Mode of delivery and the risk of delivery-related perinatal death among twins at term: a retrospective cohort study of 8073 births Gordon C.S. Smith, a Imran Shah, a Ian R. White, b Jill P. Pell, c Richard Dobbie d Objective To determine the risk of perinatal death among twins born at term in relation to mode of delivery. Design Retrospective cohort study. Setting Scotland 1985 – 2001. Population All twin births at or after 36 weeks of gestation, excluding antepartum stillbirths and perinatal deaths due to congenital abnormality (n ¼ 8073). Methods The outcome of first and second twins was compared using McNemar’s test and the outcome of twin pairs in relation to mode of delivery was compared using exact logistic regression. Main outcome measures Intrapartum stillbirth or neonatal death of either twin. Results Overall, there were six deaths of first twins and 30 deaths of second twins (OR for second twin 5.00, 95% CI 2.00 – 14.70). The odds ratio for death of the second twin due to intrapartum anoxia was 21 (95% CI 3.4–868.5). The associations were similar for twins delivered following induction of labour and for sex discordant twins. However, there was no association between birth order and the risk of death among 1472 deliveries by planned caesarean section. There was death of either twin among 2 of 1472 (0.14%) deliveries by planned caesarean section and 34 of 6601 (0.52%) deliveries by other means (P ¼ 0.05, odds ratio for planned caesarean section 0.26 [95% CI 0.03–1.03]). The association was similar when adjusted for potential confounders. Assuming causality, we estimate that 264 caesarean deliveries (95% CI 158–808) would be required to prevent each death. Conclusion Planned caesarean section may reduce the risk of perinatal death of twins at term by approx- imately 75% compared with attempting vaginal birth. This is principally due to reducing the risk of death of the second twin due to intrapartum anoxia. INTRODUCTION The effect of birth order on the risk of perinatal death among twins has been the subject of study for some years. Clinically, it is well recognised that the second twin is at increased risk of complications during labour due to diffi- culties in fetal monitoring and the possibility of traumatic delivery following vaginal birth of the first twin. 1 However, large scale epidemiological studies have generally failed to confirm increased perinatal mortality in relation to birth order. 2–4 We recently re-examined this and found a marked excess of delivery-related perinatal deaths among second twins compared with first twins. 5 The failure of previous studies to show this was due to lack of details regarding the timing and cause of perinatal death, failure to stratify by gestational age and the use of unpaired statistical tests to compare the outcome of first and second twins. Our findings suggested that delivery of all twins by planned caesarean section might reduce the risk of perinatal death. However, our own and other analyses lacked sufficient numbers to make this comparison. In the present study, we analysed data from over 8000 twin pairs born at or after 36 weeks of gestation and sought to determine the associa- tion between mode of delivery and the risk of perinatal death. METHODS The Scottish Morbidity Record (SMR2) collects infor- mation on clinical and demographic characteristics and outcomes for all patients admitted to Scottish maternity hospitals. The register is subjected to regular quality assurance checks and has been greater than 99% complete since the late 1970s. 6 A quality assurance exercise in 1996/ 97 demonstrated that mode of delivery was free of major or minor errors for 99.2% of cases in singletons or first twins and 95.2% of second twins [Dr Jim Chalmers MB ChB, BJOG: an International Journal of Obstetrics and Gynaecology August 2005, Vol. 112, pp. 1139–1144 D RCOG 2005 BJOG: an International Journal of Obstetrics and Gynaecology www.blackwellpublishing.com/bjog a Department of Obstetrics and Gynaecology, Cambridge University, UK b Medical Research Council Biostatistics Unit, Institute of Public Health, Robinson Way, Cambridge, UK c Department of Public Health, Greater Glasgow NHS Board, Glasgow, UK d Information and Statistics Division, Common Services Agency, Edinburgh, UK Correspondence: Professor G. C. S. Smith, Cambridge University, Rosie Maternity Hospital, Cambridge CB2 2SW, UK. DOI:10.1111/j.1471-0528.2005.00631.x