M. Perske
Cooking The Books

A Restaurant Study

Return to main page
Return to Researchers' page


Cooking the Books: A Restaurant Study

Date of original release: 5/12/97

Link to the Speakeasy Forum Return to main page
Return to main page

What happens when a researcher "cherry-picks" from the published works of other researchers, ignores warnings from those researchers about the limitations and uncertainties of their data, and misleads the public by grossly exaggerating his findings to the press"
Unfortunately, if the researcher's work is anti-smoking, it receives widespread media attention and is cited as "gospel" by anti-smoking groups and public health officials as they continue their efforts to ban smoking in restaurants and bars.

Such is the case with Michael Siegel, MD, MPH, of Boston University's School of Public Health and his 1993 study, "Involuntary Smoking in the Restaurant Workplace: A Review of Employee Exposure and Health Effects." [1] Conducted while Siegel was a graduate student at the University of California, Berkeley, the study claims that "epidemiologic evidence" shows that exposure to environmental tobacco smoke (ETS) is a "significant occupational health hazard for foodservice workers." Siegel's "evidence" was based on data he took from six studies conducted by various other researchers. These six studies contained information on lung cancer risk in various occupations (e.g., clerical workers, barbers, farmers, coal miners, restaurant workers, etc.) and had absolutely nothing to do with exposure to ETS in restaurants or bars (or anyplace else).

Siegel, for example, extracted data from these studies that showed small increased risks for lung cancer for food-service workers, and implied that the increased risks were due to ETS, even though there was no evidence in the studies that food-service workers were even exposed to ETS-let alone for how long or at what levels. He also omitted data in the studies that showed a decreased lung cancer risk for food-service workers.

Despite the facts, Siegel concluded-in what amounts to a monumental scientific hedge: "The epidemiologic evidence [from the six studies] SUGGESTED that there MAY BE a 50 percent increase in lung cancer risk among foodservice workers that is IN PART attributable to tobacco smoke exposure in the workplace." (emphasis added) Siegel went even further out on a limb when he told USA Today that, based on his study results, secondhand smoke has a "devastating effect" on restaurant workers, and "is now a life-and-death issue for restaurant workers." His conclusions also were reported in numerous newspapers nationwide.

What has been ignored by the media and public officials is the fact that even if Siegel's 50 percent increased risk were on solid ground, it means only that the relative risk was 1.50. According to the National Cancer Institute, relative risks below 2.00 are considered "small" and could be due simply to chance, statistical bias or confounders. [2] Lynn Rosenberg of the Boston University School of Medicine called another study with a relative risk of 1.50 "far from conclusive," since epidemiologists normally take seriously only risk factors of 3.0 or greater. [3] And the U.S. Environmental Protection Agency has refused to classify electromagnetic fields as a cause of cancer "largely because the relative risks ... have seldom exceeded 3.0." [4]

Clearly, it isn't ETS that has a "devastating effect," but rather, those researchers who, through scientific misconduct, seriously mislead the public in order to advance a political agenda (in this case, smoking bans). Even more troubling is the fact that this misconduct goes unchallenged by the media and the so-called scientific community.

The following is a discussion of the six studies "misused" by Siegel in preparing his study:
Keller and Howe Study (Based on 1985-1987 data pertaining to lung cancer and occupation among Illinois non-smokers.)

The study stated that it had no information on ETS exposure and, therefore, it was "not possible" to do an evaluation. Nonetheless, Siegel used an increased risk reported on one of the study's tables for white females currently employed in eating and drinking places and implied that it was due to ETS.

Singleton and Beaumont Study (Based on California data from 1979-1981 pertaining to occupation and death due to various causes, including accidents and suicides.)

Siegel cherry-picked from this study in that he presented, among other things, lung cancer results for MALES in the category of "other food-service workers," and ignored the significant decreased lung cancer risk for FEMALES in the same category.

Schoenberg Study (Based on 1980-1981 data pertaining to occupation and lung cancer risk among New Jersey white males.)

From this study, Siegel selectively used increased risks reported for bartenders and other food-service workers, but ignored the decreased risks for cooks and food-counter workers. He also ignored the fact that the study classified bartenders, cooks, food-counter workers and other food-service workers as "non-high-risk" for lung cancer.

Zahm Study (Based on 1980-1985 data pertaining to occupation and type of lung cancer among Missouri white males.)

The authors of this study said their results may be "statistically unstable" and that the "small numbers, the large proportion of unknown values, and the other limitations suggest that the study's results be interpreted cautiously." Siegel ignored these statements and used the study's lung cancer relative risk of 1.8 (based on only 24 cases) for male food-service workers. Siegel also claimed that data from this study included smoking histories of the study subjects. However, the study stated that information on smoking represented smoking status at the time of diagnosis only and "may be inaccurate for the time period relevant for carcinogenesis." (emphasis added)

Williams Study (Based on cancer mortality data from the Third National Cancer Survey.)

This study warned that a "cautious and conservative approach should be followed in interpreting all of these data ... Generally, these results should be used only as a research resource for the formation of hypotheses and planning of studies to follow up leads suggested." (emphasis in original) Siegel disregarded these warnings and used the lung cancer relative risk of 1.88 (not statistically significant and based on only 12 cases) for female food-service workers as if it were real and attributable to ETS.

Lerchen Study (Based on 1980-1982 data pertaining to lung cancer and occupation in New Mexico.)

From this study, Siegel used the relative risk of 1.6 (not statistically significant and based on only 26 cases), even though the authors noted that it would no doubt take a study of 10,000 lung cancer cases "to evaluate fully occupation and other risk factors for lung cancer."  


[1] Siegel, Michael, MD, MPH. Involuntary Smoking in the Restaurant Workplace: A Review of Employee Exposure and Health Effects. Journal of the American Medical Association. July 28,1993. Vol. 270, No. 4.

[2] National Cancer Institute release, Oct. 26, 1994. As reported in the Competitive Enterprise Institute newsletter, February 1995, p. 8.

[3] Cancer Risks for Thee, but Not for Me. Jerry Taylor, Cato Institute. Wall Street Journal, Jan. 3, 1995.

[4] Evaluation of the Potential Carcinogenicity of Electromagnetic Fields. U.S. Environmental Protection Agency. Review Draft, October 1990, p. 6-2.

FORCES is supported solely by the efforts of the readers. Please become a member or donate what you can.

Contact Info
Forces Contacts
Media Contacts
Links To Archived Categories

The Evidence
Inside Forces
About Forces
Book case