Everybody likes babies…and football…

Posted on August 14, 2014


343702786_hayden_shocked_face_resized_answer_1_xlargeSometimes you read something and you just know that cannot be correct. I am not talking about the Daily Mail again (that’s part of its entertainment value); I am talking about peer-reviewed scientific publications. Some of those have been covered previously on this blog, but I found another nice one (well, was pointed to it to by an unnamed individual; to give credit where credit is due). This one has to do with small babies. Also not the first time this has come up here, because we all know small babies is not a good thing.

low-birth-weightAnyway, the title of this paper is “Reduced infant birth weight in the North West of England consequent upon ‘maternal exposure’ to 7/7 terrorist attacks on central London” and it was published in the Journal of Obstetrics and Gynaecology in 2011 (link). When you read this title you may think something along the lines of “Ah right, so pregnant women from the North West of England who happened to be in London on July 7, 2005 and were involved in the horrific bombing events gave birth some time later and their babies were on average smaller than babies from women in the North West who gave birth during the same period but were not in London on 7/7.” That would make sense, and actually something like this has been shown for women from New York who were in the city during the 9/11 attacks (link). This would be an interesting study and would show a clear link from external stress leading to an internal reaction (increased cortisol has been suggested) and subsequently to reduced fetal growth.

Unfortunately, that is not quite what was done. Actually, I am a bit surprised this got published. What was really done was that the researchers had a database of all women who gave birth in the North Western region of the UK (So Manchester and surroundings basically) between January 1st, 2004, and December 31st, 2006. So far so good, but then see how involvement this event, or experience of a traumatic life event (eg the bombings or “the exposure”) was defined:

  • Women who were pregnant on 7 July 2005
  • Women whose pregnancy started during the 6 months following 7 July 2005

So basically two hypotheses are investigated here: (1) if a traumatic event during pregnancy could affect fetal growth, and (2) a bit more unlikely, if a traumatic life event prior to getting pregnant could lead to reduced birth weight. What doesn’t make this any better is that the researcher assume that everybody would have been exposed to, and affected by, this event. So to check if there was some kind of association they compared this group to the births that happened outside of this period, which in practice are the births between January 1st 2004 and roughly 9 months before 7/7/’05 (so basically 1-1-’04 to 1-10-’04) and births from 6 months (for hypothesis 2) or roughly 9 months (for hypothesis 1) after the event; so births between February 2006 and December 31st 2006 basically. Moreover, the researchers assumed that any birth within the “exposure period” would be directly affected by the 7/7 bombings! So if I phrase that differently; any birth between October 2004 and February 2006 in the North West was affected by the bombings.

That doesn’t make any sense.

Especially not when we slightly rephrase the hypothesis to: “was the average birth weight of babies born in the Northwest between October 2004 and February 2006 statistically significantly different from that of newborns before or after that period?” Using the study design in this paper, there is absolutely no reason to assume that the 7/7 bombings have anything to do with any differences between both periods, other than that July 7, 2005 just happens to lie roughly in the middle of the time period classified as “exposed”.

Anyway, it turns out that the “exposed babies” are a little bit smaller on average; 16 grams to be precise (range between 11 and 21 grams smaller). Some differences are observed between the trimesters on July 7, but the bottom line is that all confidence intervals overlap and these differences can easily be ascribed to random variation (I am not a clinician, but I don’t think 12 grams is going to make a massive difference on future health and wellbeing). I would further argue that the fact that births of which the pregnancy period started up to six months after the event have a similarly reduced birth weight points towards another reason why births in this period were a little smaller.

But what else could it be?

It could be down to natural variation and it is just bad luck that the researchers chose this particular period and found an association, or maybe there is a period in the year that babies are a little bit smaller. I am sure this is fairly straightforward to figure out, but I will leave that to you.  Babies in 2005 could also have been a little smaller than in surrounding years; due to natural variation or due to some other reason I don’t know (again, the Office of National Statistics can surely tell you this). It could of course indeed be some traumatic event, but would an event somewhere else in the country to which most, if not all, of the women were not directly exposed (other than through the news) have such an effect? It seems to make a lot more sense if a traumatic event had happened in the Northwest and to which these pregnant women would therefore have been directly exposed. Luckily Wikipedia has a solution to everything, and indeed they do have a page describing all the noteworthy events that happened in the UK in 2005 (link)!

So that’s quite a lot of candidate events, but if we follow the same rationale as the authors in this peer-reviewed scientific publication, it should be possible to narrow this down. I would argue in favour of the following set of a priori criteria:

  1. The event had to have happened in Manchester or its vicinity so that the pregnant women were truly and directly exposed
  2. It should be a traumatic event. However, data from New York women after 9/11 indicate that a truly horrific and traumatic event would have led to average reduction in birth weight of more than 100 grams (link), so we are looking for a minor traumatic event or a traumatic event that doesn’t interest women that much and would lead to a reduction of about 16 grams.
  3. It would be most plausible if this event happened in the immediate period surrounding July 2005 (following the rationale of the authors of the original paper); so basically sometime between May 2005 and October 2005.

If you are still with me and this seems an acceptable hypothesis, you’ll be pleased to know that I followed this up by searching for such events on the Wikipedia page. I scanned the May-October 2005 period looking for events in Manchester. I’ve given you the link, so you can check the approach, but really the results are obvious.

There is only one significant event that would affect people in and around Manchester directly, leading to increased stress for everyone, but would have a significantly smaller impact on women than on men (and hence explain this small decrease in average birth weight of only 16 grams, or about 10% of what happened after 9/11):


On May 21, 2005 Arsenal become the first team to win the FA Cup on penalties after they defeat Manchester United in a shoot-out that follows a goalless draw.


Now seriously, what is more likely? The explanation published in the Journal of Obstetrics and Gynaecology or the explanation I just gave you? I can tell you that the results will be same if one was actually to redo the statistical analysis.

Posted in: Public Health