I was eight years old on the first Earth Day, April 22, 1970. “Give a hoot, don’t pollute!” was the slogan for us kids. When we’d see a newscast with factory stacks spewing thick gray smoke we’d say “yuck.” We’d hold our noses when tailpipes of junker cars belched exhaust. In our minds, air pollution was a bad thing because of what we could see and smell. We sure didn’t think about it as something that was cutting short people’s lives.
One of the first prospective U.S. studies to demonstrate an association between air pollutants and premature mortality was published in the New England Journal of Medicine in December 1993. The researchers, however, began their work 20 years earlier during the heydays of the “Give a hoot, don’t pollute” slogan. The study was based on a proposal developed by Harvard University based professors Benjamin Ferris, MD, a pediatrician by training, and physician Frank Speizer, MD.
Often referred to as the “Harvard Six Cities Study”—the paper “An association between air pollution and mortality in six U.S. cities” —reported findings of a prospective cohort mortality study involving 8,111 randomly selected residents of six U.S. cities. The objective was to estimate the effects of air pollution on mortality while controlling for other risk factors such as the individuals’ smoking status and age. The residents were recruited between 1974-1977, completed questionnaires about their medical and occupational history, and underwent lung function (spirometry) tests. In addition to data provided by ambient air monitoring stations situated in their towns, some participants wore air sampling devices to assess personal exposures to air contaminants, and others had air monitoring equipment set-up in their homes.
“It has always amazed me how interested and cooperative people were,” said Douglas Dockery, ScD in an interview for the Harvard Public Health Review. “We would go into people’s homes and set up equipment; we would have kinds of boxes and noisy pumps, and they were still very happy to participate. There were people who would drill holes in the side of their houses just to make our sampling easier…”
John Spengler, PhD and some of his engineering students designed the sampling equipment. In the same interview he added:
“…we put the batteries, electronics, and pumping system into a box about the size of a Tupperware container and added some shoulder straps. [The equipment] was very quiet, and it had to be because people took them to church, work, their bedrooms, everywhere they went.”
Annually, from 1974 until 1991, the researchers contacted the residents to determine their vital status. The death certificates of the deceased were obtained to ascertain their cause of death. Of the 8,111 subjects, about 18 percent were deceased by mid-1991, the end of the follow-up period.
The six cities (Harriman, Tennessee; Portage, Wisconsin; St. Louis, Missouri; Steubenville, Ohio; Topeka, Kansas; and Watertown, Massachusetts) represented locations with different concentrations of air pollutants. The researchers assembled data on total suspended particulate matter, sulfur dioxide, nitrogen dioxide, and ozone, among others. Based on more than a dozen years of exposure data, Portage and Topeka had the lowest concentrations of particulate matter pollution, but the highest levels of ozone. Steubenville and Harriman had the highest concentrations of particulate matter and levels of aerosol acidity.
The researchers’ findings on the association between mortality and air pollution was attention grabbing. After controlling for smoking and other risk factors, the authors reported the following:
- Mortality was most strongly associated with the levels of inhalable, fine and sulfate particles than with the other air pollutants.
- Air pollution was positively associated with mortality due to lung cancer and cardiopulmonary disease, but not with mortality from all other causes.
The table below summarizes the results of some of the authors’ regression models comparing mortality risk between the most air polluted city (Steubenville, Ohio) and the least air polluted city (Portage, Wisconsin). The most basic model includes the air pollution data alone and the results show a statistically significant 31 percent increase risk of mortality among the Steubenville subjects. As additional variables were added to the model (i.e., smoking status, education, BMI, etc.) the elevated risk associated with the air pollution variables remained strong and statistically significant.
The authors acknowledged that the excess mortality could be attributed to the subjects’ past exposure to air contaminants, not necessarily the exposures measured during the study period. Their results were however consistent with previously published time-series and cross-sectional studies which also examined mortality and air pollution. They concluded their article with this recommendation:
“This study provides additional impetus to the development of strategies to reduce urban air pollution.”
Thirteen years later, some of the original researchers, plus Francine Laden, ScD and Joel Schwartz, PhD, published “Reduction in Fine Particulate Air Pollution and Mortality: Extended Follow-up of the Harvard Six Cities Study,” in the American Journal of Respiratory & Critical Care Medicine (2006; 173: 667-672.) They reported on eight additional years of follow-up (through 1998) on the Six Cities cohort.
Laden and colleagues noted that concentrations of particulate matter pollution (PM2.5) declined over the study period in the six cities. The improvements were most pronounced in the most polluted cities, especially Steubenville, Ohio. (Recall the Clean Air Act of 1970 and amendments in 1977 and 1990 were driving down air contaminants during this period of time.)
Using data from the EPA’s Aerometric Information Retrieval System from monitors within 80 km of the six cities, the researchers calculated more specific exposure measures for particulate matter 2.5 microns in size (PM2.5).
Of the 8,111 individuals in the cohort who were originally recruited in 1974-1977, 34 percent (2,732 participants) were deceased from any cause. The authors’ analysis of the association between PM2.5 exposure and mortality found:
- Death from cardiovascular disease was positively associated with average PM2.5 over the entire follow-up (RR=1.28, (95% CI: 1.13-1.44)) based on 1,196 cases (ICD-9 codes 400-440).
- Death from lung cancer was positively associated with average PM2.5 over the entire follow-up period (RR=1.27, (95% CI: 0.96-1.69)) based on 226 cases.
- Death from non-malignant respiratory diseases was positively associated with average PM2.5 over the entire follow-up period (RR=1.08, (95% CI: 0.79-1.49)) based on 195 cases.
Moreover, the authors reported that as the concentrations of PM2.5 declined in the six cities, so did the risk of death from cardiovascular and respiratory diseases. The decline in mortality risk was particularly pronounced in the most polluted cities. These findings were like icing on the Six Cities’ cake. Not only did the researchers reinforce with even more robust data the mortality risk associated with PM2.5, they demonstrated that the risk of death could be reversed by lowering concentrations of the pollutant. That evidence was very good news for public health advocates and other proponents of stronger regulation of air contaminants.
In the discussion section of Laden and her colleagues’ 2006 paper, the authors noted:
“The original Six Cities Study mortality analysis has undergone an extensive reanalysis performed by an independent group of researchers.”
They don’t hint however that the “extensive reanalysis” came about because of intense attacks and a multi-million campaign to discredit the Harvard Six Cities study. Powerful economic interests, including those funding the group “Citizens for a Sound Economy,” were threatened by the public health findings. They knew that the U.S. Environmental Protection Agency would use the study, in part, to support more health protective air pollution regulations.
The industry critics of the Six Cities study had friends on Capitol Hill who demanded that the Harvard researchers release their raw data. As project biostatistician James H. Ware, PhD said in an interview, the researchers declined because when the Six City participants were recruited for the study in the 1970’s, they were promised their personal information would not be made public. The authors also could predict how the study’s critics would use the raw data.
“We knew that if we released the data, it would be endless aggravation and defending against attacks. To have a hostile group combing through your data looking for anything to attack you about was not something any of us relished,” Ware explained.
EPA wanted to rely on the Six Cities study for its proposed regulation, but like the Harvard researchers, the agency was also under fierce congressional pressure from some anti-regulatory lawmakers, such as Senator Richard Shelby (R-AL), to scrutinize the study or forego relying on it. Ultimately in 1997, EPA and the authors found a solution. They contracted with a team of independent scientist operating under the Health Effects Institute and led by Daniel Krewski, PhD to re-analyze the data. After three years of work, the independent panel reaffirmed the quality and the findings of the original Six Cities study.
Scientists from across the country and around the globe have been inspired to model their own air pollution research using the Harvard Six Cities design. They’ve examined air contaminants and health effects in the Pearl River delta of southern China (here), in Denmark (here), in France (here), in London, Ontario (here) in New Zealand (here), and among the California Teachers’ study cohort (here), the Nurses’ Health study cohort (here), and in Medicare populations (here), just to name a few. The Harvard Six Cities study is a classic for laying a bedrock foundation for these subsequent studies, and for providing the first unimpeachable evidence linking certain air contaminants with premature death.