Computer simulation of randomized cancer screening trials to compare methods of estimating lead time and benefit time

Abstract

Screening studies provide information on lead time of the screening test (time by which diagnosis is advanced) and benefit (defined either as a reduction in overall mortality, or as the time by which survival has been extended by virtue of the screen), particularly when these studies are randomized trials. Several methods have been proposed to estimate average lead time and average benefit time for the population which is offered screening. When applied to data from actual screening trials, the performance of these methods is uncertain, since the true averages are unknown. A simulated randomized screening trial offers flexibility in choosing the parameters of an actual trial, including distributions of preclinical and clinical disease, test sensitivity, actual benefit time, and correlation between any two of these parameters. By adjusting these parameters, we gain information on their joint effects on proposed estimators. This paper demonstrates this method of evaluation for estimators of average lead time and average benefit time and proposes avenues for generalizing these types of investigations.