6. Health effects of exposure to ozone

6. Health effects of exposure to ozone

6.1 Health significance of exposure to ozone

As ozone is a powerful oxidant it can react with a wide range of cellular components and biological materials, and may affect tissues of the respiratory tract or lung. The accumulated evidence on health effects is reviewed in WHO (1987) and WHO (1995). Most of the evidence on effects of ozone in concentrations common in the troposphere relates to short term (hours) exposures. Impacts in populations residing in areas where ozone levels are elevated for prolonged periods are more difficult to be detected and are still uncertain.

The relationship that is relatively best established and quantified is that of pulmonary function to ozone exposure (Rombout, 1997). Studies indicate that exposures to ozone concentrations in the range 160 - 360 µg.m^-3 for a period of 1-8 hours reduces various pulmonary function parameters. Concentrations in the lower part of this range are often observed in Europe, whereas those in the upper part are detected only occasionally. After cessation of exposure, lung function returns to pre-exposure level. The ozone concentration and the dose rate (depending on the ventilation rate) are more important determinants of the effect than duration of exposure.

Impacts on incidence of respiratory symptoms (including cough, throat irritation, chest tightness) have also been reported at the same range of exposures as those affecting pulmonary function though the effects are rarely correlated on an individual basis. Ozone also induces inflammatory response in the lung. Epidemiological studies indicate that the rates of asthma attacks and medication usage increase on days with higher ozone concentrations. The rates of visits to hospital emergency rooms and hospital admissions for asthma and other respiratory conditions are also increased on such days.

The recently revised WHO Air Quality Guidelines have established a guideline value for ambient air of 120 µg.m^-3 for a maximum period of 8 hours per day as a level at which acute effects on public health are likely to be small (WHO, 1997). However, as the available data show, a small part of the population may experience some impacts on their health even at ozone concentrations around the guideline value. The risk of these outcomes can be estimated on the basis of the tables provided with the guidelines. For example, 8 hour exposure to ozone at concentration of 120 µg.m^-3 is expected to induce 5% (transient) decrement in FEV1 (pulmonary function parameter) in the active, healthy, most sensitive 10% of young adults and children; at 160 µg.m^-3 this decrement is expected to be over 10% of normal level, which might be of health significance. Furthermore, an increase in 8-hour average ozone concentration by 100 µg.m^-3 is expected to induce:

• 25% increase in symptom exacerbation among adults and asthmatics involved in normal activities;
• 10% increase in hospital admissions for respiratory conditions.

The above increases in risk of health effects assume linear relationships between the ozone concentrations and health effects in the range of concentration currently observed in ambient air. It must however be recognised that a great deal of uncertainty exists in respect to the shape of these relationships. It is likely that the health benefits of reduction of exposures to high concentrations are greater than the same absolute reduction of exposures in the lower concentration range.

A limited number of studies have analysed relations of short-term changes in mortality to ozone concentration. Anderson et al. (1996) reported a 3% increase in all cause mortality associated with a 50 µg.m^-3 increase of 8-hour mean ozone in London, in the warm season. Similar association, also for a warm season only, is shown in a study in Barcelona (Sunyer et al., 1996). An analysis of mortality in the elderly during a hot summer in Belgium indicates that the observed relationship depends on the range of temperatures and can be seen mainly on hot days (Sator et al., 1997). Several other studies have not, however, confirmed these associations, especially after consideration of confounding effects of pollution with inhalable particles. Due to the lack of consistency between the studies and no agreement concerning causality of the observed associations, it was decided to omit possible effects of ozone on daily mortality rates in the analysis which is presented in further sections.

Although chronic exposure to ozone can cause effects, quantitative information from humans is inadequate to estimate the extent of impacts of long-term exposure.

6.2 Estimation of health impact of exposure to ozone in 15 EU countries

All health outcomes which can follow exposure to ozone at the concentrations encountered in Europe are rather non-specific and, in most cases, are also caused by a variety of other causes than air pollution. Therefore, a direct determination of the magnitude of the pollution impact is not possible. However, the proportion of the cases which can be attributed to the pollution can be estimated on the basis of information on population exposure to the pollution and using information from epidemiological studies on exposure - response relationships.

Estimates of population exposure to ozone

For the present assessment, the 1995 data on 8-h average ozone concentrations collected in the framework of the European Council Directive (92/72/EEC) have been used. Unfortunately, according to the Directive, only the data from days with ozone concentration exceeding 110 µg.m^-3 are collected, and the information on the frequency of days in lower concentration ranges was not available for this analysis. It has been assumed that each 'urban/street' monitoring location (400 sites) is spatially representative of a circle with a 10 km radius. Overall, some 11.7% of the 15 EU countries' population (ca. 41 million people) live within such circles. This proportion varies from 0.3% in France to 35.4% in Germany. Based on the population density in the circles, the 8-h average concentration data has been used to estimate the frequency distribution of person-days exceeding 110 µg.m^-3 using 10 µg.m^-3 ranges of O₃ concentration. In the period March - October 1995, in all 15 countries, the 8-h average ozone concentration exceeded 110 and 120 µg.m^-3 for 11.2% and 8.2% of person-days respectively. The level of 160 µg.m^-3 was exceeded during 1.6% of person-days. Figure 20 illustrates an estimate of the frequency distribution of the percentage of person-days in certain concentration ranges in the summer of 1995 in the EU15 countries.

The EU wide ozone network being used in this assessment is composed from various national networks. Unfortunately, these national networks were not designed using homogeneous station siting criteria. Error! Reference source not found. and section 5.2 illustrate that some countries use a large fraction of 'kerb-side' street sites to monitor urban ozone. Due to local emissions suppressing ozone levels at these sites it is likely that the real level of population exposure is underestimated, in the south of Europe in particular. This, in turn, leads to underestimation of effects.

Figure 20: The distribution of person-days with 8-h average concentration ranges above 110 mg.m^-3 in EU15 estimated over March-October 1995.

Estimation of health impact of ozone in 15 EU countries

The exposure to ozone in concentrations observed in Europe may lead to a range of health effects. In a sensitive part of population, transient decrements of lung function and exacerbation of respiratory symptoms such as cough and wheezing can be expected. As mentioned in section 6.1, at least 10% decrement in lung function is expected to occur in 10% of the most sensitive subject on days with 8-hour mean concentrations exceeding 160 µg.m^-3. According to the exposure estimates, 1.58% of person-days belonged to that range in March-October 1995 in the areas covered by ozone monitoring. Therefore, one can estimate that some 65 thousand cases could have occurred in the monitored areas, or more than 500,000 if the estimate is generalised to the whole population of the 15 EU countries. While the 10% transient decrement of FEV1 should not create problems for most people, the capacity for normal activities may be decreased in those subjects whose respiratory capacity has been already reduced due to age, disease or other factors.

In some cases, the symptoms may lead to use of medication or even to hospitalisation. The association between daily changes in the number of hospital admissions and concentration of ozone have been observed in a number of studies, including the APHEA study analysing data from five large European cities (Anderson et al., 1997). For the present analysis, the estimates of risk from that study have been used to calculate the proportion of hospital admissions attributable to ozone exposure. According to the definitions used by the APHEA study, we have considered emergency hospital admissions for bronchitis, emphysema and chronic airways obstruction. Although these admissions are not labelled as "emergency" in all countries, they are, in most cases, caused by symptoms requiring immediate medical assistance.

Based on the estimated exposure distribution and relative risk estimates from the APHEA study, we calculated that 0.3% (3 in 1000) of the admissions could be attributed to the ozone exposure exceeding 110 µg.m^-3 in all 15 EU countries. More than 80% of the excess cases could be attributed to ozone concentrations in the range of 110-170 µg.m^-3 (Figure 21). The proportion of admissions attributed to the ozone exposure was estimated to be highest in Belgium, France and Greece, where it exceeded 0.5% of the admissions (Figure 22).

Figure 21: Cumulative distribution of the percentage of excess hospitalisations due to ozone exposure in the EU15 (March-October, 1995)

To estimate the absolute number of admissions attributable to ozone exposure, it is necessary to assume the average frequency of hospitalisation in the population. Hospital admission is used here as an index of deterioration in health requiring medical treatment. The need to use this form of assistance in case of acute respiratory symptoms depends on the system of medical services and medication use practices. This may vary between populations and, certainly, between countries. Therefore, the underlying rates of emergency hospital admissions may vary substantially. However, such data is not readily available for all countries considered and, for the present analysis, we used the frequency observed in London (20 cases per day in a 7.3 million population, i.e. 2.74 admissions per million per day).

In all populations covered by ozone monitoring (i.e. living within 10 km circles around the monitoring locations), the number of additional hospital admissions attributable to ozone exposure in concentrations exceeding 110 µg.m^-3 is estimated to be just over 80 cases in all 15 EU countries, in the period March - October 1995. Most of the attributable cases would have been registered in Germany where the greatest number of people has been covered by O₃ monitoring. If it is assumed that the exposure situation around the monitors represents the overall distribution of exposure in the country, the estimates for the monitored populations can be extrapolated to all country population, and to all 15 EU countries. Such extrapolation indicates that almost 700 hospital admissions could have been attributed to the ozone in concentrations exceeding 110 µg.m^-3 in all 15 EU countries in the period March - October 1995. More than 75% of all those additional cases would have occurred in three countries: France, Italy and Germany.

Effects of the exceedance of 110 µg/m³ vs. total effects of exposure to ozone

The available data from ozone concentration monitoring allow to estimate the effects of the exceedance of 110 µg.m^-3. However, epidemiological studies have registered an increment in the frequency of hospital admissions with ozone concentrations well below this level. Similar observations were made by the APHEA study (Ponce de Leon, 1996). Therefore, it can be assumed that the real impact of the pollution is greater than that estimated above. A conservative estimate, based on typical distribution of ozone concentrations, assuming that for 20-40% person-days ozone concentration was in the range 60-110 µg.m^-3, suggests that the proportion of hospital admissions attributable to ozone concentration over 60 µg/m3 can reach 1.5% of all admissions (i.e. amount to 400 admissions in the populations covered by the monitoring, or to more than 3000 admissions in all population of 15 EU countries).

Figure 22: Proportion of hospital admissions attributable to 8-hour average ozone concentrations above 110 mg.m^-3 in each of the EU15 Member States (March-October, 1995)

Uncertainties in the estimates

There are several uncertainties in the above estimates. Probably the most reliable estimates are the attributable proportions when applied to the populations living in areas surrounding the monitors providing ozone concentration data. The APHEA study, the source of relative risk estimates, has analysed data from five different cities, and the combined estimate should reflect average exposure - response association in other populations than those considered in the present analysis. The tables and graphs presented indicate the magnitude of possible error related to the uncertainty of the relative risk estimate obtained by the APHEA study. Considering this, the proportion of hospital admission attributable to ozone in concentration over 110 µg/m3 should be in the range from 0.15% to 0.45%. However, the uncertainty concerning the exposure pattern within the (arbitrarily selected) 10 km circle around the monitoring station is not reflected in the present calculations. If the number of person-days with ozone concentrations over 110 µg/m3 (at each concentration category) is twice as big as the number estimated above, the attributable proportion would double as well.

A further important limitation of this analysis is the fact that a large part of the air quality data has been obtained from 'kerb-side' street sites. This data may provide very inaccurate estimates for the quality of air that people breathe and severely underestimate real level of exposure to ozone. Secondly, the extrapolation of the exposure distribution from the populations living close to the monitors to the whole country is an uncertain step. It is unlikely that the scarce monitoring data represents the total population exposure. There are also other errors of estimation. Therefore interpretation of data for individual countries must be done with caution, especially for these countries with small proportions of population covered by air quality monitoring.

Transient decrements in lung function and increases in hospital admission rates are the health outcomes for which the associations with exposure to ozone in concentrations observed in Europe are relatively the best established and quantified. Therefore the estimation of the magnitude of the impacts was possible, and performed, in this analysis. Other health effects, such as respiratory symptoms or short term changes in mortality rates could also result from ozone exposure in the period analysed but the information necessary for the estimation of effect is not sufficient for the quantification of the impacts.