Study Design and Baseline Profiles of Participants in the Tianjin Birth Cohort (TJBC) in China

Background To investigate the causal link between early-life exposures and long-term health consequences, we established the Tianjin Birth Cohort (TJBC), a large-scale prospective cohort in northern China. Methods TJBC aims to enroll 10,000 families with follow-ups from pregnancy until children’s six year-old. Pregnant women and their spouses were recruited through a three-tier antenatal healthcare system at early pregnancy, with follow-ups at mid-pregnancy, late pregnancy, delivery, 42 days after delivery, 6 months after delivery, and each year until 6 years old. Antenatal/neonatal examination, biological samples and questionnaires were collected. Results From August 2017 to January 2019, a total of 3,924 pregnant women have already been enrolled, and 1,697 women have given birth. We observed the prevalence of gestational diabetes mellitus as 18.1%, anemia as 20.4%, and thyroid hypofunction as 2.0%. In singleton live births, 5.6% were preterm birth (PTB), 3.7% were low birth weight, and 7.3% were macrosomia. Based on current data, we also identified maternal/paternal factors which increased the risk of PTB, including paternal age (OR 1.07; 95% CI, 1.01–1.14 for each year increase), vaginal bleeding during pregnancy (OR 2.82; 95% CI, 1.54–5.17) and maternal early-pregnancy BMI (OR 1.08; 95% CI, 1.01–1.15 for each kg/m2 increase). Conclusion TJBC has the strength of collecting comprehensive maternal, paternal, and childhood information. With a diverse range of biological samples, we are also engaging with emerging new technologies for multi-omics research. The study would provide new insight into the causal link between macro/micro-environmental exposures of early life and short/long-term health consequences.


INTRODUCTION
In the past decades, China experienced great improvement in health conditions as a result of enormous social and economic development. The mortality rate under 5 years old and the maternal mortality rate reduced with an average annual rate of 8.2% 1 and 8.9% from 1996 to 2015. 2 However, rapid economic growth was also accompanied with new health challenges, including exposing to changing factors such as lifestyle patterns (especially dietary behaviors), 3 pollution, late marriage and childbearing, smoking and drinking, and birth policies. Correspondingly, those exposures might lead to changes in omics biomarkers, such as metabolic molecules, gut microbes, cell-free RNA, and cell-free DNA. Eventually, there has been a great transition of disease spectrums, including the surge of chronic non-communicable diseases, 4 infertility, 5 and increased growth and development problems in children. 6 According to the theory of the Developmental Origins of Health and Disease (DOHaD), maternal=paternal exposures during pregnancy can have important roles in the change of the short-term and long-term health consequences of the offspring. 7,8 It can be foreseen that the drastic social and economic development in China, and the resulting changes in people's behaviors and psychology, will have a profound impact on the health of mother and child in China in the next several decades. To better understand the social and economic impacts on maternal and child health, and identify key exposures affecting children long-term development, we intend to establish a large prospective birth cohort in Tianjin, a northern metropolitan city of China, that continuously collects longitudinal epidemiological data, biological samples and phenotype information from early pregnancy to children at 6 years old.
We established the Tianjin Birth Cohort (TJBC) based on the following reasons: (1) Tianjin has been exposed to complex environmental and social factors that may have significant impacts on maternal and child health. For instance, Tianjin is one of the cities most affected by air pollution as a result of rapid industrialization and urbanization. It is also a typical city in China facing the problems of population aging and advanced maternal age; (2) Tianjin has developed a three-tier antenatal healthcare system, which is effective in organizing population-based cohort projects; (3) Although several birth cohorts have been set up in Guangzhou and several other cities [9][10][11]20 in China, there is a lack of a large cohort in the northern part of China. Tianjin is 120 kilometers away from Beijing, with a relatively developed economy 12 and large population, which provides a good foundation for large-scale population studies in northern China; (4) Tianjin has regional health problems, such as high prevalence of gestational diabetes mellitus (GDM; 9.3% in 2012) 13 and childhood obesity (16.3% in 2014) comparing with other Chinese cities. 14 There is an imminent demand to determine public health policies to cope with future health challenges. Therefore, Tianjin is a unique and important place to investigate the health impacts of early-life exposures on fetal development, childhood growth, and long-term disease onset.
Latest research showed that multi-omics technology was of great significance in disease prediction and precision medicine. A longitudinal big data study conducted at Stanford University School of Medicine showed that multi-omics research greatly improved the predictive performance of diabetes, cardiovascular disease, tumors, and other diseases compared with clinical data alone. 22 Researchers at Baylor College of Medicine also found that multi-omics technology may have the potential to extend healthy life among active adults through improved prevention and early detection of age-related chronic diseases. 23 At present, most studies on pregnant complications and childhood diseases were focused on single-level omics, 24,25 but for those complex diseases each technology performed only limited analyses for the etiology of the disease. Therefore, in the current study, we hypothesized that longitudinal multi-omics profiling combined with early exposures may improve disease risk prediction of pregnant complications and childhood diseases such as childhood obesity, hypertension, autism, and allergy.
The research interests of the TJBC will focus on identifying the biomarkers of severe pregnant complications (such as GDM, preeclampsia, pre-term birth, and stillbirth) and birth defects using multi-omics technology (eg, genome, transcriptome, metabolome, and microbiome), following the association study between disease biomarkers and clinical data (including the height, weight, clinical test results, ultrasound measurements, pregnancy complications, and medication). More importantly, we will use the TJBC to explore the roles of early exposures during pregnancy in determining children developmental disorders (eg, mental retardation, autism, and allergy) and children's early-onset chronic diseases (eg, childhood obesity, hypertension, diabetes, and dyslipidemia).

Participants and follow-up
The TJBC was a population-based birth cohort, aiming to recruit 10,000 families with follow-ups from early pregnancy until children are 6 years old. The study recruited women and their spouses at early pregnancy from six central urban districts (Heping, Nankai, Hexi, Hedong, Hebei, and Hongqiao) and three suburban districts (Beichen, Dongli, and Jinnan) of Tianjin, covering annual birth of 55,670 infants ( Figure 1). Pregnant women and their spouses were recruited through an integrated three-tier antenatal healthcare system, consisting of (1) Community-level primary healthcare centers; (2) District-level Women and Children's Health Centres (WCHC) and other secondary obstetric hospitals; and (3) A city-level Tianjin WCHC (TWCHC) and other tertiary hospitals, which has been described in detail previously. 21 First, all pregnant women who set up pregnancy records in the nine districts were informed about the cohort verbally or through propaganda posters; then, eligible individuals (together with their spouses) who were willing to participate would be referred to district-level WCHCs for recruitment and baseline investigation. And TWCHC administrates and provides training, guidelines, and cohort operation protocols to WCHCs, secondary=tertiary obstetric hospitals, and community healthcare centers. The study was approved by the Ethic Committee of Tianjin Women and Children's Health Center (No. 201706012-1), and was performed in accordance with the ethical standards as laid down in the Declaration of Helsinki and its later amendments or comparable ethical standards. All individuals were recruited with their voluntary participation. Before enrollment, each participant was given detailed explanation about the study by a professional clinician, and written informed consent was obtained from each participant.
Participant enrollment commenced in August 2017 and planned to complete the recruitment of 10,000 families within 4 years. The recruitment criteria included: (1) pregnant women older than 18 years; (2) at ≤14 +6 gestational week (pregnant women who registered at primary healthcare centers at ≤12 +6 gestational week were informed about the cohort, and women who were willing to participate the study were required to come to the WCHCs for formal recruitment and baseline investigation within 2 weeks); (3) women and their spouses need to be registered as Chinese citizens, and have permanent residences in Tianjin; (4) intend to complete antenatal care and give birth in Tianjin, and intend to stay in Tianjin at least until the children are 3 years old; and (5) agree to participate the TJBC study and fulfill the requirement of follow-up medical assessment and biological sample collection.
Follow-up strategies are illustrated in Figure 2. Baseline investigation was conducted at <15 gestational weeks, with biological samples and questionnaires collected from the pregnant women and their spouses, and antenatal examination records collected from each woman. Follow-up investigation was conducted at the mid-and late-pregnancy, delivery, 42 days after birth, 6 months after birth, and each year until 6 years old. Women's biological samples, questionnaires and antenatal examination records were collected at 15-27 gestational weeks and at 28-41 gestational weeks, respectively, with intervals greater than 4 weeks. At delivery, neonatal samples and healthcare records were collected. At each stage after birth, children samples and physical examination results were collected, and questionnaires were obtained from their parents.
The flow chart of cohort participant recruitment and follow-up is illustrated in Figure 3 (as of January 27, 2019). Among 3,924 pregnant women enrolled in the TJBC, a total of 264 women (6.7%) were lost to follow-up. From August 2017 to December 2017, 9 women were lost to follow-up, 211 women were lost to follow-up in 2018, and 44 women were lost to follow-up in January 2019; the cumulative rate of lost to follow-up was 2.2%, 6.1%, and 6.7% in 2017, 2018, and 2019, respectively. As shown in Table 1, comparing with those who remained in the study, women who opted out were more likely to have smaller prepregnancy body mass index (BMI), higher level of education, and higher proportion of being technicians and living in urban areas.

Sample size and power calculations
The overall TJBC sample size target is 10,000 families, which provides adequate power to detect moderately strong causal effects of common environmental, social and biological exposures. According to previous reports in the Chinese population, we assumed the incidence rates of GDM was 9.3%, 13 preterm birth was 7.2%, 26 birth defects was 5.6%, 27 childhood obesity was 16.3%, 14 and childhood hypertension was 9.0%. 28 With those incidence rates, we estimated the smallest detectable odds ratios based on 80% power at a testing level of 0.05 in the TJBC cohort (as shown in eTable 1).

Data collection and measurement
Semi-structured questionnaires were designed to collect sociodemographics, environmental exposures, lifestyle behaviors, diet, psychological evaluation, and childhood development, once or repetitively ( Table 2). All questionnaires were completed with face-to-face guidance by designated clinicians to ensure the integrity and credibility of answers. Marital Satisfaction Scale, part of the ENRICH marital inventory, was used to judge the happiness of marriage. 15 The individual's perceived social support within the family and outside the family was evaluated using the Chinese versions of the Perceived Social Support Scale (PSSS). 16 Mental health (depression and anxiety) during pregnancy was assessed using a 20-item self-rating depression scale (SDS) and a 20-item self-rating anxiety scale (SAS). 17,18 Clinical information of antenatal and postnatal healthcare was collected from the hospital healthcare system, including antenatal screening test results, ultrasound findings, obstetric complications, clinical treatments and medications, and anthropometry results at each follow-up phase (Table 3).
Newborn birth data was obtained from the delivery records. In early childhood, we collected clinical assessment, physical measurements (eg, weight, body length=height), clinical examination using routine state health checks at 6 weeks and 6, 12, and 36 months of age, all of which were conducted by experienced medical staff. Further clinical information including clinical test results, intellectual development assessment, medical history, and disease diagnosis were collected through record linkage to children's medical information.
A variety of biological samples, including blood, urine, feces, and breast milk, were collected at each follow-up phase from the families participating in this cohort by clinicians (Table 4). At delivery, placental tissues, umbilical cord blood, and umbilical cord were collected from the two selected hospitals with good conditions of sample collection and quality control. Pregnant women were closely followed in case of adverse pregnancy outcomes such as miscarriage, stillbirth, and birth defects. For such cases, products of conception (POC) were collected when possible. All samples were collected with individual tubes, each sample tube was supplied with a unique barcode associating with the sample owner's personal identity. After collection, whole blood was separated as plasma and buffy coat within 8 hours.
Umbilical cord blood was aliquoted in 24 hours. Breast milk samples were aliquoted within 8 hours. All samples were temporarily stored at −80°C in Tianjin, and periodically transported to the CNGB, which is certified to ISO 9001, ISO 14001, ISO=OHSAS 18001, and ISO 27001, for long-term storage at −80°C. Beijing Genomics Institute at Shenzhen (BGI-Shenzhen) was responsible for providing technical assistance, IT support, and genomics research facilities for the TJBC study. Samples collection progress is shown in eTable 2.

Data analysis
Descriptive statistics were used to summarize baseline characteristics. Mann-Whitney U-test was conducted as univariate analysis to identify continuous variables influencing follow-up. Categorical variables were presented as numeric and were compared using chi-square test. Confidence intervals (CIs) were presented as 95% CIs and P-values less than 0.05 were deemed statistically significant. Statistical analysis was performed using SPSS 24.0 (IBM, Chicago, IL, USA).

RESULTS
As of January 2019, a total of 3,924 pregnant women were enrolled. As compared with the target population of 34,924 pregnant women (summarized by the number of women who set up pregnancy records in each district, from the specific date when enrollment started, to January 2019), the participation rate was  Table 1. Maternal age was 30.3 (standard deviation [SD], 3.9) years, and earlypregnancy BMI was 23.1 (SD, 3.9) kg=m 2 . Around 30% of women work as technicians, and 28.7% as office workers. In terms of education level, 58.6% of women had undergraduate or higher degrees, and 26.9% had vocational or high school education. A total of 72.9% participants was living in urban areas. Very few women smoked (4.1% of pre-pregnancy vs 0.9% of current) or drank (2.4%) during pregnancy. Nearly half of women passively smoked (36.3%), drank tea (53.4%), and drank coffee (47.1%) before pregnancy, but significantly decreased during pregnancy. Among 3,924 pregnant women, there were 119 (3.0%) miscarriages. Among 1,665 women who had live birth (Table 5), the prevalence of gestational diabetes mellitus (GDM) was 18.1%, anemia was 20.4%, thyroid hypofunction was 2.0%, uterine fibroids was 5.9%, hypertension was 1.9%, ovarian cyst was 2.9%, scarred uterus was 1.7%, vulvovaginal candidiasis was 1.1%, urinary system infection was 0.9%, hepatitis B virus was 0.8%, and maternal-child blood incompatibility as 0.2%. Among the 1,645 singleton births, 92 (5.6%) were preterm birth (PTB), 61 (3.7%) were low birth weight, and 120 (7.3%) were macrosomia.
The PTB rate in the current study was lower than the global average rate (10.6%) 19 but similar with other studies in China. 9,10

DISCUSSION
The main strengths of the TJBC include: (1) The TJBC study has the strength of collecting comprehensive maternal, paternal, and childhood information (not only maternal but also paternal data is collected, which overcomes the shortcomings of traditional maternal-child birth cohorts), with the potential for causal link studies between early-life exposures and later health conditions.
(2) Additionally, the TJBC study has the strength of a diverse range of biological samples, and we are actively engaging with technical advances and rapid development of emerging new technologies, including metabolomics, epigenomics, metagenomics, and immunomics. Several limitations of this study should be noted. (1) The cohort covers urban and suburban populations of Tianjin local residents, but with almost no rural participants included. Thus the findings of the cohort may have difficulty in extrapolating to rural populations. (2) The sample size is large for causal link researches of early-life exposures with health consequences, but not as large as several mega birth cohorts. The TJBC will continue to recruit participants and the follow-up will be extended to adolescence.
In conclusion, the current paper provides a profile of a large-scale prospective birth cohort in northern China. The TJBC collects comprehensive maternal, paternal, and childhood information, including a diverse range of biological samples, with the potential for multi-omics research, which would provide new insight on the causal link between early-life exposures and later health conditions.