Judith Hatton

Lies, Damned Lies And ... Statistics

We are told that 120,000 people in Britain die a year as a result of their smoking. Where does this figure come from? It doesn't come from death certificates, which don't include any information about 'smoking status'. It doesn't come from inquiry among relatives of the deceased, except in certain exceptional circumstances when a few of these people are asked a series of questions which may include 'did she eat chicken with the skin on?' and 'how many men did your mother/sister/wife/daughter sleep with, and did they smoke?' Imagine the sort of answer you'd have got if you'd asked that question in her lifetime...
 It doesn't, except in very exceptional circumstances, come from doctors. So where does it come from?

Let us look at the authority on this: the OFFICE OF NATIONAL STATISTICS 1993 (revised) and 1994 MORTALITY STATISTICS."From 1993 an automated coding system has been used to assign the cause of death...Most of the changes in numbers of deaths for particular conditions arise from a revised interpretation of WHO coding rule 3, and from the absence of medical enquiries(emphasis added)...

"In the coding of underlying the cause of death certain conditions, as reported on the death certificate, can be assigned only to very broad categories. Up to 1992 OPCS handled these cases by applying to the certifier" [the doctor signing the death certificate] "for further information, in order to assign a more definite code. This procedure has not been used from 1993 because we were unable to deal with these deaths in a timely way...

"Coders used a standard list of certain diseases and conditions which would generate a medical enquiry if found on a death certificate. No follow-up was sent if the ME produced no reply(emphasis added)...

"Coders were not generally required to send Mes concerning deaths at ages 75 or over.(NB: the largest number of deaths from cancer and heart disease, the two chief "smoking related" diseases, occur in this age group. So there has never been any check on the diagnoses in these deaths, nor now it seems, on any others, because "we were unable to deal with these deaths...").

A joint report by the Royal Colleges of Pathologists Surgeons and Physicians ("The Autopsy and Audit", 1991), says: "In autopsies (post-mortems) performed on patients thought to have died of malignant disease (cancer) there was only 75% agreement that malignancy was the cause of the death and in only 56% was the primary site identified correctly." (So if you are told you have cancer there is a one in four chance that you haven't, and even if you have there is almost a fifty-fifty chance that you're being treated for one in the wrong place).

The report ended: "Such high levels of discordance mean that mortality statistics which are not supported by autopsy examinations must be viewed with caution."The rate of post-mortems in England and Wales is 27%.

The risk of misdiagnosis is heightened by the fact that it has been found that there are 38% of un-detected lung cancers in non-smokers, 20% in moderate smokers, and only 10% in heavy smokers.

A survey in Hungary, which has a very high rate of postmortems, showed that even when they'd cut you up pathologists couldn't be dead sure of what had killed you in almost 20% of the cases.

Professor Alvan Feinstein, of Yale, a world authority on epidemiology (the study of the causes of disease), has said firmly that death certificates are merely "passports to burial", and for more than 50 years, every time someone has studied the causes of death listed on the death certificates, the conclusion has been that the information is 'grossly inaccurate and unreliable".

So the Health Education Authority, which has done so much for the health of the nation by producing a series of brochures on very high quality paper which no-one any of us knows, or has ever heard of, has been known to have read, produced one in 1996, called "The Smoking Epidemic", in which the figure of 120,000 deaths a year appeared for the first time, as far as can be made out. This researcher, possibly the only person who has ever read it without being paid to do so, first counted up the number of times the word "estimate' or a derivative appeared in its 65 pages: 121 times. She may have missed a few. And of these 65 pages, nearly half are taken up with graphs and charts.

We all know what "estimate" means. "I estimate; you're guessing; he's making it up." If a brochure on a new car was written in similar terms, would you buy that car?

"The Smoking Epidemic" (a catchpenny title), gives its figures as coming from "estimates" drawn from two major surveys, one by the American Cancer Society, and the other the famous doctors' study, by Doll and Hill, in Britain.

Both of these were self-selected, in that some groups of people were asked to give information about their smoking habits, and those who didn't want to didn't reply, these being more than 30% of the total in each case. A very distinguished scientist, Professor Eysenck, has said roundly:

"No relevant conclusions regarding causality can be drawn from studies of this kind."

Obviously, unless you know why one in three of your subjects hasn't bothered to answer, your information will be inadequate. And in the US Surgeon General's report on smoking, a very strange fact emerged: it appeared that in the 5 studies that mentioned the number of people who didn't reply, the death-rate among the non-smokers who didn't reply was more than 38 times as high as that among those who did, but among smokers it was only 1.8 times as high. What on earth this means, apart from smokers being 20 times more likely to politely answer correspondence, no-one could imagine, but it is plain to anyone that it would skew the results to the point of making them meaningless.

There is another point to make about these American studies: it was perfectly plain what the subject of each was. The ideal survey is 'double-blind'; that is, neither the researcher nor the subject knows what it is about. Unless this is so, the temptation is for the researcher to come up with the expected result, and for the subject to provide it. And in two of the surveys, one of which was that used by the HEA, there was a further deep flaw. Volunteers from cancer societies, who might be expected to hold politically correct views on the subject, were each asked to enroll about 10 families. It is not necessary to suggest that these volunteers were any more dishonest than any other interviewers in order to recognize that among them there would be some who would make up their numbers in any way that occurred to them. The chances of being discovered, if they were prudent and didn't invent the whole thing would be very low.

The Doll and Hill survey is always quoted here as the first, though there had been others, notably those in Nazi Germany, which may have reflected Hitler's obsessive loathing of smoking. Of this survey, Professor Vandenbroecke of Leiden said:

Often I have wondered why medical opinion was so suddenly and massively swayed into accepting the lung cancer - smoking hypothesis...The original case control studies by Wynder & Graham and by Doll & Hill are still used in a famous epidemiological exercise...where they serve as examples of what can go wrong (emphasis added): biased ascertainment of exposure, selection of cases and controls from different source populations, poor ascertainment of 'caseness', etc. ... I am afraid that the sudden and total acceptance of smoking as a cause of lung cancer... is a phenomenon that belongs rather to the sociology of medical science (emphasis added).

By biased ascertainment of exposure is meant, it seems, that the subjects were merely asked whether they smoked, and no check of any kind was used - indeed, it would have been difficult in such a large survey to check anything. 'Selection of cases and controls from different source populations' means 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, who differ from doctors in many ways, particularly in having less money, being on the whole less well educated, and unable to get better, or at least more medical treatment, because, unlike the medical profession, they don't know where to go and how to demand it.

A survey used as an example of what can go wrong - others that have extraordinary unexplained 38 times higher death-rate among certain of their subjects, and may well have suffered from a certain bias from the beginning - these are surely poor foundations on which to base your conclusions, even if these are frankly admitted to be "estimates'.

"The Smoking Epidemic", perhaps wisely, doesn't try to go into the curious affair of the different rates of "smoking related" diseases among other nations. The Japanese are the second heaviest smoking nation in the world, and have a low rate of lung cancer (at one time the lowest in the world) and the world's longest life expectancy. And no, this isn't because they didn't start smoking until 1948, as has been said by an anti-smoker expert; they started in 1542. The American Indians smoke a good deal more than other Americans, and have half the rate of lung cancer. There are many more such anomalies.

The Greeks, with Cyprus, have the highest rate of cigarette smoking in the world, and one of the world's lowest cancer rates. They also had for years the highest life expectancy in Europe, and are still up there in the first five. It is also interesting that they have one of the world's lowest rates in respiratory disease in general, and Singapore, with a third of their smoking rate, has the world's known highest. Very few Chinese women smoke, perhaps no more than 2%, but they have one of the highest lung cancer rates in the world. Asians living in the US smoke more than the other races, in some cases up to 90% of the men, but on average they live 7 years longer than these others.

Sir Kenneth Calman, formerly the government's chief medical officer, approved for publication a report stating that 30% to 70% of all cancer cases are linked with diet. He said:

"Diet is 10 times more important than the effect of occupational causes and of smoking on all cancers."

Since we are always told that 90% of lung cancer cases are caused by smoking, does this mean that 900% are caused by diet?

We are also told, by the American Environmental Protection Agency, that radon gas could be responsible for up to 30% of all lung cancer deaths in the US. It was recently "estimated" that 40% of these deaths could be due to occupation. An unspecified number have been put down to diesel exhaust, and some more to the keeping of pet birds, besides whatever figure emerges from the percentages given as caused by diet.

So these "coded", unable to be dealt with, "estimated" numbers of deaths, based in part on a study used as an example to students on how it should not be done, perhaps on death certificates that are "grossly inaccurate and unreliable", and otherwise on studies with a colossal, mysterious unexplained 38 times discrepancy, and the rest of it, are described to us as utterly hard and fast figures, to doubt which would be the first heresy. Not to mention the unexplained differences between different nations, which cannot be explained simply on genetic differences, since the Greeks are racially very similar to other European nations. Or indeed the extraordinary question of the percentages assigned to various causes, and what they add up to. Perhaps the believers in all this go by the old saying:

'I believe because it is impossible.'


FORCES is supported solely by the efforts of the readers. Please become a member or donate what you can.



Contact Info
Forces Contacts
Media Contacts
Advertisers
Links To Archived Categories

The Evidence
Inside Forces
About Forces
Research
Writers
Book case