Monday, 2 May 2011

Passive smoking lowers blood pressure in girls, study reveals

Before retiring to bed last night, I cast a weary, cynical eye on a brief story from The Independent which reported:

Boys who inhale second-hand tobacco smoke at home may experience significant levels of raised blood pressure, a study has found.

But in girls, passive smoking appeared to be associated with a lowering of blood pressure.

These results, if true, suggest that there are dramatic differences between the male and female bodies that make them respond very differently to secondhand smoke. They also imply that there is a new and unlikely form of treatment for girls with high blood pressure: blow smoke at them.

I'm joking, of course, although it is as valid to draw that policy conclusion from the data as it is for the study's author to claim that her work provides "further incentive for governments to support smoking bans". Actually, neither policy would have any effect because the changes in blood pressure were negligible in both cases—a 1% increase/decrease either way, which even the author admits is meaningless for individual health.

An alternative interpretation is that the study found no effect of passive smoking on blood pressure but that by stratifying the results, the researcher was able to find one by chance (in this case, by splitting the group into genders). Or it may be something else; the study only compared cotinine readings with blood pressure readings after all.

The study's author naturally favours the first interpretation:

"These findings support studies suggesting that something about female gender may provide protection from harmful vascular change," said Dr Jill Baumgartner.

Whether there are any biological reasons to explain these contradictory findings, I don't know (... paging #carlvphillips ...). I couldn't possibly comment because—once again—this study has not been published  and it has only been presented at a conference before being press released around the world. It's a PhD dissertation by a student of Population Health and Environment & Resources.

There is an abstract available, however, which shows the increase in boys is not significant while the decrease in girls is significant. Take both genders together and the overall effect of passive smoking on blood pressure is to lower it, although it is unclear whether that association is significant. The abstract also shows a complete lack of a dose-response relationship which is a bit surprising if tobacco smoke was the true cause of the correlation.

What I do know is that it is misleading to report this study under the headline used by both the Daily Mail and The Guardian:

Passive smoking raises blood pressure in boys, study reveals

In fairness to the Mail, they mention the conflicting finding for girls towards the top of the article. The Guardian, to its shame, mentions it towards the bottom. Ben Goldacre recently mentioned an interesting lay experiment which showed how few people bother reading the whole article before making their minds up. Not very many, apparently, which gives The Guardian the horrible distinction of being even worse at reporting science news than the Daily Mail in this instance. (Neither were helped by the study's press release, which only briefly mentions the findings for females.)

We know that the great majority of newspaper readers only ever read the headline, so let's hope the same is not true of blog readers, otherwise the title of this post—which is as technically correct as the Mail and Guardian headlines*—will also give people a mischievously one-sided view of the story.

* Actually, rather more correct since I have at least focused on the statistically significant finding.

(Michael Siegel and Taking Liberties have more on this story.) 


Carl V. Phillips has a definitive critique on this study over at Ep-ology. Well worth reading.


Mark Wadsworth said...

"These findings support studies suggesting that something about female gender may provide protection from harmful vascular change," said Dr Jill Baumgartner."

Well, whoopie do. Isn't it common knowledge that men are about five times more likely to have heart attacks etc than women?

Dick Puddlecote said...

I'm sure Ben Goldacre will be on this like a flash to report how shoddy it is.

Nah, course not. Some bad science is OK for him and his readers.

Michael J. McFadden said...

I have noticed a related, and probably deliberate, failing several times in the last year or two where headlines (and even formal study titles) claim "Exposure to tobaco smoke" does something...

But then you read the study and find out that SMOKING is what produces the effect. Of course the great majority of readers will only have absorbed the title/headline and it adds to mythology that "mountains of studies" support ETS claims.

- Michael

Dick Puddlecote said...

Hilarious comment at the Mail.

"negative comments by people seem to all come from people who don't understand how these studies are done. any published study worth its salt controls for other things that would impact health. socioeconomic status, rural/urban, other health conditions/obesity. so discounting this based on those isn't valid. i haven't read the study either, but if its being promoted by the scientific community, one has to assume it passes some basic rules of science."

-, atlanta, USA, 1/5/2011 17:36

Paraphrase: I believe everything I'm told. ;)

Carl V Phillips said...

I had actually already written today's EP-ology Unhealthful News post about this before I saw your post. If you would like to see the draft before I post it tonight, to see if there other other points you think I might address, email me.

Gary K. said...

I believe the standard error of measurement for such things is plus or minus 1-3 mm Hg.

It would appear the readings are a wash...nothing found.

Snowdon said...


Funnily enough, I just started reading Bad Science (the book) last night. I've never had anything against the guy. I think he writes in a clear and somewhat humourous way, even if he does assume his readers know nothing about the most basic scientific methods.

I'm not that interested in homeopaths and nutritionists - I kinda worked out they were shysters without having to get a book about it - otherwise I would have read it sooner. I suppose it's good - if depressing - that someone can be perceived as a super-sceptic just by debunking such obvious nonsense, but there's clearly a need for it and he does it pretty well. I do think he has a number of blind spots when it comes to less PC scepticism however.

He mentions smoking a few times in the book but only in passing. He thinks Austin Bradford Hill was called Austin Bradford-Hill and confuses Richard Doll with Richard Peto. He says that smoking causes 95% of lung cancer and, later, that it causes "almost all" lung cancer (the real figure is about 85%).

These mistakes aren't cardinal sins, and they don't relate to today's junk tobacco science, but they don't give me the impression that he has looked at the issue in any depth, nor is inclined to. Maybe he will if he ever writes a book called 'Fucking Appalling Science'.

Carl V Phillips said...

As requested, my analysis is up:
I agree with your assessment and add a few more observations about two problems I consider just as bad. Don't forget to give me a retweet for helping out :-)

Junican said...

One sometimes wonders if it is desirable for the researchers to draw the conclusions. Perhaps the researchers should just do the research and publish the figures. Perhaps better statisticians than the researchers should draw the conclusions.

Fat chance.

Ivan D said...

I fear that you may have missed out The Sunday Times Chris. I haven’t paid to read the “article” but reveals that Emily Payne wrote something titled “Passive smoking has higher impact on boys” which has a lot of text in common with the AAP press release. Possibly Ms Payne is another journalist who through laziness or lack of training is unable to understand the significance of the result in females and therefore the illogical and misleading nature of the explanation for gender differences proposed in the AAP press release.

Fredrik Eich said...

First 'liquor liability' trial over hotel guest death.
I don't know if this is a recent law but I would be interested to see if it catches on.