Doctors cite statistics to "prove" tobacco may kill you. But C. Harcourt Kitchin, author of the widely discussed You May Smoke; takes a hard look at these figures—and finds they tell a different story

December 17 1966


Doctors cite statistics to "prove" tobacco may kill you. But C. Harcourt Kitchin, author of the widely discussed You May Smoke; takes a hard look at these figures—and finds they tell a different story

December 17 1966


Doctors cite statistics to "prove" tobacco may kill you. But C. Harcourt Kitchin, author of the widely discussed You May Smoke; takes a hard look at these figures—and finds they tell a different story

MOST CANADIANS, in common with a good majority in other countries, have now accepted as established fact that smoking, especially cigarette smoking, is the chief cause of lung cancer.

The foundations for this belief lie in three reports of the medical profession — one from the British Medical Research Council, later elaborated in a second from the British Royal College of Physicians, which was shortly followed by a third, longer and couched in more technical language which few laymen readily understand, from the United States Surgeon General’s department.

All these reports were widely reviewed in newspapers when they were published, not only in their countries of origin but almost throughout the world, and their conclusions have since been quoted in speeches and articles at frequent intervals. Indeed, it would be more accurate to say that the general belief that smoking causes lung cancer is based on what people read in newspapers rather than on the reports themselves, which comparatively few have read and fewer still have studied.

Those who have read the reports will have seen that their conclusions rest almost entirely upon the results of certain statistical studies — notably

one by Doll and Hill of British doctors in 1956 and another by Hammond and Horn in the U. S. in 1958—which show a considerably higher lungcancer death rate among smokers than among nonsmokers. But they claim that the statistical association, which we shall examine shortly, is supported by other evidence. This is how the British Royal College of Physicians puts it:

“The strong statistical association between smoking, especially of cigarettes, and lung cancer is most simply [my italics] explained on a causal basis. This is supported by compatible, though not conclusive, laboratory and pathological evidence, namely (a) the presence of several substances known to be capable of producing cancer in tobacco smoke; (b) the production of cancer of the skin in animals by repeated application of tobacco tar; and (c) the finding, in the bronchial epithelium of smokers, of microscopic changes of the kind which may precede the development of cancer.”

Let’s look at the supporting evidence first. All the reports agree that the real murderer is the cigarette. The U. S. study showed the same mortality among cigar smokers as among nonsmokers (there are not enough regular cigar smokers in

Britain to show figures), and both studies show the death rate of pipe smokers as less than one third that of cigarette smokers.

The U. S. report details the quantities of each of four cancer-producing polycyclic hydrocarbons, in micrograms per thousand grams of tobacco consumed. Here they are:


Benzpyrene 9 34 85

Acenaphthylene 50 16 291

Anthracene 109 119 1,100

Pyrene 125 176 755

With the sole exception of the quantity of acenaphthylene in cigar smoke, there are far more cancer-producing substances in the smoke from pipes and cigars than in cigarette smoke. All the medical authorities agree that benzpyrene is much the most potent and dangerous. Cigar smoke has nearly four times as much of that as cigarette smoke, and pipe smoke more than nine times as much. Something seems to have gone wrong with this as supporting evidence against cigarettes.

On another page of the British RCP report we find a more modest claim about both the carcinogens in smoke and / continued on pape 25

continued on pape 25


continued from pupe 21

“It is improper to conclude smoking is a cause of anything”

experiments on animals: “Skin cancer can be produced in mice by applications of tar condensed from tobacco smoke, but the results obtained by various investigators have not been uniform and exposure of animals to tobacco smoke in inhaled air has failed to produce lung cancers. Moreover, the amount of cancer-producing substances in the smoke itself does not seem likely to be sufficient to account for the large number of cases of cancer associated with the habit.”

This is quite a different story. But it is a fact that mice have developed cancer of the skin after being painted with “tar.” or condensate, or cigarette smoke, in quite unrealistic quantities concentrated on the tiny areas of their backs. The tar left in the lungs and bronchi of an inhaling human is distributed over an area said to be as big as a tennis court. And you may wonder why. if cigarette smoke causes cancer of the lung, it doesn't also cause cancer of the tongue or throat. Another strange thing that has never been explained is that studies have shown no higher death rate from lung cancer among inhalers than among non-inhalers. Early studies showed a lower rate for inhalers.

The report doesn’t specify what sort of microscopic changes heralding lung cancer are to be found in the bronchial epithelium of smokers. The scientific director of the. U. S. Council For Tobacco Research tells of a study by 1 2 pathologists in different urban and rural areas.

After examining 3,000 lungs of patients, they reported that “the type ot lesion (atypical metaplasia) other than the tumor itself most frequently found in the lungs of lung-cancer patients was not more frequent in smokers than among nonsmokers.”

Does this “supporting evidence” really support anything? Or must we fall back upon the statistical association?

It is an old gibe that statistics can be made to prove anything. Statisticians insist that they prove nothing, but only indicate trends to be followed up by further research.

The smoking and lung-cancer statistics have been interpreted not by statisticians but by doctors. The most forceful critics of their conclusions have been statisticians, notably the late Sir Ronald Fisher in Britain, known as the father of modern statistics, and Dr. Joseph Berkson in the United States, who has been called the acknowledged dean of American medical statisticians. Berkson pointed out that no professional statistical association had ever issued a report, or had even been consulted, on the subject.

“I suggest to you,” said University of Toronto mathematician D. B. De Lury, “that it is quite improper, no matter what the data may seem to indicate, to reach any conclusion to the effect that smoking is a cause of anything.”

De Lury’s dry comment would likely be endorsed by most statisticians, for they hate to see people misguidedly using statistics to discover “facts” that really don’t exist. Often,

people will reach a conclusion through statistical association which is clearly questionable the moment you reduce it to absurdity. For instance, in the years when large quantities of apples were imported into Britain there was also a large number of divorces, but Britain didn’t ban the import of apples to reduce the divorce rate. Or, to cite

a second example: there is a strong “statistical association” between the rise in lung cancer and the rise in the sales of nylon stockings.

Besides, it is only too easy to choose statistics to suit your thesis, and leave alone others less convenient. If. for instance, smoking causes lung cancer in Canada, you would expect it to

cause lung cancer in Finland or Japan. The U.S. report chose six countries and gave the lung-cancer death rate for each, per 100,000 population. They are:



United States




But it didn't give the equivalent figures of cigarette consumption for the

SMOKING continued

Have doctors ignored contradictions in their own evidence?

same year. Here they are, per adult:

United States 3,900'

Canada 3,140

England & Wales 2,680

Finland 2,160

Japan 2,090

Norway 540

So we find that Canada and the United States, though smoking more cigarettes than Britain, have only about half the lung-cancer mortality. Within Canada, there is no reason to think that the inhabitants of Hamilton, Ont., smoke more than those in Regina; but Hamilton has 10 times the lung-cancer death rate of Regina. Dr. O. Shaefer, writing in the Canadian Medical Association Journal, reported that although Eskimos and Indians smoked heavily, mostly cigarettes nowadays, no bronchogenic cancer had been found in northern natives by X-ray survey teams.

Britain has the highest lung-cancer death rate in the world. British emigrants to South Africa smoke 75 percent more than those who stay at home, but their lung-cancer death rate is 21 percent lower. Can the British reduce their risk by going to South Africa and smoking a whole lot more?

Again, in countries where records are available, cigarette consumption by women has risen in the last 20 years or so much more steeply than has men’s, but their share of the lungcancer death rate is proportionately

not nearly so high now as it was when they hardly smoked at all.

In Britain, for instance, in 1920-24, women smoked 1.4 percent of the national cigarette consumption; in 1960-64 they smoked 32.4 percent. In 1920-24 they suffered 33.6 percent of the national lung-cancer death rate: in 1960-64 it was 14.8 percent.

The reports dismiss the idea that the recorded rise in lung-cancer mortality may be in any substantial part due to better diagnosis, but the Canadian tobacco industry has done some rcsarch on this. They analyzed published studies from 1900 to 1961. covering some of the best reputed hospitals in central Europe, Britain and the United States, noting the cases of death found by postmortem examination to be from lung cancer, which had not been diagnosed as lung cancer before death. They found that the error in diagnosis at the beginning of the century was 71 percent. By the early 1960s it had been reduced to seven percent. Dr. A. J. Phillips, reporting a recent study in Canada, put the error below five percent.

Maybe the real increase is not so formidable as it looks.

There is no doubt, however, that it is too high, and there must be a cause. Can we find it in the higher death rate in industrialized countries and areas?

Nearly 200 years ago, one Percivall Potts noticed that chimney sweeps

were getting more than their share of skin cancer. A century later certain carcinogens, the chief of which was benzpyrene, were established in soot and coal smoke. It is now possible to measure the quantities of it over cities and industrial areas.

Professor F. C. Pybus, of Newcastle, England, recently compared the total quantity of benzpyrene in the smoke from all the smoke-producing coal burned in Britain in a year with that in the smoke of all the tobacco consumed in Britain in a year. From coal smoke 375 tons of benzpyrene: from tobacco smoke — eight pounds of benzyprene.

Without doubt the doctors are honestly convinced that their conclusion that smoking is the chief cause of lung cancer is the right one. Few of us would dare assert that excessive smoking is harmless, or encourage youth to start smoking. But a study of the reports, as distinct from the newspaper stories, must raise doubts whether the cause of lung cancer has so simply been found. Haven’t the doctors overlooked too many contradictory pointers in their own evidence?

Dr. H. A. Shapiro, editor of three authoritative medical journals in South Africa, declared his sympathies to be with the woman who said she had read so much frightening information about smoking and lung cancer that the only sensible thing left for her to do was to give up reading.★