Predict disease-specific mortality for lung cancer in NLST sub-groups
Using the NLST data, we first create a parametric scoring system as a function of multiple baseline covariates to estimate mortality difference in comparing two screening tests in sub-groups. Based on this scoring system, we specify a desired level of this difference and obtain a subgroup of patients, defined as those whose estimated scores exceed this threshold. An empirically calibrated threshold-specific mortality difference curve across a range of score values is constructed. The subpopulation of patients satisfying any given level of screening benefit can then be identified accordingly. To avoid bias due to over-optimism, we utilize a cross-training-evaluation method for implementing the above two-step procedure. We then show how to select the best scoring system among all competing models. Furthermore, for cases in which only a single pre-specified working model is involved, inference procedures are proposed for the average mortality difference over a range of score values using the entire data set, and are justified theoretically and numerically. Note that the new procedure created by Zhao, Wei, etc (2012) can be quite useful for the management of future participants in screening trials, so that screening may be targeted towards those who would receive nontrivial benefits to compensate for the risk or cost of the new screening.
The aims of this study is to investigate whether low-dose CT is effective for reducing lung cancer mortality in comparing with standard chest X-ray for sub-group analysis, such as gender, age, smoking and prior lung disease; and did lung cancer specific mortality vary by these co-variates?
Christine Berg, MD, NLST project officer
LJ Wei, PhD, Harvard University
Lihui Zhao, PhD, Northwestern University