The Association between Daily PM₂.₅ Exposure during Forest Fire Period and Inpatient Admissions for espiratory Diseases in Northern Thailand: A Case-crossover Analysis

Abstract

For decades, PM₂.₅ has been one of the most concerned public health problems in many countries. PM₂.₅ exposure from traffic and industry contributes the risk of developing respiratory diseases but still lack of epidemiological knowledge in areas where biomass burning fire are dominant. This research analyzed an acute association between PM₂.₅ and respiratory disease hospital admission in forest fire areas of the northern Thailand. This study covered 816,139 cases of respiratory system inpatients (codes J00-J99) from 118 hospitals in 9 provinces from 2016 to 2020. The data was obtained from the Ministry of Public Health. PM₂.₅, co-pollutants and meteorological data were obtained from 15 monitoring stations provided by the Pollution Control Department. Pollutants and weather exposure were assigned to hospital locations using their nearest monitoring station, performed by SAS®. A time- stratified case-crossover approach with conditional logistic regression used to estimate the association and odd ratio and confidence interval (OR and CI) per an interquartile range (IQR) increase of PM₂.₅. The case crossover design controlled personal and time-dependent confounders by matching while the multivariate regression model was fit to control other confounders (PM₁₀, ozone, temperature, and relative humidity) in R® program. Results showed PM₂.₅ well correlated with PM₁₀ (ρ= 0.89, p < 0.001) likely from same sources of traffic and forest fire and ozone also correlated with PM₂.₅ (ρ = 0.78, p < 0.001) as its precursors possibly originated from traffic sources. The IQR of PM₂.₅ was 25.87 μg/m³ (9.63 - 35.50 μg/m³) with a max of daily mean of 398.13 μg/m³, a max of daily PM10 mean of 438.88 μg/m³ and a max of 8-hour ozone of 128.50 ppb, all far exceeding their standards. During the hazard fire period (mid of Febuary to mid of May), 5-year daily means of PM₂.₅, PM₁₀ and ozone were observed high at 56.18 μg/m³, 81.73 μg/m³and 62.70 ppb respectively. The PM₂.₅ IQR rise was found associated with the increases of the inpatient admission for respiratory diseases. In total period (fire and non-fire), ORs were found increased but not significant at lag3 (OR = 1.002, CI: 0.994 to 1.011, p < 1), for female at lag1 (OR = 1.013, CI: 1.000 to 1.026, p < 0.1), and for male showing no increased risk. In hazard fire period, ORs were increased significantly at higher level at lag1 (OR = 1.058, CI: 1.041 to 1.075, p < 0.001), for male at lag2 (OR = 1.057, CI: 1.035 to 1.079, p < 0.001) and for female at lag1 (OR = 1.082, CI: 1.055 to 1.108, p < 0.001).For subgroup respiratory conditions, increased signifiant risks were observed for acute upper respiratory infections at lag3 (OR = 1.121, CI: 1.059 to 1.186, p < 0.001), influenza and pneumonia at lag0 (OR = 1.048, CI: 1.019 to 1.078, p < 0.01) and other acute lower respiratory infections at lag7 (OR = 1.069, CI: 1.015 to 1.127, p < 0.05) and other diseases of respiratory system at lag3 (OR = 1.060, CI: 1.012 to 1.111, p < 0.05). ORs for age groups were found highest in elderly ≥ 60 yr at lag1 (OR = 1.070, CI: 1.047 to 1.093, p < 0.001). and in children ≤ 6 yr. at lag7 (OR = 1.055, CI: 1.018 to 1.093, p < 0.01). For age-sub disease risk, in eldery it was siginificantly increased in influenza and pneumonia at lag0 (OR = 1.061, CI: 1.020 to 1.103, p < 0.01) and in other acute lower respiratory infections at lag7 (OR = 1.129, CI: 1.013 to 1.258, p < 0.05) and in children in acute upper respiratory infections at lag3 (OR = 1.116, CI: 1.031 to 1.208, p < 0.01) in influenza and pneumonia at lag7 (OR = 1.059, CI: 1.004 to 1.117, p < 0.05). These reported statistically significant risks were only detected in the forest fire period and can not be noticed in total period. These significant risks during the fire period by sex, sub-disease, and age have confirmed that elevated PM₂.₅ played an important role in increasing risk of hospitalization for respiratory diseases in northern Thailand. These significant group-specific risks suggested that before the hazard fire period, central and local authorities need extra specific PM₂.₅ fire abatement, more efficient precaution and respiratory protection for community and greater hospital preparation. Future research could improve this analysis by controlling more inpatient confounders and more pollutants as well as combining cardiopulmonary diseases.