shows unadjusted characteristics by gender and ethnicity for the main ethnic groups. Ethnic minority groups had lower unadjusted FEV1 and FVC than whites. Pakistani/Bangladeshi girls had higher FEV1/FVC. Compared with whites, Black Caribbeans and Black Africans were taller, mainly due to disproportionately longer LL. With the exception of Pakistani/Bangladeshi boys, South Asians were shorter, mainly due to shorter UBS. Black African, Indian, and Pakistani/ Bangladeshi girls were less likely to report asthma than white girls. Black Caribbean boys and girls and Black African girls were more likely to be in late puberty than whites.
BASELINE STATISTICS OF THE DETERMINANTS OF ADOLESCENT SOCIAL WELL-BEING AND HEALTH SAMPLE
Compared with whites, Black African and Pakistani/Bangladeshi boys were more disadvantaged by the standard of living measure, and boys and girls from these groups were more likely to live in an overcrowded household. Indian boys and girls and Black African girls were less likely to be smokers than whites. Reported exposure to passive smoking at home was generally relatively lower in ethnic minorities. Exceptions included Pakistani/Bangladeshi boys and girls and Black Caribbean and Indian girls, among whom the reported levels of paternal smoking were similar to that for whites.
show ethnic differences in FEV1 and FVC associated with the different height measures, adjusted for age, asthma status, room temperature, and pubertal stage. Regardless of which measure was adjusted for in the models, ethnic minority groups had significantly lower FEV1 and FVC relative to whites. The largest reduction in ethnic differences in FEV1 and FVC was achieved by adjusting for differences in UBS. On adjustment for SH alone, the largest ethnic differences in FEV1 and FVC were seen for Black Africans and Black Caribbeans, and the smallest ethnic differences were seen for Pakistani/Bangladeshis. With adjustment for UBS instead of SH, these differences attenuated, more so for the Black African origin groups. Compared with adjustment for SH, adjusting for UBS led to a further reduction in ethnic differences in FEV1 for boys of 48% for Black Caribbeans, 51% for Black Africans, 29% for Indians, and 39% for Pakistani/Bangladeshis. The corresponding figures for girls were 48% for Black Caribbeans, 41% for Black Africans, 26% for Indians, and 32% for Pakistani/Bangladeshis. Similar sized further reductions were observed for FVC after adjustment for UBS rather than SH (). The R2 values were also highest for models that included UBS or SH. Adjustment for LL was associated with an increase in ethnic differences in lung function.
EFFECT OF ADJUSTMENT OF DIFFERENT HEIGHT MEASURES ON ETHNIC DIFFERENCES IN FEV1 AND FVC
Late pubertal stage was a significant correlate of FEV1 and FVC in models with or without height components, but the effect was reduced with the addition of SH or UBS. For example, in a model with age, room temperature, and asthma status, FEV1 in boys in late puberty was greater by 4.9% (95% CI, 5.9–9.8) compared with those in pre- or early puberty. When UBS was added to the model, this advantage was reduced to 1.8% (95% CI, 0.4–3.2), indicating a partial effect via UBS. BMI was a significant correlate of FEV1 (coefficient 0.05%; 95% CI, 0.03–0.07 for boys; coefficent, 0.07%; 95% CI, 0.05–0.10 for girls, also adjusted for UBS) and FVC (coefficient, 0.07%; 95% CI, 0.050.09 for boys; coefficient, 0.09%; 95% CI, 0.07–0.12 for girls). The addition of BMI to each of the models presented in did not alter the size of the ethnic differences in FEV1 and FVC and did not change the R2 values.
To assess whether social exposures could operate via height to affect lung function, we examined their impact on the different height measures. Several significant relationships were observed, with some distinct differences by gender and measure of height. For example, overcrowding was associated with shorter SH (coefficient, −1.1 cm; 95% CI, −1.9 to −0.3), UBS (coefficient, −0.5; 95% CI, −0.9 to −0.1), and LL (coefficient, −0.5; 95% CI, −1.0 to −0.1) among boys and with UBS among girls (coefficient, −0.4; 95% CI, −0.8 to −0.003). Perceived relationship with parents was inversely related to measures of height among girls: SH (coefficient, −1.2; 95% CI, −2.0 to −0.4), UBS (coefficient, −0.6; 95% CI, −1.0 to −0.2), and LL (coefficient, −0.6; 95% CI, −1.1 to −0.1). Compared with those born in the United Kingdom, boys (coefficient, −0.6; 95% CI, −1.2 to −0.1) and girls (coefficient, −1.2; 95% CI, −1.7 to −0.7) born abroad had shorter LL. Girls born abroad also had shorter SH than those born in the United Kingdom (coefficient, −1.1; 95% CI, −2.0 to −0.3). Tobacco exposure was not associated with any of the height components.
Given that UBS seemed to explain more of the ethnic differences in FEV1 and FVC than other height measures, all subsequent multivariable analyses were based on UBS. shows the additional effect of adjustment for generational status, tobacco exposure, SES, and psychosocial variables on FEV1 and FVC. The addition of generational status, tobacco exposure, and SES to the baseline model with UBS generally led to some reduction in ethnic differences in FEV1 and FVC, although the R2 values changed little. Compared with a baseline model with SH, the adjustments for generational status, tobacco exposure, and SES reduced ethnic differences in FEV1 further for Black Carribean boys (6%), Black Caribbean girls (8%), Black African boys (3%), Black African girls (10%), Indian girls (7%), and Pakistani/Bangladeshi girls (9%). This reduction was mainly due to overcrowding. The addition of psychosocial variables did not lead to further reduction in ethnic differences. There were no significant interactions between ethnicity and height, pubertal, or social variables in these models.
EFFECT OF STEPWISE ADJUSTMENT FOR SOCIAL AND PSYCHOSOCIAL VARIABLES ON ETHNIC DIFFERENCES IN FEV1 AND FVC
In the fully adjusted models in , overcrowding was an independent correlate of FEV1 (coefficient, −1.6%; 95% CI, −3.0 to −0.3) and FVC (coefficient, −1.7%; 95% CI, −3.1 to −0.3) among boys. Poor psychological well-being was also an independent correlate of FEV1 in boys (coefficient, −0.1%; 95% CI, −0.3 to −0.02) and girls (coefficient, −0.1%; 95% CI, −0.3 to −0.000) and with FVC in boys (coefficient, −0.2%; 95% CI, −0.3 to −0.1).