Multiple logistic regression was used to adjust the estimates for maternal age, ethnicity, deprivation, placenta previa at first birth, inter-birth interval and pregnancy complications. In addition, we conducted a meta-analysis of the reported results in peer-reviewed articles since 1980.