Gaseous Air Pollutants and Lung Function in Fibrotic Interstitial Lung Disease (fILD): Evaluation of Different Spatial Analysis Approaches

Abstract

Gaseous pollutants such as CO, NO₂, O₃, and SO₂ are linked to adverse clinical outcomes in patients with fibrotic interstitial lung diseases (fILDs), particularly idiopathic pulmonary fibrosis. However, the effect of various exposure estimation methods on these findings remains unclear. This study aims to evaluate three spatial approaches─nearest neighbor (NN), inverse distance weighting (IDW), and Kriging─for estimating gaseous pollutant exposures and to assess how these methods affect health outcome estimates in fILD patients. A 10-fold cross-validation showed that Kriging had the lowest prediction error compared to NN and IDW, with RMSE for CO = 0.43 ppm (11%), O₃ = 5.9 ppb (14%), SO₂ = 2.7 ppb (12%), and NO₂ = 7.6 ppb (9%), respectively. Kriging also excelled over other methods across wide spatial and temporal ranges, showing the highest spatial R² for CO and O₃ and the highest temporal R² for SO₂ and NO₂. In a large cohort of patients with fILD, higher levels of CO, SO₂, and NO₂ exposure were associated with lower pulmonary function. The magnitude of association and its precision were higher in SO₂ and CO estimated by the Kriging method. This study underscores Kriging as a robust method for estimating gaseous pollutant levels and offers valuable insights for future epidemiological studies.