We conclude that observed confounder control was insufficient to address confounding by indication. Our study supports the idea that access to higher-volume hospitals could improve perinatal outcomes at the population level.

Third, we compared women living in neighbouring municipalities but in different hospital areas. This approach assumes that access to obstetric and neonatal care would not influence the decision of where to settle among two neighbouring municipalities.

Second, we compared outcomes for births among women who moved between hospital areas between births using a sibling design. This approach assumes that access to obstetric and neonatal care did not influence the mothers change of residence.

First, we compared outcomes for women giving birth at hospitals of different sizes and with varying travel times, adjusted for observed confounding factors. This approach assumes adjustment for all confounding between women attending high and low-volume hospitals.

This is not straightforward to study since there are differences in which women give birth at smaller and larger hospitals. We used three analytical designs based on different assumptions to triangulate evidence.

We have investigated how hospital volume and travel time affect various outcomes, including perinatal mortality, risk of delivery in transit, induction of labour, cesarean section, and instrumental delivery.