The Six Cities and ACS studies were both prospective cohort studies, meaning that they enrolled subjects and watched their health change over time. (Prospective studies are better than "retrospective" studies, which rely on subjects' memory of their health and behavior.) But there's another way to look at the effects of air pollution: the time-series study. In this approach, researchers track how day-by-day changes in pollution affect health statistics -- on everything from emergency-room visits to deaths.
The idea for this approach grew from the 1952 London "pea soup" fog, an airborne soup of particles that killed 4,000 people in a few days and proved that bad air can kill quickly.
© Purdue University Multicultural Home Page
A recent time-series study was performed in Sao Paulo, Brazil, a metropolis with 16 million people, 3 million cars and 300,000 trucks and buses. In heavily industrialized Sao Paulo, the average particulate level is 82.4 micrograms per cubic meter. (For comparison, the annual average PM-10 in the largest 50 U.S. metropolitan areas ranges between 31 and 60.4 micrograms per cubic meter.)
Researchers found that when particulates in Sao Paulo increased by 100 micrograms per cubic meter, the overall death rate increased by about 13 percent. In addition, particulates were more closely linked to the death rate than nitrogen oxides, sulfur dioxide or carbon monoxide. (See "Air Pollution and Mortality " in the bibliography)
Worse than it seems?
Together, the time series and cohort studies make a strong case for blaming disease on particulate levels, Pope says. "Causation should not be pinned on any one study. You have to look at the overall picture -- at the consistency and coherence of the various studies." Looking at that data, he says, "I'd argue that combustion-source particulates increase the risk of cardiopulmonary disease and mortality, and thus the EPA focus on fine combustion-source particles is very reasonable. The evidence is quite compelling that they are the most serious threat to human health."
Or is it?
Indeed, the Sao Paulo data probably understate the problem, notes C. Arden Pope III, an economist at Brigham Young University, because a time series only counts immediate deaths, not any long-term deaths due to chronic exposure to pollution. "These acute episodes [of increased pollution] tend to be real transient. What are the effects of repeated exposure for many decades?" That's where cohort studies are more useful, he notes.
