The Big Kill -- A Big Lie?
Chapter 4 of Murder a Cigarette: the Smoking Debate
It is only to be expected that once the anti-smoker movement got going, it would make use of statistics. Statistics as we know them are made for just such movements.
Let us start with a publication that came out in 1985, The Big Kill, published by the Health Education Council jointly with the British Medical Association. This document of 15 volumes, one for each regional health authority and one for Wales, declared that smoking killed 77,774 people a year and put 108,218 people in hospital.
The first question, of course, if you have read the previous chapters, is "How could they possibly be so certain?" Since we know that a substantial number of people are not diagnosed accurately in their lifetimes, and that there can be some doubt even if an autopsy is carried out, it seems extraordinary that they can pin-point the cause of death down to single units. On this subject, it was said by the Royal College of Physicians, as quoted by Professor Burch in his 'Can Epidemiology Become a Rigorous Science? How Big is the Big Kill?' (British Medical Journal 290, 2 March 1985):
'Excess deaths' -- a pretty concept -- more than their fair share? The Royal College goes on to say:
So in a dozen lines we have three 'estimates', one 'assumption', one 'probably', and a curiously imprecise: 'It would not be unreasonable'. When it comes to women, we have:
Four lines and another assumption and a 'may well be'. Science shouldn't draw dogmatic conclusions leading to far-reaching policy decisions from assumptions and 'may well bes'. We wouldn't care to travel in an aircraft if we were assured that it was assumed that we would land safely, and it may well be that the wings wouldn't fall off in flight.The late Professor Burch, who criticised this report in the above mentioned article, was a geneticist, biologist, statistician and authority on radiation, and was distinguished in all these fields. He was never afraid of being controversial.
Note that until about 1950 there was a very wide field for the practice of epidemiology. Tuberculosis, smallpox, plague, typhus, cholera and the rest of the terrible list were widespread. Partly because of immunisation programmes and antibiotics, and perhaps more because of improved standards in public cleanliness with water, sewage and housing, these curses have withdrawn within the borders of countries whose governments cannot or will not use the counter-measures available. Since such government do not usually encourage the work of epidemiologists, or anyone else who may give their people ideas about a better life, there might seem to be an epidemic of epidemiologists who are running short of diseases suitable for their investigations.
It is very understandable that the World Health Organisation (WHO) dubbed the widespread custom of smoking an 'epidemic', there were many people eager to agree. Smoking is more widespread than any infectious disease, and tends to be most common in rather nice countries with good standards of living and five-star hotels to stay in for their conferences (which are never held in places like Phnom or Novosibirsk).
Nor are you likely to find yourself having to look at unpleasant places full of sick people, and risk catching something nasty. Instead, you spend much time at the aforesaid conferences, with your expenses paid.
Obtain and manipulate
Professor Burch, in the MBJ article mentioned above, says:
He goes on to quote someone who does combine both skills, that distinguished epidemiologist we have already met, who is also an expert statistician, Professor Feinstein, who points out that a 'licensed' epidemiologist:
However, with admirable patience he goes on to discuss the American Surgeon General's 1982 report, which includes a passage from his first report on smoking and health published in 1964. It puts over a methodology-cum-philosophy that, Professor Burch says, 'enjoys wide support among epidemiologists'. As such, it's worth a careful look. It states:
Burch remarks on 'the inadequacy of these poorly defined criteria', which is a relief to all those of us who have failed to make any sense of them. He goes on to quote a well-know statistician, Professor Brownlee, who says: 'the way it [the 1964 report] claims the facts are in conformity with the criterion is to flatly ignore the facts.'
Burch also makes a point of the fact that subjective judgement, on which the Surgeon General places repeated emphasis, should play as a limited role as possible in epidemiology a in other sciences. 'For how do we distinguish between judgement and prejudice?' asks Professor Burch.
Judgement or prejudice?
Scientific analysis should surely aim to replace subjective judgement ('I feel it must be so') by objective testing. ('I don't care what you or I or anyone else feels: I'm going to try it out'), and to do away with prejudice as far as humanly possible.
Of course there are few, if any, people, entirely without prejudice, but at least they can declare their prejudice, so we know where they are coming from, and what their agenda is. Why should only the smokers and those who provide for them be assumed to be acting out of prejudice? Most anti-smokers may have very good practical or psychological reasons for their stand, but they have learnt to try to mask them with scientific or pseudo-scientific language (laced with a great deal of 'assume', 'may', 'might' 'possibly', etc.), and with plenty of talk about their noble task of saving humanity from itself, whether it wants to be saved or not.
In his Principles of Medical Statistics, published in 1937, Professor Sir Austin Bradford Hill stated:
Hill later collaborated with Sir Richard Doll on an early piece of research into the causes of lung cancer, the Doctors' Survey, which linked smoking with the disease. Faced with their very complex task, how did they go about it?
First of all, the selected group of people who were in no way representative of the population in general: that is, doctors. Doctors are much better off to start with. They are better educated than the population as a whole, and above all they receive better, or at least certainly more, medical attention than the general public, because they know where to go and how to demand it. It has been said that doctors are twice as likely to be involved in accidents, three times more likely to be alcoholics or to commit suicide, and many times more likely to take drugs. They may also be more likely to be represented in the Chamber of Horrors than any other comparable group, according top a personal inquiry at Madame Tussaud's.
Hill and Doll sent out a very simple questionnaire to this group which included all the doctors named in the current British Medical Register, and 69% replied: that is, slightly more than two in three.
Of self-selected studies in general, Professor Eysenck has stated roundly:
Only 'randomised' studies (of which more later), in which two matched groups selected by the researchers are compared, can be taken seriously, he says.
What the professor meant by 'selection of cases and controls from different source populations' is that the comparison of death-rates in doctors who had given up smoking was not with doctors who had not given up, but with members of the general public, whom we have already seen differed from doctors in many significant ways (particularly in having less money).
In May 1978 Professor Burch read, before the Royal Statistical Society, a paper: Smoking and Lung Cancer: the Problem of Inferring Cause. In the discussion that followed, it was mentioned that in the Doll & Hill study, the doctors were asked only their name, address, and age, and their smoking habits. Dr. Seltzer of Harvard University remarked on this, saying:
This should remind us of what Professor Feinstein said about the 'traditional practices of epidemiologists'.
Later in the discussion Dr. I. D. Hill of the MRC Clinical Research Centre, perhaps with unconscious pathos, said that he hadn't taken any part in this but he had grown up with it going on all around him (he is the son of the late Sir Austin). He went on:
Now these figures quoted above must first of all be seen in the light of the knowledge we already have about the difficulty in establishing any unquestionable figures for any kind of death. And at the time when this study was going on, Heasman and Lipworth surveyed reports from 75 hospitals of the National Health Service in England and Wales, comparing the doctors' diagnoses made in the patients' lifetimes with the result of postmortem examinations. For cancer of the lung, they found that doctors had diagnosed 338 cases when the pathologists had found 417; in only 227 cases did the doctors and the pathologists agree.
In fact 33% of cases diagnosed as lung cancer were wrongly diagnosed, and 46% that were supposed to be something else were actually lung cancer.
Another study by Feinstein and Wells (Trans. Assoc. American Physicians 87) found that heavy smokers have a 90% chance of getting their lung cancer diagnosed (not necessarily correctly, of course), while non-smokers had a 62%chance. Doll and Hill stated that they sought confirmation of the cause of death from the doctor certifying the death, who, as we have seen had an almost 50% chance of being wrong, and when necessary from the consultant to whom the patient had been referred, who presumably had a good chance of being wrong too. Presumably a fair number of the patients whose diagnosis was later proved to have been wrong had seen consultants.
Doll and Hill did not mention any postmortems except very briefly and curiously inconclusively:
And remember that according to the Hungarian survey there was a very good chance that pathologists would disagree about the actual details of the disease, which presumably would include the primary site. If the cancer has spread to the lung from another primary site, the cause cannot be automatically put down to smoking.
In 1980 a lengthy report was compiled on rate of cancer in the USA. Enormous sums had been spent on 'the war on cancer', yet, as Professor Samuel Epstein later wrote in 'Losing the War Against Cancer', published in the International Journal of Health Services (1990), claims that overall cancer survival rates had improved dramatically in recent years were questionable. These claims, Epstein said, were based on 'rubber numbers', and ignored such factors as earlier diagnosis of cancer, which led to apparent longer survival rates, and the over-diagnosis of benign [non-cancerous] tumors.
Epstein went on to describe the work done on this report. He showed that it excluded from analysis those over the age of 65 and blacks, who had the highest, and increasing, cancer rates.
Epstein called this badly 'manipulation', and it's difficult to see what else he could have called it. He also pointed out that they had claimed that occupation was responsible for some 4% of cancers, which he called 'a wild 4% guess'. They also decided that diet was determinant in some 35% of all cancers.
Now we come to an interesting example of the sort of thing you can do with statistics. On 8 June 1997, the Sunday Times published an article stating that the British government's chief medical officer, Sir Kenneth Calman, had approved for publication a report on Nutritional Aspects of the Development of Cancer, which stated that 30 to 70% (a very wide spread, but let it pass) of all cancer cases are linked with diet, and further that diet is ten times more important than the effect of occupational causes and of smoking on all cancers.
We know that 90% of all lung cancer deaths are attributed to smoking. In 1986 the US Environmental Agency estimated that radon gas, naturally present in soil containing certain rock formations, could be responsible for up to 30% of all lung cancer deaths. Another recent estimate was that 40% of lung cancer deaths could be attributable to occupation. An unspecified percentage has been put down to diesel exhaust, and more to the keeping of pet birds, drinking coffee and the rest of it. There is also strong evidence suggesting a distinct ethnic factor in the incidence of the disease.
American Indians smoke considerably more than hites or blacks, yet have about half the rate of the disease. Only about 2% of Chinese women smoke at all, but they have one of the highest lung cancer rates in the world.
So we have 90% plus 30% plus 40% plus 30 to 70% plus further percentages cause by diesel, pet birds, ethnic factors? Many cancer cases must have several different causes -- not so much 'being kill' as over-kill.
This sort of thing is reminiscent of the story of the epidemiologist in the US who added up the numbers of people supposed to be suffering from various diseases. The total exceeded the whole population of the country be so much it seemed that not only was everyone ill, but most of them had several diseases at once.
On the question of ethnic factors, there are some very curious disparities. That of the Japanese, the second heaviest smoking nation, with the world's longest life expectancy and a low rate of lung cancer, has already been mentioned. Then there are the Greeks, the world's heaviest cigarette smokers, who for years had the longest life expectancy in Europe, very little below that of the Japanese. Their record has now been taken over by the Swiss (the second highest smoking nation in Europe), and the Spaniards, also heavy smokers, and the Swedes, who don't smoke many cigarettes, but do smoke cigars.
Does the anti-smoker live longer?
The Americans, in spite of their intense and often successful anti-smoking activities, are at 21 on the life expectancy list, with Central American Costa Rica, the original banana republic. As already mentioned, American Indians have half the lung cancer rate of the less heavy-smoking American whites; and Asians domiciled in the US, who smoke more than the whites (in some groups smoking rates among the men reach 90%) can nonetheless expect to live seven years longer than other Americans.
In Smoking and Society, Professor Eysenck discusses this question in great detail. He points out that relative risks in different populations vary from 1.2 to 36.0 on men, and 0.2 to 5.3 in women. (The 0.2 figure indicates that smokers were less likely to get lung cancer.)
Eysenck quotes the US Surgeon General's 1982 report as saying:
and points out that this isn't true. The risk ratio only gives an evaluation of the strength of a risk factor once its causal effect has been proved -- it can't be used by itself to prove the causal relationship. The storks certainly nest on the roofs of the houses where large families live, but you need some more evidence to show this proves that they bring babies.
Eysenck points out that this extraordinary variation can't be explained by the simple smoking-causes-lung-cancer theory. He discusses the possible role of genetic factors as the determining variables, for which he makes a good case.
These wide differences would also fit the virus theory of the disease, in that different populations could be exposed to the infection in differing degrees. American Indians tend to live apart in their reservations, and many Asians in the US keep very much within their own communities, by necessity or choice. It is possible that they are less exposed to infection by the general population.
Blood or breeding?
In the endless debate on 'nature or nurture' it is now generally accepted that genetic factors are of very great importance in all individual human manifestations. This was denied for many years by many people, and the controversy rumbles on. Freudians were obliged to deny the power of heredity, because everything had to be explained by upbringing. Babies, as described by an eighteenth-century pedant, were 'little lumps of flesh': according to Freud early upbringing was all that shaped character. Karl Marx preached much the same; only instead of the family, he blamed capitalist society.
The opposite view has been strengthened by the studies of identical twins separated at birth. They truly share an identical genetic heritage, and they are often almost comically similar in habits and attitudes. What evidence there is does suggest that the habit of smoking itself may have a genetic element.
Eysenck discusses the apparent enormous increase in cancer, and reminds us that cancer is primarily a disease of the old. Life expectancy in the developed world has risen very sharply, but people still have to die of something, and the older they are the more likely they are to die of cancer. He also calculates that if you remove this age factor from the statistical computations, the increase disappears. Methods of diagnosis, too, have changed drastically (though not enough, you might think). Cancer, particularly of the lung, was probably greatly under-diagnosed a hundred years ago. Certainly in the past people who coughed and lost weight and were generally poorly were very likely to be told hey were 'consumptive'.
What about heart disease, then? It's on the cigarette packet in capital letters: SMOKING CAUSES HEART DISEASE. The most authoritative study on this is certainly the Framingham Heart Study, which is known as the Rolls Royce of studies. In this town in Massachusetts, 5,127 men and women have been studied since 1948. They have had the fullest details taken on their health and life-style, and have been checked every two years. Dr. Seltzer of Harvard University discusses this study at length in 'Framingham Study Data and "Established Wisdom" abut Cigarette Smoking and Coronary Heart Disease', Journal of Critical Epidemiology 42, no. 8 (1989).
The results of the study show that there is no relationship between smoking and heart disease in women except a very slight favorable one (women who smoke have a very slightly lower rate of angina, not statistically significant).
For men, the relative risk starts at 1.3 in smokers of forty or more cigarettes a day. Remember, the risk ratio of 2 has been designated the lower boundary of a weak association, so this means in fact a non-significant association. This risk went down to exactly one, that is, no risk at all, as the subjects aged. When information about certain of the other 300 risk factors for heart disease were taken into account, the relationship between smoking and heart disease was lost. Dr. Seltzer asks: 'What use did the Surgeon General's Report in 1983 make of these results?' and quotes the report as follows:
It starts by declaring that 'cigarette smoking is a major cause of heart disease (CHD) in the US for both men and women.
One can only admire such a creative usage of statistics and epidemiology.
Middle-aged at 69?
In a letter to The Times in July 1994, a much-quoted popular statistician stated that 50,000 people per year in Britain dies 20 or 25 year before their time because of their smoking.
In that year the average expectation of life for men was 72 years, and for women 78. If smokers were losing '20 to 25 years of life', the men would have dies at ages between 47 and 52, and the women between 53 and 58. The official statistics gave the latest (1992) available figures, which are given at five-year intervals, so that the deaths were in a longer period of time. Deaths in England and Wales from all causes between the ages of 45 and 54 for men were 13,117, and for women between 50 and 59 were 12,305, a total of 25,422. Even when the much smaller figures for Scotland and Northern Ireland were added, the total is nearer half the statistician's, and remember that this is of all deaths, for a longer period of time. The specific figures for cancer of the lung, trachea and bronchus were 1,164 for the men, and 984 for women. (Remember that a certain number of these were certainly misdiagnosed, though we don't know whether that would leave us with more or fewer genuine cases.) For all heart and circulatory system diseases, the numbers were 5,294 for men and 2,996 for women. Even if all these deaths were of smokers, and all can be attributed to smoking and nothing else, the total is just over 10,000.
However, it must be remembered that expectation of life rises as you age, so that a man who reaches the age of 60 in one piece can expect to live another 18 years, and a woman 22. It is said that on average people who die between 35 and 69 do so at 62, so you can say they've lost 18 years of life. But it s also necessary to take into account the fact that many people who die prematurely do so because their parents did; that is, there is an inherited family predisposition. Many diseases have at least a strong hereditary factor in them, and so do psychological traits that, as Professor Eysenck has shown, can certainly affect your health.
In a report published by Lancet in 1992, it was said that half of smokers died in middle age, which was defined at ending at the age of 69. The question of the definition of ages has shifted so much with the longer life expectancies in this century that there is some variation in the figure usually given. People asked tactfully what they'd call the end of middle age give 55 or 60. Pushing it to the extreme, one might say 65, men's retirement age. Politician, and some others, tend to see 45 as still 'young'. WHO figures state that at 65 you can expect 13 more years of life, which would give us a 'youth' of 45 years, middle age of 20 years, and old age of 13 years. This looks much out of proportion, and 70 as the onset of age makes it still more so.
Much as some of us may be flattered by the thought, that we're only a few years removed from lusty middle-age, surely there are not many 68 or 69-year-olds who would have the face to describe themselves as middle aged.
There is also the figure put out by anti-smoker bodies in the UK in recent years, that 300 smokers died a day. This sounds awful enough, until you ask how many people die a day anyhow. The figure for the year in which this was being said was some 620,00, i.e. about 1,700 a day. But smokers are about 30% of the population. The 300 only represented about 17%. The figure as given could mean only that 13% of smokers weren't dying at all.
When that was kindly pointed out, the anti-smoker organisations concerned switched, without explanation or apology, to saying that the 300 died of 'smoking related diseases'. But there are a very large number of diseases described as 'smoking-related'. Cancer and heart disease accounted for some 400,000 of the 620,00 total. Many non-smokers must have died of them too.
To return to the heart disease question, Eysenck, in Smoking and Society, deals with this at some length. He discusses, as with cancer, the genetic possibilities, and the large part played by stress and other psychological factors. His reasoning appears to be common-sensical, and difficult to refute, and in fact, has never been adequately responded to at all. There are also the 300 risk factors previously described, and recently some work has been done on the role of an infection, a bacterium causing damage in childhood that results in heart trouble later in life.
But surely, you may say, this question could be settled easily enough. With all these studies being done, couldn't someone do an 'intervention trial', as described in Science without Sense, comparing two groups of smoking and non-smoking people. The answer is that this has indeed been done. There have been a number of studies that have done something like that. But you've never heard of them? When you hear about the results obtained you will see why.
There has been only one that has solely dealt with smoking. This was the first 'Whitehall' study, starting in 1968, which recruited 1,445 British civil servants. Half were encouraged to give up smoking, the others were left alone. After a year smoking in the intervention group (the nagged) was down by 75%. After ten years, 17.2% of this group was dead, as against 17.5% of the control group. This difference of percentage is not statistically significant.
There was no difference in deaths from lung cancer or heart disease, and the only other unexpected result was that the intervention group had 28 deaths from cancer other than lung cancer, compared with the control in which the number of deaths from such cancers was 12. This is statistically significant.
Another study, with a wider range, was the 'Multiple Risk Factor Intervention Trial' (MRFIT) in the US. In this there were 12,866 subjects. They were all shown to be at risk of heart disease because of their lifestyle and general health. (With 300 risk factors that's not surprising.) One group was given drugs for high blood pressure, encouraged to eat more healthily, and to stop smoking. The other was left alone, as in the Whitehall study.
These were not self-selected studies, and seem to have been conducted competently. At the end of the MRFIT study, 41.2 per thousand of the 'healthy' group were dead, as against the 40.4 per thousand of the other.
Scientists investigating the study didn't like the results, and went over them again. They found that the drugs to reduce high blood pressure had in fact increased the death rate among the men given them, and were forced to conclude that the risk factors had nothing to do with the actual risks.
Professor Burch, in a letter to the British Medical Journal (March 1985) pointed out that in these two studies:
Findings for cancer other than those of the lung were even more surprising.
Professor Burch adds, in Can Epidemiology Become a Rigorous Science?
'Many personal enemies' for pointing out the obvious?
Squirming on the hook
The Finnish businessmen's study in the 1980s took 612 48-year-old businessmen and got them to do what was done in MRFIT: change their diet, give up smoking, and take various drugs to reduce blood pressure. They also made them take more exercise.
The control group was of 610 similar men, all 48-year-old businessmen with as far as possible similar habits and life-styles. This is what is called 'randomization'. It doesn't sound random: in essence it means that the subjects and the controls are chosen from similar people by the investigators; unlike the self-selected smokers and non-smokers in the Hill & Doll study.
After the allotted period of 15 years, it was found that the healthy-livers had totted up 67 deaths, and the others only 46. There was no squirming on the hook about this because it was ignored. And that is the method now used with any evidence that conflicts with the accepted version. The deadly effects of smoking have now entered folklore. There is no need for the medicine men to debate anything. The sun goes round the earth, and if you dare to disagree, nobody's going to speak to you, so there. And you're not going o be asked to serve on any official bodies, either.
Another very good example of this sort of attitude was the Australian government's health survey (1989-90 National Health Survey Lifestyle and Health Australia, Australian Bureau of Statistics, Catalogue No. 4366.0, Ian Castles, Australian Statistician). We tend to think of the Australians as manly and tough, a view put across in such amusing films as Crocodile Dundee, but about smoking they have become almost Californian in their alarm about the habit. But then, we once thought the Americans were manly and tough.
Yet, in this government survey, in which there was no question of finance by an interested parties, or any pro-smoking bias, and in which 22,000 families were studied, it came over quite clearly, that smokers were on the whole, in better health than non-smokers, and definitely better than ex-smokers.
This surely sensational news was not mentioned in the British media at all, except for two articles in the quality press, one by Professor Eysenck in the Sunday Telegraph and the other by Lord (Woodrow) Wyatt in The Times.
There was a similar result to a similar survey in France (CREDES study 1988-90), described in Tabac: l'histoire d'une imposture, an admirable little work, witty as only the French can (still) be. This too, found smokers in better health. So, very interestingly indeed, did the US government study (Cigarette Smoking and Health Characteristics, from the National Center for Health Statistics, July 1964-June 1965), at least as far as moderate smokers were concerned.
This last study is particularly noteworthy because it came at the start of the anti-smoker movement in the US, and though its tone generally is commendably fair, it does make the most of such figures as seem to support the PC case. Nevertheless, it is one of the few such studies that takes into account the amount smoked, and shows that people who smoke what anyone would call heavily have more problems than those who smoke moderately, which indeed is what might be expected. It is reasonable to think that anyone who smokes 50 or 60 a day may have more problems to start off with.
You haven't heard of any of these? You won't have The only reference, apart from the two authoritative articles mentioned above, was a grouchy remark, on the Australian study, that the people were all 'self-diagnosed': that is their word was taken for it whether they were ill or not. But after what you know now about doctors' diagnoses, surely the patients had just as much chance of being right. In any case, they would know whether they felt ill, and what their doctors had said about it, and whether they themselves thought they were ill or not. If the favorable result proved only that smokers were less likely to be hypochondriacs, that would surely say something positive about 'the bastard'. And it should be remembered that not only are hypochondriacs a nuisance to their doctors, their families, their friends, and the NHS; they can also be a danger to themselves.
There was a study in Heidelberg, described by Professor Eysenck in Psychological Reports (1989) in which 528 men were asked whether they, as smokers, were convinced that they would be very likely to develop lung cancer, heart disease, or other 'smoking related diseases'. The 72 who answered 'yes', while admitting that their views were taken from information in the media, had an almost three times higher death rate at the end of 13 years than those who were not so influenced.
Fear can kill. This has been known since disease was first studied. We are entitled to wonder how many people have been killed more by the fear of 'smoking related diseases' than by any actual disease itself.
Certain efforts have been made to find out where some of the figures on 'smoking related deaths' have come from. When Baroness Jay reported to the House of Lords (24 July 1997) that the number of smoking-related deaths in the UK was 120,000, she admitted that 'smoking is not recorded on the death certificate and analysis is not available on individual death level'. It is difficult to see how any real estimate of the actual deaths can be made. It is also odd that the Royal College of Physician gave the number of deaths at 50,000 in 1984, since when smoking has decreased considerably, while 'smoking related deaths' have more than doubled. Ah, the experts say, the deaths relate to the higher smoking rates of years before, since these diseases take a long time to develop (in Japan, remember, some 49 years). Yet the smoking rate was dropping for years before 1984. Why didn't the deaths in that year reflect the earlier higher rate of smoking?
Ask a silly question
In the US, as in Britain, death certificates do not record anything like smoking (except in very few states where about 3% of deaths are described as 'smoking related'). The official figure of 'smoking related deaths' there is 450,000. Or 470,000, or 500,000, according to whom you believe.
Dr. Bernard M. Wagner, editor of Modern Pathology, in May 1996 told of the experiences of a Detroit News reporter, Nickie McWhirter, who dares to ask: 'Is it true that 435,000 Americans die every year from smoking related illnesses?' (This was the number circulating in her circles, apparently.) She was told to contact the local American Lung Association office, but they didn't know. Then she was told to contact the National Center of Health Statistics, a branch of the National Center for Disease Control, and given a telephone number, but it didn't help because they didn't know. Several telephone calls round several different wrong civil service numbers later, she contacted someone in Statistical Resources at yet another department, who said that his office collected mortality based on death certificates. The data is categorised by race, sex, age and place, but not by smoking. So he didn't know. Our heroine refused to give up. She tracked the Office of Smoking and Health to its office in Atlanta. The public information officer said the 435,000 figure came from its computers based on formulas specially programmed for 'smoking related stuff', as he put it. She asked him how, if no lifestyle data on individual patients and their medical histories are collected, the computer can possibly decide whether deaths are 'smoking related'. He didn't know. She was given the number of the Operations Manager of SAMMEC, the computer programme Smoking Attributable Morbidity, Mortality, and Economic Cost, who explained that the computer is fed raw data and SAMMEC uses various intricate mathematical formulas to determine how many people in various age groups, locations and other categories are likely to get sick or die from what diseases and how many of these can be assumed to be smoking related.
We're back to 'assumed'. The manager confirmed that no real people, living or dead, are studied, no doctors consulted, no environmental factor , or presumably any others, are considered. He was lyrical about SAMMEC and what it could do, provided it is fed the appropriate SAFs. These are the 'smoking attributable fraction' for each disease or group of people studied; they are derived from a mathematical formula. The reporter demanded to know whether at any time some human being looked at other human beings, talked to their doctors, somehow gathered enough information from reality to begin to devise a mathematical formula that might be applied to large groups of people much later, without needing to study these people. The manager didn't know.
He thought the original work was done by A.M. and D.E. Lilienfeld in their Foundations of Epidemiology (OUP, 1980). Mr. Lauren A. Colby, an American lawyer, author of an interesting short-study, In Defense of Smoking, says it isn't.
No real people, living or dead, are studied. Figures are studied (and estimated, and assumed, and cause-coded, and the rest of it): people never.
Heil the computer....
FORCES INTERNATIONAL (Forces, Inc.) is a non-profit educational corporation organized under the laws of the Commonwealth of Virginia, USA. Forces, Inc. has received a charitable tax exemption under Internal Revenue Code 501(c)3. Your contribution is tax deductible. FORCES INTERNATIONAL © trademark, logo and original material copyright © 1995-2004 is property of FORCES INTERNATIONAL unless otherwise indicated. Original FORCES material may be reprinted or used in any non-commercial form if proper credit is given to FORCES and original intent remains intact. All other material remains property of its creator/owner.