Olli Saarela

PhD

University of Toronto

Assistant Professor

NLST (Learn more about this study)

NLST-158

Sep 10, 2015

Time-specific measures of benefits due to screening

The proportional reduction in cancer-specific mortality has been recognized as the definitive measure of benefits due to cancer screening, and this is also how the results from randomized screening trials are commonly reported. However, the cumulative mortality reduction measure is not entirely unproblematic as the object of inference, since it depends on the number and schedule of the screening examinations used in a particular trial, and the length of the follow-up period. Also, it has been argued that the mortality impact in a randomized screening trial is time-dependent; it appears after some delay from the initiation of the screening and gradually wanes after the screening has been discontinued. However, detection of such a delay, if indeed present, is complicated by the small number of deaths early on in the trial, where the participants are asymptomatic at the outset. Instead of mortality reduction, an alternative approach to model the screening impact would be to focus on the gained lifetime. In this work we explore time-specific measures for the impact of screening in randomized screening trials, and statistical methods for estimating and visualizing these. In particular, we focus on statistical modeling of the gained lifetime due to screening. We apply the methods to the NLST data to investigate the time-specific effect of screening on lung cancer mortality.