A mathematical model of case-ascertainment bias: Applied to case-control studies nested within a randomized screening...

Area of Science:

• Epidemiology
• Biostatistics
• Medical Screening

Background:

• Screening can lead to differential case-ascertainment probabilities between screened and unscreened individuals.
• This lead-time-biased case-ascertainment (LTBCA) can occur in risk-factor studies within randomized screening trials.
• Uptake of screening may vary across risk-factor strata, even with random allocation of the screening intervention.

Purpose of the Study:

• To simulate and compare underlying versus observed cumulative incidence under LTBCA.
• To evaluate the potential for screening bias in observational lung cancer studies using a derived mathematical model.
• To quantify the impact of LTBCA due to chest x-ray screening on lung cancer risk estimates associated with smoking.

Main Methods:

• Simulated studies based on the Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial.
• Developed a mathematical model to account for LTBCA under specific assumptions.
• Compared simulated cumulative incidence with and without LTBCA.

Main Results:

• LTBCA due to chest x-ray screening can significantly increase estimated lung cancer risk due to smoking (1% to 50%).
• Simulations demonstrated differential screening impacts observational study estimates.
• Traditional adjustment methods are insufficient to correct for LTBCA.

Conclusions:

• LTBCA is a significant bias in observational studies nested within screening trials.
• Screening's influence on risk estimates extends beyond the study observation window.
• Accurate risk assessment in observational studies requires accounting for LTBCA.