Trying to make sense of nonsense

Food for thought: causation vs. association

If there’s one thing about ‘science news’ that should get your alarm bells ringing, it’s the mention of a ‘Press Association‘ as the author of the piece in question.  This generally means that the story in question is blindly copied from a press release without even taking a look at the research it’s reporting on. Tuesday’s articles commenting on the supposed link between poor diet and IQ in children can be listed in that very category.

The study by Northstone and others that made the headlines had followed 3,966 children from birth up to 8.5 years of age and on the way collected data on what these children ate, their IQ and on lots of other things that might have something to do with either your IQ or your diet. Those other things, or confounders as epidemiologists like to call them, included stuff like whether the children had been breastfed (Anderson in 1999 found that breastfeeding is associated with a 3.16 point increase in your IQ), mothers educational level (as children do tend to take after their parents) and social class.

After taking these and other factors into account, they found that, on average, children whose diet contained lots of food with high sugar and fat levels – or ‘processed’ foods – when they were 3 years old, had a slightly lower IQ when they were 8 years old. In fact, it was on average 1.67 points lower, about half of the effect that breastfeeding seems to have on IQ. Surprisingly, snacking seemed to increase the IQ of these same children, by almost 1 point.

The news jumped onto these numbers and started to proclaim that a poor diet causes a lower IQ. Although it might be tempting to conclude exactly that from the study’s results, you wouldn’t be right in doing so. As the title of the study implies, the authors have found an association, rather than causation, between the two.

When two things happen consecutively in time, it doesn’t necessarily mean that one causes the other. I for one believed for years that the fireworks on my birthday were for that special occasion. I was completely oblivious to the fact that it also just so happened to be NYE. The same is true for most cold medicines and them curing you: people tend to resort to this sort of stuff at the height of their cold. And seeing as a cold generally resolves itself, you tend to believe that it was the tablets you took – rather than the disease’s natural course – that ‘cured’ you (I also believe the buttons to open the doors on the tube are only there to please tourists’ eager fingers).

For causation, a lot more evidence is needed. This paper is just a first step to finding out whether children’s diet actually influences their eventual. The authors of the study are very clear about this – hell, they’ve put it in the title! – but for journalists/lazy editors who like to fill their pages with press releases it appeared to be a bridge too far.

(And for the record: a 1.67 point decrease in IQ is hardly a difference at all: make the same kid take the same test twice and you’ll find a larger difference (or in more posh wording: it’s statistically, but not clinically significant); the study only takes maternal education into account and completely ignores dad – I would imagine dad to have at least some influence on his kid; and finally, there is no correction for concentration on the test – as shown by Jamie Oliver’s venture to the States, processed foods to tend to lead to a lack of concentration in kids around this age)


One response to “Food for thought: causation vs. association

  1. Pingback: Tweets that mention Food for thought: causation vs. association « epidemiologista -- Topsy.com

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: