Potential effects of time-varying characteristics and discrepancies in covariates distribution upon interpreting the results from the association of air pollution and physical activity with cardiovascular disease among young adults.