Air pollution and hospital admissions for cardiorespiratory diseases in Iran: artificial neural network versus conditional logistic regression|
Shakerkhatibi, M.; Dianat, I.; Asghari Jafarabadi, M.; Azak, R. & Kousha, A.
This study was conducted to evaluate the
relationship between air pollutants (including nitrogen
oxides [NO, NO2, NOX], sulfur dioxide [SO2], carbon
monoxide [CO], ozone [O3], and particulate matter of
median aerometric diameter<10 µm [PM10]) and hospital
admissions for cardiovascular and respiratory diseases. The
study had a case–crossover design which was conducted in
Tabriz, Iran. Daily hospital admissions and air quality data
from March 2009 to March 2011 were analyzed using the
artificial neural networks (ANNs) and conditional logistic
regression modeling. The results showed significant associations
between gaseous air pollutants including NO2, O3,
and NO and hospital admissions for cardiovascular disease.
Gaseous air pollutants of NO2, NO, and CO were associated
with hospital admissions for chronic obstructive pulmonary
disease, while PM10 was associated with
hospitalizations due to respiratory infections. PM10 and O3
were also associated with asthmatic hospital admissions.
There was no significant association between SO2 and
studied health outcomes. Comparing the results of logistic
regressions and ANNs confirmed the optimality of the
ANNs for detection of the best predictors of hospital
admissions caused by air pollution. Further research is
required to investigate the effects of seasonal variations on
air pollution-related health outcomes.
Case–crossover analysis; Cardiorespiratory health effects; Hospital admissions; Air pollution