Service configuration, unit characteristics and variation in intervention rates in a national sample of obstetric units in England: an exploratory analysis

OBJECTIVES To explore whether service configuration and obstetric unit (OU) characteristics explain variation in OU intervention rates in 'low-risk' women. DESIGN Ecological study using funnel plots to explore unit-level variations in adjusted intervention rates and simple linear regression, stratified by parity, to investigate possible associations between unit characteristics/configuration and adjusted intervention rates in planned OU births. Characteristics considered: OU size, presence of an alongside midwifery unit (AMU), proportion of births in the National Health Service (NHS) trust planned in midwifery units or at home and midwifery 'under' staffing. SETTING 36 OUs in England. PARTICIPANTS 'Low-risk' women with a 'term' pregnancy planning vaginal birth in a stratified, random sample of 36 OUs. MAIN OUTCOME MEASURES Adjusted rates of intrapartum caesarean section, instrumental delivery and two composite measures capturing birth without intervention ('straightforward' and 'normal' birth). RESULTS Funnel plots showed unexplained variation in adjusted intervention rates. In NHS trusts where proportionately more non-OU births were planned, adjusted intrapartum caesarean section rates in the planned OU births were significantly higher (nulliparous: R(2)=31.8%, coefficient=0.31, p=0.02; multiparous: R(2)=43.2%, coefficient=0.23, p=0.01), and for multiparous women, rates of 'straightforward' (R(2)=26.3%, coefficient=-0.22, p=0.01) and 'normal' birth (R(2)=17.5%, coefficient=0.24, p=0.01) were lower. The size of the OU (number of births), midwifery 'under' staffing levels (the proportion of shifts where there were more women than midwives) and the presence of an AMU were associated with significant variation in some interventions. CONCLUSIONS Trusts with greater provision of non-OU intrapartum care may have higher intervention rates in planned 'low-risk' OU births, but at a trust level this is likely to be more than offset by lower intervention rates in planned non-OU births. Further research using high quality data on unit characteristics and outcomes in a larger sample of OUs and trusts is required.