Skip to main content
. Author manuscript; available in PMC: 2026 Jul 10.
Published in final edited form as: Environ Int. 2026 May 9;212:110297. doi: 10.1016/j.envint.2026.110297

Fig. 3.

Fig. 3.

Age-varying average marginal effects (AMEs) of prenatal and postnatal HAP on HAZ. Curves were derived from linear mixed-effects single-pollutant models with quadratic time terms, and exposure-by-time interactions terms. Prenatal CO was modeled on the log2 scale; prenatal PM2.5 using a natural spline (df = 3); and postnatal exposures on the linear scale. Marginal effects represent the change in HAZ per unit increase in exposure (or per doubling for log2-transformed CO), evaluated at the population level. For the spline model, effects were evaluated at the median exposure level. Models were adjusted for asset index, maternal age, maternal BMI, ethnicity, second-hand tobacco smoke exposure, parity, season of birth, child sex and prenatal CO (for postnatal CO models) or prenatal PM2.5 (for postnatal PM2.5 model). Curves were averaged over categorical covariates (frequency-weighted); continuous covariates were fixed at the mean or median. Shaded areas indicate 95% confidence intervals. CO – carbon monoxide; PM2.5 – fine particulate matter; HAZ – height-for-age z-score.