The BiB study is a prospective birth cohort study that recruited women during pregnancy, full details of the study methodology have been previously reported . To be eligible, women had to attend booking clinic between March 2007 and December 2010 and be booked to give birth in Bradford. Bradford is a city in the North of England with high levels of socioeconomic deprivation and ethnic diversity. Approximately half of the births in the city are to mothers of South Asian origin most of whom originate from Pakistan. Women were recruited to BiB at their 75 g oral glucose tolerance test (OGTT) appointment which is routinely offered at around 26–28 weeks gestation to all women booked for delivery in Bradford. Those who attended this appointment and agreed to take part in the study consented to the use of theirs and their child’s medical records, had their height and weight recorded and completed an interviewer administered questionnaire. The questionnaire included questions relating to ethnicity, social and economic circumstances, smoking, alcohol, diet, education, employment and place of birth. Interviews were conducted in a range of South Asian languages (including Mirpuri, Bengali, Punjabi). Mirpuri is the most commonly spoken Asian language in Bradford but has no written script therefore questionnaires were transliterated, that is translated verbally to Mirpuri and then written phonetically, precisely as spoken to ensure that all interpreters translated it in the same way. A total of 12,453 women who gave birth to 13,818 liveborn children were recruited to the study. For these analyses, multiple births, children born to parents of ethnic origin other than White British or Pakistani, children of mothers who did not complete a baseline questionnaire at recruitment, children with missing birthweight (for example those who were born outside the Bradford area) and children who could not be matched to their primary care record were all excluded (Fig. 1). Thus 8850 participants are included (4119 White British; 4731 Pakistani). Ethics approval for the study was provided by Bradford Local Research Ethics Committee (ref 06/Q1202/48).
The number of general practice consultations and prescription data were derived from electronic records. Primary care electronic health records were obtained for BiB participants registered with GP surgeries that use the SystmOne platform. SystmOne has 100% coverage in Bradford and high coverage in surrounding areas. Records were extracted when NHS number, surname, date of birth and gender were an exact match in SystmOne. From the full BiB cohort of children, 99.0% were matched to their primary care records. Hospital episode statistics (HES) were obtained from the Health and Social Care Information Centre (HSCIC), matched to participants using the same process. Hospital admissions were categorised as Hospital Emergency (any emergency admission including to accident and emergency or direct to paediatric departments) and Hospital Elective which describes any elective admission either as an outpatient or inpatient (90% were outpatient episodes).
Birthweight was obtained from hospital birth records and in all participants was recorded immediately following birth using SECA digital scales. We identified children as being born small using the World Health Organization (WHO) criteria for low birthweight as a weight at birth of below 2500 g. As a low birthweight can be the result of either premature birth or restricted growth in utero and because here we are primarily interested in restricted growth or low term birthweight, we included gestation as a covariable in the analyses of low birthweight (< 2500 g). We separately calculated customised birth weight centiles that take into account gestational age, maternal height, maternal pre-pregnancy or weight at booking, ethnicity, parity and neonatal sex (Gardosi 2004) and are recommended by the UK Royal College of Obstetrics and Gynaecology (RCOG) for assessment of birth weight . SGA was defined as less than the tenth customised birth weight centile and all gestations were included in the customised chart analyses. Duration of gestation was obtained from hospital birth records and was based on the date of the mother’s last menstrual period which was confirmed by a dating ultrasound at around 12 weeks.
Assessment of ethnicity
Ethnicity was self-reported at the mother’s questionnaire interview and based on UK Office of National Statistics guidance details of which have been previously reported . For these analyses, children were defined as White British or Pakistani origin.
A priori we considered maternal parity, infant sex, gestational age, maternal age, social economic information (maternal education, housing tenure, means-tested benefits) and smoking as characteristics that might confound any associations. Maternal parity, gestational age (to the last completed week) and infant sex were all obtained from obstetric medical records. Customised birthweight charts account for gestation, parity and infant sex therefore these variables were only added to low birthweight analyses. Maternal age, social economic information (maternal education, housing tenure, means-tested benefits) and smoking data were obtained from the interviewer administered mother’s questionnaire completed at recruitment. We equivalised the mother’s highest educational qualifications (based on the qualification received and the country obtained) into one of several categories using UK NARIC (
): < 5 GCSE equivalent, ≥5 GCSE equivalent, ‘A’ level equivalent, Higher than A-level equivalent, Other qualifications (e.g. City and Guilds, RSA/OCR, BTEC), Don’t know, Foreign Unknown. Don’t know relates to the mother responding “don’t know” during interview. Foreign Unknown relates to a qualification listed in the free text response but no level of qualification is given or the qualification listed cannot be equivalised to one of the above categories. For these analyses, women were categorised as having been educated beyond the age of 18 or not (i.e. Higher than A-level equivalent, Other qualifications (e.g. City and Guilds, RSA/OCR, BTEC),university undergraduate courses). Don’t know and Unknown were categorized as not educated beyond the age of 18. Receipt of means tested benefits was based on the mother or her household receiving any of: Income Support, Job Seekers Allowance, Working Tax Credit or Housing Benefit. Housing tenure was categorised according to whether the woman lived in a household where the home was either part-owned (i.e. mortgaged) or owned outright, or not (i.e. rented). Maternal smoking was categorised as never, past (but not during this index pregnancy), current/during the index pregnancy.
All analyses were performed using Stata (version 13). Negative binomial regression models were employed as the outcome measures (counts of GP consultation rates, prescriptions and hospital episodes) were over dispersed and did not fit a Poisson distribution well. Models were constructed for each outcome and used to predict the incidence of GP consultations, number of prescriptions and hospital episodes for children based on their ethnicity, low birthweight and SGA categories: after adjusting for the covariables described above and taking into account individual exposure time (the proportion of the study period that a child is registered with a GP practice using SystemOne). Incidence rate ratios (IRR), the ratio of predicted events for Pakistani children compared to White British children, with 95% confidence intervals (CI) were also derived to aid the substantive interpretation of ethnic differences.