Summary of main results

In this update we included only one trial, involving 942 women and 942 neonates. There are fewer trials in this update compared to the previous version of the review (Sotiriadis 2018) because we have assessed all the previously included trials against the Cochrane Pregnancy and Childbirth trustworthiness tool. Our results are therefore based only on data that we have judged to be trustworthy.

Antenatal corticosteroids probably reduce the risk of admission to neonatal special care for respiratory complications compared with usual care (summary of findings Table 1).

Compared with usual care, it is uncertain if antenatal corticosteroids have any effect on the risk of RDS, TTN or the risk of needing mechanical ventilation because the certainty of the evidence is low (for the outcomes of RDS and TTN) or very low (for the outcome of mechanical ventilation) and the 95% CI is consistent with possible benefit and possible harm. It is uncertain if antenatal steroids have any effect on the risk of maternal development of postpartum infection/pyrexia (there were zero events) (summary of findings Table 1).

In terms of our secondary outcomes, it is uncertain if antenatal corticosteroids have any effect on the risk of admission to neonatal special care, sepsis, perinatal death, readmission for respiratory problems, cognitive impairment, or emotional and behavioural problems. Antenatal corticosteroids may reduce the length of stay in the NICU. There is limited to no outcome data on substantial maternal health outcomes.

Overall completeness and applicability of evidence

Every effort was made to identify all published and unpublished randomised trial data for the use of prophylactic corticosteroids prior to elective caesarean section at term. Prophylactic antenatal corticosteroids may reduce the rates of admission to special care or NICU for respiratory complications after elective caesarean section (Stutchfield 2005). It is unclear if antenatal corticosteroids have any effect on the risk of RDS, TTN or mechanical ventilation. There is insufficient evidence to investigate if there are differences for gestational age at birth (37 + 0 to 37 + 6 weeks; 38 + 0 to 38 + 6 weeks; 39 + 0 weeks or later) in neonatal respiratory outcomes.

Currently, there are only a limited number of trials that evaluate the effects of prophylactic antenatal corticosteroid administration prior to caesarean section at term. Only one trial is included in this review (Stutchfield 2005). Therefore, there are insufficient data to draw any firm conclusions. Furthermore, this trial was undertaken over 16 years ago, and neonatal practice is likely to have changed, which could impact upon the clinical outcomes included in this review. This trial was a multicentre pragmatic study, and therefore the favourable results demonstrated may also reflect the true effect in regular clinical practice. However, as participants and outcome assessors were not blinded, and admission to a special care or intensive care unit is largely a subjective choice and may be affected by knowledge of treatment assignment, this outcome should be interpreted with caution.

Another systematic review published in 2016 included six trials comparing the use of antenatal corticosteroids with placebo or no treatment in women with a singleton pregnancy at 34 or more weeks' gestation (Saccone 2016). The population included both women expected to deliver in the late preterm period (34 + 0 to 36 + 6 weeks' gestation) and women before planned caesarean delivery at term (37 weeks' gestation or more). The authors found that administration of corticosteroids reduced neonatal respiratory morbidity. Interestingly, two of the three trials (which included women at 37 weeks' gestation or more) included in Saccone 2016 were considered for inclusion in our review, but unfortunately we were unable to confirm trustworthiness, therefore they remain in Studies awaiting classification.

Much larger numbers of participants are needed to adequately power clinical trials to evaluate the true impact of prophylactic antenatal corticosteroid administration on respiratory outcomes with associated morbidity (e.g. RDS). There is also a need to adequately assess for any maternal or neonatal harm associated with corticosteroid administration at term gestations. This is particularly important in view of the low rates of RDS and admission to NICU in this population, and therefore clinicians need to carefully consider the balance of statistical significance versus the clinical significance of the outcomes and the impact on the mother or baby. 

The balance of risk is particularly important, as currently there is only limited evidence on any long‐term follow‐up of these infants. In the cohort of infants from the ASTECS trial (Stutchfield 2005), nearly twice as many betamethasone‐exposed children did not attain English proficiency (13% versus 7%), and they were twice as likely to be ranked in the lower quartile of academic ability by teachers (18% versus 9%) (Stutchfield 2013). Although this follow‐up study suffered from a high attrition rate (49%), it highlights the need for caution and for future trials to consider potential harms, both short‐ and long‐term, in their study outcomes. Further data on maternal health outcomes are also needed to provide assertions that the treatment is not harmful to women in terms of adverse effects or increased infectious morbidity. 

There are currently nine studies that are ongoing and will be considered for inclusion in future updates. Encouragingly, one of the registered trials includes key clinical outcomes for both the infant and mother and has potential plans for long‐term follow‐up (Groom 2020).

Quality of the evidence

Current evidence comes from only one study. In an effort to ensure that all studies included in the review are sufficiently credible we have applied the trustworthy screening tool developed by the Cochrane Pregnancy and Childbirth Group (Figure 1). Therefore, three of the studies (Ahmed 2015; Nada 2016; Nooh 2018) included in the last version of this systematic review (Sotiriadis 2018) are awaiting classification and are not included in this update. We are awaiting further information from the trial authors particularly in relation to prospective registration, a requirement for all trials published after 2010. 

The data we present here come from the only trial that met our prespecified criteria for trial trustworthiness; therefore we can be confident that our findings are based on rigorous clinical trial evidence. Our decision not to include data from three trials that were included in the previous version of this review is based on concerns about the rigour and conduct of those trials. Similar concerns were raised in a recent article specifically examining the trustworthiness of the same trials (Mol 2021). Mol and colleagues have approached the investigators and their institutions and have received no documentation to confirm the findings presented in the published reports of the trials. Our own approaches to the study investigators have also failed to produce a response.

Furthermore, we decided not to include a further nine studies that met our inclusion criteria in terms of population, intervention, comparator and study design. The investigators of these studies have not yet responded to our attempts to seek clarification regarding various aspects of study conduct.

Unresolved queries about data integrity and lack of prospective trial registration are our main concerns about the trials whose data we have not used in this review. Where we have asked for further information about data presented in a conference abstract or in a full trial report, for example the cause of death of infants who died in a study, we are still waiting for a response. In the absence of adequate assurance about data provenance, we deemed it appropriate not to use these data in our analysis. Similarly, data from trials that were not registered before the investigators began to enrol participants may be problematic. Despite the long‐standing recognition of the important role of prospective trial registration in reducing the risk of selective outcome reporting (De Angelis 2004; Simes 1986), compliance with prospective registration remains at only 41.7% according to a recent analysis of 10,500 randomised controlled trials (Al‐Durra 2020). Without prospective registration it is very difficult to assess whether completed, published trials have reported their results in full. Systematic reviews and clinical guidelines that include evidence from trials whose integrity is uncertain can in turn lead to patients being put at risk if interventions are implemented without reliable evidence of their effectiveness.

The overall certainty of the evidence for the primary outcomes was found to be low and very low (summary of findings Table 1), apart from the outcome of admission to neonatal special care (all levels) for respiratory morbidity, which was of moderate certainty. The main concern with the included trial is the high risk of performance bias, as both participants and health professionals were aware of their group allocation. Although all outcomes may be affected, it is likely that more subjective management decisions or assessments of outcome are more susceptible to bias arising from lack of blinding, as compared to more objective ones (Higgins 2008). Admission of neonates to special care is one such example; clinicians potentially biased in favour of the intervention may have been more likely to organise admission to NICU for babies not exposed to corticosteroids. Such bias was possible, but less likely, for the radiological diagnosis of RDS or TTN by independent specialists. Similarly, admission to lower levels of special care may be more prone to bias than referral to NICU, as the latter would require more clinical and laboratory findings.

The limitations of the evidence are reflected in its assessment according to GRADE. As shown in summary of findings Table 1, for the outcomes of RDS and TTN, the certainty of the evidence was downgraded by one level for study limitations and one level for imprecision. Therefore, the certainty of the evidence for these outcomes was judged as low. For the outcome of NICU admission, we downgraded by one level for study limitations, therefore the certainty of the evidence is judged as moderate. We additionally downgraded by one level for imprecision for the outcome of need for mechanical ventilation, therefore the certainty for this outcome was judged as very low. In practical terms, this means that the true effect may be (or it is likely to be, for the outcome of need for mechanical ventilation) substantially different from our estimates, and the results should therefore be interpreted with caution.

A further potential concern is the lack of long‐term outcomes in the studies. This could result in preferentially favouring antenatal corticosteroid administration, when potential long‐term harms remain unknown.

Potential biases in the review process

We endeavoured to minimise the potential bias of the review process by ensuring an up‐to‐date search was undertaken to try and identify all relevant trials pertinent to the review. Two or more review authors independently appraised and extracted the trial data required and the data extraction process was complete and without missing data that would potentially exclude eligible studies. In the rare case where data were missing, or in order to adequately assess the risk of bias of the trial, we sought additional data from the trial authors who responded on average in a timely and clear manner.

Two of the review authors independently conducted the GRADE assessment of the certainty of evidence and four of the review authors independently applied the trustworthiness criteria to the studies that met our inclusion criteria. We acknowledge that there may be an element of subjectivity in both the GRADE assessment and in assessing trustworthiness, but we made every effort to minimise any risk of bias in this respect by ensuring that we carried out these steps independently and with referral to a Cochrane Pregnancy and Childbirth editor, where necessary, in order to reach consensus.

Agreements and disagreements with other studies or reviews

The favourable effect of antenatal corticosteroids on fetuses at risk for preterm birth has been reported in multiple trials and systematic reviews. Steroids were first tried as a means to accelerate fetal lung maturation and reduce perinatal respiratory morbidity; the latest update of the relevant Cochrane Review reaffirms that antenatal steroids reduce the incidence of RDS (RR 0.71, 95% CI 0.65 to 0.78) and moderate/severe RDS (RR 0.70, 95% CI 0.59 to 0.83) for preterm birth. In contrast, the evidence is uncertain with regard to the risk of chronic lung disease (RR 0.86, 95% CI 0.41 to 1.79) (McGoldrick 2020). It remains uncertain whether antenatal corticosteroids are equally effective in reducing neonatal respiratory morbidity in term infants delivered by elective caesarean section. 

Apart from respiratory morbidity, antenatal corticosteroids in fetuses at risk for preterm birth also reduce neonatal mortality and severe neonatal morbidity, including necrotising enterocolitis and intraventricular haemorrhage), without increasing maternal and perinatal complications (McGoldrick 2020; Sotiriadis 2015). Moreover, there is some evidence, partly derived from non‐randomised trials, that steroids in fetuses at risk for preterm birth can also reduce severe neurological morbidity in childhood (Sotiriadis 2015; McGoldrick 2020). 

In contrast, a recent population‐based cohort study has raised some concern that administration of antenatal corticosteroids at term gestations is associated with mental and behavioural disorders in the exposed children (Raikkőnen 2020). Despite some inherent limitations with this study (Raikkőnen 2020), this concurs with the cohort of infants from the ASTECS trial (Stutchfield 2005), who were followed up into childhood and were more likely to be ranked in the lower quartile of academic ability by their teachers (Stutchfield 2013). This evidence is certainly not conclusive and merits further investigation and consideration in any future randomised trials. 

A common clinical question is whether either betamethasone or dexamethasone is superior to the other regarding their effectiveness and safety profile. The paradigm of fetuses at risk for preterm birth shows that there are no substantial differences between the two agents and the sample sizes for dexamethasone are usually much smaller than those for betamethasone, resulting in wide confidence intervals for the former drug (ASTEROID 2019). As only betamethasone was tested in our review, we cannot draw conclusions for comparative effectiveness. This comparison is also included in another Cochrane Review (Brownfoot 2013).