Background Antenatal corticosteroids (ACS) have been shown to reduce neonatal morbidity and mortality associated with preterm birth in singletons, particularly respiratory distress syndrome (RDS) and intraventricular haemorrhage. The evidence regarding the beneficial effects of ACS in twin pregnancies is less clear. The objective of this study was to assess the effect of ACS on neonatal mortality and morbidity in twins at risk of preterm (<37 weeks gestation) birth.

Methods A cohort study design was used to analyse anonymised routinely collected data on all twin deliveries between 1996 and 2014 as recorded in the Aberdeen Maternity and Neonatal Databank. Those who received ACS (dexamethasone or betamethasone) and those who did not were classified as exposed and unexposed respectively. Outcomes included: stillbirth, neonatal death, admission to neonatal intensive care unit, as well as specific neonatal events, namely, RDS, bronchopulmonary dysplasia, intraventricular haemorrhage, neonatal seizure and necrotising enterocolitis. A composite adverse outcome variable was created for neonates with one or more of the pre-specified neonatal events. Multilevel mixed-effect generalized linear models were used to evaluate effect of ACS on neonatal morbidity and mortality. Associations are presented as relative risks (RR) with 95% Confidence Intervals (CI). Propensity score matched fixed effects models were used to account for confounding by indication.

Results The study included 3312 babies; 1686 babies whose mothers had received ACS and 1626 babies whose mothers had not. The propensity score matched fixed effects model showed that ACS administration was not significantly associated with RDS [RR 0.92 (95% CI 0.67, 1.27)], bronchopulmonary dysplasia [RR 1.05 (0.78, 1.41)] or the composite adverse neonatal outcome [RR 1.11 (0.92, 1.34)]. ACS administration was associated with a reduction in still-birth rate [RR 0.46 (0.30, 0.69)] but associated with increased admission to neonatal unit [RR 1.23 (1.07, 1.40)], intraventricular haemorrhage [RR 2.4 (1.15, 5.02)] and neonatal deaths [RR 3.15 (1.34, 7.39)].

Conclusion This study suggests that ACS (using the same administration protocols as for singleton pregnancy) is associated with a significant reduction in stillbirth rate in twins but with an increased rate of neonatal death, intraventricular haemorrhage and neonatal unit admission. There was no effect on the rates of RDS and bronchopulmonary dysplasia. It is possible that twin pregnancies respond to ACS differently from singletons. There is a need for a multi-centre randomised controlled trial to evaluate the effects of ACS in multiple pregnancies and to find the optimum dose and timing for this to be effective.