Gestational age at delivery of twins and perinatal outcomes: a cohort study in Aberdeen, Scotland

Background: Twin pregnancy is associated with a threefold increase in perinatal death compared to singletons. The objective of this study was to determine the risk of perinatal death in twins by week of gestation and to quantify the effect of known risk factors. Methods: A cohort analysis was performed using data from the Aberdeen Maternity and Neonatal Databank (AMND). The exposure was gestational age at delivery and the primary outcome was perinatal death. Adjusted hazard ratios (aHRs) for perinatal death according to gestational age at delivery were determined by multivariate Cox proportional hazards regression modelling with robust standard errors to account for clustering in the twin infants. Confounders and risk factors quantified and adjusted for in the model included maternal age, smoking, parity, marital status and year of birth. Kaplan-Meier time to event analysis was used to determine the differences in survival according to chorionicity and assisted reproduction technologies (ART) conception status. Results: The population comprised of 7,420 twin babies born between 1950 and 2013 in the Grampian area of Northern Scotland. There were 272 stillbirths in the cohort (3.67%) and 273 neonatal deaths (3.68%). Compared to delivery at 37-38 weeks, delivery before 37 weeks was associated with a 2-fold increase in perinatal death. Monochorionic twins had a 2-fold increase in perinatal death compared to dichorionic twins (aHR 2.15, 95% CI 1.60-2.90). Twins conceived by ART did not have a greater risk of perinatal death compared to those naturally conceived (aHR 1.21, 95% CI 0.87-1.68) Conclusion: This study suggests that delivery of twins at 37-38 weeks is associated with the lowest risk of perinatal death.