68 Chapter 3 place of birth was defined as that at the onset of labour. Therefore, women who were referred, during pregnancy and before labour, to obstetrician-led care could not have a birth planned in midwife-led care. Interventions The following primary outcome variables concerning interventions were examined: the rates of episiotomy in vaginal births; AROM, and postpartum administration of oxytocin. Data about AROM and postpartum administration of oxytocin were only available in the midwifery part of the perinatal database and were therefore not described for the obstetrician-led care group. Adverse outcomes Secondary neonatal and maternal outcomes were: antepartum and intrapartum stillbirth; neonatal mortality up to seven days; Apgar score below seven at five minutes; third or fourth degree perineal tear among vaginal births; and postpartum haemorrhages (PPH) of more than 1,000 ml. Data analysis The differences between primary outcomes per region were analysed initially as a whole and then in subgroups of women in midwife-led or in obstetricianled care, at the onset of labour and at the time of birth. Crude rates were given separately for nulliparous and multiparous women since rates of interventions are not comparable for these groups. The analyses were conducted on the level of the women, with the region as independent variable, and the primary or secondary outcome as dependent variable. In the logistic regression analyses, the overall rate of the outcome we investigated, weighted for the number of women per region, was considered as a reference. Univariable analyses were conducted in order to gain insight into the variations in the rates of primary and secondary outcomes among the twelve regions. These were followed by multivariable logistic regression analyses with adjustments for: maternal age (<40 years, 40 years or older); ethnic background (Dutch, non-Dutch); socioeconomic position (low, medium or high) - based on postal code and education, employment and level of income; and the degree of urbanisation (rural, intermediate or urban) - based on women’s residential postal code (Statistics Netherlands). A confidence interval of 99% was chosen to account for multiple testing within a large dataset. The aim of these analyses was to explore differences between the regions, which are not explained by maternal characteristics, but may be explained by variations between care professionals and/or care settings such as midwifery practices and hospitals. Therefore, we did not perform multilevel analyses. Figures with maps of the regions and boxplots with adjusted odds ratios (ORs) and confidence intervals were used to visualise the results. To test whether variables with a significant regional variation correlated with each other, a Spearman’s rank correlation coefficient was calculated. Correlations in the 12 regions with rho ≥ 0.57 or ≤ - 0.57 corresponded with a p-value of 0.05, and
RkJQdWJsaXNoZXIy MTk4NDMw