Causal relationship between outdoor atmospheric quality and pediatric asthma visits in hangzhou

Many air pollutants and climate variables have proven to be significantly associated with pediatric asthma and have worsened asthma symptoms. However, their exact causal effects remain unclear. We explored the causality between air pollutants, climate, and daily pediatric asthma patient visits with a short-term lag effect. Based on eight years of daily environmental data and daily pediatric asthma patient visits, Spearman correlation analysis was used to select the air pollutants and climate variables that correlated with daily pediatric asthma patient visits at any time (with a lag of 1-6 days). We regarded these environmental variables as treatments and built multiple- and single-treatment causal inference models using the Dowhy library (a Python library for causal inference by graphing the model, quantitatively evaluating causal effects, and validating the causal assumptions) to estimate the quantitative causal effect between these correlated variables and daily pediatric asthma patient visits in lag time. The multiple-treatment causal inference model was a model with 8 treatments (Visibility, Precipitation, PM₁₀, PM_2.5, SO₂, NO₂, AQI and CO), 1 outcome (daily pediatric asthma patients visits), and 5 confounders (Humidity, Temperature, Sea level pressure, wind speed and unobserved confounders "U"). Single-treatment causal inference models were 8 models, and each model has 1 treatment, 1 outcome and 12 confounders. Spearman correlation analysis showed that precipitation, wind speed, visibility, air quality index, PM_2.5, PM₁₀, SO₂, NO₂, and CO were significantly associated variables at all times (p < 0.05). The multiple-treatment model showed that pooled treatments had significant causality for the short-term lag (lag1-lag6; p < 0.05). Causality was mainly due to SO₂. In the single-treatment models, visibility, SO₂, NO₂, and CO exhibited significant causal effects at any one time (p < 0.05). SO₂ and CO exhibited stronger positive causal effects. The causal effect of SO₂ reached its maxima (causal effect = 11.41, p < 0.05) at lag5. The greatest causal effect of CO appeared at lag3 (causal effect = 10.67, p < 0.05). During the eight year-period, the improvements in SO₂, CO, and NO₂ in Hangzhou were estimated to reduce asthma visits by 8478.03, 3131.08, and 1341.39 per year, respectively. SO₂, NO₂, CO, and visibility exhibited causal effects on daily pediatric asthma patient visits; SO₂ was the most crucial causative variable with a relatively higher causal effect, followed by CO. Improvements in atmospheric quality in the Hangzhou area have effectively reduced the incidence of asthma.