Primary repair of obstetric uterine rupture can be safely undertaken by non-specialist clinicians in rural Ethiopia: a case series of 386 women

Authors


Correspondence: Prof. J. Wright, Faculty of Health Sciences, University of Surrey, GU2 7WG, UK. Email jeremy.wright@surrey.ac.uk

Abstract

We report 386 consecutive cases of uterine rupture over a 10-year period, revealing a case fatality rate of 4.8%, which is significantly less than the 16–37% fatality rate published in the literature. Primary uterine repair, which was carried out by non-specialised doctors in 43% of cases, was the treatment for 98% of women, and appears to be simple and effective. The main predictor of death from uterine rupture was a treatment delay of more than 12 hours from the presumed time of rupture (OR 7.7; 95% CI 1.02–58.4).

Introduction

Over the past decade, the maternal mortality ratio in Ethiopia has remained static at 676 per 100 000 live births,[1] which equates to a lifetime risk of death from maternal complications of almost 4%. Uterine rupture and obstructed labour account for around 10% of these maternal deaths,[2] although this figure may be even higher if deaths from sepsis and haemorrhage, which often lead on from obstructed labour, are considered.

Treatment of uterine rupture involves either primary repair of the uterus (sometimes accompanied by tubal ligation) or partial/total removal of the uterus. There is limited evidence, however, regarding the optimal treatment, and there are conflicting views surrounding the need for tubal ligation. One exception to this is a case series of almost 130 000 deliveries in one Indian hospital, reported almost 50 years ago, on the treatment and outcome of 110 women with uterine rupture.[3] The 43 women who were treated by hysterectomy had a mortality rate of 35% and the 66 women treated with primary repair of the uterus had a mortality rate of 9% (one women died before treatment). Although the non-randomised nature of this study raises questions of potential biases, it nevertheless provides the best evidence to date.

A further concern surrounding the use of primary uterine repair is the incidence of re-rupture in subsequent pregnancies. Again, the evidence is limited, but a small study in Saudi Arabia reported a re-rupture rate of just 5%.[4] Although not technically a ‘ruptured’ uterus, the rate of rupture following caesarean section has been shown to be just 1%,[5] suggesting that repair of a ruptured uterus might be a relatively safe option for women who have not yet completed their family.

Our study adds to the limited body of literature by reporting on the immediate outcome of primary repair of uterine rupture in women presenting at Aira Hospital, a Lutheran German missionary hospital in rural West Wollega, Ethiopia.

Methods

The study setting

Ethiopia has a population of 80 million, with 84% living in rural areas.[1] Life expectancy at birth is 59 years, and access to health care is limited by poor transportation on non-metalled roads and a low doctor to population ratio. Although maternity care is largely free of charge, 90% of deliveries occur at home.[1]

Aira Hospital is approximately 60 km from a market town, which is reached by a limited bus service that may not be operable in the wet season when roads are impassable. People in neighbouring villages largely access the hospital by walking 2 or 3 hours, or are carried on the village stretcher. The hospital has an 11–bed maternity ward and a six–bed maternity waiting house for women considered to be at high risk of a complicated delivery. There is no regular obstetric cover, and complex maternity care is provided by a resident surgeon, recently qualified doctors on a training programme (1–2 years), or expatriate voluntary doctors visiting for 8–12 weeks. Like most hospitals in Ethiopia, there are very few clinical guidelines and decisions are made by the attending surgeon/physician based on their experience and perceived competence. Primary repair of a ruptured uterus is often selected in preference to a hysterectomy as it is considered to be a more straightforward procedure for surgeons with limited experience.The hospital does not have a blood bank, medication supplies and operative materials are limited, and there is no trained anaesthetist on site, with surgery being carried out using a ketamine infusion under the supervision of an anaesthetic nurse.

The study

A retrospective case record analysis was carried out over a 10–year period from January 2000 to December 2009. Cases of uterine rupture were identified from the labour ward, and then the theatre and case records were reviewed. Data were entered onto a structured data collection sheet by two study midwives, and were then checked for accuracy by the principal investigator before being entered into SPSS 18 for subsequent analysis.

The primary outcome measure was death before discharge from hospital. Where possible, data were confirmed with admission notes and the operating theatre records. The data were explored for associations between mortality and factors recognised as possible predictors of outcome, such as delay in diagnosis, delay in seeking help and previous uterine rupture. Univariable analysis was carried out using chi-square tests and multivariable analysis using logistic regression with a forward conditional model. The study received approval from the Board of Directors at the Aira Hospital.

Results

There were 10 270 deliveries over the 10–year period. Four case notes could not be found and there were two immediate deaths: one on admission and one just as surgery started, and these are not included in the analysis. Uterine rupture occurred in 386 (3.7%) women and was surgically treated. A total of 19 women (4.8%) died as a result of their uterine rupture. Only eight women ruptured their uterus in their first pregnancy, 15 had had a previous caesarean section, and eight had a previous scar rupture. Three hundred and eighty-two (98%) women had not received any antenatal care during the pregnancy. Two hundred and forty-two women (63%) reported that their labour had lasted for more than 24 hours. The fetal heart could be heard on admission in five cases, but it was not possible from the notes to ascertain if the babies survived. Only five of the 386 women (1.2%) underwent hysterectomy, the remainder being treated by primary repair.

All of the women in this series were admitted as an emergency from home. Prior to surgery, women (or the accompanying family member) with three or more children were strongly encouraged to consent to tubal interruption, and 258 (66.8%) agreed to this. Those who did not have a tubal interruption were advised to attend the hospital in any future pregnancy, making use of the attached ‘waiting house’, which offers cheap accommodation in which they can stay in the last few weeks of their pregnancy.

There were no intraoperative deaths and the deaths that occurred were in the ensuing 10 days, the result of sepsis or untreated anaemia. There were five complications in the survivors requiring a further laparotomy: four pelvic abscesses and one urinoma. Although only ten nulliparous women ruptured their uterus, all reached hospital within 12 hours of the cessation of labour: eight had a labour lasting more than 24 hours and two (25%) died. It is not known whether any of these nulliparous women received any intervention such as uterine massage prior to admission.

Twenty-four women (6%) had previous uterine surgery, and only one of these died. None of the eight women who received antenatal care had had previous uterine surgery.

When the uterus ruptures, contractions cease, making it possible to estimate the time from uterine rupture to presentation for treatment: of the 111 women who arrived at the hospital within 12 hours of the cessation of labour, one (0.9%) died; 14 of 239 (5.9%) women who arrived between 12 and 24 hours of the cessation of labour died; and four of the 36 women (11.1%) in whom there was a delay of more than 24 hours died.

Surgery was carried out by a surgeon or visiting obstetrician on 218 occasions (56%), and by general duty physicians in the remainder of cases. Posterior ruptures, which are technically more difficult to deal with, accounted for 66 of the cases, and six (9%) of these women died. Only five (1%) women underwent subtotal hysterectomy, and all of these survived. Resuscitation was prompt and only ten women waited more than 3 hours for surgery: none died.

Univariable analysis revealed that women were more likely to die if their labour lasted more than 24 hours (= 0.05), if their labour ceased more than 12 hours before treatment commenced (P = 0.02), if they lived more than 100 km from the hospital (= 0.03), and if their admission blood pressure was less than 90/60 mmHg (= 0.05; Table 1). There was a non-significant (= 0.09) trend towards women with posterior ruptures being more likely to die than those with other ruptures. Multivariable analysis revealed that women who waited more than 12 hours from the cessation of labour to commencement of treatment were almost eight times as likely to die from their uterine rupture than women who received treatment within 12 hours of the cessation of labour (OR 7.7; 95% CI 1.02–58.4; Nagelkerke's R2 0.056.

Table 1. Mortality associated with uterine rupture in 386 women admitted to Aira Hospital, Western Ethiopia
  DiedSurvived% diedOR (95% CI)
Parity<4142275.8P = 0.290.6 (0.2–1.64)
≥551403.4
Duration of labour<24 hours31412.1P = 0.053.3 (0.95–11.62)
≥24 hours162266.6
Time from cessation of labour to treatment commencing<12 hours11100.9P = 0.027.70 (1.02–58.43)
≥12 hours182576.5
Distance from hospital<100 km41712.3P = 0.033.3 (1.07–10.05)
≥100 km151967.1
Admission blood pressure<90/60 mmHg152076.8P = 0.052.9 (0.94–8.90)
≥90/60 mmHg41602.4
Time from admission to start of surgery<1 hour2523.7P = 0.651.4 (0.31–6.25)
≥1 hour173155.1
Duration of surgery (hours)<1 hour91455.8P = 0.491.38 (0.55–3.47)
≥1 hour102224.3
Type of rupturePosterior6609.1P = 0.092.4 (0.863–6.46)
Other133074.1
Specialty of surgeonSurg/obs142046.4P = 0.122.3 (0.79–6.34)
General51633.0

Discussion

This data set is one of the largest and most comprehensive published reports on uterine rupture, and confirms the established aetiologic features of uterine rupture: specifically multiparity and prolonged labour. We have shown that uterine repair is a relatively safe procedure and can be performed by non-specialist clinicians with limited surgical equipment.

The case fatality rate of almost 5% in our data set is lower than the rates of 11 and 19% reported from other Ethiopian centres.[6, 7] It is difficult to make comparisons between these figures as the hospital settings vary considerably. Indeed, the hospital with the 19% mortality rate is very different from our hospital: it is one of very few university hospitals, it has trained obstetricians to undertake the surgery, and the majority of women with uterine rupture are treated by hysterectomy. Moreover, the authors judged that 66% of the women in their series were moribund on admission, which suggests that better transportation to this hospital allowed women who were more sick to access care. By comparison the transportation to Aira Hospital is poor, possibly resulting in women who are more seriously sick dying from catastrophic haemorrhage before reaching hospital. Despite these obvious differences, however, our low mortality rate with primary uterine repair does appear to reflect that reported almost 50 years ago in India.[3]

Our data show that a delay in reaching hospital after labour had ceased (i.e. reflecting the time since the rupture of the uterus occurred) was the most important factor increasing the chances of dying. It is not possible to say why women did not reach hospital until much later, although the majority of women who died travelled more than 100 km. Two-thirds of the women in our series were in labour for longer than 24 hours before the rupture occurred, which may also reflect difficulties in reaching hospital facilities.

Whereas in the past women who suffered from a uterine rupture were discouraged from further pregnancies, a recent study has shown that, if managed carefully, maternal and neonatal outcome is reasonable, with 50% achieving a live baby.[4] The women in our study, who were predominantly multiparous, were strongly encouraged to have a tubal interruption, and two-thirds accepted this advice. Aira Hospital is fortunate in having a waiting house beside the hospital, where women considered to be at high risk of a complicated delivery are encouraged to ‘wait’ for 2 or 3 weeks at the end of their pregnancy. In a country where transport to health facilities is extremely poor, a waiting house offers women the opportunity for safe subsequent deliveries.

Conclusion

Our series suggests that simple resuscitation with crystalloid fluids followed by primary repair surgery can be safely carried out by non-specialist clinicians with varying experience, and has a low mortality rate in women who reach a suitable facility within 12 hours of uterine rupture. Women at risk of uterine rupture, particularly those who have had previous uterine surgery, should deliver in hospital, and in order to achieve this, would ideally stay close to a hospital in the 3 weeks prior to delivery.

Disclosure of interests

None of the authors have any interests to disclose.

Contribution to authorship

The original data were collected under the supervision of WA Subsequent detailed analysis was undertaken by KB and JW. Both JW and KB were responsible for drafting the article.

Details of ethics approval

This work was undertaken while JW & KB were undertaking voluntary work for Maternity Worldwide UK. Approval for the project was given by the management committee of Aira hospital Res No 186/1379/11.

Funding

None.

Acknowledgements

These data were collected by WA as part of his specialised training dissertation, and he would like to acknowledge his supervisors Dr Dereje Nigussie and Dr Mirkuzie Woldie from Jimma University for their help and support. JW and KB travelled to Ethiopia with Maternity World Wide UK. KB was on a sabbatical year granted by the University of Surrey.

Ancillary