The rising caesarean section rate: A loss of obstetric skill?

Caesarean delivery rates in the US have increased each year since 1965. Yet there is no evidence that maternal and child health has improved as a result of this increase. (DHHS 1991)

In general maternity units, no benefit is apparent from a CSR above 10 – 12% on the singleton population as a whole. The case against a CSR below 10% rests on a single variable, onset of respiration at one minute. (Joffe et al. 1994)

In the last 35 years, the caesarean section rate (CSR) has risen all over the Western world and also in some less-developed countries (Churchill et al. 2006). The quotations above come from public health physicians. They (not obstetricians as a group) were the first in the USA to raise the alarm about the rising rate of caesarean section almost 30 years ago when the US rate had almost tripled from 5.5% in 1970 to 15.2% in 1978 (US Task Force 1981). Joffe et al. (1994) looked at a large data set from the North West Thames Region to reach their conclusion, but despite this paper; despite the WHO consensus conference (WHO 1985) suggesting that no WHO region should have a rate above 10 – 15%; despite the National Sentinel Caesarean Audit (Thomas and Paranjothy 2001); and despite the NICE Guidelines published in 2004, the rate continues to rise in the UK as it does in the USA (MacDorman et al. 2006), South America (Villar et al. 2006) and most of Europe. Italy, at 33% in 2000, leads with the Netherlands and Nordic countries as lowest at under 15%. China had a rate of 40% in 1999 as did Chile, and Brazil followed with 36%. Canada, New Zealand and Australia were 22 – 23% (Churchill et al. 2006). The USA rose to 29% in 2004 (MacDorman et al. 2006) after a fall in the early 1990s as the vaginal birth after caesarean section (VBAC) policy recommended in Healthy People 2000 took effect (DHHS 1991) and halted the rise. Since 1970, Medline lists almost 8,000 articles with the keywords ‘Caesarean section’ and over 27,000 with ‘Cesarean section’ but few of these address the issue of how to reduce the rate.

Although there have been difficulties in obtaining accurate national figures about the use of caesarean section in England and Wales (Macfarlane 1998; HoC 2003), there have been three perinatal surveys of all births in 1 week in the spring, which gave a CSR of 1.1% in 1946, 2.7% in 1958 and 4.5% in 1970 (RCOG 1948; Butler and Bonham 1963; Chamberlain et al. 1975). The perinatal mortality rate fell from 44 in 1946 to 33.2 in 1958 and to 23/1,000 births in 1970. Deaths from intracranial birth trauma, many of which followed difficult high forceps or breech births, had virtually been eliminated, falling from 3.0/1,000 in 1958 to 0.4/1,000 in 1970 (Chamberlain et al. 1975). This survey and that in 1958 (Butler and Bonham 1963) are important in that accurate post-mortem data on babies were obtained and analysed and a wealth of information about labour and delivery collected. Sadly, in my experience, few trainees have even heard of these surveys, let alone read them, and yet they are important sources of data, which should be read and understood by all those entering our specialty.

In 2005, the CSR had risen to 22.9% of births in England, 24.5% in Wales, 24.9% in Scotland and in 2004 to 26% in Northern Ireland (Birthchoice 2006). The NHS Institute for Innovation and Improvement (2006) give a figure of 24.1% in England for 2005 – 2006. In 1989 – 1990 it was 11.3% in England (DoH 1997) but in our survey of births in 1989, we found a rate of 12.1% (Savage and Francome 1993), although for Wales and Scotland we obtained the similar rates of 13.5 compared with 13.8 and 14.2 compared with 14.0. The Scottish data have always been more reliable than those collected originally in England and Wales.

Can it be right that almost one-quarter of women in the UK are being delivered surgically? What has caused this extraordinary and seemingly relentless rise in the CSR? From the mid-30 s, when the maternal mortality had remained much the same for a century, obstetricians were focussed on reducing the unacceptable loss of young women's lives and the confidential enquiries which took place from 1952, were an early and useful audit tool for achieving this aim (Turnbull et al. 1989). The rise in caesarean section from 1% in 1946 to 2.7% in 1958 and 4.5% in 1970 was an appropriate response to the recognition that with antibiotics, new anaesthetic agents and blood transfusion available, women were less at risk from surgery, so caesareans were performed more readily. By 1980, maternal mortality had fallen to about 1 in 10,000 maternities and obstetricians became increasingly focussed on the fetus. The improvements in neonatal intensive care in the 1960s and 1970s changed the perception of the balance between maternal and fetal survival. However, if one looks at the perinatal mortality rates since 1985, over 99% of babies born in England at ≥28 weeks have been born alive and survive the first week of life, yet the CSR had more than doubled by 2001 to 21.5%.

The CSR was not collected in 1985 because of the introduction of the new Korner information system, but was probably around 10 – 11%, so it does not look as if saving babies was responsible for the rise above the 10% that the majority of the members of the WHO consensus conference considered a reasonable rate (WHO 1985; Wagner 1986, pers comm). So what happened in the 1980s to drive the rate up?

When the US Task Force looked at the reasons for the tripling of the rate in 15 years, they listed four main reasons: previous caesarean section (27%), dystocia (29%), breech (15%) and fetal distress (15%). The importance of repeat caesarean section, which in the USA was always done if a woman had had one caesarean section, meant that the UK rate where a trial of scar was allowed, rose more slowly at this time. The National Institutes of Health (NIH) sponsored Task Force, which included lay women, recommended that vaginal birth after caesarean section should be encouraged, but it took another decade and the CSR reaching a peak of 24.7% in 1988 before this was set as a target (DHHS 1991). When this was done, the CSR dropped slightly to 22.8% in 1993 and 21.6% in 2002, as the proportion of births following VBAC rose to 25.4% in 1993 (Clarke and Taffel 1995; Hamilton et al. 2004). The rate has now begun to rise again after a paper published in the influential New England Journal of Medicine reported twice the number of serious maternal complications in women undergoing a trial of labour compared with those having an elective caesarean section. Perinatal morbidity was not significantly different (McMahon 1996). The ACOG guidelines (ACOG 1999) recommending that VBAC be used only in ‘institutions equipped to respond to emergencies with physicians immediately available to provide emergency care’, meant that smaller community hospitals could no longer carry out a trial of labour and the rate rose from 20.7% in 1996 to 29.1% in 2004 (MacDorman et al. (2006). That this is not due only to changes in the availability of VBAC is shown by the fact that the primary CSR rose by 6% between 2002 and 2003, while VBAC rates fell by 16% (Hamilton et al. 2004).

If one looks at the reported rate of what is called ‘uterine rupture’ in the papers in the literature comparing VBAC with elective caesarean section (ECS), what is striking is the widely differing rates found in the studies (NICE 2004). Many US studies have found higher rates of ‘uterine rupture’ in those who have an ECS and presumably what we are seeing here is a silent scar dehiscence, which is not the serious emergency that a ruptured uterus is. In the US studies, these events tend to be reported more frequently than in the British, Irish or Australian ones and this may be related to the skill with which the first caesarean section is performed. With the high proportion of births in private practice in the USA, a large number of obstetricians who do not work in teaching hospitals have small numbers of patients per year and this may affect the maintenance of skill. It is important to differentiate between dehiscence, in which if prompt action is taken, the baby will survive unscathed; and uterine rupture, which may kill the baby instantly as it is expelled into the uterine cavity and is a serious danger for the woman, as those of us who have worked in Africa know only too well. Describing the risk of dehiscence, which is probably about 1/1,000 if labour starts spontaneously, in the dramatic terms of a uterine rupture, may well frighten the woman from attempting a vaginal birth. In the 14,000 women with a previous caesarean section reported in the National Sentinel CS Audit (Thomas and Paranjothy 2001), the rate of ‘rupture’ was reported as 3/1,000 in women having an ECS and 1/1,000 in those having a trial of labour.

What is the reason for the supposed rise in the incidence of ‘dystocia’ over the last 30 years? We know that shorter women have a higher rate of feto-pelvic disproportion than those over 5 feet 2 inches in the British population, but with better childhood nutrition, women today are taller than they were. The proportion of babies that weigh over 4,500 g has not increased significantly and while obesity is rising, there is no evidence that women are less able to withstand labour now than their mothers or grandmothers were. Could it be that the way that labour is conducted – in large, harshly lit, noisy, impersonal labour wards with gangs of doctors coming round at intervals and constantly changing staff – affects the way that women labour? Is that why women who book for a home birth, even if they do not succeed, have only half the risk of requiring a caesarean section? (Chamberlain et al. 1997). Midwifery studies suggest that a CSR under 10% can be obtained with good perinatal outcomes and these studies rely on a close supportive relationship between the woman and her midwives (Armstrong and Feldman 1986; Durand 1992; van Alten et al. 1989). An Austrian obstetrician also reported very low CSR and attributed this to avoidance of technology and careful psychological preparation for childbirth (Rockenschaub 1990).

The other soft indicator for caesarean section is fetal distress. The introduction of electronic fetal monitoring (EFM) was hailed as an important advance in the mid-1960s and like ultrasound, became part or routine care without proper evaluation of its risks and benefits (Banta and Thacker 1979). Small studies soon showed that the rate of false positives was such that CSR rose while the perinatal mortality and morbidity remained the same, and these data were confirmed in a Cochrane review (Thacker and Stroup 2001). It was not until the early 1980s that a randomised controlled trial (RCT) of sufficient power comparing EFM with routine auscultation of the fetal heart by a midwife was conducted (McDonald et al. 1985). After a time lag of a decade or more and the fetal monitoring guidelines (RCOG 2001a) this has had a considerable influence on the use of routine monitoring in the UK (Savage and Francome 2007a). There was a small but significant difference in the frequency of fits in the first few days of life in the routine group but examination at 1 year showed no significant difference in neurological status between the two groups (Grant et al. 1989). Those babies in the EFM group had higher rates of infection, although this was not serious. Following the publication of this study, it was recommended that routine EFM should not be used for healthy women with an uncomplicated pregnancy but there is no doubt that much of the rise in fetal distress reported was due to the inappropriate use of EFM. The Scottish audit of caesarean sections (Wilkinson et al. 1998) recorded prospectively the indications for 87% of the 9,573 caesarean sections done in 1994. The rate for singleton pregnancies was 15.7% and 60% were done in healthy women. A total of 16% of all caesarean sections were emergency sections done for ‘failure to progress’ (FTP) or dystocia in US terms. Another 5.7% for FTP and fetal distress and 8.6% were for fetal distress; 10.7% were done electively for breech presentation and 2.8% for fetal distress or suspected intrauterine growth restriction (IUGR) before labour. In over two-thirds of the cases of fetal distress, a fetal blood sample (FBS) was not obtained, despite recommendations that this is done to reduce false positive results using CTG findings. A total of 51% of the caesarean sections were in primigravid patients and their CSR was 19%. The publication of this audit with its recommendations did nothing to reduce the CSR in Scotland, which rose slightly faster after 1998, nor did it affect the rest of the UK.

The National Sentinel Caesarean Section Audit (NSCSA), which surveyed births in 3 months in 2000 for England and Wales and for 3 months to the end of February 2001 in Northern Ireland and the Channel Isles, found that in the 50% of cases where caesarean section was done for presumed fetal distress, where the CTG was abnormal and fetal blood sampling was possible, it had not been done in the majority of cases (56%). The proportion where it was done varied from one-third to two-thirds regionally and from 11% to 100% in individual hospitals. Those cases where FBS was not done contributed 4.6% to the overall CSR. Surely, there is no reason why all hospitals cannot attempt a FBS in 100% of suitable cases? In our 1989 survey 45% did routinely monitor women, but this had fallen to 11% in the 2005 survey, although 50% did perform a 20-min ‘monitor strip’, a practice which crept in after the RCT results were published and routine monitoring was not recommended for healthy women as there was no evidence of its usefulness. It was done ‘just to be sure everything is alright’. As the only one of five obstetricians in the late 1990s not to recommend routine EFM, I had to write this on the labour ward board to remind the midwives about my policy and just before I retired in 2000, I found it erased frequently as some of the agency midwives were unhappy about not having this ‘reassurance’. However, if questioned, most had not read the evidence or not understood the concept of false positives. Anecdotally, it is reported that this practice may lead to a cycle of intervention, as the woman becomes anxious or is forced to lie on her back, which may effect the CTG if she suffers from supine hypotension. The fetal heart rate (FHR) rises, the staff decide to continue the monitoring then find changes, which lead to it being continued and the woman confined to bed. This may decrease the intensity of her contractions. Then syntocinon is used to augment labour and the woman finds the contractions more painful and requests an epidural. Sometimes the FHR drops after an epidural and an emergency caesarean section ensues. Afterwards, the woman may feel dissatisfied as she no longer felt in control of her labour and yet feeling in control is associated with increased satisfaction about the birth (Green et al. 1988).

Finally, there is the question of breech presentation, which occurs in 3 – 4% of pregnancies at term. Its incidence at term can be reduced by a policy of external cephalic version (ECV), which fell out of favour in the 1970s after some fetal deaths had occurred but has been revived since the CSR began to rise (RCOG 2001b). In the first survey that Colin Francome did of obstetricians' views as to why the CSR was rising, the changed attitude to vaginal breech and forceps delivery were leading reasons (Boyd and Francome 1981). In 1958, over 80% of breech births were vaginal and only 5.6% had an elective caesarean section. The perinatal mortality ratio (PMR) was doubled in the vaginal births and overall breech babies were seven times as likely to die as those presenting head first. It has been difficult to disentangle which aspects of this raised PMR were due to the mode of delivery when babies are more likely to present by the breech early in pregnancy and also have a higher rate of congenital abnormalities but in the next 40 years an increasing proportion of breech births were by caesarean section. Obstetricians were surprised to find that some women still chose to try and deliver vaginally even when given the results of the analysis of breech births done using the North West Thames dataset where there was a nine times higher risk of the baby dying if vaginal delivery was compared with elective caesarean section (Thorpe-Beeston et al. 1992). This paper generated a number of letters questioning the validity of the findings and who had performed the deliveries (Savage et al. 1992). By the time of the NSCSA in 2000, 88.4% of breech babies in England and Wales were delivered by caesarean section and over half of these were elective operations (56%). By 2000 – 2001 the rate had risen to 90.3% (on a smaller sample), and the rate in Northern Ireland was 86.4% and in the Channel Islands, 100%.

The multicentre RCT (Hannah et al. 2000) was designed to definitively answer the question as to whether a planned elective caesarean section was safer for the baby than a policy of planned vaginal delivery in otherwise uncomplicated cases. The trial was ended prematurely because of concern about some of the interim results, and after its publication, the NICE guidelines recommended caesarean section for breech and at first, the obstetric profession accepted the findings uncritically. However, questions began to be raised, initially – and unsurprisingly – by Dutch obstetricians (van Roosmalen and Rosendaal 2002), who pointed out that some of the 13 perinatal deaths were not related to vaginal breech delivery, one being a twin and another cephalic. A Canadian obstetrician (Kotaska 2004) drew attention to the lack of experience of those participating as the majority of the 131 centres were in the USA, which had a 13% vaginal breech delivery rate. In the trial, the rate rose to 57%, which suggested that breech delivery was being pushed to the limit. He also pointed out the odd criteria of 0.5 cm/h for progress and 3.5 h in the second stage whereas normally, 1 cm/h and 2 h for a primipara would be considered normal progress and less than that an indication that there might be a large baby. He questioned the validity of RCTs to evaluate complex procedures such as breech delivery where skill is important.

In addition, I have another problem with the recruitment of obstetricians. One cannot ethically enter a patient into an RCT unless one genuinely believes that there is uncertainty about the methods of treatment, i.e. there is equipoise. If like myself with many years of a policy of accepting that vaginal breech delivery is safe if patients are correctly selected and the labour supervised by an experienced obstetrician, then one cannot ethically subject women to a trial where they are randomly allocated to the ECS arm. I had a successful vaginal breech delivery rate of 60%. If other experienced obstetricians declined to take part on these grounds, that suggests that those taking part were probably not skilled in breech delivery. One fact that suggests this in my view was that of the four babies who died after supposedly difficult vaginal deliveries the weights were 2,400, 2,550, 3,000 and 3,500 g, respectively, hardly large babies. An Irish obstetrician involved in the planning of the trial has recently discussed this (Turner 2006). He pointed out that the analysis of the perinatal deaths was not done by parity and if this is done, there were no deaths directly attributable to breech delivery in the multigravid population. Yet now, most women in Britain are finding it difficult to find an obstetrician who will conduct a trial of breech delivery and a whole generation of trainees are being deprived of the opportunity to learn how to safely conduct a vaginal breech delivery, although some independent midwives are continuing to offer this service (Cronk 1998). The flaws in the execution and analysis of the trial invalidate the recommendations, although one study from Holland using routinely collected data does suggest that a change in policy there has had benefit in terms of perinatal mortality. Before the trial, 50% of women delivered vaginally with a PMR of 3.5/1,000. This fell to 1.8/1,000 in term breeches after the policy change. Half of the women are still offered the option of a trial of vaginal delivery and 40% of those deliver vaginally, making an 80% CSR overall (Reitberg et al. 2005). Last year a prospective observational study from France and Belgium achieved a 71% vaginal breech delivery rate in the 2,526 planned vaginal delivery group with a perinatal mortality of 1.8/1,000 which was not significantly different from that of the babies delivered by caesarean section (Goffinet et al. 2006). I think the jury is still out and that the NICE guidelines should be changed with regard to automatic elective caesarean section to allow an attempt to deliver a breech vaginally in selected cases. I am disappointed that the RCOG Green-top Breech guideline updated for 2006 still states that the perinatal mortality and morbidity of breech babies delivered vaginally is higher than if they have a planned caesarean section, relying principally on the Hannah trial, although it mentions the Goffinet study. The latest ACOG guidelines accept that vaginal delivery of singleton breeches is permissible for an experienced physician within hospital guidelines and with the woman's informed consent with regard to perinatal outcome (ACOG 2006).

A systematic review of the strategies to reduce the CSR concluded that a policy of routine ECV and providing continuous support in labour had been shown to reduce the CSR and listed the studies where obstetricians had managed to reduce the CSR in their units (Walker et al. 2002). Demott and Sandmire in 1990, after analysing different obstetricians' rates in Wisconsin concluded that the most important factor in determining the CSR was ‘physician style’ and the fact that while caseload midwifery schemes, where the woman knew her midwife, reduced the need for epidural analgesia, it had no effect on the CSR, tends to confirm this. Our study of obstetricians' attitudes (Savage and Francome 2007b) showed a wide spectrum of opinion among those obstetricians surveyed, but to me, the most telling observation was that the idea of what is the optimal CSR had changed so radically in 15 years from 81% thinking it should be under 14% in 1989, to 86% thinking it should be 15% or over in 2005. The estimates ranged from under 10% to 40% in 2005.

The RCOG has played a major role in attempting to reduce the CSR with publication of guidelines about ECV, EFM, and most recently caesarean section, organising and analysing the NSCSA and yearly conferences about caesarean section and normal birth with the Royal College of Midwives and the National Childbirth Trust, although only a handful of obstetricians have attended these. These initiatives seem to have had no impact on the continuing rise. The NICE guidelines about caesarean section were said to be accepted fully by 32% and partially by 61% of those answering our survey and only 10 (7%) did not accept them. However, Bloomfield has made some trenchant criticisms of these (Bloomfield 2006), and while the literature review and summaries are well done, the conclusions are less convincing. As one of our survey respondents said ‘the evidence was evaluated very thoroughly but evidently false conclusions were drawn in many cases’. Another commented ‘NICE not very helpful. Show patients the risks and then do what they want – that will not reduce the rate’. A third (who has done research in the field and shown how to reduce the CSR) commented, ‘I feel that they are totally inadequate/poor’. Another commented that ‘the authors did not seem to have spent much time in the labour ward’ and some of the recommendations seem unrealistic in terms of the time required to implement them. To recommend that women who request VBAC should be ‘supported’ rather than actively encouraged seems unlikely to reduce the number of unnecessary repeat operations. While it is true that overall there was a one in five chance of a woman having a caesarean section, for individual low risk women, where the risk for a primipara is about 1 in 20 and a multipara 1%, how much information should be given about caesarean section? Might it not increase a woman's anxiety rather than help her enjoy a normal pregnancy if she is told there is a 20% chance of having a caesarean section?

Does the existence of protocols in virtually all hospitals these days lead to a lowest common denominator approach to the practice of obstetrics and a loss of skilled decision making? The reduced working hours and forced introduction of shift working to comply with the demands of the postgraduate Deans and the European Working Time Directive, coupled with the reduced labour ward and home delivery experience of medical students means that trainees are used to a high CSR and have virtually no experience of normal labour and birth. Trainees are increasingly concerned about the adequacy of their training (Ghaem-Maghami et al. 2006) and consultant obstetricians cited poor skills as the third most common reason for the continued rise in 2005 – not a reason that featured in our earlier study.

Does it matter if the CSR continues to rise? I believe that it does. Although birth is safer than it has ever been, vaginal birth is safer for the woman than even ECS. Looking after a newborn baby after a major operation with a scar in the abdomen is not the best way to start life as a parent or deal with other children, and increasingly doctors do not see women after the first 2 days – and with the shift system, may not even see them at all following delivery, so there is a perception that caesarean section is an easy option for the woman as well as a pleasant occasion for those in theatre. Breast-feeding is more difficult and may not be established. The long-term effects of caesarean section are poorly studied and understood and this was one factor in the conclusion of the FIGO Ethics Committee that non-medically indicated caesarean section was unethical. The risk of placenta accrete, leading to hysterectomy and with a substantial mortality is well documented and rises with each caesarean section. Voluntary and involuntary infertility has been reported and there are also subtle psychological effects and changes in family dynamics, which require further study. This seminal statement (FIGO 1999) had only been read by 13% of the respondents, 14 of whom agreed with the statement and five of whom did not. Should the College not ensure that all consultants read this statement and act upon it? Though the contribution of caesarean section on request to the overall rise is probably small, as despite some obstetricians' views and media reports of women being ‘too posh to push’, it is not true that many women are demanding a caesarean section (Churchill 1997; Gamble et al. 2000; Bewley and Cockburn 2002; Churchill et al. 2006). Only 5.3% in the NSCSA expressed a preference for delivery by caesarean section, 3.5% of primiparas and 7.0% of multiparas, whereas 8.7% were told they must have a caesarean section (6.7% of primiparas and 10.3% of multiparas), which seems high. Most of those who do request a caesarean section have had a negative birth experience the first time round and in primigravida. Fear of pregnancy and birth can be treated by counselling and at least half of women will then attempt a vaginal delivery (Ryding 1993; Nerum et al. 2006). The FIGO statement was quoted in the NICE guidelines, but the fact that the second most common reason cited by consultants in our survey was ‘patient request, demand or pressure’ (45% of reasons), with another 22% citing ‘patient expectation’, suggests that communication between women and their obstetricians has been skewed by the professionals' fear of litigation, desire to respect women's autonomy and inability to control the standards in the labour ward.

The addition of maternity services to the National Service Framework for Children and Young People was overdue (DoH 2004) but lays down good principles for promoting normal birth. The second edition of Modernising Maternity Care produced jointly by the NCT RCM and RCOG in 2006, also has clear policies and examples of good practice. The newly formed NHS Institute for Innovation and Improvement has produced a paper focussing on caesarean section and plans a toolkit in 2007 to help hospitals to reduce their CSR. They have looked at hospitals which have a CSR which is lower than the average, and have ‘good quality’ care and provide ‘value for money’. They conclude that these organisations share a common ethos, work in a multidisciplinary way, have ‘robust clinical governance’ and ‘effective communication’ and ‘empower women to make informed choices about their maternity care’. Despite the jargon ‘over-arching, stakeholder, optimising, aspiring to optimal care’, etc., the targets for one-to-one midwifery care (100%), vaginal birth rate over 80%, normal birth rate over 70% and suggestions to improve care ‘pathways’ are still sound. Will commissioning work, where professional debate and hand wringing have failed? Reams of paper have been used in writing protocols, guidelines, policies for change and yet, the CSR continues to rise. Maybe the government focus on efficiency in order to save money in the NHS will mean that things will change in the future. However, my reading of the literature is that a reduction in a hospital CSR only occurs where there are strong leaders who are determined to reduce that rate (Myers et al. 1993; Johnson and Ansell 1995; Robson et al. 1996; Flamm et al. 1998). Swedish obstetricians reduced their national CSR after agreeing it was too high (Nielsen et al. 1994). As one of our respondents said, ‘there is no passion for vaginal delivery’ – and that is what is needed.

So how might we achieve a reduction in the CSR while maintaining good perinatal and maternal outcomes? Organisational factors are important as are training and management of labour.

First, the loss of trained midwives to the profession should be stemmed by allowing midwives to work in the way they wish, i.e. be ‘with women’. Excellent results have been achieved by the South East London Midwifery Group which has now moved to Kings College Hospital (Sandall et al. 2001) and the independent Midwives Association (IMA 2006). In order to achieve this on a large scale, I believe that midwives should accept having primary care midwives working in the community in small groups caring for the majority of women who do not need to see obstetricians and secondary care midwives who work in hospitals and can still provide the one-to-one care women need but would learn other skills appropriate to high risk labour and delivery. This need not split the profession. GPs and hospital consultants all belong to the BMA and see themselves as one profession. If there were more home births and midwife-led units, this might fit in with Mr Nicholson's desire to close 60 maternity units (Carvel 2006). The remaining ‘high tech’ units would have fewer women to care for and obstetricians would have more time to listen, talk to women and train the next generation of obstetricians.

Second, the obstetric training should start with SHOs being supernumerary (as has been done in Dundee for some years) and allocated to work as a pupil midwife with experienced home-birth midwives and hospital midwives following normal women though a complete labour and birth, so they understand the progress of normal labour. (At present, they spend too much time arranging operations and checking abnormal traces for which they are frequently ill-trained.) After 3 – 6 months of this, they would start working with consultant obstetricians who would ensure that their clinical abilities, abdominal palpation and vaginal examinations were sound before being expected to assess women in labour. They should learn how to do FBS before the registrar grade, so they will be confident in carrying this out when they progress further up the ladder.

New registrars should be assessed and their findings physically checked when they join a unit, by the consultant, senior specialist registrar or consultant midwife. They should be taught by the consultant how to perform ventouse extraction, forceps delivery, twin delivery and breech delivery and this should be recorded in their logbooks by the supervising consultant who would of course not certify these deliveries unless s/he had been present. Even if there is no 24 h consultant presence on the labour ward, which is likely in smaller units, the consultant should come in to supervise out of hours until s/he is satisfied that the trainee is competent. Studies of second stage caesarean section show one area where skills need improvement (Olah 2005; Spencer et al. 2006). It is not enough to say that trainees' skills are poor or that CSR declines with experience (Joyce et al. 2002; Griffiths et al. 2005), it is our job as professionals to ensure young doctors are adequately trained.

Induction of labour should be a consultant decision and never delegated to midwives who are the experts in normal birth but not trained to make decisions about induction. Consideration should be given to allowing pregnancy to proceed after 42 weeks with careful monitoring if the cervix is unfavourable (Vause 2006). If the membranes rupture early and there is a hind water leak, do not induce labour but manage conservatively with prophylactic antibiotics. If the baby is thought to be small, a distinction should be made between those who have IUGR and the small normal baby (Kiwanuka and Moore (1987).

Women should not be admitted to the labour ward until labour is established and should be encouraged to go home, walk around and eat light food until that happens. Assessment by a midwife at home and a decision taken only then about the place of delivery has shown to increase the home-birth rate and may prevent unnecessary admissions. From the 1958 survey, it was shown that the perinatal mortality rate did not increase in a primigravida until labour had lasted for over 48 h and in a multigravida over 24 h (Butler and Bonham, Table 51a, p. 157). This was in the days when synthetic oxytocin was being researched as a way of augmenting labour but had not yet been incorporated into routine practice, so the natural history of labour is well documented. Less than one-fifth of primiparas had labours this long and <4% of multiparas, which makes the high rates of augmentation in many hospitals seem excessive today. Patience and treating women as individuals are essential if one is to reduce the number of women diagnosed as ‘failing to progress’. Occipito-posterior presentations can be missed but the diagnosis is important as women may need more analgesia and augmentation with syntocinon.

Women need to know that in a first labour, the cervix thins out before it starts to dilate otherwise they become demoralised if the effacement of the cervix is not explained as a positive sign of progress. Progress on vaginal examination should include not only the dilatation of the cervix and the station of the head but the position of the head, i.e. rotation and flexion and the presence or absence of caput. Membranes should not be ruptured unnecessarily.

In breech presentation use ECV but also allow selective attempts at vaginal breech delivery rather than ECS. Do not use epidural routinely, or forceps to the aftercoming head and consider standing or kneeling positions for delivery. Watch the clock when the head enters the pelvis to avoid unnecessary and premature intervention.

Routine fetal monitoring should not be done in healthy women nor a 20 min monitor strip. If EFM is indicated by the presence of meconium or abnormalities of the FH on auscultation, the woman should be able to walk around or adopt any position that she wishes, not be laid on her back.

No caesarean section should be carried out for fetal distress unless a FBS has been done if this is physically possible, or there is a dramatic and sustained fall in the FHR. It requires great skill in communication to reassure the woman and her partner that the baby is doing well. Obstetricians must be patient and contain the parents' anxiety.

There is a tendency to perform a caesarean section when a woman has mild pre-eclampsia, which is not justified by the evidence. Most of these women will labour successfully after induction if the blood pressure is rising slowly and the baby in good condition.

Almost all women who have had a previous caesarean section should be encouraged to attempt VBAC and if, as is so often the case, there has been a badly managed OP position in the first labour, this should be explained to the woman and the likelihood of a different shorter labour this time around put forward. I always put a note in the woman's notes for the registrar to do all the VEs and ring me after the first one so we could discuss her fears and I think it helped her to know that someone who knew her was aware she was labouring. Smith has worked out a way of predicting the likely success rate (Smith et al. 2004) but whether this information would be helpful to the woman has yet to be tested.

Audit caesarean sections weekly and use Robson's groups to analyse your statistics; identify the major contribution to the CSR; deal with the problem and monitor progress (Robson et al. 1996).

We as obstetricians should regain control of the way we teach and train young doctors and work with midwives to ensure that the management of labour is done well and assisted vaginal delivery and breech skills are learned and perfected. Older obstetricians had to learn to contain their anxiety and wait patiently until the woman delivered vaginally and this is something younger doctors need to learn, rather than reaching for the scalpel. The fact that litigation was given as the leading reason for the rising CSR is a shameful commentary on our professional standing. As Marsden Wagner, the former WHO European Director of Maternal and Child Health in Europe says, ‘it is a chilling thought that the doctor will pull or cut your baby out because he is frightened of being sued’. In practice, it is not as stark a choice as that and as a majority of the respondents in our survey felt the CSR was too high in their hospital and wanted to reduce it, I think the will is there, if only we can harness that energy and change the system. If we all work together to treat each woman as an individual and each labour as different, we could accomplish a major fall in the CSR within 5 years, but perhaps we need an RCOG Task Force to make this happen.