Results
There were 9420 White British and South Asian women who had a BMI between 11 and 80 kg/m2. Following exclusions for missing GHQ data (n=1596, 17%), a total sample size of 7824 women was available for analysis. The final sample included 3514 White British and 4310 South Asian women (3780 Pakistani, 343 Indian and 187 Bangladeshi). There was no statistically significant difference between the included and excluded group in terms of age (27.3 ± 0.1 versus 27.3 ± 0.14) or deprivation level (Low deprivation 34.5% vs 31.3%, high deprivation 65.5% vs 68.7%) (Table S2) . However, the included population had significantly higher education levels than those excluded (>A-level 24.6% vs 21.1%, A-level 15.8% vs 8.9%, p<0.0001) and were more likely to smoke (17.6% vs 10.2%, p<0.0001) (Table S2) .
Table 1 shows the summary of characteristics of the included sample of pregnant women, stratified by ethnicity, with tests for association between the ethnic groups. In the included population, South Asian women were significantly older than White British women, had a higher percentage with >A-level education level, were more likely to live in the highest deprivation areas, less likely to smoke, had a higher proportion with depression, but a lower proportion with anxiety, independent of early pregnancy BMI. There was a significant association between BMI and ethnicity, with a higher proportion of South Asian than White British women in the overweight and obese categories(Table 1) .
Within the White British ethnic group, women with obesity were significantly older, had a lower percentage with >A-level education level, were more deprived and less likely to smoke, compared to women with a recommended BMI (Table 2) . Overall, 36% of White British women had depression and 48% had anxiety. Within South Asian women, those with obesity were significantly older and had a lower percentage with >A-level education level compared with women with a recommended BMI. There was no significant association between deprivation level, smoking status, depression or anxiety and early pregnancy BMI among South Asian women. In both the South Asian and White British groups, women with an obese BMI had higher proportions with depression and anxiety, compared with women with a recommended BMI, although this was not statistically significant.
There was no significant association between depression and early pregnancy BMI among White British women (Table 3) . Univariate regression found that South Asian women with an overweight BMI had significantly higher odds of depression than South Asian women of recommended BMI (odds ratio (OR) 1.17, 95% confidence interval (CI) 1.00-1.36). After adjusting for age, education level, smoking and deprivation, the adjusted odds ratio (AOR) was no longer significant (AOR 1.16, 95% CI 0.99-1.36). No other associations between BMI and depression were found to be significant. Interaction analysis found no significant association between ethnicity and antenatal depression within any BMI category (Table 3).
Table 4 shows the crude and adjusted ORs for screening positive for anxiety within each ethnic group for each BMI category and AORs for interaction analysis between White British and South Asian women. Univariate analysis showed that White British women with an overweight BMI had significantly higher odds of anxiety compared with White British women of recommended BMI (OR 1.21, 95% CI 1.03-1.42). This remained significant after adjusting for age, ethnicity, education level, smoking and deprivation (AOR 1.25, 95% CI 1.05-1.47). White British women with obesity also had higher odds of anxiety compared with recommended weight (OR 1.18, 95% CI 1.00-1.39). However, after adjustments this result was no longer significant (AOR 1.13, 95% CI 0.95-1.35). There was no significant association between anxiety and early pregnancy BMI among South Asian women. Interaction analysis found no significant association between ethnicity and antenatal anxiety within any BMI category(Table 4) .
Other exposure variables were shown to be associated with antenatal depression and anxiety within White British and South Asian women(Table S3). For White British women, maternal age (AOR 0.98, 95% CI 0.97-0.99) and education (no education AOR 2.12, 95% CI 1.63-2.76) were significantly associated with depression, and smoking was significantly associated with both depression (AOR 1.32, 95% CI 1.12-1.56) and anxiety (AOR 1.34, 95% CI 1.14-1.57). For South Asian women, maternal age (AOR 1.02, 95% CI 1.01-1.04) and education (no education AOR 0.65, 95% CI 0.55-0.78) were significantly associated with anxiety, and smoking was significantly associated with both depression (AOR 2.08, 95% CI 1.49-2.91) and anxiety (AOR 2.87, 95% CI 2.02-4.07).