Twenty-seven RCTs were included (3260 infants).
1. Overall mortality rate.
The latest age of reported mortality varied from 28 days to 3 years of age.
There was a non-significant difference between treatment groups favouring corticosteroids. ERD = -0.1% (95% CI: -2.9%, 2.8%). In no individual study was the reported mortality significantly different between groups. ERD ranged from -26.5% to + 16.7%.
After excluding two RCTs (early dexamethasone compared with selective dexamethasone at a later stage and dexamethasone started at 14 days compared with selective treatment at 28 days): difference between treatment groups was not significant. ERD = -0.7% (95% CI: -4.0%, 2.6%).
Mortality sensitivity analyses.
Postnatal age: mortality rate was non-significantly lower only in the subgroup with treatment starting between 7 and 14 days. Age < 7 days: ERD = 1.3% (95% CI: -2.3%, 4.9%). Age 7 to 14 days: ERD = -5.0 (95% CI: -11.0, 1.1). Age > 14 days: ERD = 0.3% (95% CI: -6.4%, 7.1%).
Rate of antenatal corticosteroids > 50%: there was a non-significant difference favouring the control group. ERD = 4.9% (95% CI: -1.5%, 11.4%).
There were no major differences in subgroup analyses restricted to higher dose of corticosteroids, where surfactant was given, or where contamination rate was low. Results were presented.
2. Motor dysfunction (5 RCTs reported this outcome beyond infancy).
Ages of follow-up ranged from 1 to 3 years. Follow-up rates were > 90% in all but one study. The rate of motor dysfunction overall was significantly higher in the corticosteroid group. ERD favouring controls = 11.9% (95% CI: 4.6%, 19.2%). In two studies (including one with no contamination) the rate was significantly higher in the corticosteroid groups.
Excluding one study where treatment started early (< 4 days postnatal age): there was no significant difference in the rate between treatment groups. ERD = 7.7% (95% CI: -0.6%, 16%).
Cerebral palsy (3 RCTs with treatment with postnatal steroids started after 14 days postnatal age): the overall rate was significantly higher in the corticosteroid group. ERD = 8.7% (95% CI: 0.3%, 17.0%).
3. Survival free of motor dysfunction (5 RCTs).
After excluding children lost to follow-up the rate of survival free of motor dysfunction was significantly higher in the control groups. ERD = 7.8% favouring controls (95% CI: 0.5%, 15.1%).
Assuming all those survivors lost to follow-up to have a bad outcome, the difference remained statistically significant. ERD = 9.5% (95% CI: 2.5%, 16.5%).
Assuming all those survivors lost to follow-up to have a good outcome, the difference became non-significant. ERD = 5.4% (95% CI: -1.7%, 12.4%).