Acute effects of air pollution in Brindisi (Italy): a case-crossover analysis

Objective: to evaluate the association between daily air pollutant concentration and daily data regarding mortality and hospital admissions among residents of Brindisi (Southern Italy) in the years 2003-2006 Design: the association between the time series of daily mortality (2003-2005) and hospital admissions (2003-2006) and the time series of daily pollutant concentration were analyzed using a case-crossover method and a conditional logistic regression. Bi-directional control periods were selected using a time-stratified approach. Models include mean temperature, relative humidity, influence of epidemics, summer decrease of resident population and holidays as confounders. Specific models with the following variables: cause of death or hospital admission, gender, age and season have been fitted. As hazard periods the following lags have been considered. single lag (from 0 to five) and cumulative lag (lag 0-1 for mortality and 0-3 for hospital admissions). Main outcome measures: a total of 1,792 subjects deceased of all natural causes (including cardiovascular and respiratory causes) and 6,925 hospital admissions for acute conditions (cardiac, cerebrovascular and respiratory diseases) were considered. PM10, NO2 and CO daily pollutant concentration series were examined. Results: PM10 was associated with mortality from all natural causes (10.36%; 95% CI 1.83-19.61 at lag 0-1). The risk was more pronounced for cardiovascular mortality (14.35%; 95% CI 2.11; 28.07 at lag 1). The association with hospitalization for cerebro vascular diseases was statistically significant for PM10 among females (13.4%; 95% CI 1.7; 264 at lag 4) and elderly over 75 years old (13.6%; 95% CI 0.4; 28.6 at lag 4). In specific population groups, increased mortality and hospital admissions have been associated with NO2. Conclusion: this study found strong and consistent associations between outdoor air pollution (coining from both industrial emissions and urban traffic) and short-term increases in both mortality and morbidity. Precautionary measures should be taken. (Epidemiol Prev 2010; 34(3): 100-107)