Association between parity and macrosomia in Shaanxi Province of Northwest China

The cross-sectional study was executed in Shaanxi province of Northwest China from August to November 2013. The infants born during 2010–2013 and their mothers were the objects of the research. Because of the different population density and fertility rates between rural and urban areas in the whole province, a hierarchical, polystage, probability-proportional-to-size sampling method was used in the present research. In China, administrative organization was divided into 3-hierarchy frames. Counties, townships and villages constitute the rural areas. Independent of rural areas, districts, streets and communities constitute the urban areas. In the first place, 20 counties and 10 districts were randomly selected from the whole province. Then, six townships and three streets were randomly sampled in the chosen counties and districts. Afterwards we selected six villages from per chosen township and six communities from per chosen street randomly. A random sampling method was used to select 30 babies born during 2010–2013 and their mothers in every chosen village, and, 60 in every sampled community. Expected sample size of our study was approximately 32,400 infants and their mothers. But 2373 subjects were unwilling to join in the study (response rate: 92.68%). Therefore 28,644 single live infants were chosen for this project. And 481 objects were removed for unknown birth weight and childbearing history. Moreover, 812 subjects were removed who had more than 3 children. In the end, a total of 27,351 singleton live infants were selected.

All data was stated by the mothers of the chosen children, including socio-demographical information and information on maternal lifestyles during pregnancy. Xi’an Jiaotong University Health Science Center devised all questionnaires. Ten field teams that every team comprised 10–12 members were formed for these counties or districts. As soon as completing every questionnaire, the supervisors were responsible for detecting any errors and/or imperfect information. All data collection was completed in the local village clinics and community health service centers. Our study was sustained by the local hospitals and health administrative departments as well as the Shaanxi province Ministry of Health.

Controlling for potential confounding factors was necessary when determining the relationship parity and macrosomia. Based on the currently available body of knowledge and the nature of our data, we selected potential confounding factors from three groups of variables: children, family and mothers. The factors included within the children group were the child’s sex and gestational age. The factors included within the family group were economic conditions, region and residence. The factors included within the mother group were maternal education level, age at the child’s birth, occupation and gestational diabetes.

Primipara: A woman who has borne only one living child.

Multipara: A woman who has given birth to 2 living children.

A database was designed by EpiData version 3.02, and data entry was duplicated. Firstly, the characteristics of participants were summarized using means±SDs for normally distributed continuous variable. The categorical variables were described using count and proportions. The χ² test was used to prove differences in proportions between groups. On account of the multilevel hierarchical structure of the data, the generalized linear mixed model approach was used, which is a good method for analyzing data with a hierarchical structure and can be applied in sampling investigations. Ultimately, a 2-level analysis was performed to adjust for the effect of randomization by counties/districts and to analyze the associations between parity and macrosomia with county/districts to level 2 and individual to level 1 by nine potential confounding factors. Model 1 adjusted for gestational age and sex of infants. Model 2 adjusted for the variables in model 1 plus the relevant maternal characteristics, including maternal education level, age at the child’s birth, occupation and gestational diabetes. Model 3 adjusted for the variables in model 2 and the status of family characteristics (including economic conditions, region and residence). All statistical analyses were performed using SAS 9.3 (SAS Institute Inc., Cary, NC). Two-tailed P < 0.05 was considered statistically significant.

Of the infants, boys accounted for 53.92% of total infants. Amongst the region of the infants, infants in Central Shaanxi, Northern Shaanxi and Southern Shaanxi accounted for 54.41, 25.06 and 20.53% respectively. The childbearing age of mothers was 26.92 ± 4.65 years and approximately 39.45% of them were educated beyond senior high school. The other details of the sample and distribution of the major demographic variables are shown in Table 1.

The average birth weight of 27,351 newborns was 3267.21 ± 455.90 g. The neonatal average birth weight of the primipara was 3262.18 ± 452.36 g, and that of the multipara was 3275.07 ± 461.29 g. Neonatal birth weight of the multipara was higher than that of the primipara, and the difference was statistically significant. The incidence of low birth weight infants was 3.38%, and the incidence rate of macrosomia was 6.79%. Among the 27,351 women of childbearing age surveyed, 16,677(60.97%) women were primiparas, and 10,675(39.03%) were multiparas. The birth weight status of the children for primiparas and multiparas were shown in Table 2. The incidence of macrosomia in multiparas was higher than that in primiparas, and the difference was statistically significant (P < 0.001).

Univariate analysis results showed that maternal age < 25 years, peasant/housework, living in rural areas and female infants were the protective factors of macrosomia. And the possibility of having a macrosomic infant also increased with gestational age, maternal education level, household wealth index, living in Central Shaanxi and gestational diabetes. Compared with primipara, multipara was also associated with an increased risk of delivering a macrosomic infant (Table 3).

In Fig. 1, the GLMM results obtained for macrosomia as an outcome variable represent the association between parity and macrosomia. After adjusting for statistically significant factors in univariate analysis from model 1 to model 3, the odds ratio (OR) showed that the risk of macrosomia in multipara is higher than in primipara. Furthermore, the 3 odds ratios were similar, which indicated that the models were stable. Specifically, a statistically significant association between parity and macrosomia was found. The risk of being born macrosomia was 1.26 times higher for a multipara compared to that for a primipara.

We found that parity was related to the occurrence of macrosomia. The incidence of macrosomia of multipara was higher than primipara, and the difference was statistically significant. After adjusting for statistically significant factors in univariate analysis, analysis based on generalized linear mixed models revealed that the risk of macrosomia was 1.26 times higher for a secondborn child compared to a first born.

The association between parity and macrosomia has been previously investigated in a few studies conducted elsewhere, and increased parity is associated with higher risk of fetal macrosomia [23, 24]. Multiparity is one of the most important risk factors for macrosomia, according to the American College of Obstetricians and Gynecologists’ Committee on Practice Bulletins—Obstetrics [26]. The rate of fetal macrosomia in multiparous women has been shown to be 2–3 times higher than that in control group in the majority of studies [27]. Faith Agbozo. etal noted that the parity of two to three children was related to raised risk for macrosomia [28]. The aforementioned studies illuminated the relationship between parity and macrosomia and provided some references and evidences for our research.

The present cross-sectional study indicated parity of two children was associated with increased risk for fetal macrosomia. Compared to primiparas, multiparas should far strengthen the pre-pregnancy education and the guidance during pregnancy to control pre-pregnancy body mass index and pregnancy weight, and keep the appropriate exercise and balanced diet in order to reduce the incidence of macrosomia.

The primary strength of the present analysis is the large sample size (27,351 single live births occurring from 2010 to 2013), which accounted for ~ 9% of neonates in Shaanxi Province [29]. Therefore, our results can be generalized to the entire province as well as Northwest China. Another strength of this study is that the birth weight data collected from birth certificates was accurate to the nearest 10 g. Moreover, the generalized linear mixed models adjusted for relevant covariates were generated to further elucidate the association between parity and macrosomia. Limitations of our data should also be noted. Some major confounders, including pre-pregnancy BMI, diet, weight gain during pregnancy and so on, were not adjusted for because we lacked these data [30, 31]. Nevertheless, the current study is the first and largest survey that has presently been conducted in Northwest China, and provides the best information on the relationship between parity and macrosomia in this geographical region.