Abstract
Introduction:
Assessing the effects of air-pollution is a significant problem in the field of modern environmental epidemiology. When modeling these effects it is important that the models must be epidemiologically meaningful and robust (that is, insensitive to variations in the model parameters). The objective of this paper is to propose a methodology for the assessment of the health impact of air pollution. The proposed methodology involves the construction of models for complex dynamic hierarchical systems in environmental epidemiology and their problem-oriented interpretation.
Methods:
The principal stages of the proposed methodology are:
Creation of a multivariate hierarchical structural model based on system analysis.
Generation of a mathematical formalization for this model.
Development of a statistical model for a particular study case based on the mathematical formalization, using the generalized estimating equations technique and time-series analysis. At this stage, for a dichotomized dependent variable, a special fuzzy algorithm was used. The algorithm employed fuzzy membership functions instead of the binary variable to obtain robust regression models.
Use of the “multi-layered” approach for model interpretation developed by the authors. This approach involved the creation of special functional time-dependent coefficients that reflect the effect of air pollutants at a given time. These coefficients allow an epidemiological meaningful model interpretation. Thus, they can be used for air-pollution health effects assessment.
Results:
The proposed methodology was used to analyze data collected from lung function measurements in 165 children from February-September 2002 (about 4000 individual daily records). The subject variables were age, gender, body-mass index, and place of residency. The meteorological variables included daily maximum temperature, average humidity and barometric pressure. The air-pollutant variables were suspended particles, ozone, nitrogen and sulphur oxides. In addition, the effects were studied up to a 3-day lag. The results demonstrated a statistically significant effect of air-pollution on lung function. Changes in most of the pollutants did not cause a critical decrease in lung function. However, for the observed period, a 10 mkg/m3 increase in ozone was associated with a mean decrease in lung function of 6 units for a one-day delay.
Discussion and Conclusions:
The assessment of the health effects of air pollution and their interpretation make epidemiological sense, lending support to the correctness of the proposed methodology. Testing the models by changing the dichotomization cutoff for the lung function variability shows that the models based on the proposed fuzzy algorithm are robust.
Assessing the effects of air-pollution is a significant problem in the field of modern environmental epidemiology. When modeling these effects it is important that the models must be epidemiologically meaningful and robust (that is, insensitive to variations in the model parameters). The objective of this paper is to propose a methodology for the assessment of the health impact of air pollution. The proposed methodology involves the construction of models for complex dynamic hierarchical systems in environmental epidemiology and their problem-oriented interpretation.
Methods:
The principal stages of the proposed methodology are:
Creation of a multivariate hierarchical structural model based on system analysis.
Generation of a mathematical formalization for this model.
Development of a statistical model for a particular study case based on the mathematical formalization, using the generalized estimating equations technique and time-series analysis. At this stage, for a dichotomized dependent variable, a special fuzzy algorithm was used. The algorithm employed fuzzy membership functions instead of the binary variable to obtain robust regression models.
Use of the “multi-layered” approach for model interpretation developed by the authors. This approach involved the creation of special functional time-dependent coefficients that reflect the effect of air pollutants at a given time. These coefficients allow an epidemiological meaningful model interpretation. Thus, they can be used for air-pollution health effects assessment.
Results:
The proposed methodology was used to analyze data collected from lung function measurements in 165 children from February-September 2002 (about 4000 individual daily records). The subject variables were age, gender, body-mass index, and place of residency. The meteorological variables included daily maximum temperature, average humidity and barometric pressure. The air-pollutant variables were suspended particles, ozone, nitrogen and sulphur oxides. In addition, the effects were studied up to a 3-day lag. The results demonstrated a statistically significant effect of air-pollution on lung function. Changes in most of the pollutants did not cause a critical decrease in lung function. However, for the observed period, a 10 mkg/m3 increase in ozone was associated with a mean decrease in lung function of 6 units for a one-day delay.
Discussion and Conclusions:
The assessment of the health effects of air pollution and their interpretation make epidemiological sense, lending support to the correctness of the proposed methodology. Testing the models by changing the dichotomization cutoff for the lung function variability shows that the models based on the proposed fuzzy algorithm are robust.
| Original language | English |
|---|---|
| Pages (from-to) | S497 |
| Journal | Epidemiology |
| Volume | 17 |
| Issue number | 6 |
| DOIs | |
| State | Published - 1 Nov 2006 |
