Scottish smoking ban miracle touches the unborn

The miracles keep on coming in Scotland, soon there will be pilgrimages.

Drop in pregnancy complications after smoking ban

Complications in pregnancy have fallen as a result of the ban on smoking in public places, according to a new study.

Researchers found the ban, introduced almost six years ago, has led to a drop in the number of babies being born before they reach full term.

It has also reduced the number of infants being born underweight.

My word, this post hoc ergo propter hoc junk science sounds like the kind of rubbish Jill Pell keeps coming out with.

The research team, led by Professor Jill Pell...

Ah, Professor Pell, we meet again and under such similar circumstances. You may recall Jill "Pinocchio" Pell from her signature piece claiming that the heart rate plummeted after the Scottish smoking ban, but for sheer effrontery in the face of rock solid evidence, her subsequent article claiming that the asthma rate fell after the smoking ban takes the cake. Rarely has science met fiction so brazenly.

...looked at more than 700,000 single-baby births before and after the introduction of the ban.

The number of mothers who smoked fell from 25.4% to 18.8% after the new law was brought in, researchers discovered.

This, as you might expect from Pell, is a distortion of the truth. The 25.4% figure relates to 2001, some five years before the ban was introduced. Any honest researcher would surely use the figure for 2005 (22.5%) as the pre-ban measure. We already know from a previous Pell study that the ban had no effect on the smoking rate in the general population. Looking at the ISD figures, it is difficult to see any effect on expectant mothers as well. There is a general downward trend which continued after 2006.

The graph above might actually exaggerate the decline. I was interested to see, upon studying the ISD data, that the smoking ban coincided with (I shall not say caused as I am not a charlatan) a large increase in number of expectant mothers for whom no information on smoking status was available. In other words, more pregnant women are refusing to tell the NHS whether or not they smoke.

The likelihood is that many of these women are smokers but do not wish to be chastisted by the denormalisers of Scotland's health service. This suspicion is supported by the fact that the proportion of pregnant women who are lifelong non-smokers has barely moved for a decade.

Back to the news story...

Experts further found there was a drop of more than 10% in the overall number of babies born "pre-term", which is defined as delivery before 37 weeks' gestation.

There was also a 5% drop in the number of infants born under the expected weight, and a fall of 8% in babies born "very small for gestational size".

This is the meat of the research. As so often, the study has been press released before publication so we cannot see which statistical tricks Pell has employed, but we can use the official NHS records to see how her claims stand up. The data are available here.

The graph below shows preterm births (ie. less than 37 weeks gestation) as a percentage of all live births recorded in Scottish hospitals between 1996 and 2010 (the period that Pell claims to have studied).

The proportion of babies born prematurely in this period remained very constant (between 6.8% and 7.9%). The post-smoking ban years were unremarkable, with percentages of 7.3, 7.4, 7.6 and 7.2 (2007, 2008, 2009 and 2010 respectively). The lowest rate was in 1996. There appears to be no relationship with the general smoking rate, maternal smoking rate or the smoking ban.

I suspect what Pell has done here is taken the highest pre-ban figure (7.9%) and compared it with the lowest post-ban figure (7.2%). The difference between these figures in percentage terms is a little under 9% which, with a bit of statistical massaging, could become "a drop of more than 10% in the overall number of babies born 'pre-term'".

Since preterm births are the major driver behind low birth weights, it should be no surprise that there has been no major change in the number of babies with a low birth weight. Between 97.0% and 97.4% of all full-term pregnancies in Scotland in this period resulted in a baby of normal weight (2500 gm+). I can see no evidence of any 'smoking ban effect' in any of the ISD data. There are moderate random variations and nothing more.

Dr Pell said the research highlighted the positive health benefits which can stem from tobacco control legislation.

To paraphrase Mandy Rice-Davis, she would, wouldn't she?

She said: "These findings add to the growing evidence of the wide-ranging health benefits of smoke-free legislation and support the adoption of such legislation in other countries which have yet to implement smoking bans.

"These reductions occurred both in mothers who smoked and those who had never smoked."

Sorry, what?

"These reductions occurred both in mothers who smoked and those who had never smoked."

Doesn't that tell you something then, Pell? If you are claiming that the smoking ban reduced preterm births because it made people give up smoking, the fact that you found the same result with nonsmokers rather gives the game away, does it not? If, on the other hand, you're suggesting that reducing secondhand smoke miraculously reduces preterm births (I haven't read the study yet, but I wouldn't put it past you to indulge in such superstition), the findings for smokers strongly suggest that this is nonsense as well. Or perhaps you are going to claim that smokers somehow feel the benefit of secondhand smoke reductions as well. Nothing would surprise me at this stage.

"The potential for tobacco control legislation to have a positive effect on health is becoming increasingly clear."

Yes, yes. We understand why you keep producing this garbage. Why don't you go find yourself a street corner to shout from?

Researchers looked at data for babies born between January 1996 and December 2009, taken from the Scottish Morbidity Record, which collected information on all women discharged from Scottish maternity hospitals.

Which is exactly what I have shown above. Feel free to check the data yourself.


UPDATE: Michelle Roberts—easily the worst of the BBC's appalling health reporting team—has been suckered by this story. Her entry on Journalisted is an A-Z of pointless epidemiology. I notice the national press have ignored the story, presumably on the basis of 'once bitten'.

Pell's study has appeared on PLoS here. It's short on data but this is her killer graph...

This barely resembles the actual data from Scottish hospitals, but even so it takes a massive leap of faith to attribute the smoking ban to any part of it. The hard line represents the smoking ban, but Pell prefers to use the dotted line because "the Akaike information criterion statistics suggested that using 1 January 2006 as the breakpoint produced a marginally superior model fit than using 26 March 2006." Hey, whatever fits your a priori conclusion the best, Jill.

Even having moved that goalpost, it's plain to see that the fall in preterm births began around ten months before the smoking ban came in. In fact, it came well over a year before, because the timeline Pell is using is the date of conception, not birth. It must have been pretty galling for her to see that the largest drop in her graph preceded the ban and came to an end as soon as the ban came in. Moving the date back to January does not help her much in that respect. Furthermore, even if it had happened after the ban, it would hardly have been proof of anything. There are two little peaks in the graph (as there are in my graph above) followed by two drops. Peaks do tend to be followed by drops, y'know. Maybe Jill Pell should look up 'regression to the mean'.