Authors: Richard Kluger
The Surgeon General’s panel had three choices. It could go through the motions but essentially just rubber-stamp the British report—plainly an inadequate procedure as this was the first full-fledged inquiry into the subject by appointees of the U.S. government. Or the SGAC members could farm out the review work to the Public Health Service staff and other expert consultants outside the federal bureaucracy, letting them work up reports that the committee members could then sit in judgment of and approve or modify as they saw fit. Or somehow, in a manner hard for such busy and already heavily committed
men to conjure, the committee could itself do the deep thinking and investigating, albeit with ample staff support. “Many of us were naive about the amount of time and work it would consume,” said LeMaistre. It would take them six months to grasp that to do the task right—
i.e.
, not delegating the intensive scrutiny of evidence to others—would amount to what LeMaistre characterized as “a massive and overwhelming job.” Yet, after considerable foot-dragging, the committee concluded that this was the only scientifically justifiable course.
The result was a consuming thirteen-month voyage of self-education and discovery. The committee sat as a whole for nine sessions, running from two to four days each, and in between there were numerous subcommittee gatherings, endless telephone communication, and many specially convened conferences, seminars, and commissioned reports involving 155 outside consultants. The immense task was necessarily parceled out to the committee members and the resulting assessments shared with the whole group. They also entertained with equal attention the findings on smoking and ciliatoxicity of industry-supported researcher Charles Kensler, “a first-rate pharmacologist who gave us a first-rate report,” in Hamill’s words, and the American Cancer Society-backed investigations of Oscar Auerbach, whose laboratory was visited by several SGAC members for a firsthand look at his slides of cell changes found in cigarette smokers.
The locus of the SGAC’s efforts was the new National Library of Medicine, on the sprawling campus of the National Institutes of Health in Bethesda, just north of Washington. Full committee sessions were occasionally held at the Public Health Service headquarters downtown in the capital, but more typically the ten men gathered on the mezzanine of the Library of Medicine, an ultramodern structure with an odd, pagoda-like roof, or in the third subbasement of the library, where, because construction on the building had not been completed, it was possible to arrange a makeshift compound of desks, tables, filing cabinets, and partitions that enclosed a temporary clerical staff of from fifteen to thirty while typewriters clattered and copying machines hummed at all hours of the day and night. That subterranean redoubt, though inelegant and characterless—also windowless but well ventilated and thus effectively removing telltale cigarette smoke from the premises—served to isolate the committee and its working papers from the rest of the world. As committeeman Schuman remembered, they were “so removed that nobody could ever find us down there. Sometimes we felt like moles.” Armed guards were stationed at the library entrances and patrolled the corridors; all committee members, their staff, and consultants had to obtain government clearance, and all the materials and internal documents generated by the study were under lock and key. These measures were aimed at preventing leaks to Wall Street, where the manipulation
of tobacco stock prices might result. The hush-hush atmosphere added to the building awareness that the committee’s business was of surpassing international significance.
VIII
THE
proceedings of the Surgeon General’s Advisory Committee on Smoking and Health were notable on three accounts—for the care its members took in examining the evidence, the quest for explicitly stated criteria in reaching their collaborative judgment, and their willingness to engage the counterarguments put forth by spokesmen for the tobacco industry. The last of the three was essential, if any truly fair appraisal was to be reached, because even the most outspoken critics of smoking had to admit that the health case against regular tobacco use was not a single, clear scientific proposition, like a geometry theorem, but an untidy mosaic of evidence.
The SGAC staff, while striving to keep the members insulated from the industry, nevertheless welcomed the vast amounts of material that the cigarette companies provided in historical and scientific mitigation of the alleged health perils of smoking. But the committee’s sense of duty went further and was perhaps best exemplified by the attentive ear that staff director Hamill gave to the industry’s chief scientific representative, Clarence Little, who at seventy-five was performing his final well-paid services for the tobacco manufacturers. He was seeking assurance that the Surgeon General’s panelists would take into account all the possibly confounding variables in the evidence before them. He and Hamill proceeded as wary performers, if not open antagonists, in a high drama, with Hamill recognizing what he called Little’s “blind spot”—an almost dismissive attitude toward the laws of statistics and probability. “You had to see a thing happen in the laboratory to convince him it was true,” Hamill recalled of their exchanges. And as a certified expert in genetics, Little carried great weight with committeeman Burdette, who was inclined at the outset to believe that smokers contracted lung cancer mainly because they were genetically programmed to do so.
The most articulate and persistent proponent of this “constitutional” theory of lung cancer causation was the British mathematician Ronald A. Fisher, who had argued throughout the ’Fifties that heavy smokers were of a distinctly different genotype from nonsmokers, whose genes in effect immunized them from the disease. In the view of the Surgeon General’s panel, such a theory logically dictated that those who quit smoking or smoked a pipe or cigar—categories that registered sharply reduced rates of lung cancer—had to make up still another genotype, closer to nonsmokers than smokers in their predisposition
to both the habit and the disease. Indeed, Fisher when challenged would so argue and be joined by fellow Briton Hans Eysenck, the psychologist who claimed that those who quit smoking were different psychologically and emotionally from continuing smokers. But given the glacial rate of human mutation, was it plausible that such genotypes could have evolved within just two generations, prior to which cigarettes were not in common usage and lung cancer was scarcely known among the list of diseases fatal to mankind? If lung cancer was not greatly exacerbated by the inhalation of tobacco smoke, why was the disease—and the genotype promoting it—all but nonexistent before the worldwide mass-merchandising of cigarettes?
Advocates of the constitutional theory were reduced to answering these questions by claiming that prior to the twentieth century, most people simply did not live long enough to contract lung cancer, essentially a degenerative disease, and that diagnostic techniques did not exist to recognize the disease in those who did contract it. Far more plausible to the SGAC members was the likelihood that some smokers were indeed genetically more susceptible than the rest of the population to the carcinogenic and toxic substances in tobacco smoke, but this unfortunate proneness in no way reduced the culpability of cigarettes in inducing disease—quite the opposite in the case of those who, not knowing their genetic makeups, indulged in the habit. By the same token, those whose genetically determined personalities made them likely to take up smoking may indeed have been betrayed by their genetic constitutions—for it was the smoke, not the genes, that served as the pathological agent.
Far more resistant to dismissal than the theories of Fisher and Eysenck, contrarians who availed themselves of research grants and consulting fees from the tobacco industry, were the relentless objections raised over the course of a dozen years to the findings of the population studies on smoking by the chief of biostatistics at the famed Mayo Clinic in Minnesota. Joseph V. Berkson, a bouncy, combative, highly excitable man whose wiry red hair seemed to grow more vivid as he charged that there were methodological flaws in the findings of heightened risk of disease and death from smoking, had to be taken seriously because he was widely recognized as a first-rate mathematician on the staff of one of the nation’s most highly regarded diagnostic centers. Hospital patients, Berkson argued, were self-selected populations, choosing their medical-care facilities in accord with their symptoms and presumed causes; the diagnoses they received were thus far more likely to be compromised by circumstances. A thoracic surgeon, therefore, confronted by lung cancer in a smoker was prone to assume that the tumor was primary when in fact it might well have metastasized from another site, where its origin might in no way have been linked to the victim’s smoking habit. In fact, such a misdiagnosis occurred perhaps in one case out of twenty, and the reverse mistaken diagnosis
—
i.e.
, labeling as primary a tumor at a distant site that had actually originated in the lung—likely occurred about as often, thus nullifying the argument. But this did not dissuade Berkson.
When, partly in response to Berkson’s constructive carping, the prospective studies on smoking were launched, drawing on a vast population of subjects who were not only outside of hospitals but also ostensibly in good health at the time, Berkson continued to rail against the dependability of the results. The cohorts were not randomly selected, he insisted, and reflected the biases of the volunteer interviewers, who may unwittingly have signed up a disproportionately high number of sick smokers or of well nonsmokers, thus skewing the results. He also insisted that the massive study undertaken by Cuyler Hammond at the American Cancer Society was of limited value because it failed to take into consideration other health habits and factors in the subjects’ histories besides smoking. Some in the biostatistical field whispered that Berkson was envious of the acclaim greeting Hammond, perhaps a less gifted mathematician, who had been a subordinate of his while both were working for Army Air Force intelligence during World War II. Whatever the motive, Berkson sustained a drumbeat of naysaying as the statistical evidence piled ever higher. At Mayo Clinic staff meetings he would rail that the ACS smoking studies and others like them “proved too much”
(i.e.
, blamed smoking for being causally linked with too many diseases) and were based on “a spurious statistical phenomenon.” In the
British Medical Journal
in 1959, Berkson began sounding like the tobacco industry’s scientific spokesman, Clarence Little, as he argued that before the case against smoking could be conclusively established, it was essential for scientists first to understand “the physiobiological factors which are involved and how they operate.” By mid-1963, as the Surgeon General’s committee was hard at its task, Berkson was still writing in
The Cancer Bulletin
that the case against smoking was built on “a statistical fallacy;” he then added, almost spitefully in the face of massive evidence to the contrary, “Even strong contenders of the smoking-lung cancer theory no longer believe that the cancer is produced by carcinogens … contained in tobacco smoke. … The whole fabric of the original theory has broken down and virtually been abandoned.”
Berkson tirelessly tried to press these views on the Surgeon General’s committee, asking repeatedly to appear before the panel and frequently telephoning its nearest member, Leonard Schuman, eighty miles away from the Mayo Clinic at his University of Minnesota office in Minneapolis. “I’d listen to his tirades for a while and then drop a hint that this was costing a lot of money,” Schuman recalled, “but Joe would say, ‘That’s okay, I’ve got a WATS line.’” Berkson’s views were so well known to the SGAC and had been so satisfactorily repudiated by others that the committee members had no further interest in having him lobby them in person. “Berkson and Fisher were discussed at great
length,” SGAC member Emmanuel Farber remembered, “with seriousness and respect,” but for all Berkson’s vehemence, his views were unpersuasive—except with his colleagues in the biometry department at the University of Minnesota, where he taught, and at the Mayo Clinic, where he was lionized as brilliant, entirely honorable, and right on the smoking question.
The twenty-nine major retrospective and seven prospective population studies on smoking and health that Berkson had found fatally flawed were of the essence to the investigation undertaken by the Surgeon General’s committee, and especially to its stellar mathematician, William Cochran of Harvard. The dour Scotsman intently scrutinized this mountain range of data and undertook cumulative computations after transforming their statistical findings into common coinage. The task was complicated by the widely varying sizes of the groups sampled and the categories of data collected in each study, but there was no gainsaying what SGAC staff director Hamill characterized as “the uncanny consistency” of the thirty-six studies. Collectively, they formed a large central core of data, like three dozen overlapping circles of different sizes and textures. To avoid top-weighting the biggest of the seven prospective studies in calculating a cumulative mortality ratio, Cochran gave each study equal standing, arranged their death rates in ascending order, and settled on the median, or fourth one on the list, as the standard for the lot; thus, the death rate from lung cancer for smokers was found overall to be 10.8 times as high as that for nonsmokers—a ratio of overwhelming significance.
“He generated excitement by the rigor of his discussion,” Mickey LeMaistre recalled of Cochran’s explications in the SGAC skull sessions, the Scotsman’s burr thickening as he got caught up in his presentment. “He gave us confidence,” added Emmanuel Farber. But what was of overriding importance to the committee was, as Hamill would later put it, “There is no major finding in the whole corpus of work in smoking research that is of an acceptable quality and jars or is discordant—that flies in the face of the main body of findings.” The closest thing to such an anomalous piece of evidence was a finding in the first Doll-Hill retrospective study of London hospital patients that self-reported inhalers of cigarette smoke had a slightly lower lung cancer rate than smokers who said they did not inhale. This finding was contradicted in three subsequent studies and was likely mitigated by the ambiguous wording in the original Doll-Hill questionnaire, which asked its subjects, “Did you inhale—that is, take the smoke deeply into the lungs?” That qualifying adverb “deeply,” some public-health investigators speculated, may have made the smoking habit seem foolhardy, so that many inhalers simply denied doing so. At best, calibrating the degree of inhalation is a tricky business, more subjective than objective, and at any rate, almost every smoker inhales, intentionally or not.