Review: Induction of labour at 39 weeks and adverse outcomes in low-risk pregnancies according to ethnicity, socioeconomic deprivation, and parity: A national cohort study in England

This is a review of the paper (https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1004259%20) published on July 20, 2023 by researchers at St George’s, University of London and partners.

Summary

The authors suggest that induction at 39 weeks may be of benefit to those from lower socioeconomic groups. AIMS questions the validity of this conclusion. These are our main concerns:

• Women with certain risk factors were removed from the induction group, but not the expectant management group. This introduces bias because there would be some women with these risk factors who may not have been offered, or who declined induction. This would make the expectant management group a higher risk sample than the induction group.

• The expectant management group excluded women who went into spontaneous labour between 39 and 40 weeks, which other studies have shown tends to produce results which are more favourable to induction, than if they are included.

• The research was promoted with the headline “New research indicates early induction of labour may reduce inequalities in pregnancy outcomes.” However the only groups for which there was evidence of benefit was low risk women of low socioeconomic status. The researchers found no evidence that it might reduce the inequalities of outcome suffered by women of Black or Asian heritage.

• The reduction in stillbirth rates observed in the induction group was almost exactly matched by an increase in the neonatal death rates. Early induction might avoid one perinatal (stillbirth plus neonatal) death out of every 10,000 births, but for 9,999 it would make no difference - a rather less dramatic finding than the headline implies.

• The main benefit of early induction appears to be that it would avoid the need for 42 out of every 10,000 babies to be ventilated.

A policy of offering induction at 39 weeks to all low risk women in the most deprived 40% or even the most deprived 20% of the population seems impractical. AIMS believes that this would impact severely on the ability of our already over-stretched maternity services to deliver safe care for all.

There are likely to be many reasons for the observed inequalities in maternity outcomes, but reaching for the blunt instrument of more inductions is a case of wanting to treat the symptom, not the cause. AIMS believes that what is needed is sensitive and supportive pre-pregnancy and maternity care, preferably with a Continuity of Carer model.

Methods and sample

This is a retrospective study which analysed the records of women with a ‘low-risk’ pregnancy who gave birth in an NHS hospital in England between January 2018 and March 2021. It compared the rate of adverse outcomes for the cohort (group) that had an induction of labour between 39+0 and 39+6 weeks of pregnancy and those who gave birth at 40 weeks or later. The definition of ‘adverse outcomes’ was a “composite outcome including stillbirth, neonatal death within 28 days of birth” or a range of other health problems or procedures “recorded in a baby’s admission record.”

The advantage of this type of study is that it can provide a much larger sampler for analysis than would be possible in a randomised controlled trial. However the records used may omit important information. This means that there could be differences between the groups or in the ways in which their labours were managed which are not identified but have an impact on the results. In this case the authors note that factors such as smoking, BMI, and the indication for induction were not included in the data. Smoking and high or low BMI are known to be risk factors for the outcomes being measured.

Of the original sample of 1,567 thousand records, around a quarter had to be excluded due to key information such as gestational age at birth not being recorded, or the quality of a hospital’s data being suspect. The authors also excluded the following:

• those that had an induction and gave birth before 39 weeks

• those that did not have induction but gave birth before 40 weeks

• those with pre-existing problems “that are most likely to be present and known before 39 weeks” such as diabetes, hypertension, cardiac or pulmonary disease, previous caesarean, gestational diabetes, breech presentation, placenta previa, or congenital abnormalities

The authors then also removed from the induction cohort any women who had risk factors such as premature rupture of membranes, placental abruption, pregnancy-induced hypertension or pre-eclampsia, eclampsia, or amniotic fluid abnormalities (leaving a total of just over 500 thousand records for analysis). This was intended to ensure that the induction cohort represented only induction in low-risk pregnancies. However, they did not exclude women with these risk factors from the expectant management cohort on the basis that “if these risk factors are present at 39 weeks they are not compatible with expectant management.” The authors appear to be making an unwarranted assumption that any woman who had developed one of these conditions would have had induction of labour before 40 weeks, and therefore would not appear in their expectant management cohort. This ignores the possibility that some of these women might not have been offered, or might have declined induction before 40 weeks. This introduces a risk of bias, since these conditions increase the chances of an adverse outcome. It appears that the information about risk factors was available so it should have been possible to check for and quantify how many women in their expectant management group were and were not ‘low risk.’ We are puzzled as to why the authors did not do this. In fact, we feel it would have been more meaningful to analyse results for a ‘low risk’ expectant management cohort versus their ‘low risk’ induction group.

Another issue with this type of study is that the results can be different according to which group is chosen for comparison with the induction group. This paper compares planned induction at 39 weeks with an expectantly managed group that gave birth at 40 weeks or later, so does not include those having spontaneous births between 39 and 40 weeks . Other studies using this type of comparison have found that the caesarean rate seems to be the same or lower for those having a planned induction. However, when those same studies compared planned induction with all those being expectantly managed in that week (so including those who birthed in the same week as the induced group or later) they found that the caesarean rate was higher for the planned induction group (Glantz JC. Obstet Gynecol 115:70-6. 2010; Stock SJ et al BMJ 2012; 344: e2838.)

The study used a statistical technique to generate some values that were missing from the data. In most cases the missing values were a small proportion of the total, but for ethnicity 12.4% of the values were filled in using the technique. There are many widely reported examples of biases existing in datasets comparing and relating various other features with ethnicity, and it is not clear how independent these inferred values were from the data being analysed.

Findings

The study found that approximately 3.3% of births in the induction cohort and 3.6% in the expectant management cohort had an adverse perinatal outcome. When this was adjusted for differences in maternal age, ethnicity, socioeconomic deprivation, year of birth, birthweight centile and parity, the difference was 0.28% in favour of induction. In other words, this suggests that about 28 out of every 10,000 fewer babies would experience an adverse outcome if labour was induced at 39 weeks in all low risk pregnancies, and for 9972 babies out of 10,000 it would make no difference.

The authors calculate that about 360 inductions in low-risk pregnancies would be needed to avoid one adverse outcome. This can be compared with the recent Cochrane review estimate that 544 induction would be needed at term to avoid one perinatal death (stillbirths plus neonatal deaths). Note that the Cochrane figure does not include the other, less serious, adverse outcomes which contributed to the ‘composite outcome’ measured in this study.

The authors went on to analyse how the difference in outcomes between the induction and expectant management cohorts differed between socioeconomic groups. For those in the most deprived 20% of the population (IMD 5) the difference in risk of an adverse outcome was 0.48% lower with induction (i.e. 48 fewer babies out of 10,000 had an adverse outcome if labour was induced at 39 weeks.) For those in the next most deprived 20% (IMD 4) the difference was 0.58% ( so 58 fewer babies out of 10,000.) In contrast, for those in the higher socioeconomic groups there was no significant difference in outcomes between the induction and expectant management cohorts. The authors conclude that “An increased uptake of IOL with birth at 39 weeks, especially in women from more socioeconomically deprived areas and in nulliparous women, may help reduce inequalities in perinatal outcomes.” As there is no evidence of benefit in offering induction at 39 weeks to low risk mothers with high socioeconomic status, their use of the word “especially” is misleading.

The study found “no statistically significant evidence that risk differences varied according to ethnicity.” In other words, there’s no evidence that early induction would help to reduce the serious inequalities experienced by the Black and Asian communities, only those that relate to socioeconomic status. This makes the headline the St George’s University of London website www.sgul.ac.uk/news/could-early-induction-of-labour-reduce-inequities-in-pregnancy-outcomes “New research indicates early induction of labour may reduce inequalities in pregnancy outcomes” somewhat disingenuous. To be fair, the authors note that despite the sample size it may have been too small to pick up variations in the risk for groups such as those of Black or Asian heritage that form a relatively low percentage of the total population.

As well as the overall measure of ‘adverse outcomes’ the paper provides a breakdown of the difference between induction and expectant management for the individual factors making up the composite outcome (Table 2.) This shows that although the stillbirth rate was around six in 10,000 lower in the induction cohort than in the expectant management cohort (0.01% versus 0.07%) this was almost exactly balanced by a higher neonatal death rate (0.1% versus 0.04%). Combining the data to calculate the perinatal death rate gives figures of 10 in 10,000 for the induction cohort and 11 in 10,000 for the expectant management cohort. So in other words induction at 39 weeks in a low risk pregnancy might help avoid the death of one baby in every 10,000 but for 9,999 it would make no difference - a rather less dramatic finding than the headline implies.

Of other factors making up the composite outcome, the most common is the need for artificial ventilation, required in 1.91% of the induction cohort and 2.33% of the expectant management cohort, a difference of 0.42%. That implies that early induction would avoid the need for 42 out of every 10,000 babies to be ventilated.

The authors point out a number of limitations of their research, including the fact that the data did not include the reason for induction, and could not be adjusted for some important risk factors such as BMI and smoking. They also think it likely that some pre existing maternal health problems including hypertension and diabetes were ‘under-ascertained’ - in other words that some of the women included would have had these health problems so were not in fact low risk.

Overall, this study produced similar findings to previous reviews: that early induction may lead to a small reduction in the risk of an adverse outcome, at least in certain groups, but a great many women and birthing people would need to undergo induction in order to avoid harm to one baby. The authors themselves appear to recognise the lack of evidence to support any policies of induction at 39 weeks. Professor Asma Khalil is quoted on the St George’s website as saying “Improved collection of routine data on the indication for induction and the presence of risk factors is required to corroborate the role that induction of labour at 39 weeks in women with a low-risk pregnancy can play in reducing inequalities in risk of adverse perinatal outcomes.”

This demonstrates an implicit assumption by those carrying out this study that the way to reduce ethnic and socioeconomic inequalities in outcome is to reach for the extremely blunt instrument of inducing labour early in many thousands of women and birthing people, the vast majority of whom would experience no benefit. There is no consideration for the potential unintended consequences of such a policy, such as the harm that induction at 39 weeks may cause to the physiological, emotional and mental well-being of mother and baby. Neither does there seem to be any thought for how a further increase in induction rates would damage the ability of our overstretched maternity services to provide safe, personalised and equitable care for all maternity service users and their babies.

This is a case of wanting to treat the symptom, not the cause. There may be many interlocking factors that underlie higher rates of poor outcomes for certain groups. The authors comment that “The greater benefit from IOL with birth at 39 weeks in women from more socioeconomically deprived areas may be explained by an increased prevalence of maternal risk factors” by which they probably mean risk factors such as smoking status, drug abuse and high BMI for which they had no data. These are all known to be more common amongst those of low socioeconomic status ( Adult smoking habits in the UK - Office for National Statistics (ons.gov.uk), Health Survey for England, 2021 part 1 - NDRS (digital.nhs.uk) Drug misuse in England and Wales - Office for National Statistics. That’s before any consideration of other factors such as nutrition and housing etc. or the many potential barriers to accessing care that women of low socioeconomic status may face. It has been recognised that systematic racism impacts on the maternity care of Black, Brown and mixed ethnicity women and birthing people Inquiry into racial injustice in maternity care - Birthrights. It is likely that social prejudice likewise can affect how those of low socioeconomic status are treated by staff.

The NMPA report Inequalities Sprint Audit Report 2021 made recommendations to “Target efforts for a life-course approach to improve the health of people, addressing the wider social determinants of health as well as specific health-related risk factors. Offer individualised preconception and antenatal information tailored to their circumstances, including BMI, smoking, pre-existing comorbidities (hypertension and type 2 diabetes) and whether this is their first birth or they have previously had a caesarean birth” and “Target efforts to reduce smoking.”

All of this points to a need for sensitive and supportive pre-pregnancy and maternity care - preferable with a Continuity of Carer model - for those in the lower socioeconomic groups as well as those ethnic groups that suffer worse maternity outcomes. This has the potential to improve the overall short and long-term wellbeing of women, birthing people and their babies without the need for ever more medicalised births.