Learning about the effectiveness of lung cancer screening in real-world target populations
Principal Investigator
Name
Sarah Robertson
Degrees
PhD
Institution
Dartmouth College
Position Title
Assistant Professor
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-1488
Initial CDAS Request Approval
Dec 22, 2025
Title
Learning about the effectiveness of lung cancer screening in real-world target populations
Summary
This project is supported by an NCI R00 award and aims to estimate the effects of lung cancer screening strategies in nationally representative US target populations. The project will advance methods for combining datasets to learn about screening strategies in different populations, to answer the questions of how screening strategies compare across trials, and how screening strategies that differ from those used in trials compare. It will use transportability methods, simulation modeling, and predictive modeling, to compare screening strategies in clinically relevant target populations. This project will provide new insights on the comparative effectiveness of lung cancer screening strategies. As a result, simulation models for lung cancer screening will be more applicable to target populations of interest where interventions are applied.
Aims
We will use NLST data to address the following aims:
1. Indirect comparisons (Aim 1): We will use NLST data, along with either aggregated or individual patient data from other lung cancer screening trials, such as the PLCO (using the chest x-ray arm) and NELSON, to conduct population-adjusted indirect comparisons. We will estimate the comparative effectiveness of NLST-like protocols versus NELSON-like protocols and no screening within a target population defined by NHIS survey data, which is representative of the US population.
2. Simulation Model Calibration (Aim 2): We will use the NLST data to calibrate the natural history parameters of a microsimulation model. We will use the NLST x-ray arm to adjust parameters until the model reflects the observed cancer incidence and mortality rates.
3. Predictive Modeling and Risk Stratification (Aim 3): We will use NLST data to fit and validate risk prediction models to identify heterogeneity in screening benefits and evaluate transportability methods. Some of the modeling may involve developing new models in observational cohorts and validating it in NLST.
Collaborators
Sarah Robertson Dartmouth College