Vol.1 No.1 2008

36/76

Research paper : A strategic approach for comparing different types of health risks (A. Kishimoto)−33−Synthesiology - English edition Vol.1 No.1 (2008) of medical science. The volume of QALYs is represented by the gray area in Figure 3, where the vertical axis is quality of life (QOL) from 0 (death) to 1 (normal health) and the horizontal axis is age from birth to death. When the QOL is reduced to the dashed line for some reason, the loss of QALYs corresponds to the reduced area. Using QALYs as health risk metrics in risk assessment makes it possible to consistently capture not only loss of life expectancy, but also loss of QOL.It was necessary to multiply the number of cases of various symptoms, including death, by the QOL weight from 0 (death) to 1 (normal health) to quantify human health risks resulting from exposures to chemical substances as loss of QALYs. The number of cases showing various symptoms was obtained by substituting the annual average exposure levels into the dose response functionsTerm 2 that describe the relationship between the intake of substance and the probability of symptom presentation. The dose response functions should be preferably derived from human epidemiology studies with chronic endpoints of established diseases or subjective symptoms. The individual exposure level used in the calculations was expressed as annual average, which is described as the weighted average of indoor and outdoor concentrations. In order to estimate the national distribution of atmospheric concentrations by running the atmospheric dispersion model, the distribution data for emission volumes was necessary. These values should be preferably estimated as a central tendency or average estimates (and distribution profile if possible), instead of mere upper bound estimates.On that basis, we reexamined the current methodology for each step of risk assessment. As a result, we found that almost all of the existing methodologies were not directly applicable. Therefore, following the gray arrows in Figure 2, we performed risk assessment for toluene step by step by revising slightly, making rough assumptions, or developing new methods. 3.2 Estimation of emission volumesInitial risk assessments placed less significance on estimation of emission volumes or search for emission sources since they were based on measured concentrations in the environment or human bodies. However, to conduct simulation of the effect of proposed emission control measures, it is necessary to know the emission sources and volumes. For example, suppose that the share of some emission category was assumed to dominate but the actual share was less than half. This means that the effect of the emission control measure was less than half of the assumed effect. Even if the emission sources are discovered, items with large uncertainties, such as unintended emissions and natural sources, are often ignored or underestimated, and this may lead to underestimation of the total emission volume. In case of toluene, there are no estimates of evaporation from automobile fuel tanks and extra emissions (cold-start emissions) that occur before engines are warmed up at the beginning of PRTR.3.3 Estimation of personal exposure distributionConcentrations of chemical substances tend to be measured at locations and times that the concentrations may become high. The outdoor measurement data are biased toward values measured near the emission sources, and the indoor measurement data are biased toward values measured in newly-built houses. Therefore, we did not have information on the total exposure for Japanese residents. In addition, risks from outdoor and indoor concentrations are usually evaluated separately. Since most of the indoor data are daily averages for houses, information on annual average and inter-house and within-house variabilities are not available. For outdoor concentrations, AIST-ADMER (Atmospheric Dispersion Model for Exposure and Risk Assessment) has been used to predict the regional distribution of concentrations with a horizontal resolution of 5 km for all Japan [2]. To run this model, estimated emission rates were allocated to 5 x 5 km square grids and meteorological data were collected from National Meteorological Observatories and from the Automated Meteorological Data Acquisition System (AMeDAS). The annual average concentrations within these 5 x 5 km square grids were between 0 and 67 µg/m3 and the arithmetic mean was 1.5 µg/m3. For indoor concentration, it was necessary to obtain the within-house variability in the annual average “indoor concentrations of indoor origin”. The daily average data for 207 houses were obtained through the Freedom of Information Law from the Ministry of Health and Welfare [3]. The concentrations of toluene tended to be higher in indoor environments than in outdoor environments due to multiplicity of indoor emission sources. Since indoor toluene consists of ambient toluene that infiltrates indoors, as well as toluene emitted from indoor sources, “indoor concentrations of indoor origin” were defined as indoor toluene concentrations minus outdoor toluene concentrations. We assumed that these data followed Fig. 3 Concept of quality adjusted life-years (QALYs)Quality of Life (QOL)Normal health = 1Death = 0BirthDeathAgeLoss of QALYs

※このページを正しく表示するにはFlashPlayer9以上が必要です