Objectives: In 1990 a meta-analysis of randomised trials showed 70% lower mortality after antenatal corticosteroid therapy for 24 h or more for infants born before 31 weeks gestation. We investigated whether antenatal corticosteroid therapy has increased in these infants since 1990 and studied variations in use by hospital of birth.
Design: Retrospective cohort study in 1601 infants in nine neonatal units.
Subjects: Neonatal admissions before 31 weeks of gestation from January 1988 to October 1993. MEASURE OF OUTCOME: Corticosteroids administered for 24 h or more before delivery.
Results: Data were obtained in 1579 (99%) infants. The proportion (range) in each hospital whose mothers had antenatal corticosteroids for 24 h or more was 16% (0-43) in 1988-89 and 29% (0-36) in 1990-93 (P < 0.001). In post hoc analyses, 65/347 (20%) births in district hospitals had treatment for 24 h or more compared with 354/1254 (28%) in teaching hospitals (P = 0.001).
Conclusions: Antenatal corticosteroid therapy increased but varied by hospital of birth. This may reflect varying performance, or bias from reporting, selection or referral. Ideally, corticosteroid therapy should be compared in women at risk of preterm delivery, but standardising risk or indications for delivery between hospitals and accurate ascertainment presents major difficulties. To minimise selection or referral bias, hospitals should publish, for all mothers delivering between 24 and 33 weeks and 6 days gestation 48 h or more after admission, the proportions treated 1. for 24 h or more (target: > 70%), or 2. at all (target: > 90%).