The causal effect of delivery volume on severe maternal morbidity: an instrumental variable analysis in Sichuan, China

OBJECTIVE: Findings regarding the association between delivery volume and maternal health outcomes are mixed, most of which explored their correlation. This study aims to demonstrate the causal effect of delivery volume on severe maternal morbidity (SMM) in China. METHODS: We analysed all women giving birth in the densely populated Sichuan province with 83 million residents in China, during the fourth quarters of each of 4 years (from 2016 to 2019). The routinely collected discharge data, the health institutional annual report data and road network data were used for analysis. The maternal health outcome was measured by SMM. Instrumental variable (IV) methods were applied for estimation, while the surrounding average number of delivery cases per institution was used as the instrument. RESULTS: The study included 4545 institution-years of data from 1456 distinct institutions with delivery services, reflecting 810 049 associated delivery cases. The average SMM rate was approximately 33.08 per 1000 deliveries during 2016 and 2019. More than 86% of delivery services were provided by a third of the institutions with the highest delivery volume (≥143 delivery cases quarterly). In contrast, less than 2% of delivery services were offered by a third of the institutions with the lowest delivery volume (<19 delivery cases quarterly). After adjusting the confounders in the IV-logistic models, the average marginal effect of per 1000 cases in delivery volume was −0.162 (95% CI −0.169 to –0.155), while the adjusted OR of delivery volume was 0.005 (95% CI 0.004 to 0.006). CONCLUSION: Increased delivery volume has great potential to improve maternal health outcomes, while the centralisation of delivery services might facilitate maternal health promotion in China. Our study also provides implications for other developing countries confronted with similar challenges to China.