Importance: Meta-analyses suggest that corticosteroids may be associated with increased survival without cerebral palsy in infants at high risk of bronchopulmonary dysplasia (BPD) but are associated with adverse neurologic outcomes in low-risk infants. Whether this association exists in contemporary practice is uncertain because most randomized clinical trials administered corticosteroids earlier and at higher doses than currently recommended.

Objective: To evaluate whether the pretreatment risk of death or grade 2 or 3 BPD at 36 weeks' postmenstrual age modified the association between postnatal corticosteroid therapy and death or disability at 2 years' corrected age in extremely preterm infants.

Design, setting, and participants: This cohort study analyzed data on 482 matched pairs of infants from 45 participating US hospitals in the National Institute of Child Health and Human Development Neonatal Research Network Generic Database (GDB). Infants were included in the cohort if they were born at less than 27 weeks' gestation between April 1, 2011, and March 31, 2017; survived the first 7 postnatal days; and had 2-year death or developmental follow-up data collected between January 2013 and December 2019. Corticosteroid-treated infants were propensity score matched with untreated controls. Data were analyzed from September 1, 2019, to November 30, 2022.

Exposure: Systemic corticosteroid therapy to prevent BPD that was initiated between day 8 and day 42 after birth.

Main outcomes and measures: The primary outcome was death or moderate to severe neurodevelopmental impairment at 2 years' corrected age. The secondary outcome was death or moderate to severe cerebral palsy at 2 years' corrected age.

Results: A total of 482 matched pairs of infants (mean [SD] gestational age, 24.1 [1.1] weeks]; 270 males [56.0%]) were included from 656 corticosteroid-treated infants and 2796 potential controls. Most treated infants (363 [75.3%]) received dexamethasone. The risk of death or disability associated with corticosteroid therapy was inversely associated with the estimated pretreatment probability of death or grade 2 or 3 BPD. The risk difference for death or neurodevelopmental impairment associated with corticosteroids decreased by 2.7% (95% CI, 1.9%-3.5%) for each 10% increase in the pretreatment risk of death or grade 2 or 3 BPD. This risk transitioned from estimated net harm to benefit when the pretreatment risk of death or grade 2 or 3 BPD exceeded 53% (95% CI, 44%-61%). For death or cerebral palsy, the risk difference decreased by 3.6% (95% CI, 2.9%-4.4%) for each 10% increase in the risk of death or grade 2 or 3 BPD and transitioned from estimated net harm to benefit at a pretreatment risk of 40% (95% CI, 33%-46%).

Conclusions and relevance: Results of this study suggested that corticosteroids were associated with a reduced risk of death or disability in infants at moderate to high pretreatment risk of death or grade 2 or 3 BPD but with possible harm in infants at lower risk.