Study
NLST
(Learn more about this study)
Project ID
NLST-1474
Initial CDAS Request Approval
Aug 25, 2025
Title
Endogenous Screening and Health Behaviors
Summary
The research will use detailed administrative data on patient smoking histories, smoking behavior, and lung cancer screening decisions and outcomes to identify the unique, endogenous relationships that exist between screening decisions, health information, patient health behaviors, and health outcomes. Utilization of and adherence to lung cancer screening guidelines is fundamentally different than other forms of screening, for at least two reasons: first, the clear link between behaviors such as smoking and lung cancer, and the associated stigma placed on smoking behaviors, leads to avoidance of screening for lung cancer among those most at-risk. Second, given this link, the outcomes of CT scans and other forms of screening provide patients with information about their health risk that directly influence health behaviors, such as the decision of whether or not to continue smoking. These factors complicate the analysis of ideal screening recommendations to improve patient and societal welfare. To parameterize a model lung cancer screening and cigarette smoking decisions, we need to understand transition probabilities to different lung cancer states given covariates such as age, smoking history, and current smoking behavior. We hope to use NLST data to estimate these transition probabilities.
Aims
* Formulate a dynamic model of lung cancer screening and cigarette smoking decisions. The model will follow in the structural economic tradition that captures dynamic tradeoffs between current utility and future health.
* Estimate parameters of the model using novel survey data on preferences, beliefs, and constraints; large electronic health registry data from Truveta; nationally representative data from the BRFSS; and cancer transitions using NLST data.
* Simulate the model under counterfactual policy scenarios address several important mechanisms. First, we will simulate how screening prevalence would change when we alter the out-of-pocket cost of screening. Next, we will simulate the model while changing the USPSTF guidelines that dictate insurance coverage. Then, we will simulate the model while changing the quality of screening through the false positive rate. Finally, we can use the model to assess the importance of stigma associated with screening.
Collaborators
Dr. Alex Hoagland (University of Toronto)