Friday, 26 November 2010

Overkill

The big news today is that secondhand smoke causes 600,000 deaths a year worldwide. Anyone tempted to take this figure remotely seriously should bear in mind that the authors of the associated study also believe that "smoke-free laws banning smoking in indoor workplaces rapidly reduce numbers of acute coronary events" and "fully smoke-free policies have a net positive effect on businesses." And, interestingly in the light of the Moral Maze the other night, they consider the main aim of smoking bans to be denormalisation:

Above all, [smoke-free] policies contribute decisively to denormalise smoking, and help with the approval and implementation of other policies that reduce tobacco demand, such as increased tobacco taxes and a comprehensive ban of tobacco advertising, promotion, and sponsorship.

Still, you can't fault their impeccable methodology:

To identify national data for second-hand smoke, the keywords “second hand smoke”, “environmental tobacco smoke”, and “passive smoking” were combined with names of countries or regions, by searching Google and the PubMed database

The spirit of Newton and Darwin lives on, does it not?

What can be said of the 600,000 estimate? Obviously if you extrapolate a relative risk over a larger population, you will get a larger number. That doesn't make the study or studies you are extrapolating from any better, but I won't risk waking this blog's resident troll by going into all that again.

It's interesting that this study appeared in the Lancet because back in the 1990s, Lancet columnist Petr Skrabanek* remarked on the tendency of public health campaigners to use larger and larger populations to get scarier and scarier death counts. He called them "jumbo-jet numbers" because talking about the equivalent number of jumbo-jets that would have to crash for the same death toll sounds more dramatic than explaining that you have to die of something and that most of the people being "killed" were very old. This, from a collection his articles [PDF]:

The intoxication with numbers is another characteristic of the modern crusaders against smoking. They typically use big population blocks as denominators to obtain bigger and bigger numbers. Richard Peto, a leading anti-smoking exponent and a statistician, announced that 'of all children alive today in China under the age of 20 years, 50 million will eventually be killed by tobacco.' This would put the hecatombs of the Second World War in the shade. The British Medical Journal referring to another such statistic, 'typical of the Oxford epidemiologist Richard Peto', quoted that 'about 20 million children now living in Europe will be killed by tobacco in their middle age.' And The Times reported on January 1, 1988, that according to Mrs Risk-factor Epidemiology: Science or Non-science?

Edwina Currie, 'more than a million schoolchildren and 60,000 babies born this year will die of smoking-related diseases such as lung cancer.' Surprisingly, no-one has yet used the population of the whole world as a denominator: this would produce numbers of babies, toddlers, schoolchildren, and other children, killed by tobacco in truly phenomenal ranges. Perhaps the reason for this reticence is the fear of overkill.

Well, now they have. In fact, they've missed a trick by not using the other PR tool of multiplying by years (eg. here). So, to save someone the trouble, here's my study:

'A quantitative, longitudinal assessment of worldwide mortality from environmental tobacco smoke'

Christopher J. Snowdon

Methodology: Google and PubMed of course. I'm a professional.

Results: 600,000 x 10 = 6,000,000

Conclusion: Secondhand smoke will kill 6 million people over the next ten years. Where's my cheque?


[*nb. Petr Skrabanek was a skeptic about—amongst many other things—secondhand smoke. You can probably guess what the anti-smokers' response was. That's right, they accused him of working for the tobacco industry. Which was a lie. He died of a non-smoking related disease in 1994.]

17 comments:

Anonymous said...

What amateur stuff!
Don't they know anything about astronomy?
There are estimated to be over 200 billion stars in our galaxy alone and 200 odd galaxies just in the observable universe. According to my modelling, about 1 in ten have planets that could at some point support life, including evil plants such as tobacco. So each planet may become inhabited with 6 billion intelligent life forms and have similar exposure to us on Earth. I make that 600,000 X 20 billion X 200 billion = 2,400,000,000,000,000,000,000,000,000 deaths every year from passive smoking.

Don't these people care?

Tony

Dick Puddlecote said...

"The spirit of Newton and Darwin lives on, does it not?"

I just spat half of my gin back in the glass. Thanks for that. ;)

Paul Bergen said...

Was going to blog about this as well but happy to see your most capable hands on it. My sense is that if they said any larger number most of the usual crowd would just accept it but if they had any smaller numbers there would be an outcry and a demand for a detailed proof.

Gary K said...

They include SIDS/COT deaths and anyone that claims that SHS/smoking 'causes' childhood deaths,like from SIDS, is a cruel,vicious moron................... By definition SIDS/COT death is a death for which there is NO known cause......even by forensic examination like an autopsy!!!!..............anti-smokers that make such a claim about SIDS deaths are using tragedy to advance their agendas and they are rotten,inhuman scum!!!!

Peopole that feel comfortable lying about terrible tragedies are capable of lying about anything.

Snowdon said...

Paul,

Please do write something. Increasingly, these days, I find the stupidity too big to acknowledge let alone confront. The idea that 1 in 100 people die from passive smoking is so insane I can hardly believe that anyone could believe it. It would mean, for example, that if you read a newspaper that produces 3 obituaries a day, you will read - on average - about one passive smoking death a month.

And yet the maths is right (isn't it?). It's the logical conclusion of the relative risks produced in all these low-quality studies. In that sense, by bringing relative risks into absolute death counts, it makes us face the obvious unreality those relative risks created in the first place.

Magnetic said...

Conclusion: Secondhand smoke will kill 6 million people over the next ten years. Where's my cheque?

Chris, you could have gone for the century-long “death toll”: Over the next century SHS will kill 60 million nonsmokers. :)

“Social marketing” or “choice architecture”, for example, are crypto-terms for propaganda. Mike Daub (in the Godber Blueprint) refers to “creative epidemiology”. The Globalink website provides numerous propaganda tools for TC advocates. The goal is to use the most inflammatory terms possible for their outrage/revulsion-potential in particularly nonsmokers.

“Tobacco control advocates draw on many skills and disciplines to make their messages effective, including political communications, public relations, social marketing, and public health education. Recently, advocates have also looked to new fields of academic research for tools to create better messages, including media effects research and cognitive linguistics.”

“Cigarettes kill many more people in the United States every year than would be killed by the crash of two fully loaded Boeing 747's each day of the entire year!”

This is a slogan even used by the resident troll.

http://strategyguides.globalink.org/guide01_07.htm


-------

The intoxication with numbers is another characteristic of the modern crusaders against smoking. (Skrabanek)

“The intoxication with numbers” is a key characteristic of eugenics. Biological reductionism coupled with economic reductionism/rationalism reduces life to quantification. It appeals to the shallow-minded. Many of the population-level statistics currently used were developed by eugenicists. Antismoking is also an obsession of eugenics.

Magnetic said...

The idea that 1 in 100 people die from passive smoking is so insane I can hardly believe that anyone could believe it.

You’re right, Chris. The problem is the use of the relative-risk statistic to argue causation. Population-level statistics, such as RR and as used in lifestyle epidemiology, have very poor application at the individual level. The predictive strength (in absolute terms) of SHS for any specific malady/mortality is usually barely above zero% accuracy. To then refer to SHS as a/the cause is ludicrous. Yet most of lifestyle epidemiology runs on this insanity. What we are constantly fed is the assumption of causation based on RR differences (with poor absolute predictive strength) acting in substitution for a demonstration/explanation of causation.

Once there is an acceptance of causation (e.g., causation by consensus), all sorts of statistical games can be played, claiming that SHS “causes” X, Y, and Z in particular numbers. Yet the actual requirements of causal explanation have never been met. High predictive strength reflects the pinpointing goal of scientific enquiry. This does not exist in lifestyle epidemiology: It is self-serving, institutional incompetence. Concerning the research in question, studies could be trawled for all day long with all manner of “tolls” extrapolated. Yet in none of those “studies” has causation been demonstrated. Rather, “causation” has been agreed (assumed) upon by committees of similarly-incompetent, ideologically-driven, and often greedy fools.

Magnetic said...

(cont'd)

Consider another example:
Iatrogenesis (e.g., medical errors, adverse drug reactions) is a very severe problem in medical treatment. In these cases, causation can be demonstrated in individual cases. For example, within half an hour of administering a particular drug, the patient goes into catastrophic failure resulting in death or permanent injury. The very short temporal lag between antecedent and consequent events is critical in inferring causation. Given that many in the medical establishment are not very forthcoming about such events (to protect careers and the hospital), estimates are made of the population-level prevalence of these events. So there is a coherent basis to claiming underlying causation for these events at the individual level and the extrapolation is from the individual to the population level.

In the case of smoking and SHS, the situation is the exact opposite. Causation cannot be demonstrated at the individual level because it involves lifelong exposure which is also open to a plethora of other exposures and events. So, the epidemiologists use population-level statistics, e.g., RR, to discern whether there are any differences between smokers/nonsmokers for specific disease. They then convince themselves that particular RR differences (even though there is poor absolute predictive strength) “indicate” underlying causation. The “causal argument”, therefore, proceeds improperly from population-level differences to the individual level.

Of the two circumstances, it is iatrogenesis that should be taken very seriously. Yet, the medical establishment, in self-interest and for ideological reasons, takes the smoking circumstance seriously. There are literally tens and tens of thousands of highly questionable antismoking studies. Yet there are only a handful of studies scrutinizing iatrogenesis.

Anonymous said...

Antismokers suffer from a mental disorder.
Called.
Notthinkingoutsidetheirownbox-osis.

Anonymous said...

Magnetic

Millions spent on doctor 'gagging orders' by NHS, investigation finds

http://www.independent.co.uk/life-style/health-and-families/health-news/millions-spent-on-doctor-gagging-orders-by-nhs-investigation-finds-2041209.html

I don't think you will get much help from them.

Gary K said...

600,000 deaths from SHS exposure.

Let's put that number into a little perspective.

60 times that number will die from starvation and starvation related diseases.

10 times that number of children will die from starvation and starvation related diseases.

http://en.wikipedia.org/wiki/Starvation


Hunger mortality statistics
On the average, 1 person dies every second as a result, either directly or indirectly, of hunger - 4000 every hour - 100 000 each day - 36 million each year - 58 % of all deaths (2001-2004 estimates)

On the average, 1 child dies every 5 seconds as a result, either directly or indirectly, of hunger - 700 every hour - 16 000 each day - 6 million each year - 60% of all child deaths (2002-2008 estimates).[

Anonymous said...

Anyone fancy going head to head with the loons at Bad Science?
This crap has just been posted up there and as I've got to go to work I don't have time for the full half hour argument.

http://www.badscience.net/forum/viewtopic.php?f=6&t=19319

Paul said...

Chris,

Will write on this over the next few days sometime...it is so insane that one tends to ignore it (which of course doesn't help).

Paul

Klaus K said...

WHO & anti-smoking lobby "estimate": 600,000 passive smoking deaths annually in the world"

Gary K said...

600,000 SHS exposure deaths out of 60 million deaths is 1/100.

In the USA there are a 'claimed' about 50,000 SHS deaths out of 2.4 million deaths, that is 1/48.

It would seem that the undeveloped countries have quite a ways to go in order to catch-up!!

Anonymous said...

They assume 50%+ children are exposed to second hand smoke. Considering that smokers are less than one third of the population and even many smoking parents do avoid smoking at home because of the children, that's pretty implausible number.

BTW: They cite the Pell study :-/

Anonymous said...

I've just had a look at the Badscience forum. It has very little to do with science - either good or bad. It's not worth contributing.