The Health Effects Institute
	PAPA-SAN Public Health and Air Pollution in Asia: Science Access on the Net
Study Designs Commonly Used in Epidemiologic Air Pollution Studies Case–control study In a case–control study, persons in a population who develop particular diseases, or who die from those diseases, are identified and then classified according to their exposure to air pollution and other factors related to disease occurrence, such as age and cigarette-smoking status. These are the cases. A sample of members of the study population who were free of the disease at the time that the case subjects became ill or died (the controls) is also selected and classified in terms of the same risk factors. An estimate of the relative risk of disease or death that is associated with a particular factor, often termed an odds ratio, can then be calculated. The case–control approach provides an efficient alternative to a cohort study (see description below), because the relative risk of disease or death can be estimated in a population without having to collect information on all its members. Case–crossover study In a case–crossover study, cases of, or deaths from, particular diseases that occur in a population are studied to provide an estimate of the effect of short-term exposure to air pollution. For each person who has the disease or who has died, exposure to air pollution is determined for a period near the time of diagnosis or death (the case period) and one or more periods during which the health event did not occur (control periods). The relative risk can then be estimated using standard methods of analysis for matched case–control data. Cohort study Cohort studies provide estimates of the effects of prolonged exposure to air pollution on mortality and morbidity due to chronic disease. In a cohort study, members of an entire population who are initially free of disease are observed over an extended period of time. Each member of the population is classified according to his or her exposure to air pollution, often determined on the basis of place of residence, and according to other factors related to the occurrence of disease, such as age and cigarette-smoking status. The rates of disease or death that occur among the exposed and unexposed persons are then compared in order to estimate the effects of exposure, with comparisons often presented as a ratio of rates, or relative risk. Cross-sectional study Cross-sectional studies are typically used to study the effects of long-term exposure to air pollution on the prevalence of chronic respiratory symptoms and disease or on chronic impairment of pulmonary function. In a cross-sectional study, the disease status and exposure to air pollution of population members at a fixed point in time are determined. The effect of exposure on the prevalence of disease can then be calculated. In contrast to case–control or cohort studies (see above), however, in a cross-sectional study it is often not possible to determine the temporal relationship between exposure and disease. Ecologic study In an ecologic study, also called an aggregate-level study, information is not collected on individual members of a population; instead, the relationship between average rates of disease or death and average exposure to air pollution in the population is analyzed. Such studies often make use of routinely collected data on both health outcomes and levels of air pollution. When ecologic studies compare incidence or mortality rates for whole populations in geographic regions with different average levels of air pollution, they can provide estimates of the relative risk of disease or death due to air pollution exposure. These estimates are often more difficult to interpret than those from case–control or cohort studies because they are not based on data on individual subjects. Health-impact study Health-impact studies, or assessments, are often used to estimate the potential benefits in terms of public health of actions taken to improve air quality. In a health-impact study, the amount of disease, disability, or death attributable to exposure to air pollution in a specific locale is quantified. Such analyses make use of data on the levels of air pollution and the rates of disease or death in that locale and estimates of the effect of air pollution on the incidence of disease or on mortality, such as relative risks from epidemiologic studies. The effect estimates might come from studies of the local population, but when such studies are not available, estimates from studies in other populations might also be used. Panel study Panel studies are often used to estimate the effects of short-term exposure to air pollution on acute respiratory symptoms or changes in pulmonary function or other biomarkers. In a panel study, a small group, or “panel,” of individuals is followed for a short period of time. During that period, health outcomes and exposure to air pollution are ascertained for each subject on several occasions, and the longitudinal relationship between exposure and outcome is analyzed. Time-series study A time-series study provides an estimate of the effects of short-term exposure to air pollution. Typically, daily counts of health events, such as deaths or hospital admissions, in one or more locales are analyzed in relation to contemporaneous data on daily concentrations of air pollutants and other risk factors, such as weather, that vary over time. Statistical techniques, typically Poisson regression, are then applied to these data to estimate the proportional increase in the rate of disease or death that is associated with a given increment in exposure to air pollution. Read more about pollutants and abbreviations commonly used in air pollution research.
PAPA-SAN | Methods | Results | About | Links For questions about PAPA-SAN please contact PAPA@healtheffects.org Home | About HEI | What's New? | Newsletter | HEI International | Publications Research | Funding | Meetings | Contact HEI | Employment | Links | Search | Site Map Copyright © May 2008 Health Effects Institute. Please send comments to webmaster@healtheffects.org