Multisite implementation of trained volunteer doula support for disadvantaged childbearing women: a mixed-methods evaluation

Background: The research examined an innovative volunteer doula service, established in one city and rolled out to four other sites. The initiative offers support to disadvantaged women with the aim of enhancing well-being and improving the uptake of health services. Aims: The project addressed four broad questions: implications for the NHS; health and psychosocial impacts for women; impacts on doulas; and the processes of implementing and sustaining a volunteer doula service for disadvantaged childbearing women. Design: A mixed-methods study using interviews, focus groups and questionnaires to obtain primary data from a range of stakeholders. Existing data sets were used to examine clinical and public health outcomes and contributed to a cost–consequence analysis. A realistic evaluation perspective supported investigation of a complex intervention in its real-world context. Outcomes: We assessed impacts, perceptions and experiences of women, doulas, midwives and heads of midwifery. Clinical and public health outcomes included epidural use, rates of caesarean section, low birthweight, admission to neonatal unit, smoking and breastfeeding. The costs of running a doula service and cost implications for the NHS were calculated. Data sources: Data sources included the service database at the original site; available outcomes were compared against those in reference data sets. Women completed questionnaires and a small number participated in focus groups. Doulas contributed information through focus groups, postal questionnaires and telephone interviews. Staff, commissioners and local champions of doula services provided information through interviews and focus groups. Midwives and heads of midwifery took part in focus groups and telephone interviews respectively. Results: Although doula-supported women in the original site used fewer epidurals and generally required fewer caesarean sections than women in reference groups, these differences were not statistically significant. The utility of comparisons is constrained ...