Lung Cancer Prevalence in Virginia: A Spatial Zipcode-Level Analysis via INLA

Area of Science:

• Epidemiology
• Public Health
• Spatial Analysis

Background:

• Lung cancer (LC) presents significant public health challenges in Virginia (VA).
• Understanding LC prevalence drivers requires examining demographic, environmental, and socioeconomic factors.
• Spatial analysis at the zip code level is crucial for localized insights.

Purpose of the Study:

• To investigate the underlying drivers of lung cancer prevalence in Virginia zip codes.
• To adjust for spatial associations and account for missing covariate data.
• To provide insights for targeted public health interventions and resource allocation.

Main Methods:

• Bayesian hierarchical modeling using Integrated Nested Laplace Approximation (INLA).
• Spatial Poisson and negative binomial regression models incorporating Conditional Autoregressive (CAR) priors.
• Simultaneous imputation of missing covariates under latent Gaussian Markov Random Field (GMRF) assumptions.

Main Results:

• Elevated smoking indices and Social Deprivation Index (SDI) scores correlated with higher LC counts.
• Higher LC prevalence observed in zip codes with larger White and Black populations, and elderly populations (≥ 65 years).
• Lower LC counts in zip codes with higher Hispanic populations; higher prevalence in women compared to men.

Conclusions:

• Demographic and socioeconomic factors significantly influence lung cancer disparities at the zip code level in VA.
• Findings support the need for targeted public health strategies to address LC prevalence variations.
• Implementation code is publicly available on GitHub for further research.