Authors: Richard Kluger
There was an exterior coolness to Cuyler Hammond, but below a surface courtliness lurked a nervous energy and a hungrily questing intellect that did not reatlily embrace received wisdom. He smoked three to four packs of cigarettes a day—yet, as if ruled by some divine monitor, inhaled only lightly. A perfectionist about his work who sometimes grew impatient with and a bit arrogant toward those he perceived as his intellectual inferiors, he had apprenticed before the war as a biostatistician with the National Institutes of Health; his wartime service was capped by extensive on-site gathering of data among the survivors of the first atomic bomb at Hiroshima. His training won him a part-time post teaching biometry at Yale while working at the cancer society in New York, where he was known for scrupulous attention to the accuracy of his department’s statistical computations.
The first major fruit of Hammond’s presence on the ACS staff was the 1948 publication of “Trends in Cancer Mortality,” which confirmed the startling rise in lung cancer and concluded that it had become, at about that time, the most prevalent and deadly form of the disease. Pressed by Dr. Alton Ochsner, the New Orleans surgeon who had early on pointed to cigarettes as the likeliest villain and was then serving as president of the ACS, Hammond set to work devising a comprehensive study of the problem. But he was a good deal more skeptical than Ochsner that smoking was the answer. It seemed too facile an explanation. The trouble with the 1950 studies of Drs. Wynder and Graham and Morton Levin in the U.S. and Doll and Hill in England, which
provided the weightiest evidence yet of a link between the disease and smoking, and Muller’s pathbreaking report eleven years earlier was that they were all based on the disparity between hospital patients who had already contracted lung cancer and comparable groups of patients free of the disease. Their smoking histories were collected and analyzed retrospectively, and the higher prevalence of smokers among the diseased subjects was interpreted as no accident. Hammond was too much a purist, though, not to grasp the potential weaknesses in this methodology. Hospital patients, to start with, were anything but typical of the population as a whole, and most of them were gathered in clinics or wards on the basis of symptoms diagnosed by their physicians; the underlying systemic causes of their sicknesses could vary widely and produce misleading diagnoses. The taking of smoking histories from those already suffering with lung cancer, moreover, was not the most reliable basis for scientific judgments. Beyond the fallibility of memory, especially among those of advanced age, afflicted smokers were prone to shade the truth, albeit unconsciously, when questioned about their lifelong habits. The heavy-smoking patient, hoping that his painful coughs were due only to an overindulgence he could curb, might be tempted to exaggerate his daily intake of tobacco by way of trying to ward off the dreaded diagnosis, whereas the patient already diagnosed with the disease and grieving over his foolhardy surrender to the habit might understate the amount he had smoked, as if to demonstrate that fate and not his own conduct had betrayed him. To avoid these and other such biases built into the sampling of hospital patients, Hammond realized, it was necessary to reverse the process: the sample population under study ought to consist of asymptomatic smokers and nonsmokers—that is, without lung cancer and equally free of other known symptoms of fatal diseases—who would be followed for a number of years to see which group the disease hit harder.
But how to manage such a study, especially with a massive cohort of the size Hammond had in mind—100,000 or more? Hospital populations were by definition confined, and data on them readily gathered through institutional staffs. It was a far more daunting task to monitor a healthy, at-large multitude over a number of years and across a broad geographic stage. Hammond consulted the Gallup and Roper polling organizations and learned they would charge about fifty dollars a head to undertake the four-to-five-year tracking he had in mind—a prohibitive price for a voluntary, non-profit health organization. Hammond’s wife then came up with a brainstorm. Why not enlist the ACS’s huge corps of well-educated and underutilized housewives?
Immediate objections were lodged against this inspired notion. How could untrained volunteers be relied on in the first place to conduct interviews and record answers correctly? At least as worrisome, could they be depended upon to follow up on a regular basis and over a protracted period of time? Actually,
interviewers employed by even the foremost polling companies—not to mention the U.S. Census Bureau—were often not highly trained professionals. Why couldn’t Hammond and his ACS colleague Daniel Horn, a Harvard-trained psychologist, provide what training was needed? The essential task, after all, was not terribly complex. Each of the ACS interviewers would be asked to find ten men in their community, preferably though not necessarily of their acquaintance, between the ages of fifty and sixty-nine—the span when lung cancer was most likely to appear; it did not matter whether the enlistees were smokers, only that they were not chronically ill at the time they were signed up for the study. The ACS volunteers had to do no more than fill out the original questionnaire with each subject, check back with him annually, and be aware of any major intervening medical event, in particular the subject’s death, to be reported promptly to the ACS.
The plausibility of the plan brought on an attack of cold feet in the ACS directorate. Such a massive study might be mismanaged and harm the society’s reputation as a noncontroversial entity devoted to educational ends. It did not exist to undertake research of its own—that was best left to independent experts. And why dwell on smoking as the prime suspect? There was arguably as much reason to blame the gas-guzzling automobile culture, with its ubiquitous spread of noxious fumes and asphalt dust. While automotive pollution might well be contributing to the rise in lung cancer, Hammond and Horn noted, it could not explain why the retrospective studies to date were revealing a disproportionately high incidence of the disease among smokers; if cars were at fault, smokers and nonsmokers ought to be affected equally. Unspoken in all this was the concern that many ACS contributors were, like Hammond and Horn, serious smokers who did not want their habit impugned.
“Dr. Hammond had to tread very carefully,” recalled Lawrence Garfinkel, a statistician-sociologist not long out of New York’s City College and assigned to the Hammond-Horn project. “A lot of people in the organization were nervous, and Cuyler had to get his funding one stage at a time.” Among those disinclined to make cigarettes the subject of an intellectual lynching was ACS scientific director Charles Cameron, himself a pack-a-day smoker who thought that smokers might well be in some fashion constituted differently from nonindulgers and thus prone to the ravages of lung cancer. Cameron was “not overly enthusiastic” about the proposed giant study, according to Garfinkel, who became a career epidemiologist with the ACS, but Cameron himself would insist long afterward that he never wavered in his support of Hammond—a view confirmed by Dr. Arthur I. Holleb, a successor in Cameron’s post. “It would not have happened without Cameron,” Holleb stated.
Cameron, though, was undoubtedly subjected to cautionary pressure from the society’s chairman, Elmer Bobst, the pharmaceutical heavyweight who had friends in the tobacco industry and on at least one occasion brought Cameron
along to a luncheon with Reynolds Tobacco operatives warning against the premature indictment of their product. But Bobst, by most accounts, was judicious and nonobstructionist in the use of his power at the ACS, and so Hammond, professing keen skepticism and stressing the need for unequivocal evidence before tobacco could be charged with causing lung cancer or other diseases, was given the green light. He, Horn, and Garfinkel visited ten cities, enlisted 200 volunteer interviewers, trained them, and ran them through practice sessions gathering data before everyone was satisfied that ACS volunteers could serve adequately. During the first half of 1952, the study, perhaps the largest of its kind ever undertaken by a single private group, began in earnest.
Some 22,000 volunteers spread among 394 counties in eleven states, including New York, Illinois, and California, were engaged in the effort. The sampling of white men over fifty, it would later be argued, was not representative of the whole nation, since it had a disproportionately high enlistment of subjects east of the Mississippi, their average age far exceeded that of the population as a whole, and their socioeconomic and educational level—only one-fourth of them had not finished high school—was also well above the U.S. norm. But Hammond had reasoned that the states selected for the test all had substantial rural as well as urban populations; his sample was not necessarily intended to represent the whole nation. And if he had not limited his subjects to those in the most cancer-prone age bracket, the study might have to run twenty years before it yielded enough data for a realistic judgment. The above-average socioeconomic standing of his subjects, furthermore, might arguably serve to make the study’s findings more conservative and thus reliable, not less so, since sickness and death rates were generally higher among lower-income and less educated people.
As the first data began to flow in, the need for keen vigilance was underscored. The questionnaires collected in one Minnesota county, for example, were found to have all been filled out in the same handwriting and had to be tossed out. Some of the subjects listed were under fifty or female or not white—racial susceptibility to the disease was thought to be a possibly confounding factor and so the sample base was all-white by design—and had to be eliminated. As the first subjects began to die off, Hammond’s staff wrote away for a copy of their death certificates and then followed up for as much detail as possible in the form of autopsy reports and, in the case of cancer deaths, histological classification of the form of the disease—
i.e.
, categorizing it by microscopic examination of the lethal cells. The first major test of the likely reliability of the whole enterprise came in the autumn of 1952, when all the volunteer interviewers were asked to conduct their first follow-up sessions; an astonishing 99.4 percent of the subjects were located and their health status accounted for.
As Hammond and his associates at the ACS were gearing up for their effort, a similar prospective study, smaller and far less varied in its sampling but uniquely reliable, was being undertaken by the leading British investigators on the subject, Richard Doll and Bradford Hill. They sent a “questionary” to the nearly 60,000 doctors in Great Britain, inquiring about their age, health, and smoking habits, among other things, with the intention of following up periodically on their subjects’ health status. About two-thirds of the recipients responded; Doll and Hill suspected that a substantial number among the doctors electing not to reply were already stricken with lung cancer or some other, likely terminal disease, the presence of which might skew the validity of the study. Women doctors and practitioners under thirty-five were also excluded from the sampling in order to limit the subject base to those most likely to become afflicted with lung cancer.
While their doctor study was under way, Doll and Hill published the final report on their retrospective study of London hospital patients. It so persuasively pointed the finger of suspicion at smoking that the
British Medical Journal
was prompted to editorialize late in 1952 that it was now “surely incumbent on tobacco manufacturers” to undertake intensive research on the chemical constituents of smoke, for if the carcinogenic agents could be isolated, then they might also be removed “so that smoking will become a less dangerous occupation than it appears to be now.” The journal’s blithe suppositions—that profit-making industrialists would without duress intently examine the perils inherent in their highly lucrative product, that science could neatly calibrate the potency of cancer-causing or cancer-spreading agents, and that even if investigators were able to achieve as much, the cigarette could then be rendered harmless and yet remain a cigarette—were to prove wildly wishful.
III
IF
the reports and suspicions being generated in the early ’Fifties by the scientific community on the documentable relationship between smoking and health were unwelcome to the leading cigarette manufacturers, the companies on the lower rungs of the industry ladder saw in this menacing news an opportunity to make hay, so to speak, while the clouds gathered.
Brown & Williamson, smallest of the top six cigarette makers, found a growing demand for its mentholated Kool and filter-tip Viceroy brands, both with supposedly therapeutic qualities vaguely hinted at in their advertising. But it was sleepy Lorillard that awoke first to the chance to reverse its lagging fortunes by exploiting the public’s slowly spreading apprehension that the
old wives’ tales and busybody moralizers’ warnings about cigarette smoking might be scientifically verifiable.
The 1950 breakthrough studies could not have come as a big surprise at Lorillard, where as early as mid-1946 one of its leading staff chemists, Harris B. Parmele, sent a memorandum to the secretary of the company’s committee on manufacturing in which he noted that certain scientists and medical authorities “have claimed for many years that the use of tobacco contributes to cancer development in susceptible people. Just enough evidence has been presented to justify the possibility of such a presumption.” But he was not yet alarmed by the evidence and saw no reason to believe smoking was any more implicated than many other possible causative factors. Five years later, however, by which time Dr. Parmele had been advanced to Lorillard’s director of research, his concern over the health issue had grown and was shared by the company’s tough, street-smart sales director, Lewis Gruber. The latter believed that the moment was ripe for Lorillard to market a new brand that specifically and with scientific authenticity addressed the fear factor. Since nobody knew just what it was in tobacco that might be doing the damage, let alone whether it could be selectively removed, the most plausible way to allay the health-conscious smoker’s alarm was to put a truly effective filter tip on the product that substantially reduced the total intake of smoke without entirely depriving the user of its taste and jolt. Concurring, Parmele began to study seriously the dynamics of smoke filtration.