Growing up in Bradford: protocol for the age 7–11 follow up of the Born in Bradford birth cohort

BiB was established in Bradford in 2007 [1, 2]. It quickly developed into an applied population health research platform to address the high levels of child ill-health in the city. The city includes some of the most deprived areas in the UK, with high levels of infant mortality, childhood obesity, asthma and disability [3]. The aims were fourfold: 1) to investigate early life determinants of health and ill-health amongst families; 2) to use this evidence to develop, design and evaluate interventions to promote health; 3) to provide a model for integrating research into practice; and 4) to build and strengthen local research capacity in Bradford.

With modest initial funding, BiB had to turn to its local network of midwives, obstetricians, paediatricians and health visitors to recruit families and collect data. This model provided an efficient means of recruitment, and also served to promote ownership of the study amongst local health care providers which, in turn, helped to support our model of translating research into practice. As the children grew up this collaborative approach was extended into schools and local government.

A key decision during the design of the cohort study was to consent parents to linkage of routine health (primary and secondary care) and education records, and to allow the BiB research team to access these records. In collaboration with our local provider of electronic health records across primary care (The Phoenix Partnership, SystmOne) we were able to link 99% of BiB participants to their health record. We also worked with our local authority, Bradford Metropolitan District Council, to link BiB children to their Unique Pupil Identification Number (UPN), allowing us to match 84% with their education record.

Between 2007 and 2010, the BiB cohort recruited 12,453 women with 13,776 pregnancies at 24–28 weeks gestational age, with 13,858 births (live births and stillbirths). In addition, 3449 of their partners were recruited^a.1 A rich set of data were collected during pregnancy and in the immediate postnatal period [1, 2].

Follow-up since then has included growth and blood pressure measures by primary school nurses at age 5–6, [4, 5] whole cohort record-linkage to health and educational data, e.g. [6, 7], detailed bespoke data collection with sub-samples [8‐10], and addition of biomarker data using stored biological samples, including genome-wide data, metabolomic data and Deoxyribonucleic acid (DNA) methylation [11, 12]. A summary of data collected can be seen in Fig. 1 and Additional File 1 and full details of recruitment and data collected can be found in the BiB cohort profile [1] and on the BiB website [13].

Analysis of BiB data has provided insights into the factors that affect health and wellbeing in pregnancy and early childhood, including on social and ethnic inequalities in health [14‐16], the relationships between maternal gestational adiposity, gestational diabetes mellitus (GDM) and maternal circulating glucose concentrations and fetal growth, birth outcomes, infant and childhood health in South Asian and White European families [4, 5, 17‐19], the associations of environmental stressors such as green space and air pollution on health e.g. [20‐22], exposures that may influence development of asthma and allergies [23] and the role of genetic variation in human health [12]. BiB data have underpinned more than 100 journal publications; a full list can be found at https://​borninbradford.​nhs.​uk/​.

Strengths of the BiB cohort include:

The diversity of the cohort. Forty five per cent of families are of Pakistani origin, and half of Pakistani-origin mothers and fathers were born outside the UK [24]. Half of all BiB families are living within the fifth most deprived areas of England and Wales. This provides the opportunity to study the interplay of deprivation, ethnicity, migration and cultural characteristics, and their relationship to child health and wellbeing.

A life course approach. The assessment of multiple-‘omics data (including genome-wide data, some genomic sequenced data, epigenomic, and metabolomic data) and lifestyle in parents and children, with repeat measures of some of these in subgroups and with the current follow-up in the whole cohort supports life course epidemiology. That is, understanding how social, cultural, lifestyle and biological factors interact across the life course and across generations to influence development, health and wellbeing.

Participant and community engagement. We have established an active BiB parent governors group, which meets every two months, to help co-produce our research programme. We have a range of activities focused on communicating and disseminating findings to participants and the wider Bradford community, using traditional (birthday cards, newsletters) and social media (Facebook: https://​www.​facebook.​com/​BornInBradford, twitter: @bibresearch @bibparents) approaches. We hold regular events to talk with the community about our research, including family science festivals and pop-up events in local shopping centres. We are regularly featured in the press, including a yearly BBC Radio 4 radio programme that features BiB [25].

The city focus. The focus on a single city has facilitated engagement with participating families, including traditionally ‘seldom heard’ or ‘under-served’ groups such as families living in more deprived areas. It has also facilitated engagement with key stakeholders and the use of findings in local decision making. BiB has recruited a sample that is largely representative of city’s ethnic and deprivation profile [26].

Detailed data collection and routine data linkage. Detailed data collection during pregnancy, including oral glucose tolerance testing which is routinely offered to all pregnant women in Bradford, [18] and repeat ultrasound scans [19], infancy and childhood (the latter supplemented by health and educational record linkage), together with stored biological samples on most participants (Fig. 1), provide a unique opportunity for clinical and public health translational research on maternal and perinatal health and developmental origins of health and wellbeing.

Community-based family assessments take place at home visits or in community settings (e.g. children’s centres, schools or GP practices). Participants eligible for a DEXA body scan are invited to our clinically equipped “Big BiB bus” mobile clinic. For those we are unable to visit in person or who live outside the Bradford area we are offering telephone/online follow-up. All children and parents who have previously been recruited to BiB are eligible to take part, and adults with parental responsibility for a BiB child who have not previously been recruited are invited to join the study. Each family is sent an invitation letter and three information sheets: ‘Information for families’, ‘Information for children’, and ‘How we use routine information’. Parents provide informed consent for themselves and their child, including consent to future linkage to health and education records. The following data are collected:

Questionnaires: Parents complete a self-administered questionnaire on a tablet computer and children complete a short diet and activity questionnaire administered by a BiB Research Assistant. A subsample of parents are asked to complete additional, optional questionnaire ‘modules’ (see Table 1 and Additional File 2).

Samples and measurements: We measure height, weight, subscapular and triceps skinfold thickness, waist circumference, bioimpedence, BP and pulse rate, take (non-fasting) blood samples (buccal swabs in those who do not provide a blood sample). We complete DEXA scans and collect urine in subsamples (Table 1 and Additional File 3). Assessments are completed by trained staff following standard operating procedures. We provide the option for participants to receive feedback on some of their results and letters for participants to take to their GP if any results could have clinical implications (see Additional Files 4 and 5 for examples of leaflets).

BiB children who attend a primary school within the Bradford Metropolitan District Council area (141 schools) and are in school years 3–6 (age 7–11 years) are eligible to take part in these assessments. We expect to obtain measurements from at least 90% of BiB children attending Bradford primary schools (total n = 11,711), an estimate based on previous work [47]. Information and consent forms for measurements in school are posted to the home address of eligible BiB families and are also provided via school.

For anthropometric (height, weight, skinfold, bioimpedance) and BP measurements, parents are sent an opt-out consent form (consistent with the consent process used for the National Child Measurement Programme, NCMP). For the accelerometry and blood samples, parents provide opt-in consent by completing and signing a consent form for their child which is either returned via a freepost envelope or the child’s school. Prior to each component of the child’s assessment, we ensure that they are comfortable and happy for the measure/sample to be taken. School nurse teams collect height, weight, skinfold thickness (subscapular, triceps), systolic and diastolic BP and bioimpedance, using standard operating procedures. Since 2018, trained BiB Research Assistants work alongside the school nurse team to administer accelerometers and a child-completed diet and activity questionnaire and take a non-fasting blood sample. It is currently too early to predict the response proportion to these measurements.

In this arm only, we conduct assessments with the whole class (including children who are in BiB and children who are not part of the study) at the request of teachers in the included schools. Data are collected in three academic school years (3–6, covering ages 7–11 years) in 90 primary schools that have high numbers of BiB participants attending them. Over 24,000 children are eligible to participate, with ~ 9000 of these being BiB children. After schools agree to take part, they provide information sheets and consent forms to parents; parents provide ‘opt-out’ consent.

A team of trained research assistants administer standardised assessments of sensorimotor control [48], working memory, processing speed and inhibition via tablet computers (further information is provided in Additional File 6). Children also complete an online questionnaire (‘me and my life’) which assesses aspects of child wellbeing (see Table 1). Teachers complete an assessment of each child’s behaviour and socio-emotional development using the Strengths and Difficulties Questionnaire (SDQ) [49].

Teachers are provided with a 1 page summary of cognitive assessment results for each child in their class, with a visual summary of results showing the child’s capability relative to other Bradford children in their age group (see Additional File 7).

Our goal is to obtain as high a follow-up rate as possible of our original cohort to maximize the value of our resource. Based on our previous experience we have estimated a 70% response rate, thus we hope to reach at least 9000 of the 13,354 currently active children in the cohort, 8400 mothers, and increase the number of fathers and partners in the cohort to 5000. Based on previous experience we also anticipated that this would not mean these numbers all had complete data. Numbers with blood samples and valid accelerometer data would likely be lower than those with weight and height, for example. Table 2 shows example calculations of minimal effect sizes based on this previous experience. For all calculations we provided minimal effect estimates for a pre-specified sample size in two ethnic groups. For some of these analyses the groups are White British and Pakistani participants (i.e. the two largest most homogenous groups) for others they are White European and South Asian. The rationale for presenting estimates in different subgroups is that previous publications using existing (e.g. baseline) data from BiB have compared associations between either White British and Pakistani (see for e.g. [4, 5, 17]) or White European and South Asian (e.g. [19]) and the objectives demonstrate that future studies are also likely to compare these groups. BiB has been, and will continue to be, used for research questions where the primary aim is not concerned with ethnic differences; statistical efficiency when all ethnic groups are combined in analyses will be greater and the minimal effect sizes possible to estimate smaller than those presented in Table 2. We have used known numbers where data exist (e.g. for birth characteristics) and conservative estimates where data are still being collected. For objective 2 we have determined the minimum effect size detectable with our given / estimated sample sizes to detect, with 85% power, a two-sided p-value of ≤0.01 and also of ≤0.0001 because of multiple testing with omics data in this objective. Outcomes for most questions in objective 2 are continuously measured variables and this is reflected in power calculations, which give the minimal difference in outcome means reported in standard deviation (SD) units per SD/category of exposure. For objective 1 most outcomes are binary and for objective 3 a mixture of continuous and binary. There are potentially fewer problems with multiple comparisons in these objectives, given the focus on testing existing theoretical models. For these, minimal effect sizes are presented as relative risks or difference in mean (SD units) per SD/category of exposures for a two-sided p-value of ≤0.05 with 85% power. Table 2 illustrates that we are likely to be able to detect minimal effects of clinical/public health relevance. For detecting differences in effect between two ethnic groups in additive models (i.e. differences in differences in means or in absolute risk for binary outcomes), even restricting these to the smaller groups of White British and Pakistani, we have 80% power at a p-value of 0.05 to detect a minimum effect difference of at least 0.20 SD between groups with continuous outcomes and at least 0.15 difference in absolute risk effects for binary outcomes. For most binary outcomes assessed in multiplicative models we would have 80% power at a p-value of 0.05 to detect at least 25% difference in relative risk effects between the two ethnic groups.

All data collection is administered or undertaken by trained staff using standard operating procedures. Staff collecting data receive training in information governance and safeguarding as well as the assessments/data collection they are responsible for. We have regular training updates and observation of data collection with feedback as appropriate.

During development of materials we conducted interviews and focus groups to ensure that materials are easy to understand and appropriate for our participants. We conducted eight parent focus groups to obtain feedback on the information sheets and to understand parents’ preferences on feedback of research findings. Questionnaires were tested prior to recruitment using think aloud interviews [50]. We interviewed seven parents and over 20 children as they completed draft questionnaires to check understanding of the questions and appropriateness of response options and reworded questions as required.

At baseline, 82% of questionnaires were completed in English, 13% in Urdu, 5% in Mirpuri and 1% in other languages. In the age 7–11 years follow-up we offer questionnaires in English and Urdu and members of the research team speak English, Urdu and Mirpuri. For other languages we work with interpreters to ensure no families are excluded. For the school-based assessments materials are provided in English only as the majority of Bradford primary school children have adequate English skills to understand our study materials. If children have difficulties, research assistants or teachers help with translation and record any assistance that they provide.

Ethical approval has been obtained from the National Health Service Health Research Authority Yorkshire and the Humber (Bradford Leeds) Research Ethics Committee for the community-based family assessments and school-based measures (reference: 16/YH/0320) and the school-based cognitive and wellbeing assessments (reference: 16/YH/0062).