# Associations between long-term exposure to air pollutants and mortality risk of critically ill patients: a multi-center cohort study in central China

## TL;DR

Long-term exposure to air pollutants increases mortality risk in critically ill patients, with ozone being the biggest contributor.

## Contribution

First study to examine air pollutant exposure and mortality in critically ill patients using a novel LightGBM model for pollutant estimation.

## Key findings

- PM2.5, PM10, NO2, SO2, O3, and CO were all positively associated with increased mortality risk in ICU patients.
- Ozone (O3) contributed the most (43.8%) to the joint effect of pollutants on mortality risk.
- Stronger associations were observed in subgroups like males, smokers, alcohol drinkers, and patients with higher SAPS II scores.

## Abstract

Air pollutants have been known as the most persistent environmental risk factors of all-cause mortality in general populations. However, few studies focused on such associations in critically ill patients who usually suffer from multiple comorbidities and even organ dysfunctions, and thus have lower resistance to external risk factors. For the first time, this study examined associations between long-term exposure to air pollutants and mortality risk of critically ill patients, also relative contribution of each pollutant to their joint health effect.

The 7,562 critically ill patients admitted to intensive care units (ICU) in a Hubei Province Medical Treatment Alliance in China were used in this study. Patient’s death within 28 days after ICU admission was used as the outcome. Daily concentrations of air pollutants, including PM2.5, PM10, NO2, SO2, O3 and CO, over their residence were estimated at a spatial resolution of 1 km by a newly developed multi-output LightGBM model, with better accuracy than all existing products. Logistic regression models were fit to estimate associations between individual air pollutants and mortality risk. Weighted quantity sum (WQS) regression was used to estimate relative contribution of each air pollutant to their joint effect on mortality risk.

The 7,222 patients were included in the study and had a mortality rate of 39.1%, with about half staying in ICU for ≤ 6 days. An increased risk for mortality was associated with a higher concentration of PM2.5 (OR = 1.007 [1.003, 1.011]), PM10 (OR = 1.002 [1.000, 1.004]), NO2 (OR = 1.020 [1.015, 1.024]), SO2 (OR = 1.025 [1.001, 1.050]), O3 (OR = 1.005 [1.001, 1.009]), and CO (OR = 4.336 [2.952, 6.457]). These associations varied across subgroups. For example, stronger associations were observed in males (PM2.5: OR = 1.010 [1.005, 1.015], PM10: OR = 1.004 [1.001, 1.007], NO2: OR = 1.026 [1.021, 1.032], and CO: OR = 6.224 [3.867, 10.019]), smokers (SO2: OR = 1.132 [1.078, 1.189], O3: OR = 1.014 [1.006, 1.022]), alcohol drinkers (SO2: OR = 1.147 [1.082, 1.215], O3: OR = 1.020 [1.010, 1.029]), and patients with a SAPS II of > 33 (SO2: OR = 1.168 [1.130, 1.207], CO: OR = 3.557 [2.165, 5.843]). The largest contribution to their joint effect on mortality risk was from O3 (43.8%), followed by NO2 (25.1%), CO (20.9%), PM2.5 (9.1%), SO2 (1.0%), and PM10 (0.1%).

Exposure to air pollutants was positively associated with the mortality risk of critically ill patients, with O3 being the main contributor to their joint effect. The findings would help multiple stakeholders, including researchers, physicians, and policy-makers, better understand health effects of air pollutants on critically ill patients, also serve as justifications for facilitate environmental justice and health equity.