Mechanistic studies in human cohorts are challenging due to the number of potential postnatal confounders and the time course required for prospective studies and thus remain largely observational.
These results were significant after extensive and rigorous control for multiple potential confounders, including several non-PCB contaminants (prenatal MeHg, DDE, HCB, and pre- and postnatal Pb).
These observations were robust to analytic control for multiple potential confounders including age, sex, insurance status, geography and hospital ownership.
Statistical adjustments for multiple potential confounders is needed in order to assess a potential causal relationship which would lay foundations for interventional studies.
We examined multiple potential confounders of the outcomes of interest.
The impact of azithromycin is independent of multiple potential confounders.
Adjusting for multiple potential confounders did not change the estimates.
The extension to multiple unmeasured confounders is straightforward.
Multiple potential confounders and other limitations preclude firm conclusions.
These associations were independent of multiple potential confounders and were mainly associated with severity of ulceration.
