Have a look at this recently launched public health campaign, pithily titled “Stop the Spread“. It aims to address the problem of overweight in the general population. Co-ordinated by Safefood, the statutory body responsible for the promotion of food safety in both the Republic of Ireland and Northern Ireland, the campaign has proven to be somewhat controversial in both countries. This is not least because of its implication that being overweight is somehow communicable, its promotion of waist-circumference as an uncomplicated measure of disease risk, and its apparent recommendation that all adults — including young and otherwise healthy people — should constantly strive to avoid putting on weight and thus regularly measure themselves in order to track their body shape empirically.

Safefood’s primary gimmick has been to issue 250,000 free measuring tapes for members of the public to use to see just how fat they actually are. Vividly, and without any apparent equivocation, they declare that waist circumferences greater than 32 inches for women or 37 inches for men signify significant risk of developing an obesity-related disease. This approach has been very harshly criticized by many commentators, including eating disorder advocacy and support groups, who spend most of their efforts and resources promoting precisely the opposite approach to healthy weight management. In response, Safefood have forthrightly defended their campaign as being necessary in the face of what they argue is a looming public health catastrophe. Moreover, both their website materials and public spokespersons have been adamant that the 32/37-inch threshold is consistent with the best available scientific research.

However, that particular claim is extremely questionable. It is not at all clear that the scientific research supports the details of this campaign; in fact, in some respects, the campaign directly conflicts with the very scientific research that it cites in its own defence. As a result, this campaign could actually be doing far more damage than good…

One indicator that the campaign is based on shaky scientific foundations can be found in the advert’s initial context-setting statements. A rather sombre-sounding voiceover warns us that the problem of overweight is “spreading“, that “most of us already have it“, and that “we are rapidly passing it on to others, giving them a higher risk of heart disease, diabetes, and cancer“. This last statement accompanies an otherwise pleasant image of two young women embracing on the street. Overall, these descriptions seem to imply that obesity is a communicable disease. However, the idea that overweight (or obesity) is somehow contagious is at best a metaphor. Overweight is not contagious or infectious or otherwise communicable. It seems fairly obvious that Safefood are using the contagion metaphor for its shock value. But is a public health campaign the best place to take artistic liberties? After all, it could be that not all members of the public will realise that this type of language is intended to be figurative. If so, then claiming that overweight is a disease that is “spreading” and that “we are rapidly passing it on to others” could provoke unnecessary anxiety.

But perhaps of greater scientific concern is the way Safefood handles the 32/37-inch threshold, which they say informs us about each individual’s personal health-risk status.

Frankly, they’re doing it wrong.

On the basis that you don’t feel too bad about being contagiously fat and will have enough self-esteem left to cope with being stupid as well, Safefood provide the following film to show you how to measure your waist circumference correctly (note how something that should take 10 seconds to explain takes Safefood 1 minute 22 seconds to show on YouTube; which is somewhat ironic given that the usual basis for recommending waist circumference measurement is that it is supposed to be easy):

Now, while epidemiologists agree that average waist-circumference across groups can be informative about health risks at the group level, Safefood’s assertion is that it is relevant to health at the individual level. Here is how the Safefood website presents their position on the threshold:

32 inches (for women) and 37 inches (for men) are the recommended maximum waist sizes, no matter how tall or small you are.

No matter what height or build, an increased waistline is a sign for most people that they may be at an increased health risk of conditions like heart disease, Type II diabetes and cancer. These recommendations by the World Health Organization take into account that people are different shapes and sizes.

And here’s how Dr Cliodhna Foley-Nolan, Safefood’s Director of Human Health and Nutrition, describes things in an interview with the Medical Independent:

With the waist measurement, there is very concrete science behind those numbers. Both the WHO and the National Institute of Clinical Excellence (NICE) use waist measurements as a straightforward indicator because every indicator is going to have its limitations.

Okay. So here are the errors in Safefood’s position.

1. Firstly, it is quite erroneous to state that the WHO “use waist measurements as a straightforward indicator“. In fact, the WHO consider waist measurements to be anything but straightforward. Their main policy statement on this issue — Waist Circumference and Waist-Hip Ratio: Report of a WHO Expert Consultation — which was published just this year, consistently explains that waist circumference is just one possible indicator among many, and that it may or may not be more useful than the alternatives (which include BMI, waist-hip ratio, and hip measurements).
2. Secondly, the WHO are explicit in stating that waist circumference measures (and potential cut-offs) are supposed to be used “for public health action” (see p. 2 of the Report). In other words, they do not intend that cut-offs be used for personal health action. What this means is that, according to the WHO, waist-circumference data is of use for assessing the overall health of a population (or large group of people) rather than for assessing the precise health risks faced by an individual person in isolation. In this context, it appears somewhat divergent of Safefood to translate the WHO approach into recommendations for individual diagnostic purposes, or to suggest that a single person can obtain meaningful information by comparing their waist circumference to a single threshold.
3. In fact, even at the population level, the WHO actually declines to recommend simplistic universal waist measurement cut-offs against which disease risk can be judged. As such, Safefood’s reference to their thresholds as representing “recommendations of the World Health Organization” is just spurious. Instead, in the ‘Recommendations’ section of their waist circumference Report (see p. 24), the WHO point out that, as far as they are concerned, suitable thresholds have yet to be properly established. In addition, they highlight the need for any future system to be complex rather than simple, and to be based on a multiplicity of categories (e.g., low, medium, and high risk) rather than on Safefood’s two categories (e.g., at-risk, and not-at-risk).
4. Finally, the WHO clearly disagree with Safefood’s insistence that waist circumference measures are meaningful regardless of a person’s body shape, age, medical status, or any other biometric information. In fact, the WHO Report states that “there is substantial evidence of sex and age variations in waist circumference and waist-hip ratio, and some evidence for ethnic differences.” They cite the International Diabetes Federation as having published meticulous breakdowns of waist circumference norms according to ethnic group (a similar position to that taken by the Australian Government, who point out the limitations of generalizing existing norms to many of the ethnic groups that comprise minorities within Australian citizenry). And Safefood’s insistence that waist measurements are the only thing that matters appears even more questionable when you consider the fact that some people have very good medical reasons for having large waist circumferences. Here’s one example that is not mentioned by Safefood at all: women who are heavily pregnant.

In short, it seems very strange indeed that Safefood invoke the WHO in their defence of the “Stop the Spread” campaign. Dr Foley-Nolan’s assertion that there is “very concrete science behind those numbers” is contradicted by the fact that the science does not actually support the identification of particular thresholds for universal use or for the use of any universal thresholds at the individual level. Furthermore, when the Safefood website asserts that “These recommendations by the World Health Organization take into account that people are different shapes and sizes“, this is contradicted by the fact that (a) these are in fact Safefood’s own recommendations, not recommendations from the WHO; and (b) as far as the WHO are concerned, such thresholds are indeed confounded by individual differences in body shape and size, such as those relating to ethnicity.

But from a scientific perspective, perhaps the most depressing aspect of Safefood’s “Stop the Spread” campaign is that it commits a logical error that is extremely well known to epidemiologists, biological scientists, logicians, and statisticians alike. In fact, it is so straightforward that those of us involved in education in these areas typically teach it to undergraduates. The error is known as the ecological fallacy.

The ecological fallacy is committed when people look at the averages of data collected from groups of people, and then assume that these same averages can be applied to each individual within the group. For example, the ecological fallacy is committed when town planners calculate the mean average family size in a neighbourhood and then decree that all new houses should be constructed with sufficient room for 2 parents and 1.4 children. Likewise, when epidemiologists identify waist-circumference thresholds that, on average, separate low-risk and high-risk categories within the overall population, the ecological fallacy is committed when agencies like Safefood come a long and decree that all people above the threshold are correspondingly facing high risk. The fallacy is compounded when the agency in question declares that the threshold applies “no matter what height or build” a person is, that the system “takes into account that people are different shapes and sizes“, and — most awkwardly of all — that there is “very concrete science behind those numbers“.

As disappointing as this all is, at least there is some humour to be found. In statistical terms, the ecological fallacy is committed when observers falsely emphasize one type of descriptive statistic at the expense of another. The unduly emphasized statistic is called the “central tendency” (examples of which include mean averages and medians). The humour arises from the deep irony regarding the other statistic, the one that tends to be forgotten about but which is actually just as important. This statistic is called…

…wait for it…

…the “spread“.

Really, you couldn’t ask for a better punchline than that.

Brian Hughes

Brian Hughes is an academic psychologist and university professor in Galway, Ireland, specialising in stress, health, and the application of psychology to social issues. He writes widely on the psychology of empiricism and of empirically disputable claims, especially as they pertain to science, health, medicine, and politics.