Document resource
We thank Drs de Vries, Gordon, Joseph and Hyett for their letter.1 We were pleased to learn that they recognised our efforts to reduce selection bias through excluding higher risk pregnancies. As the letter correctly notes, our study was constrained by the variables available within the International Network for Evaluating Outcomes of Neonates minimal dataset, which did not include onset of labour. In the absence of this information, mode of delivery was used as a proxy, recognising its limitations. We agree that including births with spontaneous onset of labour is another way to lower bias as provider-initiated births are often initiated due to fetal growth faltering.