Harvard Six Cities Study

The Harvard Six Cities Study was a large-scale study of the effects on human health of fossil fuel [Page 456]emissions, in particular, sulfur dioxide and respirable particulate matter (soot). It was inaugurated by Benjamin Ferris and involved many faculty members and researchers at Harvard University, including John Spengler, Douglas Dockery, and Frank Speizer. This study provided evidence that, in concert with similar results from other studies, prompted the Environmental Protection Agency to raise air quality standards.

One impetus for the study was a belief in the United States in the 1970s that, due to the Middle East oil embargo, high-sulfur coal would be used more widely as an energy source in the future. Coal was a well-known source of air pollution (e.g., the combination of pollution due to soft coal combustion and a temperature inversion was associated with a temporary doubling in the death rate in the London smog disaster of 1952), but it was uncertain whether the health effects observed were due to sulfur dioxide, particulate matter, or both. The Six Cities Study was designed in part to investigate this question. Ferris had observed methodological innovations used to study the effects of pollutant exposure on the health of coal miners in Wales and had already incorporated those methods into a study of sulfur emissions from a paper mill in Berlin, New Hampshire; he also applied those methods to the Harvard Six Cities Study.

The Harvard Six Cities Study followed a cohort of 8,111 Caucasian adults in six cities in the northeastern and Midwestern United States for 14 to 16 years, beginning in the mid-1970s. The cities included were Watertown, Massachusetts; Harriman, Tennessee; St. Louis, Missouri; Steubenville, Ohio; Portage, Wisconsin; and Topeka, Kansas. Questionnaires were used to collect data, including smoking history, educational level, age, sex, weight, height, medical history, and occupational exposure to gases, dusts, and fumes, from participants at enrollment and 3, 6, and 12 years afterward. Mortality data were collected through the National Death Index. Ambient air pollution was measured through centrally located monitors in each city.

The principal finding of the study was the positive association of air pollution and mortality. In particular, the study found that comparing residents of the most polluted to those of the least polluted city, higher ambient levels of respirable particulate matter and sulfur dioxide were associated with a 26% increase in mortality from all causes and that increased levels of respirable particles were associated with increased mortality from cardiopulmonary disease. In addition, the study found that the increase in mortality risk was directly proportional to increase in respirable particulate matter concentration. Mortality rate ratios were invariant for smokers and nonsmokers, and for people with and without occupational exposure to dusts, gases, and fumes. Increased air pollution was also associated with increase in a number of illnesses, including asthma and lung cancer.

The relationship between air pollution and mortality found in the Harvard Six Cities Study were confirmed in a much larger study, the American Cancer Society (ACS) study. Results from both studies were influential in the development of higher standards for air quality issued by the Environmental Protection Agency in 1997. These new regulations were questioned by representatives of various industries, who claimed that results from the ACS and Six Cities studies results could be explained by poor research design and data collection techniques, flawed statistical methodology, and factors not considered in the studies, such as temperature differences between cities. Because of confidentiality requirements, data from the ACS and Six Cities studies could not be released for reanalysis by industry representatives, so as a compromise, a third-party reanalysis was conducted by a nonprofit organization, the Health Effects Institute (HEI). This reanalysis, led by Daniel Krewski and Richard Burnett, included an audit of data quality, replication of the original studies, use of different statistical techniques, and inclusion of covariates such as climate, socioeconomic characteristics, and presence of other pollutants. The HEI reanalysis confirmed the original conclusions of both studies, although specific measures of risk were in some cases slightly higher or lower than in the original analyses. One new finding from the reanalysis was the association of lower levels of education with increased mortality.

...