Fig. 2: Univariate dose‒response relationship between postnatal phthalates and bisphenols and childhood allergic conditions.

Models obtained with generalised estimating equations. Dose-response relationships are displayed with continuous covariates held at their mean values and categorical covariates set to their reference category. P-values for non-linearity obtained with Wald tests on spline terms. Models adjusted for cohort membership, maternal age, ethnicity, parental education, marital status, family history of asthma, sex, prenatal tobacco smoke exposure, season of birth, breastfeeding duration, age at outcome assessment, postnatal smoke exposure and gestational age. All exposures are modelled as estimated daily intakes, except MCPP, which is modelled using biomarker concentrations.