Authors: Richard Kluger
But the two situations were not comparable. Public impressions to the contrary, no investigator had produced evidence remotely approaching in strength and consistency findings like those incriminating direct smoking by Wynder, Hammond and Horn, Doll and Hill, and Auerbach. The industry could thus retain the hope that a large-scale study might fail to show a correlation between
lung cancer occurrence and exposure to ETS among nonsmokers. Such results, however, might not find their way into scientific journals because of a phenomenon known as “publication bias;” studies that produced negative results or did not report a statistically significant relationship were generally assigned a low priority among submissions.
But in the spring of 1990, a Philip Morris scientist, Thomas J. Borelli, who bore the suggestive title of “manager of scientific issues,” was scouring about for unpublished studies on ETS and, while consulting the University Microfilms International Dissertation Information Service, struck gold. Not only did he find a doctoral dissertation done at Yale by Luis Varella, a Mexican postdoctoral physician, but the study was larger than any U.S. investigation yet reported in the scientific press. And the results were cause for joy within the tobacco industry. As Borelli noted, after making 300 copies of the dissertation and sending it to the EPA, ACS, HHS, and concerned scientists around the nation, “Varella looked at 439 lung cancer cases in nonsmokers”—implying that he had been the primary and original investigator, since no other names were cited—and found “no statistically significant increased risk of lung cancer in nonsmokers exposed to tobacco smoke in the workplace or in social situations.” Nor had Varella found any elevated risk among nonsmokers due to spousal smoking, which was the key measure of dosage in the Hirayama and other studies that had reported an elevated risk of, on average, about 30 percent above that for unexposed nonsmokers.
The size of Varella’s sampling and the Yale imprimatur gave his findings special cachet, and in mid-June, Philip Morris’s John Nelson, vice president for corporate affairs, crowed in a letter published in
The New York Times
that Varella’s study was the most prominent recent addition of evidence to the body of work on ETS in which “no study of exposure to tobacco smoke in the workplace or public places has demonstrated a scientifically significant risk to nonsmokers” and that eighteen of twenty-three studies that dealt with the effects of spousal smoking had failed to report such a risk. It followed, to Nelson’s way of thinking, that the nonsmoking majority ought to willingly accommodate smokers’ needs by providing them with ample designated areas or properly ventilated public places to carry away ETS.
The full story about Varella’s dissertation, which Borelli did not disclose, possibly because it did not alter the validity of its conclusions, was that it was derived entirely from data gathered during 1981–85 under an NCI grant to investigators headed by Dwight T. Janerich, director of epidemiology for New York State before transferring in 1986 to Yale Medical School. There Varella approached him about the possibility of using his group’s raw data for his own doctoral thesis. Since it was not unusual for students to work on part of large existing data sets, Janerich and the rest of Varella’s advisory committee approved the arrangement. Varella finished his work in 1987, when it was judged
to be of a high order, approved, and copied by University Microfilms for the benefit of mankind—if anyone was interested in looking it up. Janerich, meanwhile, immersed in new assignments at Yale and without a grant to complete the ETS study, which had produced a great deal of raw data, was able to attend to it only sporadically, and so it had languished—until Borelli drew attention to it somewhat indirectly by publicizing Varella’s treatment of a portion of the data.
The Philip Morris scientist was not the only one keenly interested in Varella’s dissertation. The year before Borelli came upon it, an EPA advisor, Kenneth G. Brown, working with the review panel involved in the ETS risk assessment, learned of its existence and the surrounding circumstances and wrote to Janerich, identifying his link to the EPA and asking for further data but without indicating any urgency or stressing the official nature of his inquiry. Janerich, taking the request to be informal and somewhat offhand, was not eager to share his group’s findings with a casual inquirer before they had been refined and published. Borelli learned from the EPA that Brown’s request to Janerich had not been complied with and, unable to understand why the Yale epidemiologist had not published anything on the timely subject, grew “concerned that the data was being ‘dredged’ in order to find a statistical relationship,” as he later put it.
This concern seemed not entirely groundless after Janerich broke into print in the September 6, 1990, issue of the
New England Journal of Medicine
with an article entitled “Lung Cancer and Exposure to Tobacco Smoke in the Household.” Its stress was entirely different from what Borelli had pulled out of Varella’s thesis and circulated to the scientific world. About the only significant added risk that Varella had reported from the Janerich group’s data was due to exposure in the subjects’ homes at the equivalent level of 150 “person-years,” meaning the number of years a nonsmoker spent in a residence multiplied by the number of smokers in that household, so that a nonsmoker who lived for fifteen years in a home with ten smokers or, say, thirty years with five smokers—a lot of other people’s smoke, indeed—ran an elevated risk of contracting lung cancer. In Janerich’s report, given wide press attention, this household exposure was broken down to show that while during adulthood nonsmokers enduring even as many as fifty to seventy-four “smoker years” (the term Janerich substituted for Varella’s “person-years”) of ETS exposure showed no increased risk of lung cancer, nonsmokers with twenty-five or more smoker years of exposure during childhood or adolescence showed a doubled risk of lung cancer. Janerich’s paper, citing nine co-authors including Varella, who had died by this time, attributed 17 percent of lung cancer cases in non-smokers to childhood and adolescent exposure to ETS, although he noted that “we know of no specific mechanism that would explain our findings.” But other studies had shown children of smokers to be significantly more susceptible
to respiratory diseases, and such exposure, Janerich
et al
. speculated, “might initiate changes that eventually lead to lung cancer when the exposed children become adults.”
Almost entirely lost sight of in the press reports on the Janerich study, no doubt because the study itself minimized the point, was that it contradicted most of the previous ETS studies with regard to the effects of spousal smoking. Not only was no heightened risk found attributable to either spousal smoking or workplace exposure, even at high levels, but “a statistically significant inverse association between ETS and lung cancer” was discovered in social settings. Janerich conceded, “The apparent protective effect of exposure in social settings is difficult to explain,” and later speculated that smoke-susceptible people perhaps avoided, consciously or otherwise, smoky social situations. The thrust of the Janerich study, then, although neither it nor those reporting on it said as much, was that only reasonably heavy exposure to ETS during a nonsmoker’s pre-adult years seemed linked to an increased risk of lung cancer. But even this finding, while potentially important and inviting follow-up studies, suffered from a serious shortcoming: the questionable accuracy of the “smoker years” figures. Most of Janerich’s subjects were close to seventy when interviewed and asked to remember who in their homes smoked for how many years during their childhood—no small feat of memory at a distance of half a century or longer, and one that was possibly further skewed by the more acute sensations and reactions to surrounding phenomena during the childhood years of discovery. Janerich owned up on that score as well: “The possibility of recall bias and other methodological problems may influence the results.” Still, he expressed confidence that his group’s findings “support the conclusion that exposure to ETS can cause lung cancer.”
If the epidemiological evidence for that judgment remained hazy and ambiguous, other studies published by 1990 were suggesting biological mechanisms that might be traceable to ETS as a cause of various pulmonary and cardiac diseases. A University of Arizona researcher, exposing rabbits to ETS for only fifteen minutes a day over a twenty-day span, found that the animals’ lung tissue had become permeable, opening the organ to microbes and toxins that promoted disease formation. Other investigators had found that exposure to unfiltered side-stream smoke made platelets, the tiny fragments in the blood that caused it to clot, grow less mobile and thus stickier and more prone to clot in plaque-clogged arteries. Toxins and carcinogens in ETS, moreover, were believed to cause abrasion, through chemical or mechanical interaction, in the inner linings of arterial walls, producing damage sites inviting to blood fats that could settle and cause blockage. Children exposed to smoke since birth were discovered to register elevated cholesterol levels and reduced HDL, a protein associated with a lowered incidence of heart disease. Based on the far greater frequency of heart ailments than of lung cancer and in view of both the
plausibility of these triggering mechanisms in ETS and several population studies showing an elevated risk of heart disease in nonsmokers exposed to ETS, comparable to the risk levels found for lung cancer, epidemiologists and other public-health investigators were now attributing between 35,000 and 40,000 coronary deaths a year to secondhand smoke.
This news served only to compound the tobacco industry’s anxiety as the EPA began to circulate a preliminary draft of its ETS risk assessment, which the Tobacco Institute, in a December 4, 1990, broadside, attacked as “an uncritical condensation of only selected studies” and accused of failing to include the lately published Janerich study, which on balance seemed to the tobacco camp largely to exonerate ETS as a carcinogen. The elevated risk of lung cancer found in the nine U.S. studies so far on nonsmokers exposed to ETS was just 8 percent when compared to unexposed nonsmokers, the TI said.
Seeming to ignore the industry’s scientific arguments, a revised draft of the EPA risk assessment was circulated in April 1991 and drew a careful, blistering fourteen-page letter, signed by U.S. Representative Thomas J. Bliley, Jr., a Republican from Richmond, manufacturing home of Philip Morris, whose hand was apparent in the congressman’s letter. Bliley had run his family’s century-old funeral home before serving as mayor of Richmond and moving on to Congress in 1981; antismoking advocates found it fitting that the tobacco industry’s latest champion was a former undertaker. A ranking minority member of the House Energy and Commerce Committee, with its oversight of Henry Waxman’s subcommittee on environment and health, Bliley protested in his letter that EPA had failed to apply a “weight of the evidence” test to the ETS question; that the agency had failed to factor in four recent major studies on the subject—two from China, one from Japan, and Janerich’s—which did not support the EPA draft’s claimed higher risk of lung cancer from ETS, and that he could not understand how spousal smoking could serve as the critical gauge of ETS exposure “when every epidemiological study published to date uses a different formula for defining” what constituted spousal exposure.
Bliley’s complaint could not be shrugged off lightly, and EPA said it would reconsider its data, thus further delaying the assessment. That many others in the public-health community had already made up their minds about the ETS danger, however, was suggested by a declaration two months after the Bliley letter by the National Institute on Occupational Safety and Health (NIOSH), part of the U.S. Public Health Service, that secondhand smoke was a workplace hazard that all employers ought to take steps to reduce to the lowest possible level, preferably barring it altogether, and offering workers who smoked classes and other incentives to help them quit.
But reports spread that at EPA itself, the industry’s steady drumfire had taken its toll. The agency was said to be under growing pressure from within the Bush administration to bury or materially soften the ETS risk appraisal.
and new personnel had reportedly been assigned to take a fresh look at the data. In 1992 three large new studies appeared that could only further complicate the ongoing assessment. In the premier January issue of the journal
Epidemiology, Biomarkers and Prevention
, investigators headed by Louisiana State University epidemiologist Elizabeth T. H. Fontham studied 420 nonsmoking women with lung cancer in five metropolitan areas—a subject group about as large as Janerich’s—and reached almost exactly the opposite findings: no elevated risk from childhood exposure to ETS but a 30 percent rise from spousal smoking and slightly higher from work-site exposure. In the September 16, 1992, number of the
Journal of the National Cancer Institute
, Florida investigator Heather Stockwell reported a small, statistically insignificant increase in lung cancer risk among 210 nonsmoking women subjects but a more than doubled risk for those with twenty-two smoker years of exposure in childhood or forty years in adulthood, partially corroborating Janerich’s data (which had shown no elevated risk from adulthood exposure). To add to the ever larger body of conflicting findings, Ross C. Brownson of the Missouri Department of Health examined 432 lifetime nonsmokers and discovered no increased lung cancer risk associated with childhood exposure to ETS, but at the high exposure level of forty or more “pack years” (meaning years of exposure to a pack-a-day smoker) found a 30 percent higher risk—yet none in a dose-related pattern at lower levels.
If this round of new reports seemed to yield more conflict than agreement regarding cancer, far more uniform—and indicting—findings were accumulating with regard to ETS and heart disease. By 1994, a dozen epidemiological studies dating back to 1985 and involving 3,131 subjects, more than the total in the thirty lung cancer studies on ETS exposure until then, showed a consistent 30 percent elevated risk of heart disease. Since there were ten times as many cardiac deaths as from lung cancer, the overall toll from ETS exposure was coming to be seen as much graver than previously supposed. The CDC was estimating it to be on the order of 12 to 15 percent of the lives claimed by direct smoking. The likely underlying mechanisms, too, were coming to be better understood in ETS-linked heart disease. The presence of endothelial cell carcasses detected in the blood of nonsmokers was attributed to carcinogenic agents found in ETS and held responsible for damage to the arterial walls where plaques were rooting. Studies found the loss of platelet activity in the blood, a condition which encouraged clotting, to be as high as 80 percent after repeated exposure to ETS. And eight weeks of exposure was enough to cut in half the efficacy of enzymes affecting the heart’s ability to convert oxygen into “energy molecules” like adenosine triphosphate, which served as a little chemical battery to fuel the heart muscle’s ability to contract.