CYWHS) Human Research Ethics Committee (REC 1600/12/09), informed written patient consent was obtained from eligible women agreeing to participate. After baseline demographic testing was completed, study participants were stratified for parity and randomised in (unspecified) blocks of 15 by a computer random number generator. Allocation concealment was assured by using a computer database assignment to one of three groups, which was only revealed after patient identifiers had been entered.
Participants in the hypnosis + CD group received antenatal hypnosis training from a doctor qualified in hypnosis and were given a CD on hypnosis after each session to supplement the intervention. Participants in the CD-only group received the HATCh intervention by listening to the relevant CD for each session, administered by a nurse with no training in hypnotherapy. After each session, participants in the hypnosis + CD and CD-only groups were provided with a CD on hypnosis designed for each session. The HATCh intervention was administered in a group setting, with attendance over three consecutive weekly sessions, and participants were asked to practise on a daily basis at home using the relevant CD. At the third antenatal session, participants allocated to the hypnosis + CD and CD-only groups were provided with CDs 3 and 4 and asked to use the fourth CD in labour. Control group participants continued with their usual preparation for childbirth, such as antenatal classes and clinic appointments, with no additional intervention. Participants unable to attend an allocated session in person were contacted by telephone, and the relevant CD was posted for them to listen to prior to their next session. All participants received usual care and follow-up, regardless of group allocation. Participants assigned to hypnosis groups commenced hypnosis training as near as possible to 37 weeks of gestation. Treatment allocations were concealed from the treating obstetricians, anaesthetists, midwives, and personnel who were collecting the data. Our primary outcome, the use of pharmacological analgesia during labour and childbirth was defined as using one or more analgesia techniques, such as: inhaled nitrous oxide in oxygen (Entonox™); a parenteral opioid such as intramuscular Pethidine; and/or epidural analgesia, such as 0.2% Ropivacaine with fentanyl, 2 mg/ml. The attending midwife documented all analgesia use and mode of delivery on the birth register. A researcher, blinded to treatment allocation, managed data collection from medical and nursing records on the ward from postnatal questionnaires after the birth, and then at 6 weeks follow-up, as detailed previously. In addition, At 6 months follow–up we asked women about treatment for depression since the birth.
The HATCh intervention was based on published scripts of suggestions, by experts in the administration of hypnosis for childbirth,[27, 28] combined with our own clinical experience. HATCh intervention scripts were used to produce three CDs that mirrored our hypnosis preparation for childbirth, with each CD lasting between 21 and 32 minutes.[29, 30] A fourth CD lasting 18 minutes was also developed for use during labour and childbirth. Women in the hypnosis + CD group were guided in to a state of hypnosis using the preferred induction method of the hypnotherapist: usually an eye fixation technique or Spiegel.[31, 32]
An audit of 100 consecutive mothers birthing at our institution in May 2004 showed an incidence of using one or more pharmacological interventions of 80%. In order to show a clinically relevant difference of 20% in the number of women requiring pharmacological analgesia, we used a two-tailed calculation (nquery advisor 5.0; Statistical Solutions Ltd., Cork, Ireland). A study power of 80% would require each group to contain 135 women to detect this difference at the 0.05 level. We planned to recruit 150 women per group, as we estimated that 10% of participants would deliver prior to receiving their allocated treatment.
Following baseline demographic data collection, which included the administration of the Edinburgh Postnatal Depression Scale (EPDS), Spielberger State–Trait Anxiety Inventory, and Creative Imagination Scale (CIS), trial participants were informed of where and when to attend their allocated sessions.
The Spielberger State–Trait Anxiety Inventory involves asking participants to score 20 statements on a four-point Likert scale, indicating the frequency with which they have felt a certain emotion over the previous 7 days. Scores > 44 have been classed as indicating a state of high anxiety.[36, 37]
The EPDS consists of ten statements that evaluate how an individual has felt over the previous 7 days. The EPDS has been recommended for routine use to identify women at risk for postnatal depression, and a score > 12 is widely used to indicate a probable depressive disorder. Study participants were left undisturbed to measure their baseline hypnotisability by listening to a CD of the CIS. This scale is self-appraising, and the maximum score is 40. The CIS scale CD was made in-house from published scripts, and contained ten scenarios that encourage test subjects to use their imagination to create suggested effects, such as arm heaviness, age regression, and time distortion. Each scenario is ranked by the participant on a five-point Likert scale to score their experience: 0, 0%; 1, 25%; 2, 50%; 3, 75%; 4, 90%.
After completing nearly 2 years of recruitment, we became aware that some women, who were ineligible for participation, had been inadvertently randomised, outside our eligibility criteria, prior to 34 weeks of gestation. We therefore continued to recruit women until our initial planned sample size of eligible women had been reached. Only women who met all eligibility criteria for inclusion were analysed (Figure 1). Participant data were analysed by a researcher who was not blinded to allocation, in accordance and with the ‘intention to treat’ (ITT) principle, and with comprehensive pre-planned subgroup analyses. Initial analyses examined baseline characteristics of all randomised participants, and any chance differences in this baseline data between treatment groups were adjusted for in subsequent analyses, and reported upon if there were any significant differences to unadjusted data.
Preplanned subgroup analyses of primary outcomes were performed for women regarding: induced versus spontaneous labour; attendance at all three sessions, whether each session CD was used at least once; women's hypnotisability, as measured by the CIS; women's beliefs of the efficacy of hypnosis prior to labour; women's expectations of requiring an epidural; women's expectations of having a normal spontaneous birth; and previous use of yoga. Descriptive statistics are presented as mean, standard deviation (SD) for normally distributed data, median and interquartile (IQR) range for skewed data, and proportions for dichotomous outcomes. Likelihood ratio chi-square statistics and relative risks (RRs) with 95% confidence intervals (95% CIs) or odds ratios (ORs) have been used. Differences at the level of P < 0.05 were considered to be significant. For ease of interpretation, relative risks were estimated wherever possible using binomial models with a log-link. In subgroup analyses where models with the log link failed to converge, odds ratios were estimated using a binomial model with a logit link.
The eligibility of 776 women for trial entry was assessed between December 2005 and July 2009, and 448 women were found to be eligible and were included in the study (Figure 1). At 6 weeks postpartum, 400 women (89.3%) returned questionnaires with data suitable for analyses. Our baseline data shows that the randomisation with stratification for parity produced comparable groups (Table 1).
|Baseline demographic data||Hypnosis + CD n = 154||CD only n = 143||Control n = 151||Total n = 448|
|Agea||30.5, 5.1||31.4, 4.4||31.2, 4.7||31.0, 4.7|
|Nulliparity||124 (80.5)||110 (76.9)||114 (75.5)||348 (77.7)|
(n = 151)
(n = 140)
(n = 150)
(n = 441)
|Gestational age (weeks)b||35.1, 1.4||35.3, 1.6||35.0, 1.3||35.1, 1.6|
|Tertiary education||86 (55.8)||84 (58.7)||83 (55.0)||253 (56.5)|
|Australian born||112 (72.7)||107 (74.8)||111 (73.5)||330 (73.7)|
|Previous caesarean section||7 (4.6)||5 (3.5)||6 (4.0)||18 (4.0)|
|Previous epidural||20 (13.0)||20 (14.0)||18 (11.9)||58 (13.0)|
|History of depressionc||43 (27.9)||26 (18.2)||27 (17.9)||96 (21.4)|
|Complementary therapy in pregnancy||86 (55.8)||75 (52.5)||83 (55.0)||244 (54.5)|
|Yoga||65 (42.2)||68 (47.6)||70 (46.4)||203 (45.3)|
|EPDS||6, 6||5, 6||4.5, 6 (n = 150)||5, 5 (n = 447)|
|EPDS > 12d||15 (9.7)||9 (6.3)||6 (4.0)||30 (6.7)|
|Spielberger state||30.0, 13 (n = 153)||29, 13||30, 11||30, 12 (n = 447)|
|Spielberger trait||32.5, 12||31, 12||30, 10||31, 12|
|CIS||23, 11 (n = 137)||23, 10.5 (n = 124)||23, 11 (n = 129)||23, 11 (n = 390)|
The attendance of HATCh trial participants according to their allocated group and self-reported compliance with listening to the CD on at least one occasion is shown according to their allocated group (Table S1). Less than 50% of women allocated to an intervention attended all three allocated hypnosis sessions, and even fewer attended the CD sessions. Hypnosis training undertaken outside the trial was reported by 17 HATCh trial participants (3.9%): two allocated to the hypnosis group (1.3%), seven allocated to the CD only group (5.0%), and eight allocated to the control group (5.6%). Trial participants were no less likely to seek additional training if they were allocated to the hypnosis group (RR 0.24, 95% CI 0.05–1.12, P = 0.07), when compared with controls.
There was no difference in our primary outcome – the use of pharmacological analgesia during childbirth – between groups when comparing hypnosis 125/154 (81.2%) with control 115/151 (76.2%) (RR 1.07, 95% CI 0.95–1.20), and when comparing CD only 110/143 (76.9%) with control (RR 1.01, 95% CI 0.89–1.15). There were also no differences found with respect to the use of epidural analgesia or other pharmacological analgesia (Table 2).
|Analgesia use during childbirth||Hypnosis n = 154||CD only n = 143||Control n = 151||Total n = 448||RR (95% CI)||P|
|Any analgesia||125 (81.2)||110 (76.9)||115 (76.2)||350 (78.1)|| |
|Analgesia, except Entonox||98 (64.1)||81 (56.6)||85 (56.3)||264 (59.1)|| |
|Epidural labour analgesia||78 (51.0)||63 (44.1)||71 (47.0)||212 (47.4)|| |
No differences were found in key secondary outcomes with respect to: mode of delivery; use of augmentation of labour; and other birth outcomes (Table 3), including median length of labour (IQR) of 8 (7.3), 8.3 (7.9), and 7.4 (6.7) hours for women in the hypnosis + CD, CD only, and control groups, respectively. Similarly, there was no difference in the number of days women stayed in hospital postpartum: 4 (2) days for each group. There was an increased use of prostaglandins for induction in women allocated to the hypnosis group when compared with controls, but no significant differences between groups on admission to the high dependency unit (HDU), incidence of episiotomy, need for blood transfusion, or number of days in hospital after the birth (Table 3). There was also no difference in the incidence of exclusive breastfeeding at hospital discharge, although more women in the control group reported that they were exclusively breast feeding at 6 weeks when compared with those in the hypnosis group (P < 0.05).
|Key outcomes||Hypnosis + CD n = 154||CD only n = 143||Control n = 151||Total n = 448||RR (95% CI)||P|
|Oxytocin||57 (37.3)||53 (37.1)||56 (37.1)||166 (37.1)|| |
|Spontaneous vaginal birth||85 (55.2)||84 (58.7)||92 (60.9)||261 (58.3)|| |
|Induced labour||63 (40.9)||43 (30.1)||47 (31.1)||153 (34.2)|| |
|Induction with prostaglandins||55 (35.7)||32 (22.4)||35 (23.2)||122 (27.2)|| |
|Augmentation||66 (42.9)||66 (46.2)||66 (43.7)||198 (44.2)|| |
|Caesarean section||38 (24.7)||25 (17.5)||29 (19.2)||92 (20.5)|| |
|Forceps/vacuum||31 (21.3)||34 (23.8)||30 (19.9)||95 (21.2)|| |
|Episiotomy||24 (15.6)||25 (17.5)||26 (17.2)||75 (16.7)|| |
|Intact perineum||94 (61.4)||71 (49.7)||91 (60.3)||256 (57.3)|| |
|PPH||24 (15.6)||22 (15.4)||14 (9.3)||60 (13.4)|| |
|Blood transfusion||4 (2.6)||7 (4.9)||1 (0.7)||12 (2.7)|| |
|Admission to HDU/ICU||3 (2.0)||7 (4.9)||2 (1.3)||12 (2.7)|| |
|Readmission to hospital|| |
n = 134
n = 133
n = 133
n = 400
|Neonatal Apgar score < 7 at 5 minutes||1 (0.7)||1 (0.7)||2 (1.3)||4 (0.9)|| |
|Baby admitted to SCBU||53 (34.4)||45 (31.5)||51 (33.8)||149 (33.3)|| |
There were no differences between groups with regard to the mother perceiving that she had received adequate pain relief, or of the birth being better than expected or a positive experience (Table 4). There were also no differences between groups regarding, perceived maternal pain intensity, satisfaction with the birth experience, sense of control during labour, ability to cope, or satisfaction with the childbirth experience (Table S2).
|Key outcomes||Hypnosis + CD n = 154||CD only n = 143||Control n = 151||Total n = 448||RR (95% CI)||P|
|Received adequate pain relief|| |
n = 136
n = 131
n = 128
n = 395
|Birth a positive experience|| |
n = 149
n = 139
n = 144
n = 432
|Birth better than expected|| |
n = 144
n = 137
n = 143
n = 424
At the 6–week assessments, there were no differences between the groups in EPDS, Spielberger, zero to 10 numerical pain scores during childbirth, or satisfaction with the birth experience (Table S2). There were no differences in the incidence of maternal re-admission to hospital or problems related to the use of hypnosis between groups (Table S3). Those women exposed to the hypnosis or CD intervention were more likely to state that they would use hypnosis in future pregnancies (Table S3).
Of the planned subgroup analyses, there were no differences between groups in any of the key outcomes. Analgesia use during labour and childbirth subgrouped by parity, induction, CIS score, yoga, previous experience of clinical hypnosis, and compliance with the intervention showed no difference between groups (Table 5). Women who used yoga and who were allocated to receive hypnosis used less analgesia than women who did not use yoga and were allocated to receive hypnosis 70.8 versus 88.8% (P = 0.005). This difference was also seen in the CD-only group, 66.2 versus 86.7% (P = 0.003), but was not seen in the women in the control group who used or did not use yoga (Table 5), 74.3% versus 77.8% (P = 0.616).
|Subgroup||Hypnosis + CD n = 154||CD only n = 143||Control n = 151||Total n = 448||OR (95%CI)|
|Parity||N||105/124 (84.7)||90/110 (81.8)||93/114 (81.6)||288/348 (82.8)|| |
|M||20/30 (66.7)||20/33 (66.6)||22/37 (59.5)||62/100 (62.0)|| |
|Induction/spontaneous||I||59/63 (93.7)||39/43 (90.7)||45/47 (95.7)||143/153 (93.5)|| |
|S||60/85 (70.6)||68/97 (70.1)||67/101 (66.3)||195/283 (68.9)|| |
|CIS score||<23||50/60 (83.3)||48/60 (80.0)||45/63 (71.4)||143/183 (78.1)|| |
|≥23||63/77 (81.8)||47/64 (73.4)||52/66 (78.8)||162/207 (78.3)|| |
|Yoga||Y||46/65 (70.8)||45/68 (66.2)||52/70 (74.3)||143/203 (70.4)|| |
|N||79/89 (88.8)||65/75 (86.7)||63/81 (77.8)||207/245 (84.5)|| |
|Previous experience of clinical hypnosis||Y||8/12 (66.7)||3/5 (60.0)||9/13 (69.2)||20/30 (66.7)|| |
|N||117/142 (82.4)||107/138 (77.5)||106/138 (76.8)||330/418 (79.0)|| |
|Three hypnosis sessions||Y||55/68 (80.9)||44/57 (77.2)||99/125 (79.2)||1.25 (0.53–2.97)a|
|N||70/86 (81.4)||66/86 (76.7)||136/172 (79.1)||1.33 (0.63–2.77)a|
|All CDs listened to||Y||45/58 (77.6)||45/66 (68.2)||90/124 (72.6)||1.62 (0.72–3.62)a|
|N||68/79 (86.1)||59/70 (84.3)||127/149 (85.2)||1.15 (0.47–2.85)c|
Approximately 15% of babies had meconium-stained liquor across the groups, and one-third of babies were admitted to the special care baby unit (SCBU). However, there were no differences in the incidence of meconium-stained liquor, babies with an Apgar score < 7 at 5 minutes, and admissions to the SCBU, between the three groups (Table 3).
Nearly 50% of women in the intervention groups believed that hypnosis was helpful during the birth. No differences were reported with respect to incidence of baby re-admissions to hospital and whether the baby was settled or not (Table S4).
The key finding of the HATCh trial is that antenatal group hypnosis training in the third trimester, using the particular three-session regimen of this study, did not influence the use of pharmacological analgesia during labour and childbirth. In addition, mode of delivery and the use of oxytocics were similarly unaffected. An unexpected finding of our study was that women allocated to the hypnosis + CD group had an increased incidence of having an induction of labour and prostaglandin administration. Hypnosis has been used as an intervention to induce labour. Whether this was a chance occurrence or not, any effect of hypnosis in reducing analgesia may have been masked by more painful, induced labour. Apart from this outcome, there was a surprising lack of statistically significant findings between groups. Of the planned subgroup analyses there were no differences between groups in any of the key outcomes; however, it was interesting to note that women who used yoga and who were allocated to receive hypnosis used less analgesia than women who did not use yoga and were allocated to receive hypnosis. Like hypnosis, yoga also involves the use of imagery, positive suggestion and ‘trance-like’ states. The familiarity with these states may have primed this subgroup of women to learn the hypnosis techniques taught in this study over a shorter period of time.
The HATCh trial was a comprehensive, high-quality, randomised trial that was designed to assess the efficacy of a short, three-session, standardised hypnosis intervention in late pregnancy. Strengths included: the health significance of the issue; the inclusion of both nulliparous and multiparous women; the comprehensive, clinically meaningful nature of the outcomes chosen; the support of preliminary studies and meta-analyses; and the investigators’ experience with the intervention. A rigorous experimental design was used and similar groups at baseline were achieved by using balanced randomisation, stratified for parity. Prior to this trial, only one study had a comparable sample size but was much less rigorous in its methodology, and made no attempt to standardise the intervention. All other previous hypnosis studies in this setting have recruited <100 participants. The use of the CDs was an innovative feature of the study, with which investigators attempted to standardise the suggestions used during the intervention and facilitate replication should the findings had shown efficacy. The psychological measures for depression and anxiety have been previously validated in this setting. The CIS used for hypnotisability assessments were standardised by using an identical CD of the scale and response sheets administered on an individual basis to trial participants at trial entry. This had the advantage of avoiding a formal hypnotic induction, which minimised the possibility of contaminating our control group with an experience of hypnosis. The planned primary outcome of reduced pharmacological analgesic intervention was a reasonable objective, as it was absolute and relatively easy to measure.
The HATCh trial had several limitations. Firstly, there were a number of post-randomisation exclusions resulting from an inadvertent error during the eligibility assessment of gestation at trial entry. We therefore continued recruiting until our initial planned sample size had been reached, a strategy consistent with a suggested approach for the management of inadvertently recruiting ineligible subjects after randomisation. Only a minority of women actually attended all three sessions and listened to all four CDs; however, our subgroup analyses do not support poor compliance as a reason for the lack of efficacy of the hypnosis intervention. Although every attempt was made to conceal treatment allocations from obstetricians, anaesthetists, midwives, and the personnel collecting data, double-blinding a hypnosis intervention is unlikely if women are not fully informed, as in earlier studies.[21-23]
About 80% of all women in our institution received pharmaceutical analgesia. A high proportion of HATCh trial participants had a post-school qualification (Table 1). These highly educated women may not be representative of the pregnant population of other studies where hypnosis training has been found to be effective.
Previous studies, where a beneficial effect has been shown, have used more than three antenatal sessions,[21, 23, 24] and administered the intervention at an earlier gestation than we did in our study.[23, 24] Interestingly, a previous randomised trial where hypnosis failed to reduce analgesia requirements during labour also had a design that involved beginning the antenatal hypnosis training late in the third trimester. The ready availability of epidural analgesia may also have influenced our study findings. If epidural analgesia is not available on demand, as appeared the case in previous studies with positive findings, the likelihood that hypnosis will reduce analgesia use may be increased. Many medical interventions in our institution are protocol driven. In this regard, there may have been rates of intervention, such as caesarean section or epidural use during labour and childbirth, in this tertiary setting that were not amenable to change by the hypnosis intervention. In another study investigating continuous support in labour provided to a low-risk population, in general greater benefits were found when epidural analgesia was not routinely available, including: decreased caesarean section rates and increased spontaneous vaginal birth. Future areas of research might include: a comparison of antenatal hypnosis training inside and outside a tertiary referral centre, where an ‘on demand’ epidural service is unavailable; the use of different methods to teach hypnosis preparation for labour analgesia; and using yoga as part of the hypnosis training regimen.
The HATCh trial represents the best-quality evidence to date investigating the effects of hypnosis training on analgesia use during childbirth.[43, 44] The negative findings may convince many that it is ineffective in the clinical environment we have explored, irrespective of timing or techniques, and it may not be worthwhile pursuing lengthier and less feasible approaches in the hope of gaining benefit in a tertiary referral setting with an ‘on demand’ epidural service. Although our findings show that the HATCh intervention in late pregnancy cannot be recommended, the rigorous methodology used provides a potentially useful template for the conduct of future investigations assessing the effectiveness of different ways of administering a hypnosis intervention for pain relief and other beneficial outcomes in childbirth.
All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author), and declare: support from the University of Adelaide; Australian Society of Anaesthetists; the Women's and Children's Foundation; and NH&MRC project grant 453446. This paper represents, in part, the PhD thesis http://digital.library.adelaide.edu.au/dspace/bitstream/2440/69216/1/02whole.pdf. Dr Allan Cyna is a co-author of the Cochrane review that includes, in part, data from the HATCh trial reported in this article. There were no other relationships or activities that could appear to have influenced the submitted work.
AMC, MIA, JSR, and CAC have contributed to the overall concept and design. AMC and MIA helped design and co-ordinate the final content of the hypnotherapeutic interventions. PB and GA organised the acquisition of the data and the analysis of the data. AMC, JSR, CAC, PB, and GA were involved in drafting the article. Each author has participated sufficiently in the work to take public responsibility for appropriate portions of the content. AMC was responsible for acquiring funding, supervised the research, and takes responsibility for it. All authors have read and approved the final version of the article.
Trial registration: ACTRN012605000018617, 18 July 2005; NCT00282204, 24 January 2006. Initial ethics approval was obtained on 13 December 2004 from the Women's & Children's Hospital Research Ethics Committee (approval number 1600/6/2007), and this approval was extended from the Child Youth and Women's Health Service (CYWHS) Human Research Ethics Committee on 13 December 2006 (approval number REC 1600/12/09).
This study received funding from: the University of Adelaide; the Australian Society of Anaesthetists; the Women's and Children's Foundation; and NH&MRC project grant 453446.
We thank, Drs Celia Whittle, Graham G Wicks, and Patrick McCarthy for their advice on the hypnosis intervention. Dr Graham Wicks gave advice regarding hypnotisability testing and Professor Deborah Turnbull advised on the psychological aspects of the study. We also acknowledge research assistance from Denise Healy, Karen Belchambers, Carmel Mercer, Louise Goodchild, Meredith Krieg, Ros Lontis, Deni Haines, Pat Ashwood, Andrea Deussen and administrative assistance form Cindy-Lee Brown.
This trial provides a good example of a multi-arm study. Multi-arm studies are becoming more prevalent in medical research. The main advantage to this type of study is that they allow the testing of more than one potential or promising intervention. In the example cited above there are two intervention arms: a group with hypnosis plus CD and a group with CD only, and the study is then completed with a third control arm that received no intervention above treatment as usual. Comparing the two intervention arms with a shared control arm reduces the required sample size by 25%, and in turn can reduce both the time and cost required to conduct the study (Freidlin et al., Clin Cancer Res 2008;14:4368–71). It has also been suggested that multi-arm studies may be more attractive to potential participants because it increases the chance that they will be randomised to a treatment arm rather than the control arm.
In general terms, multi-arm studies are designed and analysed in a similar way to the traditional two-arm study (Appraising multi-arm RCTs, www.clinicalevidence.bmj.com). Prior to starting the study the comparisons to be made should be listed and a formal sample size estimated. An interim analysis may prove valuable, after 50% of the participants have completed the study. This will allow any arms that are unlikely to show efficacy to be identified and dropped from the study (Watson and Jaki Stats Med 2012;31:4269–79). This, in turn, will reduce the required sample size, and possibly shorten the recruitment time and cost. This type of adaptive design is becoming the norm in most clinical trials.
A more controversial issue in the design and analysis of multi-arm studies is the use of multiple testing (Freidlin et al., Clin Cancer Res 2008;14:4368–71; Appraising multi-arm RCTs, www.clinicalevidence.bmj.com). Most studies are powered to detect a difference in the primary outcome at the 5% significance level. Statistically this means that there is a 5% chance of rejecting the null hypothesis when it is actually true; however, when there is more than one primary outcome being tested the risk of rejecting the null hypothesis increases as the number of tests increase. Consequently, when two intervention arms are being compared with a common control arm there could be an 8% risk of obtaining a false result. There are a number of adjustments that can be made to reduce this risk, one of the simplest and commonly used approaches is known as the Bonferroni correction, which divides the initial significance level by the number of tests being performed, so if two tests were being carried out, with an initial significance level of 5%, then the significance level of each individual test would be 2.5%. However, it is now being argued that if the two comparisons were made in separate studies then no correction would be required. Consequently, a more realistic approach may be to use the initial significance level, e.g. 5%, if the tests are independent, and to apply a Bonferroni correction or other adjustment method if the tests are related.
In summary, multi-arm studies can reduce the required sample size when comparing two or more interventions, compared with the more traditional approach of undertaking a separate study for each intervention under consideration. Reducing the sample size and ultimately the cost of studies increases the probability of obtaining definitive results.
Department of Biostatistics, University of Liverpool, L69 3GS, UK
During her routine antenatal visit, a patient in her third trimester of pregnancy was keen to discuss analgesia and relaxation techniques during pregnancy. She asked ‘One of my friends had hypnosis when she was pregnant. Does it work?’
|Participants||Women between 34+0 and 39+0 weeks gestation, planning a vaginal birth|
|Intervention||(i) Group hypnosis sessions plus hypnotherapist-guided audio CDs|
|(ii) Hypnotherapist-guided audio CDs only|
|Comparison||No additional antenatal intervention|
|Outcomes||Primary outcome: use of pharmacological analgesia during labour.|
|Secondary outcomes: selected maternal and neonatal outcomes, including use of augmentation, mode of delivery, and neonatal Apgar scores.|
|Study design||Randomised controlled trial (1:1:1 design)|
EYL Leunga & D Siassakosb
aWomen's Health Research Unit, Centre of Public Health and Primary Care, Queen Mary,University of London, Yvonne Carter Building, 58 Turner Street, E1 2AB, London, UK
bSchool of Clinical Sciences, University of Bristol, The Chilterns, Southmead Hospital, BS10 5NB, Bristol, UK
Join us at #BlueJC: You can collaborate with us to host a #BlueJC session. Find out more on our Journal Club page at www.BJOG.org. Follow @BJOGTweets to stay updated on scheduled #BlueJC sessions. For background of #BlueJC, please refer to BJOG 2013;120:657–660. http://bit.ly/10VaiRZ