Negative Binomial regression model for analysis of the relationship between hospitalization and air pollution

Özet

Air pollution is currently one of the major environmental risk to human health in the world. Several studies linkhealth to air pollution through the Poisson regression model. However, this study showed that the best method isthe regression model with Negative Binomial distribution. The analysis was made based on daily data for theperiod between 2010 and 2014 regarding hospitalizations due to circulatory and respiratory diseases, stratifiedby age due to the concentration of air pollutants: CO, PM10, NO2, SO2 and O3 and meteorological variables. Theresults with Negative Binomial distribution model indicate, for interquartile range (IQR), that CO is a risk factor(1.00259–1.19688, 95% CI) for circulatory diseases in the elderly (aged over 64) and for respiratory diseases inthe population aged 10–19 years old, while NO2 is a risk factor for the population in the range of 20–64 years(1.00313–1.25827, 95% CI) in the city of Araucaria. In Curitiba, air pollutants presenting a risk factor forrespiratory diseases were PM10 (RR = 1.05322) in the range of 0–9 years old, CO and O3 (RR = 1.05047) in therange of 20–64 years old, and CO for the elderly. It could be verified that exposure to air pollutants is harmful tohealth even when their concentrations are below the standards set out in legislation. Additionally, the city ofAraucaria presented the highest number of hospital admissions, which may be associated with the large numberof industries and lower human development index in the city