Year

2006

Degree Name

Doctor of Philosophy

Department

School of Health Sciences

Abstract

Health effects caused by air pollutants may range from subtle biochemical or physiological signs, such as mildly reduced lung function, to difficult breathing, wheezing, coughing and exacerbation of existing respiratory conditions such as asthma, and chronic obstructive pulmonary disease (COPD). These effects can lead to school absenteeism, increased medication use, increased doctor or emergency room visits and more hospital admissions.

Lung function and respiratory symptoms have been the primary focus of most studies addressing the respiratory health effects of air pollution among children. Peak expiratory flow (PEF) has been found to be significantly lower in children exposed to air pollution. These studies have also reported higher prevalence of asthma and respiratory symptoms such as cough, phlegm and wheezing.

Most of the data collected in this study are in the form of time-series. The focus of this thesis is on relationships between time-series, but explicit ‘time-series’ methods for the analysis of univariate series have not been adopted. The case-crossover method is used to examine short-term effects of air pollution. In a symmetric bi-directional design, two control times are selected, e.g. two weeks before and two weeks after phenomena of interest, so as to reduce auto-correlation effects in the exposure series. In this study, dates of high respiratory related hospital admission rate, respiratory related school absenteeism rate on two consecutive days, or the occurrence in individuals of worst lung function are defined as ‘case days’ and two weeks after or before each date defined as ‘control’. In the case definitions, either a case date or case window was used with the corresponding two control dates or windows.

Two statistical methods were used for investigating the effects of air pollution on school absenteeism, hospital admission and lung function. Poisson regression is commonly used to model responses that are counts. In this study, the response was the absenteeism or hospital admissions per day, the predictors were air pollutants concentrations, weather parameters and other related factors. Time-series methods provide a traditional analytical approach in epidemiology studies but could not be used here because of the existence of too many missing values.

For the case-crossover approach, conditional logistic regression and hazard ratios were used to investigate the relationship between absenteeism, hospital admission, poor lung function or airway obstruction and daily temperature, air pollutant concentration and other covariates such as gender and teaching shift. For both methods, SAS statistical package version 9.1 was used.

The population was potentially exposed to ambient air concentrations of SO2 in excess of the WHO limit for 304 days in 2000, for the entire year of 2001, and for 241 days in 2002. Since, there is no limit for PM10 in WHO air quality standards it is compared to World Bank and other countries’ limits. In the period 2000-2002, the population of Tehran was also potentially exposed to ambient air concentrations of PM10 in excess of the World Bank limit for 25 days in 2002 and some courtiers such as Australia, Canada, EU, Philippines and Hong Kong limits for 327 days in 2000, 352 days in 2001 and 296 days in 2002. Exposure to ambient air concentration of NO2 was also potentially in excess of WHO standards for 132 days in 2000, 0 days in 2001 and 142 days in 2002. Potentially exposure to ambient air concentration of O3 was also there in excess of WHO standards just for 3 days in 2002. Likewise, ambient air concentrations of CO exceeded the WHO limit for 314 days in 2000, 277 days of 2001 and 272 days in 2002.

Absenteeism data over 295 days were obtained from two schools in Tehran, and was found to be associated with some pollutants using both Poisson and case-crossover analyses. Specifically, the strongest association found in this study was equivalent to an increase in daily absenteeism of 0.8 for the Poisson model, and an odds ratio of 10 for the probability of daily absenteeism for case-crossover analysis. However, absenteeism was also negatively associated with concentration of seven-day moving average of PM10, NO2 and O3. Such associations are not consistent with the literature.

Hospital admission data were obtained from two hospitals in Tehran over a 2-year period. After using both Poisson regression, addressing the full data set of hospital admissions, and a case-crossover analysis of the reduced set of cases and controls, the following results were found. Temperature and O3 (same day) were the only statistically significant predictors. However, the association between ozone and hospital admission was negative, which is not consistent with the literature.

This study also assessed the results of a survey using a standard questionnaire on students at two elementary schools. This demonstrated that some respiratory prevalence rates are higher in males than females. Around 22% of students have current wheeze, and 38% of the students who have current wheeze are passive smokers. In general, the level of exposure to known risk factors, with the exception of number of smokers at home, does not seem to be extraordinarily high in this population.

The overall prevalence rate of poor lung function (percentage of students whose lung function was less than 50% of their predicted value or best blow) found in this study is 12% using predicted value and 30% using best blow.

Lung functions of students were measured for six weeks. The lung function data were analysed using the case-crossover method, and it was found that a change between mean and maximum concentration of seven-day moving average of NO is predicted to lead an increases the probability of poor lung function (OR = 19 and OR = 80 using predicted value and personal best blow respectively. The concentrations of seven-day moving average of PM10 and CO were negatively associated with the probability of poor lung function, which is not consistent with the literature.

Overall, some significant negative effects of air pollution level on respiratory health were found in this study. Case-crossover analysis tended to confirm the general pattern indicated using Poisson regression. Some puzzling positive effects were found, and these are inconsistent with the literature. While there were some limitations in the quantity of data available (eg. lung function measurements which were only available for a six week period) the major limitation was of the quality of air pollution data, with many missing values as well as problematic measurements.

02Whole.pdf (8006 kB)

Share

COinS