Study design
This study was a retrospective analysis of data from the BiB cohort. White British and South Asian pregnant women were the populations of interest. Women from other ethnic groups, or where ethnicity data were missing, were excluded from the analysis. The two outcome variables were antenatal depression and anxiety. In the BiB cohort, mental health was assessed using the GHQ. The GHQ is a validated self-report questionnaire with 28 items relating to the respondent’s current mental state, ability to carry out functions and daily activities and appearance of new and distressing phenomena.13 The 28 items are grouped into four categories each used to identify symptoms of certain psychiatric disorders. Since this study focused on depression and anxiety as the most common mental health conditions during pregnancy, the total Likert score of the subscales for depression and anxiety (D, items 22-28; B, items 8-14) were used in the analysis (Appendix S1).13There are no agreed thresholds for the subscales to indicate depression or anxiety, but studies suggest that the cut-off score should be based on the mean/median of the sample of interest.14,15 In this study, the median was used due to the non-normality of distribution. This was 0 for depression and 6 for anxiety. Therefore, a score of >0 was used to indicate depression and >6 indicated anxiety.
The main exposure variable was maternal early pregnancy BMI. Data from the BiB cohort includes information on the mother’s booking BMI calculated using measured height and weight between 10-12 weeks of pregnancy.16 A realistic lower limit of 11 kg/m2 was set as this has been shown to be the lowest BMI for survival in women.17 An upper limit of 80 kg/m2 was based on the frequency distribution in the data from the BiB cohort and a published study.18Women with a booking BMI outside this range were excluded from analysis (n=720, 6.5%). BMI was analysed as a categorical variable due to the inclusion of underweight which is also associated with increased depression and anxiety,19 therefore, a continuous analysis may skew the results. BMI was stratified by the WHO’s classification. For White British women, the categories were: underweight, <18.5 kg/m2; recommended weight, 18.5-24.9 kg/m2; overweight, 25-29.9 kg/m2; and obese class 1, 30.0-34.9 kg/m2; class 2, 35.0-39.9 kg/m2; class 3, ≥40.0 kg/m2.2 The categories used for South Asian women were: underweight, <18.5 kg/m2; recommended weight, 18.5-22.9 kg/m2; overweight, 23-27.49 kg/m2; and obese class 1, 27.5-32.49 kg/m2; class 2, 32.5-37.49 kg/m2; class 3, ≥37.5 kg/m2 .20 A secondary analysis was performed using the general population BMI criteria (Table S1)for South Asian women due to the current lack of guidance in the UK for using Asian-specific criteria in pregnancy.
Additional variables included in the adjusted models were maternal age, maternal education, area of residence deprivation (based on postcode) and maternal smoking. Maternal age (years) was analysed as a continuous variable. Maternal education was defined as mother’s highest educational qualification (equivalised) with the following categories: none, GSCE equivalent, A-level equivalent and higher than A-level (used as reference group). The Index of Multiple Deprivation (IMD) was used to categorise area of residence deprivation. The IMD is a measure of relative deprivation for small areas in England and is the most widely used tool to measure deprivation in health-related research in the UK.21 Since BiB was carried out in Bradford, the national IMD quintiles would be of limited use for this study because Bradford has a higher level of deprivation compared to most areas of the UK.12 Therefore, the deprivation data were skewed towards the most deprived quintiles for this population. A binary variable was created with quintiles 2-5 combined to represent lower levels of deprivation (used as the reference group) and quintile 1 represented the highest level of deprivation. Maternal smoking during pregnancy was a binary variable (yes/no).