Risk factors of stillbirth in rural China: A national cohort study

People living in rural China are more frequently exposed to some specific risk factors which made stillbirth rate higher than urban areas. National Free Preconception Health Examination Project was launched to investigate these risk factors and collected a representative sample of 248501 participants from 31 provinces in China from 2010 to 2013. Parental risk factors were ascertained twice before and during pregnancy respectively by questionnaires. Stillbirth or live birth were recorded by trained physicians. In the analysis, nested case-control study was conducted, and propensity score matching method was used to adjust the confounding. Multi-level logistic regression was used to fit for multi-level sampling. The overall stillbirth rate was 0.35% in rural China, it was higher in North (0.42%) and West (0.64%) areas. Maternal exposure to pesticide (OR (95%CI 1.06, 3.39)), hypertension (OR = 1.58 (95%CI 1.07, 2.34)), lack of appetite for vegetables (OR = 1.99 (95%CI 1.00, 3.93)), stress (compared with no pressure, OR of a little pressure was 1.34(95% CI 1.02, 1.76)); paternal exposure to smoking (OR = 1.22 (95% CI 1.02, 1.46)), organic solvents (OR = 1.64 (95% CI 1.01, 2.69)) were found independent risk factors of stillbirth. Folacin intake 3 months before pregnancy (OR = 0.72 (95%) CI 0.59, 0.89), folacin intake 1-2 months before pregnancy (OR = 0.71 (95% CI 0.55, 0.92)), folacin intake after pregnancy (OR = 0.81 (95% CI 0.65, 1.02) for) were protect factors of stillbirth. Maternal pesticide exposure, lack of vegetables, stress, paternal smoking and exposure to organic solvents were risk factors of stillbirth. Folic acid intake was protective factor of stillbirth, no matter when the intake began.

Definition and Assessment on variables. Stillbirth was defined as fetal death on or after 20 weeks of gestation in this study. Maternal BMI in adults was categorized into four groups: <18. 5, 18.5-23.9; between 24.0-27.9, ≥28.0 kg/m 21 . Advanced maternal age was defined as first pregnancy on or after the age of 35. Past medical history of hypertension or systolic blood pressure >140 or diastolic blood pressure >90 were considered as high blood pressure. Parental education levels were classified into illiteracy, primary school, junior middle school, senior middle school, undergraduate, postgraduate and above according to the current education system. Parental occupation was classified into farmer, physical worker, service industry laborer, businessman, house worker, office clerk, and others.
We divided the participants geographically according to the Qinling-Huaihe line, and Heihe-Tengchong line. Qinling-Huaihe line bisects China into north and south regions, culture, climate, living habits are all different between north and south, which affect people's health situation 22 . Heihe-Tengchong line divides China from Heihe in Heilongjiang province to Tengchong in Yunnan province into east and west. The east is more economically developed with higher civilization degree than west, and it takes up 96% of total Chinese population even though it only covers 36% areas of China. The economic imbalance between east and west resulted the huge differences in health resources 23 .
Environmental exposure as pesticide and new decoration were classified into two groups, no and yes, social pressure was classified to five levels, with 1 the lowest and 5 the highest. Folacin intake duration was classified into more than 3 months before pregnancy, 1-2 months before pregnancy, after pregnancy, and no. And whether they take folacin regularly was also asked.

Statistical analysis.
To control multiple confounding between the cases and controls and address severe imbalance between numbers of case and control which may cause high misclassification of interested outcome by logistic regression 24 , nested case-control study was conducted, and propensity score matching method was used to find 8 controls for each case. Logistic regression was used to calculate the propensity score, confounding factors as area 22,23 , breeding history 25,26 , education level 27,28 , occupation 29 , age group 30 were adjusted in this procedure. Parental age, education level, occupation were closed related and can't be put into the logistic regression together, we select maternal age, education level 27,28 , and paternal occupation as the dependent variables, as they were commonly used in the scoring of family social economic status, and were reported to be more closed related with the family utilization of health resources and family health status 31 . Wald χ 2 was used for the model's overall significance test, Hosmer-Lemeshow was used for the goodness test. After propensity score matching, cases and selected controls were used for further risk factors analysis.
T-test and McNemar test were conducted respectively for continuous data and categorical data in the univariate analysis. Due to differences in climate, culture and economics between different provinces, the subjects may share some homogeneity within a province and heterogeneity between provinces, individually independence required by traditional logistic regression was not matched. As a result, multi-level logistic regression was used to compare the exposures between case and control group. Province was set as the first level and individuals as the second level. Due to collinearity between some exposures between parents, models were built separately for maternal exposures and paternal exposures. Risk factors which were statistically significant in univariate analysis or were proven to be confounders in previous studies were included in the models. GLM procedure in SAS was used to build the model, parameters as Tolerance (TOL) and Variance Inflation Factor (VIF) were used to detect the collinearity between variables 32 . Intra-Class Correlation Coefficient (ICC) was used to test the independence and heterogeneity within level.
All data were expressed as mean ± SD or count (percentages), as appropriate. All statistical procedures were analyzed by SAS 9.4.

Results
The overall stillbirth rate was 0.35% in rural China, it was higher in north (0.42%) and west (0.64%) than in South (0.32%) and east (0.34%).
Before the propensity score matching, parental education level, occupation, age group, height, weight, were all significantly different between stillbirth group and live birth group (Table 1). Maternal illiteracy group had a stillbirth rate of 0.60%, while mothers with education level equal or higher than postgraduate had a stillbirth rate of 0.64%, both significantly higher than the average rate. Parental advanced age group also had higher stillbirth rate, 0.54% for paternal advanced age and 0.77% for maternal advanced age. Beside, stillbirth group had average lower parental height. By propensity score matching, 8 controls were found for each case. Breeding history, living region, maternal education level, maternal age, paternal occupation were taken into consideration in the matching process. The logistic regression used to calculate the propensity score was statistically significant, withχ 2 Wald = 127.04, P < 0.0001, and the Hosmer-Lemeshow test showed that the goodness of fit of the model is good, withχ 2 HL = 3.56, P = 0.8946. After the propensity score matching, basic characteristics were all equally distributed between the two groups ( Table 2). Following analysis were conducted with the two groups of 811 cases and 6488 controls.
Parental exposed to working or life stress was associated with stillbirth, but the relationship was not linear. Economic stress and got ready for pregnancy were not statistically significant related with stillbirth (Table 3).
We found that folacin intake was a protective factor of stillbirth in our study (P = 0.0016), the OR was similar for those who take folacin at least 3 months before pregnancy (OR = 0.70 (95%CI 0.58, 0.85)) or 1-2 months before pregnancy (OR = 0.71 (95%CI 0.56, 0.90)*), and slightly higher in those who take folacin after getting pregnant (OR = 0.78 (95%CI 0.63, 0.96)). And as long as the mothers take folacin, it's not significantly different between those who take regularly and who not (P = 0.2213) ( Table 3).
To illustrate the relationship between vegetables consumption and stillbirth, we stratified the data by folacin intake. For both who take folacin and who don't, lack of vegetables seemed to be a risk factor, with OR of 3.60(95% CI 1.04, 12.38) and 2.04(95% CI 0.94, 4.43) respectively (Table 4).
Multi-level logistic model of risk factors of maternal and paternal were analyzed independently. TOL of variables in both models were all greater than 0.1, and VIF were all less than 10, so that we assume no collinearity among variables in the two models respectively. Overall tests of the two models were statistically significant, with both P < 0.001. Tests of Random parameter was also significant, with P = 0.03045 < 0.05, which mean that multi-level models should be used.

Discussion
We found that stillbirth rate was much higher in the west, almost twice of the average rate. Due to low economic development level and sparse population, the maternal and child healthcare services provided in western areas were not as good as east. Some researches showed that the management rate, physical examination rate at early pregnancy, the visiting rate after delivery were all lower in western areas 33 . How to implement effective maternal and child health management for western areas are of vital important now. Maternal exposure to pesticide was identified as an independent risk factor of stillbirth, which is consistent with previous studies [34][35][36] . Pesticide exposure are more common with farmers with lower education level. Due to limited health literacy, they know little about the harm of pesticide and protection methods. To protect people from the harm of pesticide, regular health education on how to apply pesticide safely should be provided, and specialized teams should be organized to help with the pesticide spraying 37,38 .
Maternal high blood pressure before pregnancy was also associated with stillbirth. According to previous studies, 4-7% of stillbirth happened due to high blood pressure during pregnancy, and high blood pressure before pregnancy which was not ideally control was an important cause of high blood pressure during pregnancy [39][40][41][42] . Reinforce the management of maternal blood pressure, and provide adequate treatment for those with high blood pressure before pregnancy are very important to reduce the stillbirth rate.
Folacin intake was proved to be protective factor of stillbirth, regardless of when the woman started the intake. And for those who rarely or don't eat vegetables, the protective effect was larger. Previous studies showed that the rate of folacin intake among women in rural China is rising these years, which is a good phenomenon. It is important for basic public health services departments to increase the rate of folacin intake in a standardized way 18 .  Stress in daily life or work is also a risk factor of stillbirth, which may be a bigger problem for women in business. But it is worthwhile to notice that the stillbirth rate showed no statistically significance between those who got ready for this baby and those who not.
For paternal risk factors, paternal smoking, paternal drinking, and paternal exposed to new house decoration were all associated with stillbirth after adjustment of paternal age. In rural China, male smoking and drinking are still severe public health problems that need to be addressed 43,44 . New decorated house may be associated with higher dosage level of formaldehyde or organic solvent, which were reported to be risk factors of stillbirth 45 .
Limitations of our studies include that many of risk factors collected by our study were binary data, which made the analysis of dose-response relationship impossible, the exposing status such as smoking, drinking or pesticide were self-reported, which may not be very accurate. Also, the participants were not randomly selected, which may also cause some bias of our study.
Ethical approval. The study was approved by the institutional research review board at the National Health and Family Planning Commission and National Research Institute for Family Planning. Informed consents were obtained from all participants or their legal representatives. All research was performed in accordance with relevant guidelines.

Data Availability
The data generated cannot be made publicly available according to the Chinese law of personal data protection and also our project data management rules. However, data inquires or further suggestions for analyses can be made to the corresponding author.   Table 5. Multi-level logistic regression of parental risk factors of stillbirth. *For P < 0.05.