Study design, setting and participants

Since May 2017, we have invited all infertile couples who received medically assisted reproduction through IVF-based technologies (ie, fresh or frozen IVF protocols, including intracytoplasmic sperm injection (ICSI)) in the First Affiliated Hospital of Anhui Medical University (AHMU), Hefei City, Anhui Province, China, to take part in AMCHS. Participants’ recruitment in AMCHS is consecutive. The inclusion criteria for study participants are (1) couples who are receiving ART treatment for the first time and (2) those who have not accepted oocyte or sperm donation or intrauterine insemination.

At the beginning of the recruitment cycle, AMCHS research coordinators register sociodemographic information of eligible couples. To obtain consent from the participants, a fertility specialist would introduce the AMCHS to married couples who come to the clinic for IVF treatment consultations. Patients can fully consider from consultation to treatment. When married couples were both willing to participate in the study, they would sign an informed consent form. The discussion of the study and consent was carried out in a private room in the clinic. The couples would join in this cohort study for IVF treatment once they signed the consent form. We offered women free ultrasound monitoring throughout their IVF treatment as an incentive to taking part. Participants could withdraw at any time they wanted without giving any reasons, which would not affect their ongoing care. Researchers interviewed participants using a general health and risk factor questionnaire; these interviews were carried out in a private room in the clinic. Following participant consent, they also accessed participant’s medical records, including cycle, pregnancy and delivery data, and took biological samples for long-term storage for solely medical research purposes.

From May 2017 to December 2020, we recruited 2198 eligible couples. The participants were interviewed for baseline assessment and their fasting blood samples were taken before commencing IVF treatment. Figure 1 shows the flow chart of the participants’ recruitment and follow-up processes throughout the study. Participants are followed up from the study entry throughout their fertility care, pregnancy, birth (for those achieving pregnancy) and children (until 3 years old) or until they discontinue treatment or withdraw from the study. As of April 2021, 2042 couples have had IVF treatment, of which 1828 couples received embryo transfer surgery (figure 1). Among 1475 women who confirmed pregnancies, 138 suffered from abortion, 8 suffered from ectopic, 9 suffered from stillbirth, 263 were ongoing pregnancies and 1057 gave live birth to babies. There are 940 children followed up until delivery, 576 children completed the follow-up at age 30–42 days, 459 children completed the follow-up at age 6 months, 375 completed the follow-up at 12 months and 26 completed the follow-up at 36 months.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow chart for participants and updated outcomes in the follow-up of AMCHS. AMCHS, Anhui Maternal–Child Health Study; IVF, in vitro fertilisation.

As an ongoing cohort study, AMCHS continues to recruit the participants and we conduct follow-up and interviews with the remaining couples in the cohort (figure 1). Initially, we planned to recruit 2000 couples for the outcome analysis and close the recruitment in June 2021. The sample size was considered according to the incidence of reproductive and pregnancy outcomes. The number of participants (n=2042) in the current dataset has shown enough study power for some outcomes based on the formula of the cohort sample size, for example, current analysis for the impact of maternal age on implantation failure. From January to November 2021, AMCHS recruited and followed up 29 participants. The study powers for other complicated data analysis for rare outcomes will be ensured in ongoing recruitment and follow-up of the AMCHS cohort.

Patient and public involvement

Patients and the public involvement were considered, but neither are involved in the design, conduct, reporting or dissemination plans of our research.

Questionnaire survey

Both female and male participants completed a baseline questionnaire survey. The questionnaire includes sociodemographic information (eg, date of birth, sex, educational level, occupational class, income and residential area), lifestyle questions (eg, smoking habits, alcohol consumption, dietary intake, tea and coffee drinking habits and physical activities), environmental exposures (eg, passive smoking or indoor air pollution), reproductive history, medical history, medication history and family history. We assessed mental health via questionnaires of Chinese Perceived Stress Scale (0–40), Centre for Epidemiologic Studies Depression Scale (0–60), Self-Report Anxiety Scale (20–80) and sleep quality (Pittsburgh Sleep Quality Scale (0–20)).

We include a questionnaire recording ART medications and pregnancy complications at later visits during each per trimester. To investigate newborn health, we use questionnaires to document food diaries of newborn in food frequency questionnaires including intakes of dietary supplements, mothers’ feeding practices, breast milk frequency, sleep patterns in Brief Infant Sleep Questionnaire, anthropometry of infant, medical history and medication history.

Data collection and follow-up of the cohort

The AMCHS is designed to examine exposures from preconception period throughout once per trimester of pregnancy (for those achieving pregnancy), and up to labour and delivery. Ultrasound monitoring of ovarian response was offered as an integral part of the cycle. When the number of mature oocytes met the requirement of IVF/ICSI, couples would be arranged to retrieve oocytes and sperm. Laboratory technicians in the IVF laboratory collected follicular fluid, sperm and seminal plasma. The fasting blood used to test for pregnancy was collected after 2 weeks of embryo transfer surgery. When a positive pregnancy test was obtained, pregnant women were invited to the reproductive centre at gestational weeks 4, 24, 32 for a questionnaire interview and blood sample collection. Medical records at gestational weeks 4, 24, 32 about gestational clinical information were also accessed. The baseline questionnaire and mental health questionnaires are used at each visit.

During the birth admission, maternal blood, and cord blood and tissue were collected by the nursery midwife. The child was examined at ages 30–42 days, 6 months, 1-year and 3-year-olds (during early childhood). At age 30–42 days, 6 months, 12 and 36 months after giving birth, mothers were approached by phone for a questionnaire interview and provided information about clinical physical examination at the local community hospital.

Electronic questionnaires were used for data collection. The original questionnaire data were encrypted and saved to the cloud every day. Quality control staff used prewritten programmes to check the reliability and logic of the data. Results would feed back to the research coordinators on the same day for verification and correction. When missing and unreasonable data occurred, our research coordinators would correct them through face-to-face interviews for participants, by telephone or by checking the electronic medical record system (figure 1).

Outcomes measures

We collected data from the interview questionnaires, extracted medical information from a fertility clinic and delivery records from hospital. Blood samples and follicular fluid were collected from the female partner. Sperm and seminal plasma samples were collected from the male partner. A sample of cord blood and tissue was collected from the newborn.

Electronic medical record abstraction: cycle, pregnancy and delivery data

The AMCHS team abstracted clinical medical records during each individual fertility treatment cycle and throughout follow-up (up to the birth of an infant for those achieving pregnancy). Research coordinators abstracted clinical information in the First Affiliated Hospital of AMHU to ascertain the outcome of each cycle, including physician assigned infertility diagnosis, infertility medical history, blood pressure, body mass index (BMI), medication regimen; hormone at HCG (human chorionic gonadotropin) day, luteal phase support after IVF, scoring for fertilisation, embryo culture and transfer, biochemical pregnancy (with β-hCG measurements), clinical pregnancy (with ultrasound assessment) and relevant data. Antenatal and neonatal hospital medical records were also accessed. During the study period, antenatal, intrapartum and postpartum events were captured from hospital medical records or phone interviews.

The AMCHS Biobank

Blood samples were obtained from female partners at several times during the periods of preconception to delivery and from the male partner at preconception. We also collected blood samples from cord blood and tissue at birth. The collection of blood samples from couples to their babies enable us to explore any heredity influences and detect any potential biomarkers for disease aetiology. Follicular fluid, sperm and seminal plasma were collectedto evaluate the quality of gametes, which could be the key factors of IVF outcomes.

Plasma and blood cell samples were separated from the blood via centrifugation for 10 min at a relative centrifugal force of 3000 g at 4℃. Plasma was stored at a −80℃ refrigerator, and the blood cell was stored at −20℃ refrigerator. Cord blood and tissue were stored at 4℃ after a temporary delivery in the operation room and then transferred to a 20℃ refrigerator within 24 hours. Seminal plasma was centrifuged before it was subpackaged. Finally, the sperm (1 mL) and seminal plasma (2 mL) were stored in a nitrogen canister in a −80℃ refrigerator. Follicular fluid was collected more than 5 mL on average. After being centrifuged, the subpackage was stored in an −80℃ refrigerator.

The overview of data collection and timepoint of AMCHS are shown in table 1 and table 2.

Data presentation and statistical analysis

Data were presented as mean (SD) for continuous variables and as number (percentage) for categoric variables. The continuous variables include age, BMI, depression symptoms, anxiety symptoms, stress symptoms, sleep quality, follicle-stimulating hormone, luteinising hormone, oestradiol. The categorical variables include district (city, country, countryside). Education level (under secondary, secondary, high school, college/university, postgraduate), annual household income (<¥50 000/year, ¥50 000–¥100 000/year, 100 000–200 000/year, ¥200 000–¥300 000/year, >¥300 000/year), occupation (officers, technician, clerk, business and service personnel, agriculture, forestry, animal husbandry, fishery and water conservation personnel, production and shipping equipment operators and related personnel, solider, other practitioners who are inconvenient to be classified, unemployed), smoking status (active smoking, passive smoking), tubal status (normal, abnormal), sexual abstinence (normal, abnormal), infertility type (primary infertility, secondary infertility), antral follicle counting (0, 1–5, 6–10, 11–15, >15) and sperm volume (normal, abnormal). Multivariate adjusted regression models were used to examine the adverse outcomes in relation to baseline exposure variables. All analyses were performed using SPSS V.26.0 (IBM, Armonk, New York, USA).

Study participants and characteristics

Online supplemental table 1 shows baseline characteristics of AMCHS women and men. Most of the couples were from the urban areas with a household income of ¥50 000 to ¥1 00 000 (RMB) per year. Of 2042 women, the average age at the time of enrolment was 31.1 years old (SD 4.6) (range 20–48 years), with the majority being in the age group 30–40 years (50.4%), 60.8% are educated ≥college and 26.5% are working in office. While 5.1% of women smoked or used to smoke before pregnancy, 61.5% were exposed to passive smoking, and about one-third of women consumed alcohol. The majority of women (57.4%) had a normal prepregnancy BMI (18.5–<23.9 g/m²) (the mean BMI was 22.5 kg/m² (SD 3.1)). 6.1% of women suffered anxiety, but 36.7% of women suffered from depression. None of the women had any sleeping problems. Approximately four-fifths of women had a female infertility factor as their primary diagnosis. The main reason for infertility in these women was tubal factors (56.7%). 85.7% women never had a history of pregnancy, and 85.7% of women had never given birth. Of 291 women who used to have a pregnant history, very few participants had a history of hypertension or GDM. Of 1004 women who completed questions about pregnancy history, the rates of caesarean section history, adverse pregnancy history, artificial abortion were 47.1%, 44.4%, 63.6%, respectively, which are both higher than the average rate in China.36–38

Of 2042 men, the average age at the time of enrolment was 32.5 years old (SD 4.6) (range 22–57 years), and 55.8% were in the age group of 30–40 years. 65.1% of male participants are educated ≥college, and 26.2% are working in business and service personnel. 54.3% smoked or used to smoke, more than 80% were exposed to passive smoking and 63.3% consumed alcohol. The percentage of normal BMI (18.5–<23.9 kg/ m²) in men is the highest (%) (the mean BMI was 24.5 kg/ m² (SD 3.3)). 27.9% of men suffered from depression, 4.6% had anxiety and 0.1% had a sleeping problem. 33.8% had a male factor as their primary infertility diagnosis, with the main reason being abnormal progressive motility (37.7%).

Cycle endpoints

1828 couples have been examined for a total of 2757 cycles during the follow-up time. These cycles resulted in 1589 pregnancies, of which 5.9% (n=94) were only chemically detected by a β-hCG blood test and not clinically visualised on ultrasound (biochemical losses) (figure 1).

Among these 1475 ultrasound-confirmed pregnancies, 9.4% ended in spontaneous abortion before 20 gestational weeks, 0.58% ended in preterm birth, 9 pregnancies ended in stillbirth (loss in or after 20 weeks) (figure 1). There have been 1057 live births, of which 117 need to be followed up. The remaining 940 pregnancies resulted in 1123 live births: 758 singletons and 363 multiples (181 pairs of twins, one set of triplets). To date, the overall live birth rate per initiated cycle is 38.3% (n=940/2757), and the live birth rate among cycles achieving pregnancy is 71.6% (n=1057/1475) (figure 1).

Data analysis for the impact of female age on implantation failure

This paper has shown our preliminary analysis examining the impact of maternal age on implantation failure based on the current follow-up data in AMCHS. Table 3 shows number and percentage of participants and implantation failure across four age groups, and adjusted OR for implantation failure according to maternal age. The rate of implantation failure significantly increased with age. After adjustment for district, education and family income, OR of implantation failure in women aged 30–34 years was 1.43 (95% CI 1.16 to 1.77), 1.97 (95% CI 1.46 to 2.66) in 35–39 years and 6.52 (95% CI 3.35 to 12.68) in ≥40 years compared with those aged 20–29 years. Further adjustment analysis for including clinical variables did not substantially change these ORs (see model 2 in table 3).

Strengths and limitaions of AMCHS

The AMCHS is the first prospective reproductive cohort study conducted in Anhui, China. The main strengths of AMCHS are the data collected based on couples from preconception period to 3 years of child. This allows us to evaluate the influence of specific sensitive windows towards reproductive and pregnancy outcomes, which are less possible to observe in most pregnancy cohorts.26 27 We investigate a variety of risk factors for reproductive, pregnancy and child health outcomes. The in-person collection of multiple biospecimens from women, men and their child is over an extended time, enabling us to explore the relationship between risk factors and endpoints of interest at the molecular and genetic level. Using standardised questionnaires, our experienced interviewers have collected rich data from May 2017 to April 2021 and clinical abstraction from electricity medical records, which allows us to examine lifestyle, nutritional, environmental exposure, mental status of infertile couples and provided us with comprehensive covariate data. The resourceful data allow us to explore multiple interactions among risk factors. Furthermore, we highly focus on the paternal impact on adverse reproductive and pregnancy outcomes as the decline of sperm quality of male accounts for 30%–50% of infertility factors39 and contribute to miscarriage.40 This is often ignored in most pregnancy cohort studies.26 27

The study has limitations. First, the AMCHS cohort study was set up in only one hospital in Anhui province, which may lack external validation. While there have been more than 523 medical institutions approved to carry out human reproductive-assisted technologies across China,41 we will set up a collaboration with other similarly designed maternal–child cohort studies from different provinces to enlarge the sample size by combining data sources to increase our study external validation. Furthermore, the majority of participants in the AMCHS cohort come from Hefei and 76.5% participants lived in urban areas. They were well-educated and had a higher socioeconomic status and nutrition status, but were less physically active compared with rural residents. Thus, caution should be exercised in generalising our findings to the whole Anhui province. As an ongoing cohort study, we will recruit more participants, particularly from rural areas to examine some specific factors (eg, indoor air pollution) associated with ART adverse outcomes. Second, although the participants’ involvement is easy to sustain when they are in the middle of treatment for live birth, 11.1% (117/1057) of families that gave birth to babies are lost to follow-up, and after their babies’ birth 6.6% (62 of 940) families withdrew from the study. Although the current withdrawal rate in our study is similar to those in HICs,28 42 it may increase with the follow-up time. We will take all measures to maintain the cohort members in the follow-up as many as possible and examine the effect of those lost to follow-up on the cohort study findings.