The Global Burden of Ozone on Respiratory Mortality: No Clear Evidence for Association

Anenberg et al. (2010) estimated the global burden of respiratory mortality attributable to long-term ozone exposure based on a single observational study by Jerrett et al. (2009). Because no other study has clearly demonstrated impacts of chronic ozone exposure on deaths from respiratory-related causes, we believe that reliance on the study by Jerrett et al. to establish causality and global impact is misplaced and that the conclusions of Anenberg et al. are likely unfounded.

The Global Burden of Ozone on Respiratory Mortality: No Clear Evidence for Association doi:10.1289/ehp.1003276Anenberg et al. (2010) estimated the global burden of respiratory mortality attributable to long-term ozone exposure based on a single observational study by Jerrett et al. (2009).Because no other study has clearly demonstrated impacts of chronic ozone exposure on deaths from respiratory-related causes, we believe that reliance on the study by Jerrett et al. to establish causality and global impact is misplaced and that the conclusions of Anenberg et al. are likely unfounded. Jerrett et al. (2009) carried out a follow-up analysis of the American Cancer Society (ACS) cohort.Other ACS studies reported no associa tions between long-term ozone exposure and cardio pulmonary mortality that are robust to model inclusion of co-pollutants (e.g., Krewski et al. 2000;Pope et al. 2002).In addition, other long-term studies of ozone-related respiratory or cardio pulmonary mortality did not report positive associations (Goodman 2010;Health Effects Institute 2009).Anenberg et al. (2010) suggested that long-term respiratory mortality is plausible because short-term ozone mortality has been documented, but inconsistent evidence for an association between short-term ozone exposure and respiratory mortality indicates that this relationship is not well established.Jerrett et al. (2009) did not provide "clear" evidence of an association between long-term ozone exposure and respiratory mortality, as Anenberg et al. (2010) stated in their article.Jerrett et al. (2009) did not adequately control for potential confounding effects of particulate matter ≤ 2.5 µm in aerodynamic diameter (PM 2.5 ) for several reasons.Jerrett et al. (2009) used only 2 years of data for PM 2.5 (1999)(2000) but ozone concentrations from 1977-2000.Although ozone and PM 2.5 levels decreased considerably from 1977 to 2000, they used higher ozone levels observed in the past but only the more recent PM 2.5 levels.Furthermore, their ozone metric focused on daily maximum hourly levels in the warm seasons, whereas they used annual average PM 2.5 concentrations.As noted by Jerrett et al. (2009), this approach likely increased the potential to observe an association between ozone and mortality and decreased the ability to observe potential PM 2.5 confounding of this association.In addition, confounding by other co-pollutants (e.g., sulfur dioxide), a clear issue in earlier ACS analyses (Krewski et al. 2000), was not examined.Accordingly, Jerrett et al. did not demon strate an association between ozone and respiratory mortality that is independent of other co-pollutants.
Another aspect of the Jerrett et al. (2009) study that is inconsistent with an association between long-term ambient ozone exposure and respiratory mortality is the biologically implausible, inverse associations of ozone with cardio vascular and all-cause mortality.The magnitude of these associations is the same-although opposite in direction-as the risk estimate for respiratory mortality; thus, it is likely that associations of this magnitude are not indicative of a causal relationship.
It was inappropriate for Jerrett et al. (2009) to combine data across cities for a U.S. national risk estimate, given the known geographic hetero geneity of ozone-mortality findings (Goodman 2010).In addition, socioeconomic data (a potential confounder) was collected in 1982-1983 for the ACS study but never updated.For these reasons, the U.S. national risk estimate reported by Jerrett et al. (2009) should not be extrapolated globally.
The analysis by Anenberg et al. (2010) was based on an uncorroborated study that likely mis interpreted the findings regarding ozone effects.The utility of estimating the global burden of an effect based on a single study, for which no causal association has been established in other studies, is not apparent.Conclusions drawn from such an analysis should be interpreted with caution.
Funding for the writing of this letter was provided by the American Petroleum Institute.Prueitt and Goodman raise concerns about our use of chronic ozone mortality relative risk (RR) estimates from Jerrett et al. (2009) to estimate the global burden of outdoor ozone and fine particulate matter (< 2.5 µm in aerodynamic diameter; PM 2.5 ) on human mortality (Anenberg et al. 2010).We believe that our use of RR estimates from Jerrett et al. ( 2009) is justified and does not strongly affect our conclusions.Our goal of demonstrating the use of chemical transport models in estimating the global burden of outdoor air pollution on mortality is not affected by the choice of risk estimates.Further, using chronic RR estimates for ozone has only a minor effect on our mortality estimates, because the mortalities attributed to PM 2.5 are much greater than those for ozone.

The Global
We chose to use RR estimates from Jerrett et al. (2009) in our study (Anenberg et al. 2010) because they are consistent with the widely accepted RR estimates used for long-term PM 2.5 mortality (Krewski et al. 2009), as both are based on the American Cancer Society study cohort and capture delayed mortality effects (National Research Council 2008).
In response to particular criticisms, we note that while Jerrett et al. (2009) found the first significant positive association between chronic ozone exposure and mortality in a major cohort study, some previous smaller cohort studies have also found positive associations (National Research Council 2008).Biological plausibility for chronic ozone effects on respiratory mortality is evidenced by toxicology and human exposure studies that found that ozone affects airway inflammation, pulmonary function, and asthma induction and exacerbation (National Research Council 2008).Using earlier PM 2.5 data would be unlikely to affect confounding in the model, because using PM 2.5 data from 1979-1983 and 1999-2000 yields similar PM 2.5 mortality associations (e.g., Krewski et al. 2009).Jerrett et al. (2009) also found that socio economic data are not strong confounders and that The correspondence section is a public forum and, as such, is not peer-reviewed.EHP is not responsible for the accuracy, currency, or reliability of personal opinion expressed herein; it is the sole responsibility of the authors.EHP neither endorses nor disputes their published commentary.

Environmental Health Perspectives
A 159 using more recent data is unlikely to change that conclusion (see Appendix of Jerrett et al. 2009).Finally, national risk estimates are more applicable globally than city-specific estimates because they include larger and more diverse populations.
However, because the evidence for chronic ozone mortality is more limited than the large body of evidence demonstrating mortality associations with short-term ozone exposure, we present here estimates of the global burden of ozone on mortality using RR estimates from Bell et al. (2004), a large multi city study of short-term ozone mortality.We estimated mortalities daily using the difference between pre industrial and presentday 8-hr maximum ozone, and sum mortalities over the 1-year simu lation.We used the reported relationship for cardio pulmonary mortality and daily average ozone [0.64% (95% posterior interval, 0.31-0.98%)for a 10-ppb increase], and corrected to 8-hr ozone using the reported ratio between daily 8-hr and 24-hr average ozone associations with non accidental mortality.
Using these methods, we estimated 362,000 (95% confidence interval, 173,000-551,000) annual global pre mature cardiopulmonary deaths attributable to ozone, approximately 50% of the 700,000 pre mature deaths we calculated in our origi nal study (Anenberg et al. 2010).Since estimated deaths due to PM 2.5 (3.7 million) are an order of magnitude larger, using a short-term rather than long-term RR estimate for ozone has only a minor effect on the overall global burden of disease due to outdoor air pollution.As RRs for chronic ozone mortality are not as strongly supported as those for PM 2.5 , we expect that estimates of mortality burden will improve as research on chronic ozone exposure and mortality continues globally.Influence of Selenium and Mercury on Age-Related Cataracts in the Brazilian Amazon doi:10.1289/ehp.1003242

S.C
In their article, Lemire et al. (2010) provided important data on the frequency of agerelated cataracts among adults in the Brazilian Amazon and found a correlation between agerelated cataracts and whole-blood total mercury (Hg) concentrations and selenium (Se) levels in plasma and whole blood.However, in the "Discussion" of their paper, they stated that they "observed no adverse effects although Se concentrations were very high, reaching 1,500 µg/L for [blood]-Se and 913 µg/L for [plasma]-Se" (Lemire et al. 2010).However, they did not mention the potential and substantial adverse health effects associated with a high body Se burden.There are potentially adverse consequences to Se body burden, such as hair loss (alopecia), tooth decay, nail changes, peripheral paresthesias, weakness, skin lesions, and diabetes (Hira et al. 2004;Nuttall 2006;Shearer 1975;Stranges et al. 2010;Sutter et al. 2008;Yang et al. 1983).It would be valuable to know what Se-related adverse effects were observed by Lemire et al. (2010).Most studies, taken together, suggest a possible attenuation of Hg toxicity, probably as insoluble form of Hg selenide (Clarkson 2002;International Programme on Chemical Safety 1990).
Selenium may be able to delay the onset of toxic effects in animal models exposed to methylmercury in the diet (Clarkson 2002;Ganther et al. 1972).However, the Hg-Se inter action may not have an equivalent effect in some animals.There is evidence that coadministration of methylmercury and Se may lead to an important synergistic effect (Heinz and Hoffman 1998).In studies of persons who have been co-exposed to Hg and Se, toxic effects of Se should be taken into account to discover the potential synergistic effect between Se and Hg.
The authors declare they have no competing financial interests.