The China birth cohort study (CBCS)

Pregnant women are eligible to participate in the study if (1) they are of Chinese nationality; (2) they are 6–13⁺⁶ weeks of gestation at the time of recruitment, including both natural pregnancy and conception using assisted reproductive technologies; (3) they plan to attend routine antenatal examination and deliver in the study site, and plan to continue to live locally for more than one year; (4) they have no notifiable infectious diseases such as hepatitis B, syphilis and HIV; (5) they are able to understand the study and willing to give informed, written consent. It is possible for included women to withdraw from the study at any stage.

The large sample size of CBCS will provide adequate power to investigate not just associations but also potential causal effects of different exposures on birth defects. The sheer size of the study means that detailed planning is necessary to ensure feasibility, effectiveness of recruitment and follow-up, and standardization of data collection as well as biological sampling / processing; for this reason, CBCS has been divided into three phases.

The first two phases have now been completed, with phase 2 finishing, as planned, on 29th February 2020. On that date we had enrolled 120 377 eligible women in early pregnancy (less than 14 weeks of gestation at enrolment), accounting for just over 24.1% of the total target. This recruitment was completed at 38 research sites in 17 provinces, cities, autonomous regions and municipalities covering most areas of China (Fig. 1). Most of these sites are referral hospitals (3A hospitals), with a total of just over 300 000 deliveries annually.

In CBCS, women are enrolled in early pregnancy, at 6–13⁺⁶ weeks of gestation. At this point, all participants are asked to complete a baseline questionnaire and donate 10 ml of peripheral blood, taken before 13⁺⁶ weeks of gestation. Clinical laboratory measures are collected for each participant at recruitment (Supplementary table 1). The first and second follow-up visits are undertaken mid-pregnancy at 20–23⁺⁶ and late pregnancy at 28–33⁺⁶ weeks of gestation, respectively. For all participants, questionnaires are completed at these two follow-up visits by in-person interviews at their routine prenatal examination. Corresponding clinical laboratory measures are collected at both of these follow-up visits (Supplementary table 1). The third follow-up visit is undertaken after delivery. The clinical information is recorded for all the participants by trained researchers, doctors or nurses.

If the participant has a miscarriage in early pregnancy, or a pregnancy loss during mid- or late pregnancy, all clinical information will be recorded by trained researchers, doctors or nurses. If a birth defect is found at any of these stages, clinical information, including ultrasound scan information on fetal defects, is documented, and biological samples collected by specially trained researchers, doctors or nurses according to standard operation procedures and protocols. For each case of a birth defect, two controls are identified with maternal age ± 2 years and gestational age ± 2 weeks. Figure 2 shows the time points of the recruitment and follow-up in CBCS.

The CBCS covers a wide range of areas, including demographic characteristics, health status, reproductive status, prenatal screening, genetic factors, medication exposure and medication, pre-existing chronic disease, nutrition, lifestyle, environment factors and clinical laboratory measurements. All the data collected via the questionnaires, clinical laboratory measurements, standardised ultrasound scans at the enrolment and at follow-up are shown in Table 1. The questionnaire (Questionnaire (Q) 1: 97 questions), completed at enrolment, includes the data on demographic characteristics, outcomes of previous pregnancies, medical history before and during pregnancy, medication, family history on birth defects, occupation, exposures at home and in workplace and lifestyle habits. A food frequency questionnaire (Q2: 70 questions) is required for a representative sample of 80 000 participants. The questionnaires (Q3: 24 questions and Q4: 17 questions) used for the first two follow-up stages, cover the pregnant woman’s health status as well as any prenatal diagnoses. For Q5 (14 questions) data on mode of delivery, perinatal outcomes and birth characteristics (standardised measurement of birth weight, birth length, head circumference, new-born sex, Apgar score and the information of placenta and umbilical cord) are collected at the third follow-up, according to international standards [9]. The information about fathers was embedded in Q1 completed at enrolment, which was filled in by pregnant women, including demographic characteristics, family history of birth defects, smoking status and alcohol use.

All clinical laboratory measures include routine blood tests, as well as blood biochemistry, thyroid function, coagulation function, vitamin level tests, screening for Toxoplasma, Rubella virus, Cytomegalovirus and Herpes virus, maternal serum alpha fetoprotein, oral glucose tolerance test (OGTT) and HbA1c (Supplementary Table 1). Biological samples collected at different periods during pregnancy include peripheral blood samples donated by all participants, and may include amniocytes for those women opting to have amniocentesis, tissue, placenta, cord blood or umbilical cord tissue donated by participants with pregnancy loss or birth defects, and, where appropriate, controls.

All baseline, follow-up and outcome data, as well as clinical laboratory measurements and bio-banked samples, are managed through a secure cloud-based platform, including an electronic data capturing (EDC) and a bio-bank system. This platform can combine the information from the two systems automatically and is used to establish, manage and maintain the database. Data can be collected in different ways including on mobile phone, PC or data import. Logic check functions are programmed during data input or capturing in order to avoid data entry errors and missing items, and to ensure integrity and accuracy of the data. The logic validation is not only in each questionnaire, but also operates across different questionnaires for the same woman. Accounts have different levels of permissions for research staff, database administrators and investigators. The platform issues follow-up reminders to both administrators and participants at each follow-up stage. Overall, this cloud-based platform has been designed based on the main principles of data security; ongoing quality control; the ability to share data among different research sites; ease of usability; and extensibility.

The data, including all the measurements in the questionnaire and biological information from each participating centre, are transferred into the central CBCS Database in real time when the information is submitted. Additionally, all collected data is backed up and sent to the central server once a week. The data is managed by professional data administrators and will not be exported until the CBCS steering committee agrees.

Currently, as our study is still in progress, our data cannot be shared externally. However, the data is shared within the team and among participating centres. We will consider the time of data sharing externally when recruitment and follow-up is completed. There will be an application process for use of the data through the completion of a request form which will include the proposed analysis strategy. Applications will be evaluated and once approved by the CBCS committee, data will be shared; results from such analysis must be returned to the database, to allow further sub-analysis.

The current data are from the pre-planned interim analysis, prior to the third phase of the study, undertaken to ensure proper functioning of all processes, examine follow-up rates, establish whether recruitment of more sites is needed and highlight any problems. As the study is still ongoing, early findings are described below. Table 2 presents the current baseline characteristics of study population. A total of 120 377 pregnancy women have been enrolled, nearly half nulliparous. More than 95% of pregnancies were conceived naturally, while 4.2% used some form of assisted conception. The mean maternal age was 30.08 ± 4.27 and for paternal age it was 31.45 ± 4.93. More fathers were ≥ 36 years old (18.52%) than mothers (11.25%). The age at menarche for most women was 12–14 years, similar to the majority participants in USA, Germany and African-Americans [10‐12]. Very few women smoked or consumed alcohol regularly, and more men did so. The prevalence of chronic hypertension and diabetes mellitus before pregnancy was about 0.32% and 0.26% respectively. About 15.44% of women’s pre-pregnancy body mass index (BMI) was over 25 kg/m². However, more men (41.32%) had a BMI above 25 kg/m². The vast majority of participants and fathers were of Han ethnicity (> 93%), which is slightly higher than reported in the China Statistical Yearbook 2019 (91.5%) [13]. Roughly 65% of participants and fathers had an undergraduate or higher education, and most of the participants were employed (> 75%). Approximately 60% of annual family income was 50 000–200 000 yuan (USD7000-28 000). A small proportion (1.69% and 1.36% of mothers and fathers, respectively) reported a history of any birth defect in their biological families. The prevalence of birth defect in this study was 2.5% and the three leading birth defects were congenital heart defects, urinary system and genital organ malformations, and chromosomal abnormalities.

The CBCS has a number of strengths. Most important is the very large sample size, the detailed, comprehensive information, and multiple biological samples collected. To our knowledge, this will be the largest birth cohort in the world and will provide adequate power to investigate the causal effects of different exposures on birth defects [14]. The data collected in this study cover environment factors, genetic factors, medication exposure and medication, chronic disease, nutrition and lifestyle. Measuring this multitude of exposures in a single study, coupled with detailed follow up and outcome ascertainment mean that associations can be explored in detail. For example, the relationship between environment (atmosphere, greenness, light, noise, etc.) and teratogenesis, or the associations of the use of drugs with unclear teratogenic effect and birth defects. Moreover, extensive maternal blood samples are collected. Remarkable advances in -omics technologies, including genomics, metabolomics, proteomics and assessment of the microbiome have provided new opportunities for systematic epidemiologic research and further exploration of the mechanism of birth defects. Therefore, integrating-omics and digital technologies, and incorporating a multidisciplinary approach across the life cycle, should be most effective for understanding the factors associated with, and ultimately the prevention, of birth defects.

One weakness of this cohort is the selection bias which had been recognized. However, the main aim of this study is to explore the associations between maternal exposure and birth defects and other adverse outcomes, and the selection is balanced for maternal exposure and the outcomes since the pregnancy outcome is not known at recruitment. Another weakness of this cohort is the self-administered questionnaire at recruitment, with some data based on participant self-evaluation, including health-related variables. However, the questions are presented in simple language and have been piloted; comparison undertaken with forms completed by healthcare staff have demonstrated high concordance. Furthermore, we have collected many different types of biological samples, but we have not started to analyse these. However, genomic sequencing is planned, and we are keen to incorporate emerging new technologies.

In our study, the neonatal outcome at delivery is the final end-point, and no further follow-up of participating mothers and their children is being carried out; this is a weakness when compared to other long-standing birth cohort studies, such as the Danish National Birth Cohort (DNBC) and the Norwegian Mother and Child Cohort Study (MoBa) [14‐16]. This is due to financial constraints and a study time-line that funders have implemented; it means that longer-term conditions in the infant, that are not evident at birth, may be under-ascertained. Detailed newborn assessment is being undertaken to mitigate this risk. At the same time we plan to expand the scope of disease and health assessment in future to explore the causal relationship between early life exposures and later health status. In addition, we are aiming to extend our work in order to collect additional information about the father, establish a biological sample database, and develop a long term follow-up plan covering the entire life cycle.

The increasing number of birth cohort studies over recent decades has given opportunities for projects to integrate multiple cohorts, such as the Environmental influences on Child Health Outcomes (ECHO) program which contains 84 observational cohorts and the EU Child Cohort Network which contains 19 pregnancy and childhood cohorts [17, 18]. Combining data will make it possible to identify smaller effect estimates, and search for differences in risk factors across countries. This will enable better research into causal understanding and modelling of life course health trajectories. To maximize the benefits of our research, we are already forming links to collaborate with other birth cohort studies, and our data will ultimately be made accessible to other researchers.