Headline alert: Poor diet is not necessarily the CAUSE of allergies

The word “linked” might be a read a bit too strongly in this headline.

BBC News – Fast-food ‘linked to childhood asthma and eczema’.

Data from more than 500,000 children in more than 50 countries suggests poor diet may be to blame for rising levels of these allergy-related conditions.

Fast food often contains high levels of saturated- and trans-fatty acids, which are known to affect immunity, while fruit is rich in antioxidants and other beneficial compounds, say the researchers.

In the study, children in their early teens who ate three or more weekly servings of fast food had a 39% increased risk of severe asthma.

The study authors, Prof Innes Asher, from the University of Auckland in New Zealand, and Prof Hywel Williams, from the University of Nottingham in the UK, said: “If the associations between fast foods and the symptom prevalence of asthma, rhinoconjunctivitis and eczema is causal, then the findings have major public health significance owing to the rising consumption of fast foods globally.”

Here is the link to the paper. The conclusions say this:

Conclusions If the association between fast foods and the symptom prevalence of asthma, rhinoconjunctivitis and eczema is causal, then the findings have major public health significance owing to the rising consumption of fast foods globally.

So, they found an associated between A and B. Not a causal relationship. This type of mistake is probably THE most common error when the media reports on such a study. Sometimes even the institution that did the study puts too much emphasis on “cause” when the study didn’t establish cause at all. These types of studies CAN NOT determine cause. It was not an experiment. It was a survey. Answers were given via questionnaires, thus subjective – as in what did you eat, how is your condition today? An experiment must be designed to be far more rigorous to eliminate all other variables and narrow in on a cause. In this case, there is an association (which the piece actually notes) which means they are seen together. We don’t know how they are associated or if this is dependent on another factor such as social or economic variables.

This is why you need to be VERY doubtful about headlines and never assume a causal relationship unless the study is a very controlled experiment.

Tip: Agent J

Thanks to @thePapersBehind

  3 comments for “Headline alert: Poor diet is not necessarily the CAUSE of allergies

  1. Brian
    January 15, 2013 at 7:00 AM

    It isn’t the food. It’s all the crap put into the food that causes problems. Eating right and getting up off ones bottom works wonders. Is it 100%? No, but it is darn close.

    I also cannot help but wonder if people can’t be manipulated into thinking they have an allergy. A case I read about some time back involved a woman with multiple personalities. One had a allergy to milk, I think. Yet, another of her personalities did not, and yes, there were the associated allergy symptoms (and lack thereof)
    Think of the power of suggestion, and think how people have become gullible the past few years. Warren Jeff’s comes to mind. Look how he manipulates people, from jail yet. Or some of these church leaders who proclaim healing, and all sorts of miracles, yet while none of it can be proven, people swear up and down by them. And then of course, hand over tons of cash to these liars. A person can’t be made to scratch themselves, just by suggesting there might be an ant or bug crawling on them.

    I do not say this is true…. It’s just a theory I have had for about 2 years now.

  2. Elisabeth
    January 16, 2013 at 3:33 PM

    I totally admire your skeptical outlook and critical thinking but I would advise a little study in epidemiology to bolster your reporting on these types of studies. I am in complete agreement on the media conflating correlation with causation, but it’s not true that causation can only be determined via experiment or that studies based on surveys are incapable of providing evidence for causation. While it’s true that double-masked placebo-controlled studies are the gold standard for proving cause and effect, this is simply not ethically or logistically possible for many exposures we wish to investigate in the real world. Thus there are prospective cohort studies which may follow people in the real world going about their normal business and periodically recording or reporting on their exposures to, or consumption of, various risk factors of interest. How the recording or reporting is designed is crucial to the validity of the data but it is always subject to recall or reporting bias and good researchers will acknowledge the weaknesses in their design when they publish. They will also make attempts to statistically control for potential confounders that can lead to faulty conclusions. Anyway, this is getting longer than intended but just trying to make the point that an “experiment” is not always necessary for showing causality. Evidence from multiple well-designed cohort and case-control studies combined with some biological plausibility is often enough to show causality .
    http://en.wikipedia.org/wiki/Bradford_Hill_criteria

    • January 16, 2013 at 3:41 PM

      I do understand that. But this study was not rigorous in the ways you mention. And many brought out in the media today are also not. It is rare to see a large cohort study these days because of the nature of research. Instead, we have these types of correlation studies that can give the public a false sense of what is really going on.

      I was put in mind of this piece http://www.sciencebasedmedicine.org/index.php/everything-we-eat-causes-cancer/ that described how poor the correlational evidence was for a food and cancer effect.

Leave a Reply

Your email address will not be published. Required fields are marked *