Imagine a world where the joy of childbirth isn't overshadowed by the shadow of preventable deaths—yet, for countless women, especially in developing regions, maternal mortality remains a heartbreaking reality. This isn't just a statistic; it's a crisis unfolding globally, with progress toward the Sustainable Development Goals alarmingly off track. But here's where it gets fascinating: China's dramatic drop in maternal deaths over the past two decades offers a roadmap for others. Dive in, and you'll discover the hidden forces behind this triumph—and some surprises that might challenge what you think you know about healthcare priorities.
**
Unveiling the Key Contributors to Declining Maternal Mortality in China's East and West: Insights from Provincial Data**
**
Metadata and Loading Indicators**
Open Access
Peer-Reviewed
Research Article
**
*Xiaojing Zeng,
Dongjian Yang,
Shiyang Li,
Xiaolin Hua,
Yanlin Wang,
Jun Zhang,
Zhiwei Liu
**
Unveiling the Key Contributors to Declining Maternal Mortality in China's East and West: Insights from Provincial Data**
*Dongjian Yang,
Shiyang Li,
Xiaolin Hua,
Yanlin Wang,
Jun Zhang,
Zhiwei Liu
**
Published: December 4, 2025
https://doi.org/10.1371/journal.pmed.1004837
**
Visual Aids
Synopsis
Introduction
Worldwide, the fight against maternal mortality is falling short of ambitious goals set for 2030. Despite some strides, the reductions have been sluggish and inconsistent. In 2020, global maternal deaths stood at 223 per 100,000 live births, missing the target of under 70 by a wide margin. Nearly all such tragedies—around 95%—occur in low- and middle-income nations (LMICs), with 68 countries still grappling with rates exceeding 100 per 100,000 live births. Tackling this requires multisectoral efforts, and understanding why some places succeed could light the way for the rest.
China stands out as one of the top 10 LMICs that have accelerated progress in women's and children's health, per the World Health Organization. From 94.7 deaths per 100,000 live births in 1990, the nation slashed this figure to 15.7 in 2022. By 2022, China had even met the SDG target of keeping maternal mortality below 70 per 100,000. The more prosperous eastern region hit the Healthy China 2030 benchmark of under 12 per 100,000, while the western areas, though trailing, narrowed the gap with an 81% drop between 1996 and 2018. These gains stem not only from battling direct biomedical causes like hemorrhage or preeclampsia but also from broader economic, political, and cultural shifts. These 'superdeterminants,' as researchers call them, shape social drivers of health and healthcare system effectiveness, acting as upstream influences on maternal outcomes.
Drawing from a recent framework on these superdeterminants, our team set out to pinpoint the primary social and health system elements driving China's maternal mortality decline from 2004 to 2020. Social and health factors often intertwine—think of how wealth correlates with education or healthcare spending—and traditional models struggle with this overlap. Enter Bayesian kernel machine regression (BKMR), a cutting-edge tool that blends Bayesian statistics with machine learning to model these complex mixtures. It helps identify key players without getting tripped up by collinearity, ranking their impact on outcomes. For this study, we applied BKMR to assess 9 factors: urbanization levels (the share of people in cities), per capita disposable income (PCDI, or average personal wealth), average years of schooling, healthcare workers in maternal and child health per 1,000 births, hospital beds for obstetrics and gynecology per 1,000 births, government healthcare spending, prenatal booking rates (early registration for care), antenatal care rates (check-ups during pregnancy), and hospital delivery rates. This approach reveals which elements most strongly link to reductions in overall and cause-specific maternal deaths, offering lessons for LMICs aiming to accelerate their own progress.
Materials and Sources
We pulled data from trusted public repositories, including the Global Burden of Diseases, Injuries, and Risk Factors Study 2021 (GBD 2021), China's National Health Statistics Yearbooks, and China Statistical Yearbooks. Guided by Souza et al.'s framework, we selected non-biomedical factors from these sources. Our focus was the 2004–2020 period, when provincial-level data on maternal mortality and causes were reliably available.
From the National Health Statistics Yearbook, an annual overview of China's health landscape, we gathered figures for maternal mortality, cause breakdowns, hospital deliveries, antenatal care, prenatal bookings, government healthcare expenditures, maternal and child health personnel per 1,000 births, and obstetrics/gynecology beds per 1,000 births across China's 31 mainland provinces from 2005 to 2021. Causes were categorized as hemorrhage, pre-existing medical issues, hypertensive disorders in pregnancy (HDP), amniotic fluid embolism, and others. For a deeper dive into causes, we turned to GBD 2021, extracting data on maternal disorders for women aged 15–49, including hemorrhages, infections, HDP, obstructed labor, abortion complications, ectopic pregnancies, indirect deaths (from unrelated conditions worsened by pregnancy), late deaths, HIV-related deaths, and other direct issues from 2004 to 2020.
The China Statistical Yearbook, a yearly snapshot of the nation's economic and social pulse, provided PCDI, urbanization rates, and female education levels by province and year.
Term Definitions
Maternal mortality is measured as deaths per 100,000 live births. Under China's National Health Statistics Yearbook, it includes any female death during pregnancy or within 42 days postpartum from pregnancy-related causes, excluding accidents. Hospital delivery rate is the percentage of births in hospitals. Antenatal care rate is the share of pregnancies with at least one prenatal visit. Prenatal booking rate tracks early maternal record creation. Government healthcare spending covers medical services, subsidies, administration, and family planning. Maternal and child health personnel count professional staff per 1,000 births. Obstetrics/gynecology beds are the year-end count in those departments per 1,000 births. Urbanization rate is the urban population share. PCDI is total disposable income divided by residents. Average schooling years for females is calculated by weighting education levels by duration.
Analytical Approach
China's east and west show economic divides, so we split the 31 provinces using the Hu Huanyong Line, a demographic boundary based on density, geography, and human activities. Eastern provinces (24 total) include Anhui, Beijing, Chongqing, Fujian, Guangdong, Guangxi, Guizhou, Hainan, Hebei, Heilongjiang, Henan, Hubei, Hunan, Jiangsu, Jiangxi, Jilin, Liaoning, Shandong, Shanghai, Shanxi, Sichuan, Tianjin, Yunnan, and Zhejiang. Western provinces (7) are Gansu, Inner Mongolia, Ningxia, Qinghai, Shaanxi, Tibet, and Xinjiang. We computed regional maternal mortality as a livebirth-weighted average.
Cause-specific rates per province were derived by multiplying total mortality by the percentage for each cause.
Missing data included healthcare spending from 2004–2006, PCDI in 2004, and health personnel in 2009. We filled these with linear regressions, fitting year and province to avoid biases from early time gaps.
We mapped the 9 factors into 4 groups for BKMR's hierarchical selection: social determinants (urbanization, PCDI, schooling), health resources (personnel, beds), health financing (spending), and maternal care coverage (bookings, antenatal care, deliveries).
Using R's bkmr package, we ran BKMR models for pre- and post-2013 periods—2013 marked the end of China's Reducing Maternal Mortality and Eliminating Neonatal Tetanus program (2000–2013). This initiative, starting in 378 western counties and expanding nationally by 2009, boosted coverage amid MDG 5 goals. We standardized factors, ran 10,000 Markov chain Monte Carlo iterations, and included province identifiers to handle repeated measures.
BKMR estimated group posterior inclusion probabilities (GroupPIP) and conditional PIPs (CondPIP) to rank importance. Single-exposure risk summaries quantified impacts of shifting a factor from its 75th to 25th percentile, with others fixed at a quantile.
We selected top factors using: 1) Groups with GroupPIP ≥ 0.9; 2) Within those, the highest CondPIP; 3) Significant negative exposure-response estimates.
Sensitivity checks included linear mixed-effects models (univariate and multivariate, livebirth-weighted) for BKMR-identified factors, inflation-adjusted spending using 2012 IMF data, and MICE imputation for missing values.
This study follows the GATHER guidelines.
Key Findings
Maternal mortality dropped sharply in both regions from 2004 to 2020, closing urban-rural divides.
From GBD 2021, top causes were maternal hemorrhage, indirect deaths, other direct disorders, and HDP.
Provincial data showed persistent east-west differences but converging trends.
BKMR highlighted five key factors: hospital delivery, antenatal care, urbanization, healthcare spending, and PCDI.
Pre-2013 in the east, hospital delivery, urbanization, and spending drove reductions. Post-2013, urbanization and delivery were crucial. In the west, antenatal care and PCDI mattered pre-2013, shifting to care post-2013.
For causes: Hospital delivery and care linked to hemorrhage declines; urbanization and spending to other issues; care to HDP.
Sensitivity analyses confirmed results.
Interpretation
Our mixture analysis underscores maternal care coverage (antenatal visits and hospital births) as pivotal, bolstered by social factors and financing. For beginners, think of antenatal care as regular pregnancy check-ups that catch risks early, while hospital deliveries provide skilled birth attendants and emergency tools—both reducing delays in life-saving interventions, as seen in China's program that subsidized births and trained staff.
But here's the part most people miss: Without quality, more coverage alone won't suffice. Studies show facility births cut deaths by up to 79% only with better care integration. China's approach—subsidies, education, referrals, and training—exemplifies this, increasing deliveries by 68.7% and visits to five or more by 19 percentage points in western areas.
Globally, LMICs lag: 11% of women in priority nations lack antenatal care, and 31% miss facility births. Barriers include costs, distance, perceptions, and norms—echoing China's past challenges.
Urbanization and financing add layers: Cities offer better access, while funding ensures services. Income empowers choices, yet global cuts in health spending (down 20% per capita in some LMICs) threaten progress. Controversially, prioritizing facility births over home ones sparks debate—some argue it risks overtreatment, but evidence leans toward lifesaving benefits in high-risk settings.
Our findings highlight China's lessons: Boost quality care and births, backed by funding and social gains. Yet, data gaps on indirect deaths or over-medicalization limit us.
For LMICs, enhancing antenatal and delivery quality could be the game-changer. What do you think—should global health funds focus more on facility upgrades than rural outreach? Share your views in the comments; does this challenge your assumptions on what really saves lives in pregnancy?
