Investigation of the structure and magnitude of time-varying uncontrolled confounding in simulated cohort data analyzed using g-computation.

Time-varying uncontrolled confounding has the potential to substantially impact observed effect estimates. Given the importance of longitudinal studies in advising public health, the impact of time-varying uncontrolled confounding warrants more recognition and evaluation using quantitative bias analysis.