Many clinicians agree that an elective caesarean section at term is necessary when a woman has been delivered twice by caesarean section. It is generally assumed that there is a considerable weight of historical evidence in favour of this management but close inspection of the relevant publications reveals a plethora of unsupported personal opinion quoted, misquoted and paraphrased until the rule 'twice a section always a section' has become ingrained within our practice.
One of the earliest authors providing us with an unsupported opinion was O'Donel Browne (1951) who said that the woman who has already had several lower segment operations runs an increasingly greater risk of rupture if submitted to labour, for the unavoidable formation of scar tissue at the site of operation weakens the lower segment. Two years later Harris (1953) said that he rejected patients (from a trial of labour) who gave a history of several previous caesarean sections, believing that repeated assaults on the uterine musculature were not conducive to the formation of strong scar tissue. His opinion was quoted by Lavin et al. (1982) and Tahilramaney et al. (1984) without reference to his lack of supporting data. The opinion seemed a popular one. Kaltreider & Krone (1959) stated, again without evidence, that vaginal delivery after more than one caesarean section was 'not generally recommended'. Even as recently as the last decade personal opinions have abounded. O'Sullivan et al. (1981) laid down their unsupported criteria for allowing a trial of scar, their first being that the patient should have only one scar. Flamm et al. (1984) felt that more than one previous caesarean section was a contraindication to a trial of labour but produced no supporting evidence. Bider et al. (1990) stated that: 'We do not allow patients with two previous caesarean sections to deliver vaginally', but had to admit that this policy was quite arbitrary.
Examples exist whereby authors who made no mention of women with more than one section are later quoted as having given an authoritative opinion on the subject, eg Pauerstein et al. (1969) are reported by Lavin et al. (1982) as having said that they believed having more than one section increased the risk (of scar rupture), in fact Pauerstein et al. (1969) made no mention of such women.
McGarry (1969) is quoted by Lavin et al. (1982) in that he believed there to be an increased risk associated with more than one previous section. In fact McGarry (1969) acknowledged previous work illustrating the safety of such a labour and allowed two such women to do so. He did write: 'it seems reasonable that patients who have had two previous sections should not be submitted to the anxieties attendant on attempts at vaginal delivery especially since the attempt may be unsuccessful because of inefficient uterine action or fetal distress even if scar rupture does not occur'. His concern for the feelings of pregnant women is laudable, his paternalistic attitude questionable.
When figures have been produced, some have been interpreted in a most unusual manner: Donnelly & Franzoni (1964) reported a series of 219,755 deliveries containing an unknown number of patients with a caesarean section scar. They argued that because four out of every 10 ruptures of caesarean section scars that occurred before labour did so in women with more than one scar, 'this statistic reaffirms the necessity for elective repeat caesarean section in any patient who has had more than one section'. Later the same authors (Donnelly & Franzoni 1967) produced guidelines for the selection of patients for trial of scar. Their first was that 'any patient who has had more than one caesarean section ... should be excluded from consideration for vaginal delivery'. This was after reporting a series of 2904 patients with between one and four section scars, who suffered 13 ruptures (five of these before labour) of which three occurred in those with more than one scar. There is no logical connection between the statistics presented and the conclusions drawn.
Case et al. (1971) reported an almost identical uterine rupture rate of 0.69% in 1299 and 0.67% in 297 women with one and two caesarean section scars respectively. Porreco & Meier (1983) cited this as a comment on the safety of allowing a trial of scar to occur. Besides failing to note the practically identical incidence of scar rupture in women with two scars as compared with those with one, they failed to report that the work by Case et al. (1971) was on findings at elective caesarean section and made no reference to the safety of trials of scar.
Tahilramaney et al. (1984) quote Schmitz & Gajewski (1951) as saying that multiple scars predispose to uterine dehiscence: Schmitz & Gajewski (1951) had reported a rupture rate of 1.1% among 128 with two scars and 1.9% amongst 320 with one scar, but had made no mention of the relative proportions of each group electively sectioned. The same paper also quotes Pedowitz & Schwartz (1957) as saying the same thing, albeit to a degree that was not statistically significant, when their figures did not include trials of scar.
As the number of references containing advice has mounted, clinicians have naturally shied away from allowing these women to be delivered vaginally. Together with the intention of reporting homogeneous groups when publishing series of trials of scar this has led to researchers excluding such women from their prospective series. Later researchers have then tended to automatically assume that this was because of a proven concern about the rate of uterine rupture. For example Lavin et al. (1982) reported that Gibbs (1980) believed that the presence of more than one scar increased the risk of uterine rupture but, in fact, Gibbs merely excluded such women from his series of deliveries, he gave no reasons why.
There is no conclusive proof of an increased risk of scar dehiscence during labour after two caesarean sections and the manner in which we have come to believe that there is should be an embarrassment to all who consider obstetrics to be a scientific speciality. Enlightened authorities, albeit frequently responsible for the perpetuation of the myth in their references to the literature, have presented what is now a considerable weight of evidence that labouring with two scars is no more of a risk than with one. More importantly from the clinical viewpoint, they have failed to show any advantage from an elective abdominal delivery in these cases (Novas et al. 1989; Phelan et al. 1989; Hansell et al. 1990). Early work is marred by poor methodology and a high prevalence of classical section scars amongst study populations and previous reviewers have noted how the relevant figures are often irretrievable from series reporting overviews of vaginal birth after caesarean (VBAC)(Enkin 1989). Arriving at a completely satisfactory summary of the facts is further hampered by varying study populations, selected for vaginal delivery upon different criteria, managed differently in labour for varying amounts of time. Terminology regarding scar complications also varies.
Even so, the current literature records the paucity of scar complications arising from attempts at a vaginal delivery. Novas et al. (1989); Phelan et al. (1989) and Hansell et al. (1990) have recorded between them 557 trials of scar in women having had two caesarean sections. 11 dehiscences were diagnosed, no maternal or fetal mortality resulted. The same three series all report a higher dehiscence rate at elective caesarean section allowing one to make the inference that the presence of asymptomatic and presumably bloodless scar defects is not being diagnosed following successful vaginal delivery.
The vaginal delivery rates from trials of scar after two caesarean sections are consistently high, despite exclusion categories set by authors which vary qualitatively and hence quantitatively in the percentage of women with two scars allowed to labour. Figures range from 69% (Phelan et al. 1989) to 82% (Stovall et al. 1987).
The potential for a vaginal delivery is partially preserved even if one of the patients' previous deliveries was for cephalopelvic disproportion/failure to progress/dystocia; 78% (Porrecco & Meier 1983), 70% (Farmakides et al. 1987) and 50% (Hansell et al. 1990) are typical statistics. Phelan et al. (1989) was even able to achieve a 56% vaginal delivery rate in those who had had two caesareans sections for cephalopelvic disproportion (CPD), this figure compares favourably to the 67% vaginal delivery rate described by Rosen et al. (1990) in their meta-analysis of labours after one section for CPD. It must also throw some doubt on the very existence of CPD as a necessarily recurrent phenomenon as a proportion of these infants were larger than their predecessors.
Obstetricians should remember that to allow a patient to labour is not a treatment, it is a virtually unavoidable consequence of pregnancy. If we are to perform a surgical procedure in order to circumvent labour we should have a clear indication. The historical evidence does not provide one and current publications indicate that we do not appear to benefit our patients by delivering them electively by caesarean section.
We would do well to attend the words of Bertrand Russell (1957): 'If you have an opinion about any matter it should be based on ascertained facts, not upon hope, or fear, or prejudice'.
This was first published in the British Journal of Obstetrics and Gynaecology, December 1991, Vol 98, pp 1199-1202