Beyond utilization: measuring effective coverage of obstetric care along the quality cascade.

Objective: To determine the effective coverage of obstetric care in a rural Tanzanian region and to assess differences in effective coverage by wealth.
Design: Cross-sectional structured interviews.
Setting: Pwani Region, Tanzania.
Participants: The study includes 24 rural, government-managed, primary healthcare clinics and their catchment populations. From January-April 2016, we conducted a household survey of a census of women with recent deliveries, health worker knowledge surveys and facility audits.
Main Outcome Measures: We explored the proportion of women receiving quality care through the cascade and conducted an equity analysis by wealth.
Results: In total, 2,910 of 3,564 women (81.6%) reported delivering their most recent child in a health facility, 1,096 of whom delivered in a study facility. Using a minimum threshold of quality, the effective coverage of obstetric care was 25%. Quality was lowest in the emergency care dimensions, with the average score on the provider knowledge tests at 47% and the average provision of basic emergency obstetric services below 50%. The wealthiest 20% of women were 4.1 times as likely to deliver in facilities offering at least the minimum threshold of quality care through the cascade compared to the poorest 80% of women (95% confidence interval: 1.5-11.3).
Conclusions: Effective coverage of delivery care is very low, particularly among poorer women. Health worker knowledge caused the sharpest decline in effective coverage. Measures of effective coverage are a better performance measure of under-resourced health systems than utilization. Equity analyses can further identify important discrepancies in quality across socio-economic levels.
Trial Registration: ISRCTN 17107760.
(© The Author 2016. Published by Oxford University Press in association with the International Society for Quality in Health Care. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com)