Abstract:Purpose: While risk factors for lung cancer incidence are well established, our understanding of factors driving an increased risk of lung cancer among Black/African American populations remains incomplete. Methods: We assembled data from five U.S. lung cancer case-control studies matched on race, gender and age: the Southern Community Cohort Study, the Inflammation, Health, Ancestry and Lung Epidemiology study, the Northern California Lung Cancer Study, three Detroit area lung studies, and the NCI-Maryland Lung Cancer Study. Incident lung cancers were ascertained from state and regional cancer registries or National Death Index mortality records. Social, behavioral, and environmental factors were ascertained by in-person interviews or mailed-in questionnaires. Participant's residential addresses were geocoded and area-level area deprivation and environmental justice variables were linked at the census block. We evaluated interactions between known lung cancer risk factors and race using multilevel logistic regression analyses and novel tree-based approaches to identify potential factors contributing to racial disparities. Tree-based approaches included a conditional matched classification tree analysis designed to detect complex interactions, as well as a specialized partially recursively induced structured moderation (PRISM) analysis to study variation in disparity in lung cancer risk across levels of educational status. Results: Among 5,829 lung cancer cases and 10,671 controls (64% self-identified Black/African American and 36% White), the average age was 60 years, 50% were female, and 24% had less than a high school education. Conditional logistic regression models fit after imputation of missing data identified increased risk of lung cancer associated with personal history of cancer, family history of lung cancer, current smoking status, smoking duration, smoking pack-years, and higher neighborhood deprivation; whereas, reduced lung cancer risk was associated with higher education and increased BMI. Similar patterns were observed among Black/African Americans only, with the addition of secondhand smoke exposure. Assessment for interactions with race identified significant interactions between race and body mass index (BMI) (p-value < 0.0001), PM2.5 (p= 0.010) and gender (p=0.001) and lung cancer risk. A conditional matched classification tree analysis identified both increased and decreased risk subgroups defined by multi-way interactions of similar variables that were identified as significant main effects in the conditional logistic regression analysis. Smoking pack-years and BMI were identified as the most important determinants of heterogeneity of disparity in lung cancer risk across educational levels in our study population. Conclusions: The use of both traditional and novel tree-based based approaches provide approaches for identifying factors contributing to racial disparities. Future studies should assess the interplay of lung cancer risk factors to inform development of targeted interventions. Citation Format: Melinda C. Aldrich, Chen Zhao, Christine M Lusk, Lucie McCoy, James Jaworski, Michael Mumma, Catherine Pichardo, Michael Betti, John K Wiencke, Brid Ryan, Ann G Schwartz, Sunil Rao. A comparison of modeling approaches to assess the interplay of social, behavioral, and environmental factors on lung cancer racial disparities [abstract]. In: Proceedings of the 17th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2024 Sep 21-24; Los Angeles, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2024;33(9 Suppl) nr PR001.

Evaluating machine learning model bias and racial disparities in non-small cell lung cancer using SEER registry data

Fairness in Predicting Cancer Mortality Across Racial Subgroups

Improving lung cancer health equity by applying deep learning to low dose CT screening of minority and disadvantaged patients.

Assessing fairness in machine learning models: A study of racial bias using matched counterparts in mortality prediction for patients with chronic diseases

Racial and socioeconomic disparities in lung cancer screening in the United States: A systematic review

Healthcare disparities, screening, and molecular testing in the changing landscape of non–small cell lung cancer in the United States: a review

Racial Disparities in Young-Onset Patients with Colorectal, Breast and Testicular Cancer

Abstract PR001: A comparison of modeling approaches to assess the interplay of social, behavioral, and environmental factors on lung cancer racial disparities

Demographic bias in misdiagnosis by computational pathology models

Social and Biological Determinants in Lung Cancer Disparity

A hybrid-computer vision model to predict lung cancer in diverse patient populations

Simulating the population impact of interventions to reduce racial gaps in breast cancer treatment

Optimizing the fairness of survival prediction models for racial/ethnic subgroups: A study on predicting post-operative survival in stage IA and IB non-small cell lung cancer.

Racial and socioeconomic disparities in NSCLC molecular diagnostics uptake

Impact of Structural Racism and Social Determinants of Health on Disparities in Breast Cancer Mortality

Racial disparities in early-stage NSCLC treatment: A Call for Action

Racial and Ethnic Bias in Risk Prediction Models for Colorectal Cancer Recurrence When Race and Ethnicity Are Omitted as Predictors

Abstract C021: Racial and socioeconomic disparities in molecular diagnostics for NSCLC patients

Conceptualizing bias in EHR data: A case study in performance disparities by demographic subgroups for a pediatric obesity incidence classifier

Participation in Non-small Cell Lung Cancer Clinical Trials in the United States by Race/Ethnicity

Disparities in Lung Cancer Treatment