Can investment in quality drive use? A cluster-randomised controlled study in rural Tanzania

Background: Reduction in maternal and neonatal mortality requires women's use of high-quality facilities for childbirth. Evidence that quality influences women's selection of place of delivery suggests that an investment in quality may increase use of facilities for childbirth. We assessed the impact of a quality improvement project on facility use for childbirth among women in Tanzania. Methods: Of 24 government-managed primary care clinics in Pwani Region, Tanzania, we randomly selected 12 to receive an intervention consisting of: training, supportive supervision, infrastructure support, and peer outreach. The cluster was defined as the health clinic and the villages assigned to its catchment area by the ward. We collected household surveys of women living within the catchment area of each clinic with a birth in the past year at baseline (February–April 2012) and endline (January–April 2016). Women reported the location of birth for each of their deliveries, including the facility name. The primary outcome was report of facility use for childbirth for their most recent birth. The effect of the intervention was assessed using difference-in-difference analysis. We conducted an exploratory secondary analysis among women least likely to use the health system—ie, those whose prior delivery was a home birth. We investigated three pathways from the intervention to increased facility use: improved obstetric quality, improved antenatal care (ANC) quality, and improved links between the health system and the community. Findings: In the intervention clusters, 999 (71·7%) of 1393 women gave birth in a facility at baseline and 1165 (85·3%) of 1365 did so at endline. The corresponding figures for the control clusters were 1146 (72·3%) of 1586 at baseline and 1411 (81·1%) of 1739 at endline. The intervention thus led to an increase in facility births of 6·7 percentage points (95% CI 0·6–12·8). The intervention was substantially more successful in increasing use among women least likely to utilise the health system, giving a 18·3 percentage point increase (95% CI 10·1–26·6). Among the hypothesised mechanisms, the most likely pathway of effect was through ANC: the intervention led to an increase in ANC quality, with providers performing an additional 0·8 (95% CI 0·21–1·34) actions among the low-use population and 0·5 actions among the full population (95% CI −0·01 to 1·01). Interpretation: The quality improvement intervention led to a modest increase in facility use for childbirth and a strong increase among women whose previous delivery was at home. Our analysis provides empirical evidence that investment in quality can increase health care use. In an environment of rising use, quality may be a mechanism for encouraging remaining non-system users to engage in the health system, playing a role in achieving universal health coverage. Funding: US National Institutes of Health 1R01AI093182