Impact of third- and fourth-degree perineal tears at first birth on subsequent pregnancy outcomes: a cohort study

Authors

  • LC Edozien,

    1. Maternal and Fetal Health Research, Manchester Academic Health Science Centre, University of Manchester, Manchester, UK
    Search for more papers by this author
    • LCE and IG–U share joint first authorship of this paper.
  • I Gurol-Urganci,

    Corresponding author
    1. Department of Health Services Research and Policy, London School of Hygiene and Tropical Medicine, London, UK
    2. Office for Research and Clinical Audit, Lindsay Stewart R&D Centre, Royal College of Obstetricians and Gynaecologists (RCOG), London, UK
    • Correspondence: Dr I Gurol-Urganci, Department of Health Services Research and Policy, London School of Hygiene and Tropical Medicine, 15–17 Tavistock Place, London, WC1H 9SH, UK. Email ipek.gurol@lshtm.ac.uk

    Search for more papers by this author
    • LCE and IG–U share joint first authorship of this paper.
  • DA Cromwell,

    1. Department of Health Services Research and Policy, London School of Hygiene and Tropical Medicine, London, UK
    Search for more papers by this author
  • EJ Adams,

    1. Department of Urogynaecology, Liverpool Women's NHS Foundation Trust, Liverpool, UK
    Search for more papers by this author
  • DH Richmond,

    1. Office for Research and Clinical Audit, Lindsay Stewart R&D Centre, Royal College of Obstetricians and Gynaecologists (RCOG), London, UK
    2. Department of Urogynaecology, Liverpool Women's NHS Foundation Trust, Liverpool, UK
    Search for more papers by this author
  • TA Mahmood,

    1. Office for Research and Clinical Audit, Lindsay Stewart R&D Centre, Royal College of Obstetricians and Gynaecologists (RCOG), London, UK
    Search for more papers by this author
  • JH van der Meulen

    1. Department of Health Services Research and Policy, London School of Hygiene and Tropical Medicine, London, UK
    Search for more papers by this author

Abstract

Objective

To investigate, among women who have had a third- or fourth-degree perineal tear, the mode of delivery in subsequent pregnancies as well as the recurrence rate of third- or fourth-degree tears.

Design

A retrospective cohort study of deliveries using a national administrative database.

Setting

The English National Health Service between 1 April 2004 and 31 March 2012.

Population

A total of 639 402 primiparous women who had a singleton, term, vaginal live birth between April 2004 and March 2011, and a second birth before April 2012.

Methods

Multivariable logistic regression models were used to estimate odds ratios, adjusted for other risk factors.

Main outcome measures

Mode of delivery and recurrence of tears at second birth.

Results

The rate of elective caesarean at second birth was 24.2% for women with a third- or fourth-degree tear at first birth, and 1.5% for women without (adjusted odds ratio, aOR 18.3, 95% confidence interval, 95% CI 16.4–20.4). Among women who had a vaginal delivery at second birth, the rate of third- or fourth-degree tears was 7.2% for women with a third- or fourth-degree tear at first birth, compared with 1.3% for women without (aOR 5.5, 95% CI 5.2–5.9).

Conclusions

The risk of a severe perineal tear is increased five-fold in women who had a third- or fourth-degree tear in their first delivery. This increased risk should be taken into account when decisions about mode of delivery are made.

Introduction

Pregnant women and their obstetricians face a challenge when deciding on mode of delivery after a severe perineal tear damaging the anal sphincter (third degree) and the rectal mucosa (fourth degree). A choice has to be made between a planned caesarean section, which avoids the risk of another anal sphincter rupture, but carries its own morbidity, and a vaginal birth with the prospect of the woman having another perineal tear. The rate of reported severe perineal tears has been increasing[1, 2]: in England, it tripled from 1.8 to 5.9% between 2000 and 2012.[3]

To counsel women appropriately, local and national information should be available on the recurrence rate after anal sphincter rupture, and on the impact of the woman's age and whether or not she has had an episiotomy or instrumental delivery for previous pregnancies. Large, population-based studies based in Norway,[4, 5] Sweden,[6] and Denmark[7] reported that, compared with women who do not have rupture of the anal sphincter, women with a rupture have an up to five-fold increased risk of a third- or fourth-degree tear at vaginal delivery in the next pregnancy. A recent population-based study in Australia found no such increase, however.[8] Results from hospital-based studies range from no increase to up to eight-fold higher risks.[9-16] Apart from the size and setting of the studies, there are other factors that make comparisons between studies and applicability to maternity care in England difficult. For example, in some settings episiotomies are commonly or exclusively made in the midline (unlike the UK practice of mediolateral episiotomy). Compared with mediolateral episiotomies, midline episiotomies carry a higher risk of third- or fourth-degree perineal tear.[2, 17, 18]

Among women who have had a third- or fourth-degree perineal tear in England, this study investigates the mode of delivery in the subsequent pregnancy and the recurrence rate of third- or fourth-degree tears. The study used a large population-based database that includes all maternity admissions in National Health Service (NHS) hospitals.

Methods

We used the Hospital Episode Statistics (HES) database to identify births that have taken place in English NHS trusts (acute hospital organisations). The HES database contains patient demographics, clinical information, and administrative data for each inpatient episode of care since 1997. The records are extracted from local patient administration systems, and undergo a series of validation and cleaning processes before being made available for analysis.[19] A unique identifier links episodes of care related to the same patient, which enables studies to examine events before or after an index episode. Diagnostic information is coded using the tenth revision of the International Classification of Diseases (ICD10),[20] and operative procedures are coded using the fourth revision of the UK Office for Population Censuses and Surveys classification (OPCS4)[21]. For maternity episodes, supplementary fields known as the ‘maternity tail’ capture parity, birthweight, gestational age, method of delivery, and pregnancy outcome; however, the completeness of data in the maternity tail varies across NHS trusts in England. For example, birthweight and parity are available in 79 and 65% of the delivery episodes, respectively.

This study included primiparous women aged 16–45 years, who had a live, singleton, vaginal birth between 1 April 2004 and 31 March 2011, and who also had a second birth by 31 March 2012. For the analysis of mode of delivery at second birth, we excluded women who had a multiple pregnancy, non-cephalic presentation, or placenta praevia or abruption, as these are indications for elective caesarean section. We also excluded women who went into preterm labour because they do not have a choice about mode of delivery. Preterm labour was identified by ICD10 code ‘O60’. For the analysis of recurrence of obstetric tears, we further restricted the cohort to women who had a vaginal (including instrumental) birth.

For both the first and second birth, third- or fourth-degree perineal tears were identified by ICD10 codes ‘O70.2’ and ‘O70.3’, respectively. Mode of delivery was defined using information in the OPCS4 procedure codes, and we distinguished between non-instrumental vaginal (OPCS4 codes ‘R23’ and ‘R24’), forceps (‘R21’), and ventouse (‘R22’), or if not defined using OPCS4 codes, by the delivery method specified in the maternity tail. OPCS4 code ‘R27.1’ identified whether or not an episiotomy had been performed.

Parity was defined using historical data from the HES database because the maternity tail is incomplete. A woman was defined as primiparous if there was no evidence of a birth prior to the index delivery, using minimum 7 years of obstetric history. Recent research suggests that over 90% of women in this population have their second child within 7 years of the first delivery.[22]

We identified the following potential confounding risk factors. Maternal demographic factors were age at second birth (<20, 20–24, 25–29, 30–34, ≥35 years), ethnicity (white, Asian, Afro-Caribbean, other, unknown), and socio-economic deprivation of the mother's area of residence using the index of multiple deprivation (IMD) (quintiles of 32 480 areas in England ranked according to a measure of deprivation that combines a range of economic, social, and housing indicators).[23] Obstetric risk factors for the analysis of mode of delivery at second birth included mode of delivery, episiotomy, and birthweight at first birth, and pre-existing conditions (hypertension, diabetes) and gestational diabetes at second birth. Obstetric risk factors for the analysis of recurrence of tears at second birth included mode of delivery, episiotomy, birthweight, prolonged labour, and shoulder dystocia at second birth. The duration of labour was marked as prolonged if the delivery record included an ICD10 diagnosis code ‘O63’ (long labour), whereas shoulder dystocia was identified by ICD10 code ‘O66.0’ (obstructed labour caused by shoulder dystocia). The year of the second birth was included as a linear variable in the logistic regression model to take into account changes in clinical practice over time. The interval between the first and second birth was calculated from the date of the first birth to the date of the second birth.

We used logistic regression models to estimate odds ratios adjusted for confounding risk factors reflecting the relative risks associated with third- or fourth-degree tears at first birth, and having an elective caesarean section at the second birth. We also used logistic regression models to estimate the relative risks associated with third- or fourth-degree tears at first birth and the occurrence of third- or fourth-degree tears at second birth. To account for a lack of independence in the data of women treated in the same trust, we used the Huber sandwich estimator to calculate robust standard errors. All analyses were performed in stata/se 12.

Results

There were 1 719 539 singleton, vaginal, live births to primiparous women aged 16–45 years between April 2004 and March 2011. Of these, 707 184 (41.1%) women went on to have a second delivery within the study time frame. Using information from the second delivery record, we excluded women who had a preterm delivery (4.0%) or an indication for elective caesarean section (4.5%) (Figure 1).

Figure 1.

Flowchart.

This left 639 402 women in the cohort. The prevalence of third- or fourth-degree tears at first birth for the cohort was 3.8%. At second birth, 15 190 (2.3%) women had an elective caesarean section. Table 1 describes the rates of elective caesarean section according to maternal and obstetric risk factors. For most factors, the rate of elective caesarean was typically between 1 and 4%. In comparison, among women who had a third- or fourth-degree tear at first birth, 24.2% were delivered by elective caesarean section (adjusted odds ratio, aOR 18.3, 95% confidence interval, 95% CI 16.4–20.4). Women who had an instrumental delivery or an episiotomy were also more likely to have an elective caesarean section. Other factors that were associated with higher elective caesarean section rates were older age, white ethnicity, living in a less-deprived area, pre-existing or gestational diabetes, higher birthweights, and longer birth intervals. The rate of elective caesarean section increased during the study period.

Table 1. Impact of third- or fourth-degree perineal tears at first birth on elective caesarean section as mode of delivery at second birth (n = 654 592)
 Distribution of factor (%)Elective caesarean section rate (%)Crude OR (95% CI)Adjusted OR (95% CI) P
Maternal age
<20 years3.20.80.42 (0.36–0.48)0.57 (0.49–0.67)<0.001
20–24 years22.21.20.58 (0.55–0.62)0.71 (0.67–0.76) 
25–29 years26.52.011 
30–34 years29.32.91.46 (1.39–1.54)1.23 (1.16–1.30) 
>35 years18.83.61.86 (1.74–1.98)1.58 (1.48–1.68) 
Ethnicity
White75.22.611<0.001
Asian10.61.50.59 (0.52–0.67)0.57 (0.51–0.63) 
Afro-Caribbean5.11.30.49 (0.42–0.57)0.62 (0.53–0.73) 
Other3.91.70.67 (0.60–0.75)0.71 (0.63–0.80) 
Unknown5.21.70.64 (0.56–0.73)0.70 (0.61–0.80) 
Deprivation (quintile)
Least deprived17.93.2110.007
217.02.90.90 (0.84–0.96)1.00 (0.93–1.07) 
318.22.50.76 (0.71–0.81)0.95 (0.89–1.03) 
420.82.00.62 (0.57–0.67)0.09 (0.83–0.98) 
Most deprived26.11.50.46 (0.41–0.51)0.86 (0.78–0.96) 
Characteristics of first birth
Third- or fourth-degree tear 3.824.221.5 (19.4–23.8)18.3 (16.4–20.4)<0.001
No third- or fourth-degree tear96.21.511 
Mode of delivery
Non-instrumental74.51.511<0.001
Forceps10.46.94.82 (4.57–5.09)2.84 (2.65–3.03) 
Vacuum15.03.12.11 (2.01–2.22)1.72 (1.62–1.82) 
Episiotomy 30.03.31.78 (1.69–1.87)1.18 (1.11–1.25)<0.001
Birthweight
<2500 grams3.50.90.44 (0.39–0.50)0.70 (0.60–0.80)<0.001
2500–4000 grams67.82.011 
>4000 grams5.86.63.46 (3.28–3.66)2.40 (2.26–2.55) 
Unknown22.82.41.20 (1.11–1.31)1.18 (1.09–1.29) 
Preterm birth 3.51.40.58 (0.51–0.65)0.95 (0.83–1.10)0.512
Risk factors at second birth      
Diabetes0.27.13.24 (2.60–4.05)3.17 (2.41–4.17)<0.001
Hypertension0.32.20.95 (0.69–1.32)0.78 (0.55–1.12)0.175
Gestational diabetes1.94.72.10 (1.86–2.37)1.94 (1.69–2.22)<0.001
Interbirth interval
Less than 2 years34.22.00.82 (0.78–0.86)0.97 (0.92–1.02)<0.001
2–3 years33.92.411 
3–4 years17.82.61.06 (1.01–1.11)1.13 (1.07–1.18) 
More than 4 years14.12.61.07 (1.02–1.13)1.31 (1.23–1.38) 
Year of subsequent birth
20051.31.611 
Change per year  1.05 (1.04–1.06)0.98 (0.96–1.00)0.010

Of the women included in the cohort, 619 717 (96.9%) had a vaginal delivery. The rate of third- or fourth-degree tears at second birth was 1.5% (Table 2), less than half of the rate among primiparous women. Among women with a third- or fourth-degree tear at first birth, the unadjusted rate of recurrence was 7.2%, compared with 1.3% among women without a tear, and this increased risk remained five times higher after adjustment for potential confounding factors (aOR 5.5, 95% CI 5.2–5.9). Among the other risk factors, the factors with the highest increase in the risk of third- or fourth-degree tears at second birth were high birthweight, forceps delivery, and the presence of shoulder dystocia. The risk of a third- or fourth-degree tear was also higher in older women, in women living in the least deprived communities, and in Asian women. Women who had an episiotomy were less likely to experience a severe perineal tear. The adjusted risk of third- or fourth-degree tears increased with birthweight and shoulder dystocia, but was not associated with the duration of labour. To test the robustness of our results, we re-ran the analyses using multilevel logistic regression in which the effect of patient clustering within NHS trusts was modelled as a random coefficient. These analyses produced comparable results to those presented here (Table S1).

Table 2. Impact of third- or fourth-degree perineal tears at first birth on recurrence of tears at second birth (n = 619 717)
 Distribution of factor (%)Third- or fourth-degree tear rate (%)Crude OR (95% CI)Adjusted OR (95% CI) P
Maternal age
<20 years3.30.30.23 (0.18–0.30)0.33 (0.25–0.42)<0.001
20–24 years22.60.70.50 (0.46–0.54)0.58 (0.53–0.62) 
25–29 years26.61.411 
30–34 years29.12.01.43 (1.35–1.52)1.35 (1.28–1.42) 
>35 years18.32.01.42 (1.34–1.51)1.36 (1.28–1.44) 
Ethnicity
White75.11.411<0.001
Asian10.61.91.33 (1.23–1.44)1.59 (1.48–1.71) 
Afro-Caribbean5.11.30.87 (0.77–0.98)1.01 (0.90–1.13) 
Other3.91.30.89 (0.78–1.02)0.96 (0.85–1.09) 
Unknown5.31.20.85 (0.75–0.97)0.92 (0.81–1.04) 
Deprivation (quintile)
Least deprived17.72.011<0.001
216.91.70.83 (0.78–0.89)0.87 (0.82–0.94) 
318.21.50.75 (0.69–0.81)0.84 (0.77–0.91) 
420.91.30.67 (0.61–0.73)0.81 (0.74–0.88) 
Most deprived26.31.10.53 (0.49–0.58)0.74 (0.68–0.80) 
Characteristics of first birth
Third- or fourth-degree tear 2.87.25.92 (5.56–6.31)5.51 (5.18–5.86)<0.001
No third- or fourth-degree tear97.21.311 
Risk factors at second birth
Mode of delivery
Non-instrumental96.11.411<0.001
Forceps1.45.03.73 (3.32–4.19)4.02 (3.51–4.60) 
Vacuum2.51.91.39 (1.21–1.59)1.34 (1.16–1.55) 
Episiotomy 5.52.31.63 (1.47–1.81)0.66 (0.58–0.75)<0.001
Birthweight
<2500 grams1.60.20.15 (0.09–0.23)0.16 (0.1–0.25)<0.001
2500–4000 grams70.41.21  
>4000 grams11.83.12.58 (2.44–2.73)2.29 (2.16–2.43) 
Unknown16.21.31.09 (0.99–1.19)1.14 (1.04–1.26) 
Long labour 2.42.31.61 (1.43–1.82)0.89 (0.78–1.01)0.068
Shoulder dystocia 1.15.84.27 (3.83–4.76)2.92 (2.59–3.28)<0.001
Interbirth interval
Less than 2 years34.41.20.77 (0.73–0.82)0.91 (0.86–0.96)<0.001
2–3 years33.91.511 
3–4 years17.71.71.13 (1.07–1.20)1.11 (1.04–1.17) 
More than 4 years13.91.71.08 (1.01–1.15)1.04 (0.97–1.11) 
Year of subsequent birth
20051.30.611<0.001
Change per year  1.10 (1.08–1.12)1.06 (1.04–1.08) 

Discussion

Main findings

The rate of elective caesarean section in the subsequent pregnancy for women with a severe perineal tear in their first delivery was 24.2%. For women who had a vaginal delivery in the second pregnancy, a third- or fourth-degree tear at first birth increased the risk of recurrence of a tear by five-fold.

Strengths and limitations

This study included over 600 000 first and second births in women who delivered in an NHS hospital over a 7–year period. HES captures over 96% of all deliveries in England,[24] and provides a large sample size required for the analysis of rare outcomes.

This study represents practice in England. Recent population-based studies demonstrated an increase in the rate of reported obstetric tears in the last decade.[3] Since then, there have also been significant changes in the management of second-stage labour,[25] and a lower threshold for performing an elective caesarean section.[26, 27]

A limitation of this study is that our adjusted results may contain residual confounding because we were not able to control for some risk factors, such as intrapartum anaesthesia,[28, 29] experience of the birth attendant,[30, 31] the angle and size of an episiotomy,[32-34] or fetal head circumference,[7] which may affect the risk of third- or fourth-degree tears at second birth. It is unlikely, however, that any residual confounding caused by the absence of these risk factors could account for the observed large differences in the risk of recurrence.

Although it has been suggested that the diagnostic coding in the administrative data sets is potentially inaccurate, the majority of NHS trusts submit good-quality data to HES that conforms to national recommendations,[35-37] and the data are sufficiently robust for research and decision-making.[38] Recent publications have demonstrated that, when data completeness, consistency, and accuracy are analysed carefully,[39, 40] HES is a valuable source of data for studies exploring patterns of care and reproductive epidemiology.[3, 41-43]

Finally, we focused on primiparous women, as birth order is a risk factor for perineal tears,[4, 6, 12, 15, 44-46] but our ‘lookback’ approach to define parity may have resulted in some multiparous women whose first birth was not recorded in HES, for example because they delivered in another country, being incorrectly labelled as primiparous.[39] Sensitivity analyses using 10 years of patient history to identify primiparous status or the information in the maternity tail, instead of the current approach, yielded comparable results (Table S2).

Interpretation (findings in light of other evidence)

This is the first study of mode of delivery and recurrence rate in a pregnancy subsequent to a third- or fourth-degree perineal tear in England. The prevalence of a third- or fourth-degree perineal tear at first birth (in this population of women who had a second birth during the study period) was 3.8%. Women who have had a third- or fourth-degree perineal tear in their first birth can be advised that the chance of having a similar tear in the next birth is approximately 7 in 100. This study confirms the finding of previous studies elsewhere that there is a manifold increase in the risk of an anal sphincter rupture at delivery in women who had a third- or fourth-degree tear at the previous delivery (Norwegian 1967–1998 cohort, aOR 4.3, 95% CI 3.8–4.8;[4] Norwegian 1967–2004 cohort, aOR 4.2, 95% CI 3.9–4.5;[5] and Swedish 1973–1997 cohort, aOR 4.7, 95% CI 4.3–5.2;[6] Danish 1997–2010 cohort, aOR 5.9, 95% CI 5.4–6.5).[7] Two population-based studies, from Australia and the USA,[8, 47] did not find an increased risk of recurrence; however, the Australian study did not adjust for case mix,[8] and the US study reported on practice from more than 20 years ago.[47] Hospital-based cohort studies with comparable control groups also showed a two- to five-fold increase in the risk of recurrence.[13, 14, 16]

Mode of delivery after a third- or fourth-degree tear has been reported in few studies. In population-based studies, the rate of elective caesarean section after an anal sphincter rupture was 6.0% (Sweden),[5] 6.2% (Norway),[4] 6.2% (Australia),[8] 7.2% (USA),[47] 17.4% (Australia),[48] and 29.9% (Denmark).[7] In hospital-based studies, the rates of elective caesarean section after a prior third- or fourth-degree tear was 19.6% (Ireland),[10] 18.6% (Israel),[16] and 8.1% (USA).[8] These differences in elective caesarean section rates may reflect the time periods studied or variations in the management of pregnancies after third- or fourth-degree tears across countries. The most comparable cohorts in terms of time period and design with ours are the studies from Australia (2000–2009) and Denmark (1997–2010).[7, 48] These relatively high rates of elective caesarean section may be the result from the perceived high risk of recurrence of tear associated with vaginal birth and the lack of evidence or professional guidance on how to identify women who are at high risk of functional impairment following vaginal delivery, for whom the balance of risks and benefits favours an elective caesarean section. A survey of clinicians based in the UK found that 70% of coloproctologists and 22% of obstetricians would recommend an elective caesarean section to prevent anal incontinence following prior anal sphincter injury.[31]

Our study reports on the recurrence of tears and mode of delivery, but in the absence of large, population-based studies on functional outcomes or quality of life after a severe tear, we are unable to comment categorically on whether the relatively high caesarean section rates are justified. Although a caesarean section will prevent a recurrence of a repeat perineal tear, it is also associated with risks to the mother and the baby.[49] A study that compared outcomes after elective caesarean section versus vaginal delivery, specifically for women with a previous anal sphincter rupture, found that the prevalence of any morbid event was 11.3% in the caesarean section group versus 4.2% for vaginal deliveries (relative risk, RR 2.7, 95% CI 2.6–2.8).[50] These risks of an elective caesarean section have to be weighed against the clinical, psychological, and social burden of anal incontinence.[51] One could argue that the best approach for women with a previous tear is not to offer them an elective caesarean section but to improve delivery suite practice, for example by providing manual support of the perineum in the second stage of labour, which significantly reduces the rate of anal sphincter rupture.[52-54]

For clinicians advising pregnant women with a previous anal sphincter rupture, robust evidence on whether and under what conditions to recommend an elective caesarean section is lacking. One could consider using additional criteria to guide decision-making. For example, an elective caesarean section could be considered if there is evidence of a persisting defect after repair or if anal manometry shows reduced squeeze pressures. Unfortunately, many units do not have a dedicated perineal post-trauma clinic with endoanal ultrasound scan and anal manometry facilities, and in those units decisions on mode of delivery may have to be made solely on the basis of history and maternal preference. It should be noted, however, that a persisting defect after a repair and reduced squeeze pressure should not be considered in isolation, as they do not on their own give information about functional or long-term outcome.

Our study showed that the risk factors (other than prior severe tear) for a third- or fourth-degree perineal tear at second birth are similar to risk factors at first pregnancy,[3] such as birthweight and instrumental deliveries (in particular use of forceps), but the effects were generally lower. This is consistent with the findings of previous studies.[4, 5, 7, 12] The most likely clinical explanation is that the lower risk of recurrence at second births reflects the stretching of the perineum at the prior delivery. At first births, the effects of birthweight and instrumental delivery are complemented by relatively rigid perineal tissues.

In addition to known risk factors at second pregnancy, women who had an instrumental delivery, episiotomy, and a higher birthweight baby at first birth, and longer birth intervals, had higher rates of elective caesarean section at second birth. Similar associations were found in a population-based study in Australia.[48] It is likely that elective caesarean section is offered by clinicians or preferred by women after obstetric interventions or adverse pregnancy outcomes, because of the possibility of recurrence of risk factors (such as macrosomia) or childbirth-related distress.[55-57] Women who lived in less deprived areas were more likely to have an elective caesarean section. This result is in agreement with previous studies from England and Scotland.[58, 59]

Conclusion

A national guideline on the optimal mode of delivery for women with a prior anal sphincter rupture is needed, and the results of this study along with those of other studies can inform the development of such a guideline. The findings could also be helpful as a starting point to women and their obstetricians in discussing mode of delivery in their next pregnancy.

Disclosure of interests

None.

Contribution to authorship

IGU, LCE, TAM, LA, and JHvdM conceived the study. IGU and DAC contributed to its design and conducted the analyses. IGU and LCE wrote the article, and DAC, TAM, LA, DR, and JHvdM commented on drafts. All authors approved the final version for publication.

Details of ethics approval

The study is exempt from UK National Research Ethics Service approval because it involved the analysis of an existing data set of anonymised data for service evaluation. Approvals for the use of HES data were obtained as part of the standard Hospitals Episode Statistics approval process.

Funding

IGU is supported by the Royal College of Obstetricians and Gynaecologists.

Acknowledgements

We thank the Department of Health for providing the Hospital Episode Statistics data used in this study.

Ancillary