I have read the journal’s policy and the authors of this manuscript have the following competing interests: YG is a member of Editorial Board of PLOS Medicine. All other authors declare no competing interests.
‡ These authors are joint senior authors.
As one of its Millennium Development Goals (MDGs), China has achieved a dramatic reduction in the maternal mortality ratio (MMR), although a distinct spatial heterogeneity still persists. Evidence of the quantitative effects of determinants on MMR in China is limited. A better understanding of the spatiotemporal heterogeneity and quantifying determinants of the MMR would support evidence-based policymaking to sustainably reduce the MMR in China and other developing areas worldwide.
We used data on MMR collected by the National Maternal and Child Health Surveillance System (NMCHSS) at the county level in China from 2010 to 2013. We employed a Bayesian space–time model to investigate the spatiotemporal trends in the MMR from 2010 to 2013. We used Bayesian multivariable regression and GeoDetector models to address 3 main ecological determinants of the MMR, including per capita income (PCI), the proportion of pregnant women who delivered in hospitals (PPWDH), and the proportion of pregnant women who had at least 5 check-ups (PPWFC). Among the 2,205 counties, there were 925 (42.0%) hotspot counties, located mostly in China’s western and southwestern regions, with a higher MMR, and 764 (34.6%) coldspot counties with a lower MMR than the national level. China’s westernmost regions, including Tibet and western Xinjiang, experienced a weak downward trend over the study period. Nationwide, medical intervention was the major determinant of the change in MMR. The MMR decreased by 1.787 (95% confidence interval [CI]: 1.424–2.142,
Although China has accomplished a 75% reduction in the MMR, spatial heterogeneity still exists. In this study, we have identified 925 (hotspot) high-risk counties, mostly located in western and southwestern regions, and among which 332 counties are experiencing a slower pace of decrease than the national downward trend. Nationally, medical intervention is the major determinant. The major determinants for the MMR in western and southwestern regions, which are developing areas, are PCI and PPWFC, while that in China’s developed areas is PCI. The interactive influence of any two of the three factors, PCI, PPWDH, and PPWFC, in western and southwestern regions was up to and in excess of 80% (
Junming Li and co-workers report on trends in maternal mortality at county level across China.
Information about the spatiotemporal trends of the maternal mortality ratio is helpful in the policymaking response to reducing the maternal mortality ratio (MMR) in developing areas.
The study can help the government to preassess the effects of policy if the corresponding magnitudes of influence of the underlying determinants can be quantified.
The quantitative statistical results of national and subnational influencing effects and patterns can help the government to create policies with precision.
We employed a Bayesian space–time model to explore the spatiotemporal trends of the MMR in 2,205 Chinese counties from 2010 to 2013 and used Bayesian multivariable regression and GeoDetector models to address 3 main ecological determinants of MMR.
The major determinants of the MMR in China are medical intervention factors. The MMR will decrease by 1.787 (95% CI 1.424–2.142,
The major determinants for the MMR in the western and southwestern regions of China are per capita income and antenatal care, while in the eastern and southern coastal regions, it is per capita income.
Many countries, and particularly developing countries, may learn from China’s dramatic improvement in maternal survival rates.
This progress has profited from long-term strategies to enhance delivery care in healthcare facilities and the provision of professional maternity care in large hospitals. There are, however, a variety of policy effects that have occurred in different areas due to regional heterogeneity.
We have revealed the dominant factors and their corresponding influencing magnitudes at the national and subnational level, and this evidence may help China or other developing countries to preassess policy effects.
The data of per capita income at the county level in China are available from the China County Statistical Yearbook on the following website:
Although some countries have achieved a decrease in the maternal mortality ratio (MMR), maternal mortality remains a global public health issue [
China has more than 17 million live births each year and in 2014 reached the target of the fifth of its Millennium Development Goals (MDG 5) by achieving a 75% reduction in the MMR (the number of maternal deaths per 100,000 live births) from 1990 to 2015 [
China has a spatially stratified heterogeneity in medical technology level, economy, culture, and geographic environment. These differences affect the occurrence of and deaths from childbirth [
The target of this paper was to investigate the spatiotemporal trends in the MMR at the local (county) level in China from 2010 to 2013 and to address, nationally as well as subnationally, the main nonmedical impact factors on the MMR. Based on the maternal mortality data covering 1,832 counties across 23 provincial regions, excluding the 11 eastern coastal provincial regions, the spatiotemporal trends in the MMR for 2,205 Chinese counties were explored through a Bayesian space–time model. Among the 2,205 counties, the data surveillance occurred in 1,832 counties, while there were no survey data for the other 373 counties. Because of a full lack of sampling, the counties located in the 11 eastern coastal provincial regions of China (blank areas in
The determinants of the MMR mainly include direct obstetric causes and indirect causes [
MMR, maternal mortality ratio; PCI, per capita income; PPWDH, proportion of pregnant women who delivered in hospitals; PPWFC, proportion of pregnant women who received 5 or more maternal check-ups.
We used county-level maternal mortality data from the National Maternal and Child Health Surveillance System (NMCHSS) over the 2010–2013 period, which covers 1,832 counties of China (
MMR, maternal mortality ratio.
To reveal the spatiotemporal pattern and influencing patterns of the MMR in China, we have applied a Bayesian space–time model [
Referring to
The overall spatial random effect
The Bayesian estimation employs WinBUGS 14 [
To further explore the patterns of influence of the MMR, especially the interaction of multiple factors, this study employed the GeoDetector model [
Generally, the spatial distribution of the mean MMR in 1,832 counties in China from 2010 to 2013 (
(A) The posterior median of the spatial relative risks (exp(
We classified the 2,205 counties into 3 categories—hotspots, coldspots, and warmspots—based on the 2-stage classification rules, as previously outlined. Among the 2,205 counties, 925 (42.0%) are classified as hotspot regions and 764 (34.6%) as coldspot regions; 516 (23.4%) counties are identified as warmspot regions. Two contiguous areas are identified as hotspots, shaded in red in
The local trend, quantified by the parameter
MMR, maternal mortality ratio.
Stronger Decreasing Local Trend | Weaker Decreasing Local Trend | Approximate to the National Decreasing Trend | Total | |
---|---|---|---|---|
552 (59.7%) | 332 (35.9%) | 41 (4.4%) | 925 (100.0%) | |
66 (8.6%) | 223 (29.2%) | 475 (62.2%) | 764 (100.0%) | |
100 (19.4%) | 77 (14.9%) | 339 (65.7%) | 516 (100.0%) |
As has been noted, the explanatory variables include PCI (unit: thousands of Chinese Yuan), PPWDH (%), and PPWFC (%). To detect whether the variables are closely correlated, we calculated Pearson’s correlation coefficients between them. The statistical test probability values of all Pearson’s correlation coefficients are all less than 0.001 (
The national and subnational regression parameters can be estimated from formulas 5 and 6.
Group | PCI | PPWDH | PPWFC |
---|---|---|---|
−0.163 (−0.967, 0.656); | −1.787 (−2.142, −1.424); | −0.623 (−0.798, −0.436); | |
−0.723 (−1.279, −0.195); | 0.195 (−0.581, 1.017); | 0.058 (−0.182, 0.276); | |
0.177 (−0.281, 0.677); | 0.223 (−0.473, 0.905); | 0.042 (−0.076, 0.152); | |
−1.111 (−3.665, 1.485); | −0.081 (−0.808, 0.660); | −1.686 (−2.090, −1.275); |
Different patterns of influence are exhibited in the 3 regions of China, which include the eastern and southern coastlands, the central and northern regions, and the western and southwestern regions. The results (
In terms of the univariate, the GeoDetector results (
Group | PCI | PPWDH | PPWFC |
---|---|---|---|
2.2% ( | 24.1% ( | 21.6% ( | |
9.2% ( | 4.2% ( | 2.7% ( | |
2.6% ( | 1.0% ( | 0.9% ( | |
2.5% ( | 26.0% ( | 29.3% ( |
Group | PCI ∩ PPWDH | PCI ∩ PPWFC | PPWDH ∩ PPWFC |
---|---|---|---|
41.7% ( | 34.4% ( | 47.2% ( | |
21.0% ( | 35.1% ( | 17.3% ( | |
10.5% ( | 18.2% ( | 8.5% ( | |
82.5% ( | 80.1% ( | 82.7% ( |
In this study, we used a Bayesian space–time model integrated with the ZIP model to explore spatiotemporal trends in the MMR of 2,205 Chinese counties from 2010 to 2013. We found that, although China has decreased the MMR in recent decades, temporal and spatial heterogeneity still exists. The different patterns of influence of the 3 main ecological determinants of the MMR at national and subnational level were identified through the utilisation of a Bayesian multivariable regression model and GeoDetector
Previous research conducted by Liang and colleagues [
A major strength of our study is its use of the state-of-the-art Bayesian space–time model, which can disassemble the overall spatial relative risk, overall temporal trend, and local trend from the complex space–time coupling process to closely investigate the spatiotemporal heterogeneities in the MMR in Chinese 2,205 counties. Another strength of this research is its estimation of the quantitative influencing effects of the 3 main determinants—PCI, PPWDH, and PPWFC—at the national and subnational levels through the use of the Bayesian multivariable regression model. The explanatory power of a single factor in the MMR and the interactive explanatory power of any two of the three factors were investigated using the GeoDetector model. A better understanding of the quantified ecological determinants of the MMR would support evidence-based policymaking to sustainably reduce the MMR in China and other developing areas worldwide.
Our study has several limitations. First, the registration system does not cover the entire country because the eastern coastal counties have very low maternal mortality, and therefore, the results do not represent the total trend of maternal mortality throughout China. Second, we only have the monitoring data for the years 2010–2013, a limited time period. Although the spatiotemporal trends may remain relatively steady, the space–time variability of recent years should be investigated. Third, the influencing factors discussed in this research could be more comprehensive; besides the 3 factors of PCI, PPWDH, and PPWFC, there are certainly other influencing factors that should form the focus of future research, such as accessibility of blood banks, road conditions, and the natural environment.
There are currently still many counties in China with an MMR higher than the SDGs target; appropriate interventions at the national and subnational level should be devised to reduce avoidable maternal mortality. An in-depth study of patterns of influence in necessary to inform effective interventionist policymaking. Our study investigated the influencing factors of maternal mortality in China through 2 approaches; the results suggest that nationally, medical intervention factors, PPWDH, and PPWFC are major influencing factors. Social and economic factors are not the principal factors at the national level but are major influencing factors in the western and southwestern regions. Interestingly, the strength of influence of PPWDH nationwide is greater than that it is in the western and southwestern regions, and the strength of influence of PPWFC nationwide is less than it is in the western and southwestern regions. This indicates that antenatal care should be strongly reinforced in the western and southwestern regions, while nationwide, the percentage of hospital births should first be improved and the number of midwives increased, at which point antenatal care should also be improved. It is also important to raise the income levels of birth families in the western and southwestern regions and eastern and southern coastlands. Public medical resources and conditions are better in the eastern and southern coastlands, and therefore, the social and economic factors have a major influence on the MMR in these areas. Additionally, this study did not find a significant influencing factor for the maternal mortality in central and northern regions, where socioeconomic development and public medical conditions are at an intermediate level compared to the national standard [
Despite the generally decrease of the MMR, it nonetheless presents a highly spatial heterogeneity. Some underdeveloped areas from the western and southwestern regions not only have a higher MMR but also show a weaker downward trend. Traditional customs, poor (health) education, and inadequate medical resources are 3 major determinants of the MMR in these areas. The issue of traditional customs is linked with the proportion of minority or indigenous population. According to the Chinese Population Census data for 2010, the 4 western provincial regions with a higher MMR—Tibet, Xinjiang, Qinghai, and Yunnan—also had a higher proportion of minority population—91.8%, 59.5%, 47.0%, and 33.4%, respectively. However, the proportion of the minority or indigenous population for 4 central and northern provincial regions that were coldspot areas for the MMR—Shanxi, Hubei, Jiangxi, and Jilin—were 0.3%, 4.3%, 0.3%, and 8.0%, respectively (
The hospital delivery and prenatal examination rates in the western provinces were far lower than the national average; the average rates of hospital delivery and receipt of 5 or more maternal check-ups in Tibet were only 62.5% and 25.7%, respectively, while for Qinghai, this was 90.4% and 59.6%. The rural areas where hospital delivery rates are less than 50% were located mainly in Tibet, Sichuan, and Qinghai, while the rates of hospital delivery and receipt of 5 or more maternal check-ups were 99.6% and 91.1% in Hebei, 99.4% and 83.4% in Shanxi, and 99.6% and 90.0% in Jilin, respectively. This indicates that there is also much room for improvement in the provision of healthcare for pregnant women in the western regions. As mentioned above, because of a lack of education, many women in the western regions do not know the importance of prenatal care and rarely go to hospitals to receive it. Income growth also plays an important role; it is apparent that people with higher incomes have an advantage in accessing medical facilities and pay more attention to nutrition and health knowledge. Therefore, it is important that the government provides financial assistance to poor pregnant women and improves (health) education.
All in all, the income level of the birth family and medical intervention are the most important influencing factors, and this is further evident when considering the effect of the interaction of different factors. The results of this study suggest that intervention should focus on low-income families. Furthermore, this study provides some quantisation of influencing effects. Certainly, on account of the differing patterns of influence at the national and subnational levels, the corresponding diverse policies should be carefully devised to reduce the national and regional MMR. For those western and southwestern regions with a higher MMR risk, on the one hand, the health awareness of pregnant women should be raised through professional health education, while on the other hand, the rates of hospital delivery and receipt of 5 or more maternal check-ups should be increased by improving the conditions of county-level and township hospitals with better accessibility. Although China has achieved the SDGs target at the national level, 191 counties, mostly located in the western regions, had not reached the SDGs target in 2015 [
A distinct gradient structure with a gradually higher MMR from the east to the west has been stable. An almost continuous zone from the northwest to the southwest experienced a strong downward trend in the MMR. This study identified 925 (hotspot) high-risk counties, mostly located in the western and southwestern regions, and 332 counties among them that are experiencing a slower downward trend than the overall national downward trend. The counties in the western provinces of Tibet, Qinghai, and Xinjiang have the highest level of MMR and a weaker downward trend than the national downward trend in these regions. Nationally, medical intervention, hospital delivery, and antenatal care are the major determinants. The major ecological determinants for the MMR in western and southwestern regions, which are developing areas, are PCI and PPWFC, while for developed areas, it is PCI. The interactive effects of different factors in China’s western and southwestern regions was the strongest, and the corresponding interactive influencing power of any two of the three factors PCI, PPWDH, and PPWFC were all greater than 80%.
(DOCX)
Click here for additional data file.
(DOCX)
Click here for additional data file.
MMR, maternal mortality ratio.
(TIF)
Click here for additional data file.
(TIF)
Click here for additional data file.
The data were collected by Sichuan University and processed by the Institute of Geographic Sciences and Natural Resources Research. The authors thank all health workers in the priority counties for providing the data and investigating the cases of death.
Annual Report System on Maternal and Child Health
conditional autoregressive
confidence interval
Strengthening the Guidelines for Accurate and Transparent Health Estimates Reporting
Millennium Development Goal
maternal mortality ratio
National Maternal and Child Health Surveillance System
per capita income
proportion of pregnant women who delivered in hospitals
proportion of pregnant women who had at least 5 check-ups
Sustainable Development Goal
zero-inflation Poisson