Article Text

PDF

Use of personal exposure modelling in risk assessment of air pollutants
  1. P Sarin
  1. Center for Risk Science and Public Health, Department of Environmental & Occupational Health, George Washington University School of Public Health and Health Services, 2300 K Street, NW, Warwick #201, Washington, DC, USA; psarin{at}gwu.edu

    Statistics from Altmetric.com

    Kromhout and van Tangeren1 raise important issues regarding the papers by Cherrie2 and Harrison and colleagues.3 The major shortcomings of the paper by Harrison and colleagues3 are the small size of the sample (six subjects each) used in the extrapolation of results. The three groups studied were the children, the elderly, and subjects with preexisting disease. The sample size in the disease category used only two subjects each with chronic obstructive pulmonary disease (COPD), left ventricular failure (LVF), and severe asthma. These sample sizes are rather inadequate to draw any correlation. Thus the paper by Harrison and colleagues3 does not adequately represent a generalised level of exposure of the individuals to carbon monoxide (CO), nitrogen dioxide (NO2), and particulate matter (PM10).

    In risk assessment of ill health association with air contaminants, uniform sampling of the pollutants at home, school, work, and outdoor activities is important. In a recent article4 detailing the effect of the World Trade Center collapse, the authors emphasise that environmental monitoring for exposure assessment is a complex technical task that involves selection of pollutants for monitoring, location of monitors, sample collection methods, risk assessment standards, sampling results, and data analysis. For example, the initial approach to the environmental sampling at the World Trade Center4 was to locate monitors at the perimeter of the site, at locations where emergency and debris carrying vehicles were leaving the site, on the debris pile, and at locations in the surrounding community. Additional monitors were set up in the community to ensure a safer environment for workers and students returning to workplaces, homes, and schools. Approximately, 66 000 results were entered in a database collected between 11 September and 13 November for subsequent analysis. Substances monitored included asbestos, particulate matter (PM2.5 and PM10), dioxins, polychlorinated biphenyls, CO, heavy metals, and volatile organic compounds (VOCs).

    As pointed out by Schneider,5 deterministic models are developed from equations based on mathematical principles, while statistical models are developed by fitting to observed data.6 In the statistical modelling of inhalation exposure, mixed effect models have been useful. The linear mixed effects model with AR-1 autoregressive correlation structures has recently been used by Levy and colleagues7,8 in their studies.

    For example, due to the difficulties in the measurement of personal exposure, data on air pollution patterns in homogeneous microenvironments linked with activity data are often used as surrogates.7 In these studies7 PM2.5 indoor:outdoor ratios were found to be greater than 1 in settings with high levels of human activity. Cooking activities contributed significantly to increases in levels of PM2.5 along with other pollutants. Using linear mixed effects models with AR-1 autoregressive correlation structures, 10 minute average outdoor concentrations were generally weak predictors of indoor levels.

    As mentioned by Levy and colleagues,8 although ambient particulate matter has been associated with ill health, the health risks for individuals depend in part on their daily activities. Information about levels of PM size distributions in indoor and outdoor microenvironments can help identify high risk individuals. The authors used linear mixed effects models with an AR-1 autoregressive correlation structure to evaluate statistical significance of differences between microenvironments. Levels of larger particles were generally higher near significant human activity, and levels of smaller particles were higher near combustion sources. The indoor PM10 concentrations were reported to be significantly higher than those outdoors on buses and trolleys. Statistical models showed significant variability among some indoor microenvironments.

    As pointed out by Chang and colleagues,9 simulation of activities performed by 65+ year olds indicate substantial variability in personal exposures of PM2.5, O3, CO, and VOCs over a 12 hour period. For example, one hour personal CO exposures measured in vehicles were significantly higher than those measured in other microenvironments, and the correlation between personal PM2.5 exposures and ambient concentrations was lowest in the winter months for indoor non-residential microenvironments and highest in vehicle microenvironments.

    Thus the conclusions drawn by Harrison and colleagues3 from the data on a relatively small sample of subjects utilising microenvironment modelling of CO, NO2, and PM may not adequately reflect the overall exposure patterns. The variance observed between measured and modelled values for PM10 among the elderly, those with COPD, and children could be minimised by takingmeasurements in a larger sample, both indoors and outdoors, and during summer and winter months.

    References

    View Abstract

    Request permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.