BackgroundThis study compared government sub-district hospitals in Bangladesh without globally standard midwives, with those with recently introduced midwives, both with and without facility mentoring, to see if the introduction of midwives was associated with improved quality and availability of maternity care. In addition, it analysed the experiences of the newly deployed midwives and the maternity staff and managers that they joined.MethodsThis was a mixed-methods observational study. The six busiest hospitals from three pre-existing groups of government sub-district hospitals were studied; those with no midwives, those with midwives, and those with midwives and mentoring. For the quantitative component, observations of facility readiness (n = 18), and eight quality maternity care practices (n = 641) were carried out using three separate tools. Willing maternity staff (n = 237) also completed a survey on their knowledge, perceptions, and use of the maternity care interventions. Descriptive statistics and logistic regression were used to identify differences between the hospital types. The qualitative component comprised six focus groups and 18 interviews involving midwives, other maternity staff, and managers from the three hospital types. Data were analysed using an inductive cyclical process of immersion and iteration to draw out themes. The quantitative and qualitative methods complemented each other and were used synergistically to identify the study’s insights.ResultsQuantitative analysis found that, of the eight quality practices, hospitals with midwives but no mentors were significantly more likely than hospitals without midwives to use three: upright labour (94% vs. 63%; OR = 22.57, p = 0.001), delayed cord clamping (88% vs. 11%; OR = 140.67, p < 0.001), skin-to-skin (94% vs. 13%; OR = 91.21, p < 0.001). Hospitals with mentors were significantly more likely to use five: ANC card (84% vs. 52%; OR = 3.29, p = 0.002), partograph (97% vs. 14%; OR = 309.42, p = 0.002), upright positioning for labour (95% vs. 63%; OR = 1850, p < 0.001), delayed cord clamping (98% vs. 11%; OR = 3400, p = 0.003), and skin-to-skin contact following birth (93% vs. 13%; OR = 70.89, p < 0.001) Qualitative analysis identified overall acceptance of midwives and the transition to improved quality care; this was stronger with facility mentoring. The most resistance to quality care was expressed in facilities without midwives. In facilities with midwives and mentoring, midwives felt proud, and maternity staff conveyed the greatest acceptance of midwives.ConclusionFacilities with professional midwives had better availability and quality of maternity care across multiple components of the health system. Care quality further improved with facility mentors who created enabling environments, and facilitated supportive relationships between existing maternity staff and managers and the newly deployed midwives.