posures to large populations, and therefore high attributable risks. This has also been true for the study of air pollution, because entire populations of cities or regions are exposed. In contrast, studies of hazardous-waste sites will often have to deal with risks from multifactorial outcomes, and small populations are usually involved. Questions of analysis and sample size addressed in air pollution studies also must be addressed in any epidemiologic assessment of hazardouswaste exposure. Furthermore, exposure through air is recognized as a feature in the epidemiology of hazardous-waste sites, not only because lead dust can blow off such sites and volatile organic compounds can be encountered, but because the outgassing of volatile contaminants in domestic water is a recognized phenomenon (Andelman et al., 1986; McKone, 1987). As noted in Chapter 3, trichlorethylene, which has been shown to outgas in domestic water (Andelman et al., 1986), is the second most commonly found compound at hazardouswaste sites.
Studies of trends over time in air pollution and disease patterns have produced a growing body of literature that has associated day-to-day fluctuations in air pollution with daily fluctuations in mortality across a wide range of exposures with no evidence of thresholds. The early studies of pollution and daily mortality in London examined discrete episodes (Great Britain Minstry of Health, 1954), and recent analyses show a strong and consistent association between daily particulate concentration and daily mortality across 14 years of data (Schwartz and Marcus, 1990). Figure 4-1 illustrates that relationship. In both cases, the relationship with particulates held independently of sulfur dioxide but not vice versa. One report that analyzes mortality in Steubenville, Ohio (Schwartz and Dockery, 1990), finds a significant association between mortality and airborne particulate matter at concentrations well below the ambient air quality standard of 150 micrograms per cubic meter. The consistency of findings in these studies is complemented by a consistency in the magnitude of the effect. Assuming a log linear model for mortality counts, the Steubenville analysis associates an increase of 100 micrograms per cubic meter in particle concentration with a 3.8 percent increase in the daily rate of mortality; in London it was associated with a 4 percent increase, and in New York with about a 3 percent increase. In separate analyses of the relationship between particu-