Number of the Month
Every now and then the comments inbox at Number Watch begins to buzz, signifying a junk science event of unusual significance. So it was with that Lancet paper.
The important thing when judging such a paper is to realise that by far the greater part of it comprises misdirection (the patter and activity by which the conjuror distracts the audience from what he is actually doing). Cut away the persiflage and there are just two statements that matter. The first is a risk of death estimate (RR=1.5, 95% CI 1.1-2.3) and the second is a conversion of this into a virtual body count of nearly 100,000.
A relative risk of 1.5 is not acceptable as significant. Those who have read Sorry, wrong number! might remember that Headline Man obtained a relative risk of 1.63, when there was no underlying effect at all. So even without the major considerations of confounding factors and biases, real science is obliged to reject such small risk increments as insignificant, due to random variation.
It is the second step, however, that creates the illusion. The number of sample deaths corresponding to the unacceptable relative risk is then multiplied by the ratio of the actual population to the sample size to yield a virtual body count. It is a widely used trick in epidemiology. The EPA, for example, used it to create an annual 3,000 imaginary corpses due to passive smoking; a lie that has had world wide social consequences. The relative risk is almost never exactly one, so on this basis every such study, however banal, will produce a loss (or saving) of thousands of lives.
Further comments have been made by the likes of Michael Fumento on the fact that a once respectable British journal sought to make such a crude intervention in an election taking place in another country (further comments on which here).
Which all brings us to another mystery. What is Tim Worstall apologising about? His statistical argument is slightly confused, but not enough to warrant the donning of sackcloth and ashes.
An assiduous number watcher has pointed out that the specific relative risk (1.5) was mentioned in the very section (Headline Man) of the said book, in the following paragraph:
One of the lessons of these simulations is that a problem with an implicit risk ratio of, say, 1.5 might exist; but the fact is that it is beyond the reach of scientific statistical methodology (nothing, of course, is beyond the reach of junk science), which is not a difficult concept for scientists, who are used to such ideas as the uncertainty principle. It is, however, not so easily acceptable to the layman, a fact that is ruthlessly exploited by the villains of our piece. One of the most difficult things to get over to non-mathematicians, including most journalists, is that 90% is NOT a large number. It is less than one. What causes difficulties for the innumerate is that 90% is a large proportion (nearly all) but it not a large increase (less than a doubling). What are examples of really significant risk ratios? Well, heavy cigarette smoking has an association with lung cancer with a risk ratio of about 20, while aspirin produces a relative risk of 35 for Reye's syndrome in children. This is where epidemiology can be effective – for large risks of comparatively rare diseases.
There's one for the mystics.
Footnote 2: It is an unproductive and depressing exercise to get involved in spats in the blogosphere, but students of woolly thinking might like to pick the bones out of this.
Footnote3: (10/11/04) apparently this discussion is rambling on. For those who did not follow the link on RR above, the statement about the unacceptability of 1.5 applies to observational studies, not necessarily to properly randomised double blind surveys that produce a highly significant result.
The improver of natural knowledge absolutely refuses to acknowledge authority, as such. For him, scepticism is the highest of duties; blind faith the one unpardonable sin.
Thomas H. Huxley
On the whole, it proves wise to resist fleabites from the blogosphere, because scratching them just creates an infinite cycle of irritation and response, but sometimes the temptation to scratch becomes irresistible. Your bending author has been repeatedly accused of “insisting” that the hole in the ozone layer was always there, though how words such as “probably” and “moot” amount to insistence is something of a mystery. A particular sin was failing to comment on this particular graph from the EPA.
Well, there is going to be no comment on this graph. However, in order to show willing, below is a comment on another graph entirely.
Scientists at the Phlogiston Research Unit at the Metropolitan University of Nether Wallop have been monitoring the length of hours of daylight and have uncovered disturbing evidence that they are decreasing severely. The graph shows how the hours of daylight have steadily diminished since measurements were begun in July.
The Head of Computer Modelling within the Unit, Dr Hugh Jerrors, says that this phenomenon could have a severe effect in areas such as agricultural production and drastic action needs to be taken.
A search of the literature suggests that one preventive measure would be to sacrifice a virgin at Stonehenge, which is fortunately close to the University. Less fortunately, a search for virgins in the environs of Nether Wallop proved fruitless. Further research is needed.
A regular reader, after some kind comments, notes that many of the quotations here come from the Daily Telegraph, and wonders why other newspapers are not so frequently quoted. In fact, every day your bending author reads The Times and then the Telegraph, the former being a much more productive source of junk. However, overseas readers tend to get upset when they cannot access articles in The Times without payment so, as remarked a few months ago, links to it have been reduced. Other British newspapers do not have a comprehensive internet version. There is always the BBC, which is great for junk and has superb internet coverage.
Most of the stories now come from the global early warning system provided by number watchers, but only a few are selected to illustrate particular points. There is no point in Number Watch trying to compete with www.junkscience.com for comprehensiveness. There are so many examples from the likes of the BMJ that are amusing from the point of view of being downright silly (this one for instance) but one bit of junk epidemiology is very like another and it is important not to lose sight of the wood for the trees.
A wealthy Yorkshire man hired a stone mason to carve an edifice for his dead wife with the inscription "She were thine". When he visited the graveyard to see the result he read "She were thin". He phoned the mason and said "You left out the E" and was told "Don't worry we will soon put that right".
The next week he visited the grave and read "Ee, she were thin!"
That inbox is buzzing again. This time it is the shock horror stories about vitamin E. The vitamin boost that could cause early death yells The Times. Vitamin E may do more harm than good says CNN. High dose vitamin E death warning from the BBC. Heart attack risk from vitamin E says the Telegraph.
Going back to the original source, we find that it is yet another meta-analysis, requiring the efforts of no fewer than six authors. Only registered masochists should attempt to read the whole thing, but figure three and its associated table of data might provide some entertainment. The relative risks rise from 0.98 to a horrifying 1.08 (95% CI 1.01-1.14).
We now have the delightful prospect of a battle royal between two sides of the junk science movement. There is a billion dollar industry of dieticians, gurus and manufacturers coining money from the anti-oxidant scam versus medical academics involved in the eternal paper chase.
Of course, taking vitamin supplements for anyone on a normal balanced diet is just a stupid way of giving away your money. Furthermore, even if the “research” had produced an impressive result, there remains a causality problem – does taking a vitamin overdose cause serious illness or does serious illness cause taking a vitamin overdose?
Don’t bet on the result, even if one side has all the wealth. Look what happened to the tobacco industry. The anti-tobacco zealots in the UK have been stoking up the hysteria ready for the big push on smoking in public places. In the same issue of The Times is a lead article English pubs offer to ban all smoking in five years – surrender! Leading zealots ASH have now inched up the virtual body count to 114,000 a year. We know this is a lie, because it is pro rata considerably higher than the US equivalent of 400,000, which has been comprehensively exposed as a fraud. In more secretive Britain it is not considered necessary to publish your evidence, so there is no danger of it being exposed.
Millions are at risk from different CJD strain yelled The Times headline on November 12th. The BBC had the more restrained BSE 'linked to different CJD types'. The story emanated from an old friend of number watchers, Professor John Collinge. His attempts to keep the CJD scare in the air gave rise to outbreaks of MMC in May 2001 and WIDIMITWEED in December 2002. A year later Old Faithful came up with a thousands to die headline from the same source. The annual scare has come a bit early this year. Anyway, who is this upstart Sam Lister? Where is our Nige?
Which all serves to remind us that the annual Numby awards are almost upon us. How time flies when you are enjoying yourself! Nominations are invited and this year the best nomination with justification will earn the entrant a luxury two week holiday in Fallujah.
Following mention of the buzzing inbox, it might be illustrative just to give some flavour of the e-mails that come into Number Watch. Here is just a small representative selection of those in the last week.
Our Man in Puerto Rico followed up the Pizza People:
No cancer risk in marital status
We know (almost) all causes of cancer in young adults
Plus nothing works (not even pizza) if you don't have have familial predisposition to acute myocardial infarction.
Dig this conclusion: "Family history of colorectal cancer interacts with environmental risk factors of colon cancer."
We are doomed by natural selection: "Our study indicates that in populations with frequent wine consumption, wine per se can strongly increase the risk of cancer of the oral cavity and pharynx, and confirms that the most prevalent alcoholic beverage in each population tends to be the one with the highest risk."
Forget pizza! "Alcohol drinking during meals was inversely related with risk of acute MI, whereas alcohol drinking outside meals only was unrelated to risk."
I love this one (I have BHP)
OMIPR also discovered this modest claim about pollution.
Theresa M Rice found that chocolate had lost its charmed status:
Haven't seen this yet on NumberWatch - someone in California is suing chocolate manufacturers about the amount of the toxic metals lead (Pb) and cadmium (Cd) present in chocolate, using California's Proposition 65 as the basis.
The Attorney General of California has dismissed the lawsuit as frivolous.
John Davis noticed that the Greens are a bit confused about the conservation of matter (and they don’t seem to be aware that most of the Earth is green, either)
I noticed the other day that there was a "missing carbon sink" mentioned in a newspaper article (just the odd two billion tons or so...), so I thought I'd try and find out more about the Carbon cycle. This is informative:
As far as I can see they assume that about 200, or is that 400, (billion?)tons of CO2 flows through the atmosphere yearly, of which 6.5 is contributed by Fossil Fuels and 1 or 2 by Deforestation. Strangely, in the midst of this vast & largely unquantified flow, some of the excess CO2 we think we are generating seems to get lost...
Then in all their subsequent calculations the main flow is neglected utterly. Presumably it never ever changes, even by as much as 1% ?
The big story is, of course, the push by anti-smoking zealots, for a complete ban.
Brad Tittle observes:
The other day I heard: "500,000 deaths a year are caused by second hand smoke." A day later it was: "50,000 people die each year in the US because of second hand smoke."
How many people have died because of Vitamin E?
Show me the bloody bodies. I want a medical examiner to have a body on his slab and be able to say "This man died of second hand smoke". "These people died from arsenic poisoning." I am firmly convinced that every body that appears on an ME slab has been killed by 700 different things. It doesn't matter that there is and ax through his head, he really died of excessive caffeine consumption, second hand smoke, toluene exposure, carbon dioxide inhalation, and water consumption (he either drank too much or too little).
Dennis Ambler adds:
As a "never have" smoker, I find the accelerating clamour on cigarette smoking is becoming quite hysterical. Methinks they do Profess too much! The current claim of 100,000 deaths per year means that smoking is responsible for over 16% of all deaths in 2003, difficult to square with their own figures, unless they include all the scam claims about it causing everything, ( see the site ref: scooponsmoking, below.)
In 2003 there were 611,800 deaths registered in the UK. Deaths at age 75 and over comprised only 12 per cent of all deaths at the beginning of the last century. They rose to 39 per cent in 1951 and 65 per cent in 2001. These have to be called in the main, deaths due to old age, and comprise nearly 400,000. I wonder how many of the "100,000" are aged 75 or over?
And you really should have a look at this site:
Here's an example "fact" on the skeleton......
one group of people who had broken their legs, complete healing took an
average of 269 days in smokers, as compared to 136 days in nonsmokers. (1)|
The body system affected (by smoking)
respiratory system (lungs)
skeletal system (bones)
muscular system (muscles, joints)
circulatory system (heart, arteries)
urinary system (kidneys, bladder)
digestive system (stomach, intestines)
nervous system (brain, nerves)
endocrine system (thyroid, hormones)
female reproductive system (uterus, ovaries)
male reproductive system (penis, testes)
immune system (t-cells, anti-bodies)
the senses (eyes, ears, nose, mouth, throat)
I have pasted in their web site description........notice how they can't quite bring themselves to say it's education!
The American Council on Science and Health is a consumer eduction consortium with a board of 350 physicians, scientists, and policy advisors.
Their home site claims to de-bunk junk science and has, ironically, a link to Stephen Milloy.
Denis also found found some common sense from an unexpected source.
But with almost all the media toeing the party line, you have to go to Sp!ked to get a reasoned view of the anti-smoking campaign.
Then there’s still vitamin E. John Baltutis says
“Here's one to make your day!”
Niclas Sand Engberg found an RR of 1.67:
This may be of most interest to people in the Scandinavian countries since they are heavy consumers of so called moist snuff (a.k.a.
Copenhagen snuff). An investigation carried out at Karolinska Institutet in Stockholm has revealed that extensive use of snuff causes an increasein the risk of pancreatic cancer by whopping 67%. Immediate calls for a ban on snuff have followed. Newspaper aritcle (in Swedish)http://www.dn.se/DNet/jsp/polopoly.jsp?d=597&a=343354
The study is led by Professor Göran Pershagen. He is one of the leading figures in Swedish epidemology and he has previously shown how air-pollution has caused a 40% increase in the risk of cancer for those living in areas with heavy traffic.
Following our discussion of Gamesmanship, Ray Futrell draws attention to 38 ways to win an argument.
And that's just a few of them. It's a funny old world.
Details of this and associated events are available here.
‘In any investigation, my Bunter, it is most damnably dangerous to have a theory.’
‘I have heard you say so, my lord.’
‘Confound you – you know it as well as I do. What is wrong with the doctor's theories, Bunter?’
‘You wish me to reply, my lord, that he only sees the facts which fit into the theory.’
‘Thought-reader!’ exclaimed Lord Peter bitterly.
Dorothy L Sayers, The footsteps that ran.
10.05 pm and just come from a viewing of the channel 4 Dispatches Programme MMR: what they didn't tell you. This wisdom after the event is absolutely maddening. It is an extraordinary tale, not just of junk science, but of hidden financial interests, snake-oil salesmanship, media manipulation, arrogance, dismissal of evidence from their own laboratories and a host of other offences. What they played down was the part played by the media and completely forgot to mention THIS from the channel next door.
Some of us expressed our doubts long ago, but the anti-scare does not make a good media story. It is too much to hope that the media will have any conscience about the damage to children if the threatened outbreaks of disease do occur.
Number watcher Dr Mark Wilks wrote
Did you see these? Article and commentary. You'll love the RR values!
Hung HC, Joshipura KJ, Jiang R, Hu FB, Hunter D, Smith-Warner SA, Colditz, GA, Rosner B, Spiegelman D, Willett WC.
Fruit and vegetable intake and risk of major chronic disease.
J Natl Cancer Inst. 2004 Nov
Schatzkin A, Kipnis V.
Could exposure assessment problems give us wrong answers to nutrition and cancer questions?
J Natl Cancer Inst. 2004 Nov
Thank you! They are real collector’s items. The source is our old favorite the Harvard Nurses’ Health Study. It is the same old data dredge based on anecdotal evidence, this time being mined for the same old five-portions-of-fruit-and-veg scam. Unfortunately, they did not find any evidence of protection from cancer, contradicting the establishment “fact”, thereby launching an editorial fumble among the entrails for an explanation. However, all is not lost, as they managed to “prove” protection against coronary disease – relative risk, wait for it, 0.88. There is much to savour. This for example:
Relative risks were adjusted for total calorie intake, age, smoking status, alcohol use, body mass index, multivitamin and vitamin E supplement use, physical activity, family history of myocardial infarction, family history of colon cancer, personal history of hypertension, personal history of hypercholesterolemia, personal history of diabetes, and (for women only) family history of breast cancer, menopausal status, and use of hormone replacement therapy. Relative risks are based on a continuous measure for increments of five servings/day for all fruits and vegetables, three servings/day for all fruits and all vegetables, and one serving/day for other food groups, using median values of deciles of intake to represent the exposure of all participants in the same deciles.
No summary could do justice to this fine example of epidemiological art. It deserves to be hung up in the Tate Modern Museum along with Damien Hirst’s pickled sharks and Tracey Emin’s unmade bed. It is well worth a bit of a Google on this one, as the angles taken by the various media “experts” are a revelation.
Don’t forget the Number Watch Forum, accessible from the index page. Many contributions come in that would be of interest to other number watchers, more than your bending author can cope with.