Introduction

Birth weight is a convenient factor indicating adverse outcomes of infants. Low birthweight infant has a higher perinatal mortality and it also increases health risk in adulthood1. The definition of low birthweight(LBW) infant was recommended as infant ≤2500 g regardless of gestational age by The World Health Organization(WHO) in 19502. The incidence of LBW was 6.1% in mainland China reported by Chen Y et al.3 and 8.1% in Foshan of China by Rao J et al.4. The risk factors of LBW become important in lowering the incidence of LBW. With the support of National Health and Family Planning Commission of China(NPCP), our study covered 60% pilot counties with over 200,000 population in total. Although several studies concluded some common maternal risk factors associated with LBW, few studies included paternal factors. Some experts started to suggest the consideration of paternal information in studies aimed at the risk factors model of LBW. We are the first large study to analyze both maternal and paternal risk factors of LBW in China.

Results

Prevalence of term low birthweight infants

202,725 cases were analyzed in our study, the prevalence of term low birthweight infants(TLBW) was 4.8% (n = 9687) in rural China with the mean birthweight was 2140.54 ± 402.53 g. Among term low birthweight babies, 647 of them were very low birthweight infants. Term infants between 50th–97th (TNBW 50th–97th) accounted for 70.8%(n = 143,630) and it had the highest mean birthweight (3454.34 ± 297.76 g). The remaining term babies were 31432 with the mean birthweight was 2855 ± 147.66 g. Figure 1. showed the distribution and compared the mean birthweight of these 3 groups.

Figure 1
figure 1

The distribution of term babies among different birthweight groups and comparison of mean birthweight of 3 groups. TNBW = Term normal birthweight; TLBW = Term low birth weight; Tmacrosomia = Term macrosomia; Others included birthweight of infants between 2500 g to 50th.

Full size image

Univariate analysis

In this study, maternal weight, height, BMI, age, hypertension, history of stillbirth, contraception, irregular folic acid intake, white blood cell count(WBC), red blood cell count, positive hepatitis B surface antigen, gestational age and paternal weight, height, BMI, age, history of varicocele and female infant were found significant in univariate analysis. Table 1. compared TLBW and TNBW (50th–97th) group, maternal hypertension and smoking in our study had a negative effect on birthweight of infants. However, it was no statistically significant. Maternal positive hepatitis B surface antigen was associated with decreased birthweight in our study (RR = 1.979, P = 0.046). Birthweight in regular folic acid group was higher than irregular group (RR = 1.152, P = 0.003). Additionally, Paternal history of varicocele would significantly increase the risk of LBW (RR = 2.404, P = 0.003). Female babies had a little higher risk of low birthweight (RR = 1.072, P = 0.046) (Fig. 2).

Table 1 The distribution of LBW and NBW(50th–97th) in univariate analysis.
Full size table
Figure 2
figure 2

Independent risk factors found in multiple logistic regression analysis. RR = Relative risk. Cl = Confidence interval. The vertical bar represented RR = 1 and horizontal lines represented the range of adjusted RR of each risk factors. The reference group were: No history of varicocele, having regular folic acid intake, negative hepatitis B surface antigen, maternal non-smoking, no history of stillbirth, baby sex = male; RR were adjusted by maternal age and educational level.

Full size image

Discussion

LBW is associated with increased risk of diabetes mellitus, hypertension and cognitive dysfunction later during life5. But it is still not rare in developing countries with the overall prevalence was 15.9%6. To lower the incidence of LBW, it is very important to identify risk factors. Research have found maternal age, education level, BMI, ethnicity, socio-economic level, medical disease before and during pregnancy, health care and traffic air and noise pollution were related to LBW6,7,8. We found several important maternal and paternal risk factors which may involve in the process of LBW by a large, population-based database. Several studies have reported maternal hypertension and smoking were associated with LBW. Our current study showed the prevalence of LBW in maternal hypertension and smoking group were higher in those without hypertension and smoking before pregnancy but with no statistical significance in multiple risk analysis. Nevertheless, the number of rural smoked mothers were too small and many of them exposed to second-hand smoke which may mix the results. Interestingly, maternal exposure of hepatitis B surface antigen was found to be related to TLBW. Recent research also showed positive HBsAg may lead to a high risk of malformation and cesarean delivery but lower risk of macrosomia and non-significant higher risk of LBW in China9. The mechanism may be related to reduced blood supply of uteroplacenta due to the infection. This conclusion addressed the importance of management of maternal HBsAg carrier status. Several studies had demonstrated that folic acid supplementation was associated with decreased lowbirthweight infants. We found LBW easily happened in irregular folic acid intake group10. Maternal anemia is known to be related to folic acid supplementation, which will cause fetal hypoxia and intrauterine fetal distress11. Maternal anemia during pregnancy is an independent risk factor for low birthweight and preterm delivery. We are the first study found the relationship between low birthweight and paternal history of varicocele. Christman MS et al. tested the lower seman density and count in patients with a history of varicocele and cryptorchidism. We guess it is the main reason lead to LBW by zygotes with poor quality12. Further studies need to be done to address the mechanism. Our study also had limitations. Some of our data source were derived from questionnaire which lacked quantitative indices. Further, this study was conducted in rural area of China, so it was difficult to obtain the data covered whole China.

Conclusion

Our study demonstrated that both maternal and paternal factors could affect birth weight. Maternal smoking and hypertension would lead to a non-significant LBW. Maternal positive hepatitis B surface antigen and irregular folic acid intake and paternal history of varicocele would increase the risk of LBW. In conclusion, more effort should be made to improve pre-pregnancy checkups to prevent LBW and long-term outcomes.

Methods

Description of our database system of NPCP

Our study was a population-based, retrospective cohort study including the information from database of National Pre-pregnancy Checkups Project (NPCP) organized by The Population and Family Planning Commission and The Ministry of Finance of China between January, 2010 and December, 2013. It covered rural areas of 30 provinces of China: Beijing, Hebei, Tianjin, Shandong, Zhejiang, Jiangsu, Anhui, Jilin, Fujian, Jiangxi, Henan, Hubei, Guangdong, Sichuan, Chongqing, Yunnan, Shanxi, Shanghai, Guangxi, Hunan, Hainan, Heilongjiang, Qinghai, Tibet, Inner Mongolia, Shanxi, Liaoning, Ningxia, Guizhou and Gansu. This project was built for couples planning on getting pregnant in rural China. Each couple will receive general information and medical history inquiry, physical examination, clinical laboratory examination and imaging examination. Included couples will be followed up on 12th weeks of gestation and 6 weeks after delivery. Last menstrual period was used to calculate the gestational age. Infants at 37–42 weeks were defined as term newborn. The group of NPCP has already worked on the risk factos of adverse pregnancy outcomes13,14,15. Our study included 248,501 records, among which ectopic pregnancy occurred in 176 cases and spontaneous abortion and self-induced abortion occurred in 5714 and 2526 pregnant women, respectively. 786 of records were stillbirths. 230,190 live births at 37–42 weeks were included, twins, triplets and those infants with missed birth information were first excluded. Full-term infants whose birthweight ≥97th percentile according to Chinese birthweight percentile norm for newborn(Boy ≥4410 g; Girl ≥4212 g) were then excluded16. Finally, 202,725 single live births at 37–42 weeks were included in our study (Fig. 3).

Figure 3
figure 3

Study profile.

Full size image

Data processing

The medical staffs from NPCP were trained by professional experts before collecting data and the database was built by application developers. To ensure the accuracy of data, questionnaires were completed by two trained interviewers at the same time. Our database had items including social demographics, illness and medical history, living habits, psychological status, environmental exposure, maternal pregnancy complications and biological parameters of parents (Table 2).

Table 2 Items included in our study.
Full size table

Definition of variables and variable grouping

All included infants were divided into three groups: (1). Term low birthweight (TLBW): birthweight ≤2500 g; (2). Term normal birthweight from 50th–97th (TNBW 50th–97th): Infants between 50th–97th percentile with boy’s birthweight between 3073 g to 4410 g and girl’s birthweight between 2964 g to 4212 g. (3) The remaining infants. We included the following variable in our analysis including categorical variables and continuous variables. Table 3 showed the assignment of numeric variables.

Table 3 The assignment of numeric variables.
Full size table

Ethical Approval

The informed consent was obtained from all subjects and all methods were carried out in accordance with relevant guidelines and regulations. All experimental protocols were approved by NPCP.

Data Analysis

Continuous data were performed as the mean and standard deviation (SD) and categorical data were performed as frequency and percentage. The chi-squared test was used for qualitative data and independent sample t-test were used for quantitative data analysis. Variables with P value <0.1 was considered as significant in screening single factors. To address confounding variables, stratified analysis on all suspected confounders was used to estimate the specific risk for each stratum, stepwise multiple logistic regression was performed to correct the confounders in multi-factor analysis if the ajusted effect differs between each subgroup17. P < 0.05 was considered statistically significant. SPSS 24.0 edition was used for data analysis.