National Ambient Air Quality Objectives For Ground-Level Ozone- Summary - Science Assessment Document

9. Identification Of Reference Levels

Within the NAAQO process, Reference Levels are defined as a level or levels, above which there are demonstrated effects on human health and/or the environment. Conceptually, Reference Levels are LOAELS or NOAELs, where these can be defined. For non-threshold pollutants, for which LOAELs and NOAELs as traditionally conceived cannot be defined, statistical LOAELs can be used to help set the RLs. A statistical LOAEL represents the level (concentration) at which statistical significance in the concentration-response relationship is lost. Reference Levels may be identified for several receptors over specific time periods depending on the sensitivity of the target organism or material(s).

9.1 Materials, Birds and Mammals

There is insufficient information on the effects of ozone on materials or birds and mammals on which to base Reference Levels. Therefore it is recommended that no Reference Levels be developed for these receptors.

9.2 Vegetation

Form

It is recommended that the appropriate form of an index to capture chronic vegetation impacts be cumulative (i.e. hourly concentrations are summed over some period of time) and emphasize peak concentrations. The SUM60 index is such an index and is identified as being the preferred index for use in a Canadian regulatory context, for both short-term and chronic exposures. The SUM60 index is calculated by summing hourly ozone concentrations equal to and greater than 60 ppb during daylight hours (08:00 - 19:59) over a given period of time (3 days, 3 months). Adoption of the SUM60 index is recommended while a the same time it is acknowledged that there is no biological basis for inferring that hourly average concentrations below 60 ppb are not important in affecting vegetation. Development of an exposure index for vegetation, that better reflects the many factors that influence plant response to ozone, is a key challenge for the future.

As described in Chapter 8 (Section 8.2.3), a 10 percent biomass loss level is the minimum level of yield loss that can be confidently measured and attributed to ozone exposure under experimental conditions. Loss estimates less than this can be within the range attributable to experimental error. Therefore, the 10 percent yield loss level is selected as the relevant endpoint for assessment of chronic impacts from ozone exposure.

Given the importance of protecting vegetation from both acute and chronic foliar injuries that may negatively impact crop quality and marketability, the development of a short term index was considered. Seasonal, cumulative-indices may not adequately capture the frequency and episodicity of short-term, high concentrations. Data from field studies in Ontario on two crops, radish and white bean, were available for analysis and quantification of foliar injury impacts in response to exposure to ambient ozone. The analysis consisted of testing the exisiting one-hour ozone NAAQO (82 ppb) and a cumulative one day and three day SUM60 indices. The 3-day SUM60 was identified as the best performing index overall, based upon predictions of trace foliar injury.

Level

Agricultural Crop Yield

The U.S. NCLAN database for crops, excluding crops grown in California and those not grown in Canada, was used to identify LOAELs for effects of ozone on crop yield. Given the degree of inter- and intra-species variability in crop response to ozone, it was felt that selection of a single LOAEL, from the most sensitive of all crops evaluated, as a Canadian Reference Level for effects of ozone on vegetation, would be inappropriate. Instead, a LOAEL range has been identified that represents a conservative estimate of the range of ozone concentrations above which effects are expected to occur on Canadian crops. This range is 5,900 - 7,400 ppb-h (3 month SUM60, ozone measured during the hours of 08:00 - 19:59). This range is derived from the responses of wheat and turnip (Table 1).

Tree Growth

Identification of LOAELs for effects of ozone on tree growth were based on studies conducted by the U.S. EPA in the 1980s on the impact of ozone on forest trees of the U.S. via an OTC exposure protocol similar to that of the NCLAN studies on crop yield. SUM60 values corresponding to 10 percent loss of biomass for individual tree species were presented in Table 2. As with the LOAELs for crop species, it was felt to be inappropriate to select a single LOAEL as the Canadian Reference Level, given the degree of variability within and among species. A LOAEL range is identified from the responses of black cherry and aspen (Table 2) of 4,400 - 6,600 ppb-h (3 month SUM60, with ozone measured during the hours of 08:00 - 19:59).

Foliar Injuries

Given the limited quantitative data available for foliar injury on which this LOAEL determination is based, it was felt to be inappropriate to identify a single LOAEL value. Instead a LOAEL range for trace foliar injury is identified as follows: a 3-day SUM60 during the daylight period (08:00 - 19:59) in the range of 500-700 ppb-h. Although both the form (the 3-day SUM60 index) and the range of the LOAEL for acute effects have been developed on only two crops (radish and white bean), both of these plants are known to be sensitive to foliar injury development and to be significantly impacted as a direct result of the foliar injury.

In summary, it is recommended that the cumulative SUM60 index used in the assessment of both acute and chronic effects of ozone on vegetation in Canada. The SUM60 index is the sum of hourly ozone concentrations equal to or greater than 60 ppb over the daylight period 08:00 - 19:59. The 3-day SUM60 is used for the assessment of acute effects, whereas the 3-month SUM60 is used for the assessment of chronic effects. Given experimental uncertainties, questions around how results from experimental work apply to real-world growing conditions, and the amount of both inter- and intra-specific variability in the response of plant species to ozone exposure, it was felt that identification of Canadian Reference Levels for effects on vegetation was inappropriate. Instead, LOAEL ranges have been identified for acute effects on crops and chronic effects on crops and trees. These represent conservative estimates of ozone concentrations above which effects on Canadian vegetation are expected to occur.

The LOAEL ranges for chronic effects, derived from Tables 1 and 2 of this Summary (SAD Tables 8.9 and 8.11) are as follows, based on agricultural crops and tree species:

Period of Calculation	SUM60 range (ppb-h)
	crops	Trees
3 month daylight hours (08:00-19:59)	5,900 - 7,400	4,400 - 6,600

The LOAEL range for acute effects, derived from agricultural crops, is:

Period of Calculation	SUM60 range (ppb-h) crops
3 day daylight hours (08:00-19:59)	500 - 700

9.3 Human Health

Form

For ozone and human health effects, the Reference Levels are derived statistically from several studies and should be interpreted as a level above which there is confidence (statistical significance) in the dose-response relationship and the ability to provide some quantification of adverse endpoints. The Reference Level should therefore not be interpreted as a threshold of effects.

Based on the weight of evidence presented in Chapter 13, the strength of the epidemiological evidence for mortality and respiratory hospitalization effects at current levels of ambient ozone are significant, consistent, coherent, robust and compelling. Studies on non-accidental mortality and respiratory hospitalization provide quantitative estimates of the health risks of ozone pollution and LOAELs for these endpoints, which are the most appropriate indicators on which to base Reference Levels for ozone.

The number of studies of emergency department visits is limited (10 studies). Results from these studies were generally consistent with those from the mortality and hospitalization studies. The exposure endpoints were assessed based on 1 to 24 hour averaging times. While qualitatively sound, the quantitative aspects of these studies indicate that it would be impossible to establish a Reference Level using these data.

Measures of other respiratory health effects such as school absenteeism, days of work loss and restricted activity are usually collected through survey instruments, and thus are often subjective. Although these parameters are valid measures of respiratory health, they are not appropriate indicators for establishing a Reference Level for ozone. Small and reversible changes in lung function are measured by spirometry, which in general is a robust measure. However, clinical studies have shown that lung function changes do not correlate with tissue injury and thus lung function parameters are not suitable for deriving a Reference Level.

The acute effects are, by definition, related to peaks in ozone levels: clinical and population health studies have correlated responses with hourly or longer (up to 24 hour) exposures. A majority of North American epidemiological studies have used 1-hour daily maximum concentrations, and have found a positive association with increased mortalities, hospitalization and emergency department visits (see Chapter 12). A few European and Canadian studies have used both 1-hour and 8-hour daily maximum concentrations, and found that the results were highly consistent. Greater correlation with the health endpoints has been demonstrated for the 1-hr daily maximum average than for the 8-hr daily average, in both the Canadian epidemiological studies and those from other countries. An averaging time of 24 hours is not appropriate because of the strong diurnal pattern exhibited by ozone and the substantial year to year variation in ozone maxima at different sites across the country. Therefore, the averaging time of any target level should be 8 hours or less.

Controlled human exposure (clinical) studies have provided evidence indicating that exposure to ozone at concentrations as low as 120 ppb for 1.5 to 2 hours, or 80 ppb for 6.6 hours, caused pulmonary function decrements and airway inflammation in both asthmatic and healthy subjects. It should be noted that no clinical studies examining inflammation for 1 to 2 hours of ozone exposure at concentrations below 120 ppb have been documented.

It is noteworthy that a clinical study, that used varying doses as well as a constant dose over an 8 hour period (with the same total concentration), has demonstrated that subjects responded almost twice as markedly, in terms of pulmonary function decrements, to the varying dose regimen as the constant dose of ozone. These results suggest that the average dose value, calculated as a mean over an 8-hour exposure period, may underestimate the effect of ozone on pulmonary function induced by a peak exposure.

In view of the results from controlled human exposure and epidemiological studies, it is recommended that the ozone Reference Levels be expressed on the basis of a 1-hour daily maximum averaging time.

Level

The ozone Reference Level is defined as an estimate of the lowest ambient concentration at which statistically significant increases in health responses have been detected. In general, Reference Levels should not be interpreted as thresholds for affects. In the case of ozone, most studies indicate a continuum of effect through all ambient levels examined, and adverse effects are expected below the Reference Level. However, the analysis performed here indicates that the statistical strength of the data below the identified Reference Levels is inadequate to provide quantification of effects at lower levels.

The published literature does not provide sufficient information necessary for the derivation of Reference Levels. Instead, regression analyses were performed on mortality and respiratory hospitalization data from 13 Canadian cities over an 11-year period, and LOAELs for these two endpoints were derived (Appendix A). These LOAELs form the basis of the Reference Levels for ozone.

The Canadian data used to derive LOAELs for both total mortality and respiratory hospital admissions are consistent with the data from other published studies. These other studies were conducted in many geographic locations, including cities in North America, Europe and South America, where sources and concentrations of ozone and population composition differ from the Canadian situation. This leads to the conclusion that the data upon which Canadian studies are based are the most representative of the effects of ozone in this country while being within the range of observations across other geographical conditions, and the most appropriate for the purpose of deriving ozone Reference Levels. Furthermore, the use of raw data from 13

Canadian cities over an 11-year period (Appendix A), and the application of the same statistical treatment throughout, enhance the reliability of the results. Using the same statistical treatment and corrections for confounding factors reduces the uncertainties around any value derived as a Reference Level.

The results of the regression analyses provide assessments for two different types of responses. The LOAEL for the total mortality rate is 20 ppb (p≤0.05), and does not represent a threshold. The respiratory hospitalization rate exhibits a threshold of effect between 15 and 20 ppb of ozone, with a LOAEL at 25 ppb (p≤0.05).

The lack of evidence of a threshold for mortality precludes the possibility that a sufficiently low level of exposure will be free of any degree of impact. Therefore, the Reference Level, a level above which there are demonstrated effects on human health, is an estimate of the lowest ambient ozone level at which statistically significant increases in health responses have been detected, and not a level below which there are no health impacts.

Recommended Reference Level

Based on the Ozone Science Assessment Document, there is sufficient evidence to conclude an association exists between ambient levels of ozone and human mortality, respiratory hospitalizations and several other health endpoints. The Working Group has developed Reference Levels for the mortality and hospitalization endpoints, as sufficient data exist for the quantification of LOAELs for these endpoints only.

The Working Group notes that this is one of the first instances in which the association between ambient ozone and mortality has been made in the context of a regulatory risk assessment. Although the 1996 US EPA Criteria Document and Staff Paper for ozone did not make a link between ozone and mortality, the subsequent "Regulatory Impact Analyses for the Particulate Matter and Ozone National Ambient Air Quality Standards and Proposed Regional Haze Rule" (July 1997) noted that the literature on this issue had been evolving rapidly. The RIA report provided an extensive review of the new literature on this subject and a quantitative analysis of the mortality risk for ozone, concluding that "this new evidence suggests that substantial additional health benefits associated with reducing ozone concentrations may exist" in addition to the benefits associated with other endpoints. The database has continued to evolve, providing even stronger evidence of an association between ambient ozone and premature mortality.

Thus, based on the analysis of Canadian mortality and respiratory hospital admission data, the following Reference Levels for ambient ozone are recommended:

Period of Calculation	Level
	non-accidental mortality	respiratory hospitalization
Daily, 1 hour maximum	20 ppb	25 ppb