Long-term exposure to ambient fine particulate matter chemical composition and in-hospital case fatality among patients with stroke in China

Abstract

Background: There is little evidence on the association between PM_2.5 chemical components and fatality among hospitalized stroke patients.

Methods: This study used an inpatient discharge database from 2013 to 2019 in four provinces (Sichuan, Shanxi, Guangxi, and Guangdong) in China. Annual average exposure to PM_2.5 and its five chemical components [black carbon (BC), organic matter (OM), sulphate ( ${SO}_4^{2-}$ ), nitrate ( ${NO}_3^{-}$ ), and ammonium ( ${NH}_4^{+}$ )] were estimated using bilinear interpolation at patient's residential address. Mixed-effects logistic regression models were conducted to estimate the odds ratios (ORs). Counterfactual analyses were used to estimate the population attributable burden (PAF).

Findings: Among 3,069,093 hospitalized patients with stroke, each interquartile (IQR) increment in PM_2.5 and its chemical components was significantly associated with stroke fatality: the ORs were 1.137 [95% confidence interval (CI): 1.118-1.157; IQR: 15.14 μg/m³] for PM_2.5, 1.108 (95% CI: 1.091-1.126; IQR: 0.71 μg/m³) for BC, 1.086 (95% CI: 1.069-1.104; IQR: 3.47 μg/m³) for OM, and 1.065 (95% CI: 1.048-1.083; IQR: 2.81 μg/m³) for ${SO}_4^{2-}$ . We did not find significant associations for ${NO}_3^{-}$ (OR: 0.991, 95% CI: 0.975-1.008; IQR: 3.30 μg/m³). The associations were larger among patients with ischemic stroke than those with hemorrhagic stroke. The PAFs were 10.6% (95% CI: 9.1-12.2%) for BC, 9.9% (95% CI: 8.2-11.7%) for OM, and 6.6% (4.9-8.3%) for ${SO}_4^{2-}$ .

Interpretation: Ambient BC, OM, and ${SO}_4^{2-}$ might be important risk factors for stroke fatality. The findings advocate the need to develop tailored guidelines for PM chemical components and curb the emissions of the most hazardous chemical components.

Funding: Bill & Melinda Gates Foundation (INV-016826).

Keywords: Chemical components; China; Fine particulate matter; In-hospital mortality; Stroke.

Fig. 1

Spatial distribution of sample provinces (1a.) and patients' residential addresses in four Chinese provinces (1b-1e).

Fig. 2

A flowchart of sample selection, inclusion or exclusion criteria, and exposure measurement. API: application programming interface; TAP: tracking air pollution in China.

Fig. 3

Bilinear interpolation method to estimate ambient air pollution at patient's residential address (P) using the nearest four grids. (a) The four colored squares denote the gridded raster data for air pollution, the four black points are their center points, and G₁₁, G₁₂, G₂₁, G₂₂ are the concentration of air pollution in each grid. P (longitude: x_P; latitude y_P) is the location of patient's residential address. (b) The concentration of air pollution estimated using bilinear interpolation method within the four center points.

Fig. 4

Pairwise Pearson correlation coefficients between annual concentrations of ambient PM_2.5 and its chemical compositions [black carbon (BC), organic matter (OM), sulphate (${\mathrm{S}\mathrm{O}}_4^{2-}$), nitrate (${\mathrm{N}\mathrm{O}}_3^{-}$), and ammonium (${\mathrm{N}\mathrm{H}}_4^{+}$)] prior to the day of hospitalization.

Fig. 5

Exposure-response relationships of annual concentrations of(a)ambient PM_2.5and its chemical compositions [(b)black carbon (BC),(c)organic matter (OM),(d)sulphate ($\mathbf{S}{\mathbf{O}}_4^{2-}$),(e)nitrate ($\mathbf{N}{\mathbf{O}}_3^{-}$), and(f)ammonium ($\mathbf{N}{\mathbf{H}}_4^{+}$)] with stroke in-hospital case fatality. Annual concentrations of PM_2.5 and its chemical compositions were assessed as 365-day averages prior to the day of hospitalization using 10 ∗ 10 km grids. The solid curves represent the odds of in-hospital fatality, and the shaded bands are the associated 95% confidence intervals; the gray bars are the histogram showing the statistical distribution of PM_2.5 and its chemical compositions in the study sample. PM_2.5: particulate matter <2.5 μm in aerodynamic diameter.

Fig. 6

(a)Population attributable fraction of in-hospital case fatality attributable to ambient PM_2.5and its chemical compositions [black carbon (BC), organic matter (OM), sulphate ($\mathbf{S}{\mathbf{O}}_4^{2-}$), nitrate ($\mathbf{N}{\mathbf{O}}_3^{-}$), and ammonium ($\mathbf{N}{\mathbf{H}}_4^{+}$)] in patients with strokeand (b) the statistical distributions and the fifth percentiles of PM_2.5 and its chemical compositions.