Monday, 26 March 2018

Another evidence review fails to link fast food outlets to childhood obesity

A few weeks ago, the IEA published my review of 74 studies looking at the association between the number and proximity of fast food outlets to obesity. In short, there isn't one. The majority of studies find no association whatsoever. Of the 39 studies that look at childhood obesity, only six suggest a positive association (and five suggest an inverse relationship).

So I was interested to see a systematic review of the evidence published today in the European Journal of Public Health. It, too, fails to find an association, although the authors do not exactly go out of their way to highlight that conclusion. On the contrary, they strongly imply that there is a relationship which dozens of studies have failed to demonstrate as a result of unspecified methodological issues.

Few studies found were able to adequately quantify a correlation between the food environment surrounding schools and obesity amongst pupils attending those schools. The lack of reliable evidence found in this review is more a factor of the ability of the studies found to identify the correlation than the actual lack of a correlation between the two variables.

They recommend that fast food outlets be restricted 'despite the lack of good evidence' and use a ridiculous tobacco analogy to push the precautionary principle:

Planning policy is difficult to change; years may pass between the first inclination to change a policy and the change. Several more years may then pass before the built environment is significantly impacted by the policy. This makes the study of this impact difficult to analyse and time consuming.

This is reminiscent of the study of exposure to cigarette smoke and its impact on health. Tobacco smoking was identified as harmful to health in the 1940s and 1950s. The prevention of exposure to tobacco smoke in the working environment was a hard won change to the built environment and was legally enshrined in the Health Act 2005. Similarly the correlation between fast food retail location, fast food consumption and obesity is still disputed. This lack of evidence may however indicate the inability of many papers to measure the impact of hot food takeaway exposure accurately.

This is a laughable argument for so many reasons. For a start, the smoking ban was not introduced on the back of the evidence that emerged in the 1950s (not the 1940s) showing that it harmed smokers, but on the back of studies published several decades later claiming that it harmed others.

Admittedly, that was only the official justification. Most people realise that its true purpose was to encourage smokers to quit, but even if you think that the smoking ban was a reasonable response to the evidence that smoking is bad for smokers, the fact remains that there is a hell of a lot of evidence that smoking causes cancer whereas there is very little evidence that living near a fast food outlet causes obesity.

It is not that the evidence linking fast food outlets to obesity is 'disputed'. The bulk of the evidence simply does not support the hypothesis that living, working or going to school in areas where fast food is available has any impact on your chances of being obese. Those who argue that restricting the number of fast food outlets will have no effect on obesity rates are not disputing the evidence. They are asserting it.

The implication of the passage quoted above is that 'public health' campaigners should be allowed to act on their gut instincts regardless of what the evidence says. As I said in the little film we made to accompany the IEA report, so much for evidence-based policy.

As for the evidence review itself, it does not do what it says on the tin. My review included 74 studies. This one only includes 14, but that is because it restricts itself to the UK. It includes only two studies that appeared in my review (Harrison et al. 2011, Griffiths et al. 2015) and ignores another one (Heroux et al. 2012). It also includes two evidence reviews conducted by British researchers which mainly look at studies from the USA (Fraser et al. 2010 and Harrison and Jones 2012).

So there are ten studies in the new review that didn't get mentioned in the IEA report. Have I been cherry-picking? I assure you that I have not. Although the new review is titled 'The impact of hot food takeaways near schools in the UK on childhood obesity: a systematic review of the evidence' most of the studies in it do not look at the relationship between fast food outlet density/proximity and obesity. Many of them include neither data on obesity nor data on 'hot food takeaways'.

The authors claim in the abstract that...

Most included studies compared anthropometric measures with geographical location of hot food takeaways to find correlations between environment and childhood obesity.

But this is simply untrue. Here are the studies that are in the new review which were not included in mine. It should be pretty obvious why I didn't include them.

Briggs and Lake (2011) involved giving 24 kids cameras to photograph where they ate their lunch. It doesn't measure the number or proximity of fast food outlets, nor does it measure body weight.

Caraher et al. (2014) looks at the concentration of fast food outlets around schools but doesn't attempt to find any correlation with obesity rates or body mass index.

De Vet et al. (2013) is a survey of kids in four countries which finds that children who have access to takeaway food tend to eat more of it. It does not look at the density or proximity of fast food outlets. Interestingly, the authors do not tell us whether those who ate more takeaways were fatter, despite having the BMI data for each child. I suspect that this is because there was no association, but there is no way of telling.

Devi et al. (2010) is based entirely on an interview with staff and pupils at one school. It doesn't even mention fast food, let alone measure the number of outlets or measure the children's BMI.

Edwards et al. (2010) doesn't look at fast food outlets at all. The closest it gets is finding that obesity is correlated with 'perceived poor access to supermarkets'.

Ellaway et al. (2012) studies the number of takeaways around 'disadvantaged schools in Glasgow'. It does not measure obesity or body mass and makes no attempt to correlate these variables with the number of fast food outlets.

Estrade et al. (2014) doesn't look at obesity or the number of fast food outlets. It is just a summary of some interviews conducted with food shop owners.

Fraser et al. (2011) found that adolescents who were more 'exposed' to takeaway food at home tended to eat at fast food restaurants more frequently. It did not look at the proximity or density of fast food outlets.

Macdiarmid et al. (2015) is a survey of school kids which doesn't attempt to correlate the proximity or density of fast food outlets with obesity. It does, however, ask kids how often they buy food out of school at lunchtime and finds no association between this and obesity. Incidentally, the authors note that only ten per cent of them ate food out of school at lunchtime and say that this finding 'questions the emphasis, effort and likely impact of changing the food environment around schools on improving the overall diet of young people and tackling obesity.'

None of these studies are relevant to question of whether living or schooling near fast food outlets increases childhood obesity risk. However, there is one relevant study in the review which escaped my attention when I conducted my literature search and I am more than happy to give it some attention now. It is Gallo et al. (2014) and it found...

No significant association was observed between food outlet frequency and IMD quintile (F=1.125, p=0.627); obesity prevalence rates nationally or locally (both F = 0.370, p = 0.565) or Supergroup class (F = 2.314, p = 4 0.191).

.. No significant association was observed between obesity prevalence and any specific food outlet typology.

No surprise there. That is what most of the studies show. But it is worth noting that both sentences quoted above are followed in the text by the words 'Despite non-significant results...' and some whataboutery from the authors who are clearly desperate to find an association. There is a lot of that kind of thing in this literature and that is why it is important to focus on what the evidence says rather than what 'public health' campaigners say about it.

No comments: