Home » Ozone » Level 3 » Question 2

Air Pollution Ozone

2. How does Ozone (O3) affect human health?

  • 2.1 Effects of long-term exposure to levels of Ozone observed currently in Europe
    • 2.1.1 Chronic effects at current Ozone levels
    • 2.1.2 Effects on mortality at current Ozone levels
  • 2.2 Is Ozone per se responsible for effects on health?
  • 2.3 Are health effects of Ozone influenced by the presence of other air pollutants?
  • 2.4 Characteristics of individuals that may influence how Ozone affects them
  • 2.5 Is there a threshold below which nobody’s health is affected by Ozone?

2.1 Effects of long-term exposure to levels of Ozone observed currently in Europe

    • 2.1.1 Chronic effects at current Ozone levels
    • 2.1.2 Effects on mortality at current Ozone levels

2.1.1 Chronic effects at current Ozone levels

The source document for this Digest states:


There are few epidemiological studies on the chronic effects of ozone on human health. Incidence of asthma, a decreased lung function growth, lung cancer and total mortality are the main outcomes studied. At levels currently observed in Europe, the evidence linking O3 exposure to asthma incidence and prevalence in children and adults is not consistent. Available evidence suggests that long-term O3 exposure reduces lung function growth in children. There is little evidence for an independent long-term O3 effect on lung cancer or total mortality.

The plausibility of chronic damage to the human lung from prolonged O3 exposure is supported by the results of a series of chronic animal exposure studies.


Incidence of asthma was studied in adults (228), in California in the AHSMOG (Adventist Health Smog) study. A cohort of 3 091 non-smokers, aged 27 to 87, was followed over15 years (1977 to 1992). For males, a significant relationship between reports of doctor diagnosed asthma and 20-year mean 8-hour average ambient ozone concentration (relative risk = 2.09 (1.03–4.18)) for an inter-quartile range of 54 µg/m3 (27 ppb) was found. Use of alternative O3 metrics, such as hours above a certain level, showed the strongest effect in relation to the mean ozone concentration, followed by 8-hour average concentration and then by hours exceeding a certain threshold value. No relationship was observed in females. Adjustment for other pollutants did not diminish the strength of the relationship. The prospective nature, a small loss to follow-up and a detailed measure of the cumulative air pollution exposure (incorporating accurate individual interpolations using residence and work location (229) strengthen the validity of these results. The small number of cases (32 males and 79 females), a potential misclassification of self-reported diagnosis, an imprecise time-pattern in the measures of outcome and past-exposure (residence was measured three times in 15 years), and a lack of consistency between the two genders undermines the validity of results. Hence, low quality of outcome diagnosis might not allow a clear distinction between incidence and exacerbation. Although gender differences were not among the prior hypotheses but were a result of subgroup analyses, the lack of effect in females could be a phenomenon due to a differential exposure by gender.

In a cohort study of 3 535 children, aged 10 to 16 years with no history of asthma recruited in 12 communities in the Southern California study and followed during 5 years, the relative risk of developing asthma among children playing three or more sports (8 % of the children) was 3.3 (1.9–5.8) compared with children playing no sports in communities with high ozone concentrations (four year average of 112 µg/m3 to 138 µg/m3 (56 to 69 ppb)), but not in communities of low ozone (230). This effect modification of ozone was not seen for the other standard pollutants. The longitudinal nature of the study and the low proportion of subjects lost during follow-up strengthen these results. In the same study in Southern California, prevalence of asthma was not associated with ozone levels among the 12 studied communities (231). On the contrary, prevalence of asthma increased with average levels of O3 among the 2 445 13 to 14 year-old children of 7 communities participating in the French ISAAC study (International Study on Childhood Asthma and Allergy) (232), and among the 165 173 high school students aged 11 to 16 from 24 areas in the Taiwan ISAAC study (233), but in the French study, analysis at the individual level did not show an association. However, difficulties in diagnosis of asthma using self-reporting of symptoms and limitations of prevalence studies with no control of in/out migration could explain these differences.

Lung function growth was studied in three prospective studies with repeated measures in the same subjects. In nine areas without major industrial sites in Austria, 1 150 children aged 8 to 11 were followed during 3 years (1994–1997) performing 6 lung function tests (234). The change in lung function (FVC, FEV1 and MEF50) between the pre and post-summer test was negatively associated with the O3 mean concentration (with a personal interpolation). A 10 ppb (20 µg/m3) difference in average O3 exposure was associated with a small but significant predicted decrement of 2 %. The wintertime change in O3 was also negatively associated with the lung function change, but the association was weaker. The use of peak O3 concentrations instead of average O3 levels resulted in a non-significant association. The analysis of only those children who did not change their town of residence increased the association. Presence of asthma did not modify this association. A further analysis showed the effect of O3 to be independent of particles and nitrogen dioxide (235).

These results were not replicated by the first of the Southern California cohort studies (22). More than 3 000 children from 12 communities around Los Angeles were followed during 4 years (from 1993 to1997) and lung function tests were performed annually. A negative effect of O3 on lung growth was not observed. A low variation of O3 and a high variation in particulate matter among these Californian communities could explain the lack of the effect. However, a second study following 1 678 children of nine to ten years from 1996 to 2000 in the same 12 communities showed that exposure to O3 (expressed as the annual average of the concentration between 10 a.m. and 6 p.m.) was associated to reduced growth in peak flow rate (PEF), as well as to FVC and FEV1 growth among children spending more time outdoors (23). However, there was a greater negative association with acid vapours, NO2, and PM2.5 than for O3 in this cohort. The repeated measures among the same children give more validity to these studies than to the cross-sectional studies.

Cross-sectional studies are not fully consistent. In the same study in South California, lung function level was lower in communities with higher ozone in comparison to communities with lower O3 average levels, particularly among girls with asthma and spending more time outdoors (236). In a study on 24 communities in the United States and Canada among 10 251 children between the age of 8 and 12 a negative association with several O3 exposure metrics was found for FVC and FEV1, although the association with FVC was reduced after adjustment for strongly acidic particles. O3 and acidic particles were highly correlated in the study areas (78).

Among adults, in the 1 391 non-smokers of the AHSMOG study a decrement of FEV1 in relation to cumulative O3 exposure was observed in males whose parents had asthma (237), as well as in a sample of 130 UC Berkeley freshmen (238), while an association between O3 and lung function was not found in the 9 651 adults residing in the eight areas of the SAPALDIA study (Swiss study on air pollution and lung diseases in adults) (93). However, this study did not have adequate power to assess the O3 effect (range of long-term O3 average was 31 to 51 µg/m3 or 15.5 to 25.5 ppb).

Symptoms of bronchitis did not increase in children from communities with higher levels of O3 among the 3 676 children participating in the south California study (88), as they similarly did not increase among the 9 651 adults in the Swiss communities with higher O3 participating in the SAPALDIA study (94).

Lung cancer both incidence (239) and mortality (9) was strongly associated with long-term concentrations of ozone among males of the 6 338 non-smoking adults participating in the AHSMOG study and followed from 1977 to 1992. Differences in exposure to O3 (males in the study spent more time outdoors) could explain the gender differences. It was difficult to separate the effect from ozone and particles, since a similar association was obtained with particles and correlation between particles and ozone was high (9). The ACS cohort study (13) did not find any association of long-term O3 exposure and lung cancer or total mortality.

The plausibility of chronic damage to the human lung from prolonged O3 exposure is supported by the results of a series of chronic animal exposure studies, especially those in rats (240, 241) using a daily cycle with a 180 ppb (360µg/m3) average over nine hours superimposed on a 13-hr base of 60 ppb (120µg/m3), and those in monkeys of Hyde et al. (242) and Tyler et al. (243) applying 8 hours per day of 150 and 250 ppb (300 and 500µg/m3). The persistent cellular and morphometric changes produced by these exposures in the terminal bronchioles and proximal alveolar region and the functional changes are consistent with a stiffening of the lung reported by Raub et al, (244) and Tyler et al. (243).

Source & ©: WHO Regional Office for Europe  "Health Aspects of Air Pollution" (2003), Chapter 6 Ozone (O3), Section 6.2 Answer and rationales, Question 2

The same information on
Particulate MatterNitrogen Dioxide

2.1.2 Effects on mortality at current Ozone levels

The source document for this Digest states:

To what extent is mortality being accelerated by long and short-term exposure to Ozone?


Long-term O3 effects have been studied in two cohort studies. There is little evidence of an independent long-term O3 effect on mortality so that no major loss of years of life is expected. The issue of harvesting, i.e. the advancement of mortality by only relatively few days, has not been addressed in short-term exposure studies of O3.


For the long-term effects of O3 see the answer and rationale to Question 2. In short-term studies, the issue of harvesting, i.e., the advancement of mortality by only few days has not been studied for O3 effects. A few studies have addressed this issue for the effects of PM10 or PM2.5 and it was found that mortality displacement was substantial for most causes of death and harvesting could not explain all the excess mortality (see also answer and rationale to question 5 in the PM section). Whether there are also persistent effects of O3 as well, has not been determined.

Source & ©: WHO Regional Office for Europe  "Health Aspects of Air Pollution" (2003), Chapter 6 Ozone (O3), Section 6.2 Answer and rationales, Question 5, Question 2

2.2 Is Ozone per se responsible for effects on health?

The source document for this Digest states:


In short-term studies of pulmonary function, lung inflammation, lung permeability, respiratory symptoms, increased medication usage, morbidity and mortality, O3 appears to have independent effects (especially in the summer). For long-term effects the results are not entirely consistent. When particle acidity was studied, O3 effects were partly explained. A few studies in North America found effects of O3 on asthma incidence and functional changes independent of other classical pollutants, but acidity was not taken into account.

Experimental studies show the potential of O3 to cause these health effects.


Several short-term mortality studies adjusted O3 for particles, after including multiple pollutants in the same regression model. For a meta-analysis (210), 18 studies including O3 and particles in the same models were selected (including studies carried out in all seasons or summer and with various lag periods). Almost all estimates were positive (14 out of 18; 10 with a p<0.05), while only 2 were negative (and statistically not significant) and summarized estimates are very similar to those obtained without adjustment. These findings coincide with results from the NMMAPS study in 90 North-American cities during the summer. Adjustment for sulfate, SO22 or NO2 was done rarely, though in the few studies that incorporated these pollutants, the association of O3 was not modified after including the other pollutants in the regression model (247, 281, 282). For the effects reported to be associated with ambient O3 in population-based excess frequencies, the answer is less clear-cut, and other components in photochemical smog that elicit reactive oxygen species (ROS) in the cardiopulmonary system may also play a role. For hospital and emergency department respiratory admissions, O3 appears to be more influential than other pollutants that may have either additive or synergistic effects based on stronger association in multiple pollutant model analyses. However, emergency and hospital admissions are metrics that differ widely between countries. On the other hand, for excess daily mortality, O3 appears to have a lesser effect than fine particles (PM2.5).

In some of the long-term studies that adjusted for other classical pollutants (PM10, PM2.5, SO4, NO2 and SO22), O3 effects on incidence of asthma were independent (228, 230). Similarly, some studies on lung function growth also have adjusted for other pollutants and found that O3 had independent effects on several functional markers (22, 23, 235). In the UC Berkeley study, lifetime O3 exposure was negatively associated with mid and end-expiratory flows even after adjusting for particles and NO2 (238). In the study of the 24 cities (78) adding particle acidity concentration in a two-pollutant model reduced the effect of daily mean O3 on FVC (although a negative effect of O3 persisted). However, the study was designed to measure effects of acidity, and levels of O3 among communities were probably not sufficiently heterogeneous.

For the functional and symptom responses, which have been identified in both controlled exposure studies and in field studies at comparable concentrations of O3 the observed effects can be clearly attributed to O3 per se.

Finally, the case of lung cancer appeared different since in the AHSMOG study there was a strong correlation between particulate matter and O3, and a similar association in the single pollutant models was observed for particulate matter and for O3. The American Cancer Society Study (13) did not implicate O3 in the long-term effects on mortality.

Source & ©: WHO Regional Office for Europe  "Health Aspects of Air Pollution" (2003), Chapter 6 Ozone (O3), Section 6.2 Answer and rationales Question 6

2.3 Are health effects of Ozone influenced by the presence of other air pollutants?

The source document for this Digest states:


Epidemiological studies show that short-term effects of O3 can be enhanced by particulate matter and vice versa. Experimental evidence from studies at higher O3 concentrations shows synergistic, additive or antagonistic effects, depending on the experimental design, but their relevance for ambient exposures is unclear. O3 may act as a primer for allergen response.


Synergy between O3 and particles (or other pollutants) has been measured in very few epidemiological studies. In a follow-up of more than 2 000 children during the first 6 months of 1996 in the Southern California Children’s Health Study (283), the short-term effect of O3 on school absenteeism was stronger in periods with low particulate levels than with high particulate levels. In the APHEA2 study (284) the effect of daily particle concentrations on respiratory hospital admissions among those over 65, was stronger in areas with high O3 levels.

The evidence based on studies comparing the functional decrements induced by exposures to O3 in combination with acid aerosols and NO2 in ambient air to those induced by O3 as a single pollutant in inhalation chambers (253) suggest a synergism of O3 with the co-pollutants at levels known not to produce significant effects as single pollutants in controlled exposures. Frampton et al. (214) have shown a synergistic functional effect of O3 with H2SO4 in controlled exposures of both healthy and asthmatic subjects.

The best evidence for synergy between O3 and other pollutants like NO2 or H2SO4 comes from controlled short-term exposures in laboratory animals, that have generally been made at concentrations much higher than those occurring in recent years in ambient air. The endpoints considered include lesions in the gas-exchange region of the lungs, enzyme activities etc. However, the effects can be synergistic, additive, or antagonistic, depending on the combination of the pollutants and their concentrations (227), exposure regimen (concomitant or sequential) as well as on the health endpoints considered. The pollutant combinations studied include O3 with H2SO4, (NH4)2SO4, HNO3, HCHO or cigarette smoke.

There is evidence that O3 exposure potentiates the functional and inflammatory responses to inhaled allergen in subjects with pre-existing allergic airway disease (Jorres et al, 1996; Holz et al, 2002 (224, 285)."

Source & ©: WHO Regional Office for Europe  "Health Aspects of Air Pollution" (2003), Chapter 6 Ozone (O3), Section 6.2 Answer and rationales, Question 8

2.4 Characteristics of individuals that may influence how Ozone affects them

The source document for this Digest states:

Are effects of Ozone dependent upon the subjects’ characteristics such as age, gender, underlying disease, smoking status, atopy, education etc?

What are the critical characteristics?


Individuals vary in their O3 responsiveness for different outcomes, for reasons which remain largely unexplained but appear to be partly based on genetic differences. There is some evidence that short-term O3 effects on mortality and hospital admissions increase with age. Gender differences are not consistent. It appears that the effects of O3 exposure on symptoms are greater in asthmatic children. Lung function decrements are more consistent in asthmatic children, especially those with low birth weight. One important factor modifying the effect of O3 on lung function is ventilation rate. As tidal volume increases, O3 penetrates deeper into the lungs. Duration of exposure is also a critical factor: Ozone effects accumulate over many hours but after several days of repeated exposures there is adaptation in functional but not inflammatory responses. The effects of O3 exposure on lung function, symptoms and school absences are larger in children who exercise more or spend more time outdoors.


In two studies, the effect of O3 on daily mortality has been higher in persons older than 65 years (247, 262). There are very few time-series studies on mortality that have addressed effects in persons younger than 65 to allow confirmation of this finding. In the meta-analysis of time series studies on emergency admissions for respiratory causes, there was a pooled positive association with all respiratory admissions at all ages (mainly including old people) while there was no association for studies with asthma in children (210), probably reflecting that the risk of O3 effect increased with age. By contrast, lung function decrements attributable to O3 have been much greater in children and young adults than in older adults (263).

Data from the National Cooperative Inner-City Asthma study (NCICAS) in the United States were analysed to identify susceptible subgroups (264). In a panel of 846 asthmatic children, morning peak-flow decrease and incidence of symptom increase were associated with O3 exposure in children with low birth weight or premature birth. In another study in children in Australia, children with doctor-diagnosed asthma or bronchial hyper-reactivity were at higher risk of functional responses (265). In the meta-analysis of acute effects on lung function, the negative effect of O3 on peak flow rates was more consistent among studies only including children with asthma compared to studies including general population children or healthy children, but the size of the association in some of the latter studies was even larger than in studies including only asthmatics (Table 3). Finally, the Southern California Children’s Health study shows that children exercising more (230) and children spending more time outdoors (23) are at higher risk of an effect of O3on asthma incidence and decrease of lung function growth, respectively.

Results on gender differences are not consistent. In the AHSMOG study in adults and the NCICAS study in children, long-term effects of O3 for several outcomes were only seen in males (9, 228, 237, 264). In both studies, this effect was attributed to a larger time spent outdoors by males, which was observed in the AHSMOG study. In contrast, the geographical comparison of lung function levels among the children participating in the Southern California study showed a larger effect of O3 among girls with asthma (236) although they stayed less time outdoors and less time exercising than boys. In general, a higher exposure might explain why boys are usually at higher risk of asthma symptoms and male adults at a higher risk of all respiratory effects. The gender and racial differences, when adjusted for lung capacity, if any, are much smaller (266, 267). Overall, it is debatable if there are gender differences of O3 effects on lung function and respiratory diseases.

Persons with underlying respiratory diseases have been found to have greater responsiveness to O3 associated function changes in some controlled exposure studies (268, 269), but not in others (270, 271). However, controlled exposure studies have involved only subjects with very mild disease. “Healthy” smokers tend to be less responsive than non-smokers (272), but this effect of smoking falls over time after successful smoking cessation (273). There are no data on the influence of education or other socio-economic variables on O3 associated changes in respiratory function. Other short-term responses to the inhalation of O3 have been investigated primarily in healthy, non-smoking young adults. These effects of O3 include: increases in lung inflammation (261, 269, 274); lung permeability (275); respiratory symptoms (258, 276)); and decreases in mucociliary clearance rates (277). Little is known about the influence of age, gender, underlying disease, smoking status, atopy or education and other socio-economic variables on these responses to O3.

A critical factor affecting O3 deposition and the induction of short-term functional responses is the duration of the exposure. O3 is a lower-lung irritant and effects accumulate over many hours. However, after several days of repetitive exposures, there is an adaptation that leads to a reduced respiratory functional responsiveness which lasts for a week or more (217, 278). While there is an adaptation at least in the larger airways, it does not seem to involve the functional responsiveness at the site of maximum O3 injury, i.e. respiratory bronchioles (223), and the lung inflammation responses (274). In any case, those responsive on one occasion are fairly reproducibly responsive when similarly exposed to O3 (279).

It is also noteworthy that individuals vary in their overall functional responsiveness to O3. For a group of 20 to 30 individuals exposed to 160 to 240 g/m3 (80 to 120 ppb) of O3 for 6.6 hrs while undergoing moderate exercise for 50 min each hr, there would be significant average decrements in lung function and significant average increases in symptoms and inflammatory cells in lung lavage, ranging between individuals from little or none in some of them to major changes in others. However, those who responded in terms of one endpoint did not necessarily respond in terms of the others. This is not unexpected, when the different mechanisms that underlie the responses are taken into account (e.g., inhibition of deep inspiration related to neurokinins and alterations in small airway function related to damage to respiratory bronchioles). Thus, the critical host characteristics for functional responses remain unknown and there are, as yet, no biomarkers that can reliably predict responsiveness in humans. It has been suggested that intrinsic narrowing of the small airways may be a significant component of the functional response (280).

Source & ©: WHO Regional Office for Europe  "Health Aspects of Air Pollution" (2003), Chapter 6 Ozone (O3), Section 6.2 Answer and rationales, Question 4

2.5 Is there a threshold below which nobody’s health is affected by Ozone?

The source document for this Digest states:


There is little evidence from short-term effect epidemiological studies to suggest a threshold at the population level. It should be noted that many studies have not investigated this issue. Long- term studies on lung function do not indicate a threshold either. However, there may well be different concentration-response curves for individuals in the population, since in controlled human exposure and panel studies there is considerable individual variation in response to O3 exposure. From human controlled exposure studies, which generally do not include especially sensitive subjects, there is evidence for a threshold for lung damage and inflammation at about 60 to 80 ppb (120–160 µg/m3) for short-term exposure (6.6 hours) with intermittent moderate exercise. Where there are thresholds, they depend on the individual exercise levels.


For epidemiological studies of short-term effects, the acute association with mortality for all causes has been shown in studies carried out in places with low mean O3 levels such as London (245) suggesting a lack of a threshold effect. There is no obvious trend towards larger associations in places with higher O3. For example, the association found in Mexico City (246) was relatively small in magnitude (although still significant). This may suggest a lack of a threshold. These paradoxical phenomena of a lower association in places with higher levels could be explained by other factors such as adaptation; occurrence of protective factors such as diet; the lower levels of other pollutants or modified activity patterns which often occur when ozone concentrations are high or fewer competitive risks such as a higher mortality due to infectious diseases.

Most studies do not explicitly describe the shape of the concentration-response function. Some studies suggest a curvilinear association, including one recently conducted in Canada (247) that suggested an inflexion to a steeper slope above around 25 [µ]g/m3 (13.5 ppb) as 24-hour average. Wong et al (248) in Hong Kong found a slight increase above about 40 µg/m3 (20ppb) as 8-hour average and Hong et al (249) in Korea a steeper increase above about 46 µg/m3 (23 ppb) (also 8- hour average). Hoek et al (250) found a chi-squared test for non-linearity was not significant and that there was little change in slope until all days above 30 µg/m3 (15 ppb) as 24-hour average had been removed. However, extrapolating from single studies has limitations, in comparison with meta-analysis. In addition, many of the concentration-response functions suggesting thresholds are for single pollutant models and confounding by other pollutants may vary across the concentration range.

Concentration-response curves are also rarely described explicitly in studies on respiratory hospital admissions. A study in London suggested a threshold around 80 to 100 µg/m3 (40 to 50 ppb) as 8-hour average (251) but other studies have shown linear associations (252).

For healthy young adults, the thresholds for the short-term effects of O3, as evaluated by spirometry markers of lung damage and inflammation, lie below 160 µg/m3 (80ppb), based on the effects observed in a series of controlled 6.6 hour human exposure studies at concentrations of 160, 200, and 240 ug/m3 with intermittent moderate exercise (221, 258, 259, 260, 261).

Long term Effects: An increase in asthma incidence in the AHSMOG study occurred when comparing the effect of the O3 exposure in the lowest tertile of 70 µg/m3 (35 ppb, as 8-hr average) with the second tertile (relative risk, RR = 4.4), and the magnitude of the association did not increase when comparing the first and the third tertile (RR = 4.0) (228) again suggesting a lack of threshold, or an effect at low exposures. The 70 µg/m3 (35 ppb) level in the AHSMOG study is lower than the median level of 102 µg/m3 (51 ppb) that was observed in the South California study that separated communities with and without an effect of sport on asthma incidence (230). In an Austrian study (234) on lung function growth the annual O3 concentration was lower than in California, ranging from 36 µg/m3 (18 ppb) to 81 µg/m3 (40.7 ppb) with an average summertime O3 of 70 µg/m3 (35 ppb) and a standard deviation of 17 µg/m3 (8.7 ppb). The best fitting dose-response function on the association between O3 and lung function in the Austrian study was a linear model, suggesting again a lack of a threshold level. Hence, the decrease in FEV1 did not vary when only exposures below the median of O3 exposure of 57 µg/m3 (28.6 ppb) were selected (-0.014) compared to when only levels above the median were chosen (-0.015). In the South California study (23) only linear models were tested, and other forms of the dose-response curve or the occurrence of threshold could not be evaluated.

Source & ©: WHO Regional Office for Europe  "Health Aspects of Air Pollution" (2003), Chapter 6 Ozone (O3), Section 6.2 Answer and rationales, Question 3

See also: General Issues and Recommendations on Air Pollutants:

Themes covered
Publications A-Z