ACOG SMFM Practice Advisory: Use of Antenatal Corticosteroids at 22 Weeks of Gestation

Corticosteroid administration before preterm birth is an important antenatal therapy available to improve newborn outcomes. In the setting of periviable birth, if neonatal resuscitation is planned after appropriate counseling (see Periviable Birth and Perinatal Palliative Care), antenatal corticosteroids have been recommended at 24 0/7 weeks to 25 6/7 weeks of gestation and may be considered at 23 0/7 weeks to 23 6/7 weeks of gestation 1.

At 20 0/7 weeks to 22 6/7 weeks of gestation, previous studies demonstrated no significant reduction in neonatal death and neurodevelopmental impairment with the administration of antenatal corticosteroids 1 2. More recent retrospective, observational data suggest that antenatal corticosteroid administration, in combination with resuscitation efforts, improves rates of survival at 22 weeks of gestation, although there is little overall impact on the absolute rate of survival without major morbidities.

A 2021 systematic review and meta-analysis that included 31 retrospective, observational studies of 2,226 infants who were delivered at 22 0/7 weeks to 22 6/7 weeks of gestation found that survival among infants born to pregnant individuals receiving antenatal corticosteroids was twice that of infants born to pregnant individuals not receiving antenatal corticosteroids (39.0% versus 19.5%; P<.01) 3. One multicenter observational cohort that analyzed over 1,000 live births at 22 0/7 weeks to 22 6/7 weeks of gestation found that infants who received antenatal corticosteroids with postnatal life support were more likely to survive than infants who received postnatal life support alone [38.5% versus 17.7% (adjusted risk ratio, 2.11; 95% CI, 1.68–2.65)] 4. While survival without a major morbidity was improved with antenatal corticosteroids, the absolute rate of survival without major morbidities still remained very low [4.4% versus 1.0% (adjusted risk ratio, 4.35; 95% CI, 1.84–10.28)] 4. It is important to note that there are significant limitations in these data, including selection bias in choosing to intervene and confounding by indication.