Why traffic light labelling of foods won’t work

The policy makers are desperate to do something about obesity but traffic light labelling would achieve nothing. The nutrition criteria at the heart of the concept are wrong.

Health authorities around the world have struggled to find effective ways of addressing the obesity epidemic. One recommended strategy is the introduction of traffic light labelling of food products to make it easier for the general public to choose healthier foods in the supermarket. It is argued that healthier food choices would not only tackle the obesity issue but could also address many other diet-related chronic diseases. Sounds good in theory but it simply won’t work.

What is traffic light labelling?

Currently nutrition information on food packaging appears in the nutrition information panel where amounts of various nutrients per 100g of product or per serve are listed. It’s all very quantitative – lots of numbers – and for many people it’s just too technical and difficult to understand. The rationale behind traffic light labelling is to interpret the current nutrition information for the public and give it some meaning. Green, amber or red traffic lights would be assigned to ‘nutrients of concern’ such as fat, saturated fat, sugar and salt to indicate whether a product was low, medium or high in these nutrients. It is argued that with better communication of the nutrition information the public would make healthier food choices.

If you are interested in the argument in favour of traffic light labelling take a look at the Australian Medical Association’s position papers on the issue.

Image: source


There appear to be three assumptions underpinning the argument in favour of traffic light food labelling: (1) An unhealthy food can be defined in terms of its content of total fat, saturated fat, sugar and sodium, (2) people become overweight because they eat unhealthy food, and (3) communicating these nutrient details to the general public will change food choice. All three assumptions are debatable but let’s focus on the first.

There is a fundamental problem with the nutrition criteria that have been included in the proposed traffic light model – total fat, saturated fat, sugar and sodium. Although some of these are linked to health outcomes, others are only useful for assessing certain food groups and some are totally useless. Let’s start with a useless one.

Total fat as a criterion

The idea that the total fat content of a food is a measure of its healthiness is one of nutrition’s great anachronisms. Thirty years ago it was hypothesised that total fat intake was linked to health outcomes – high fat intakes possibly increasing the risk for breast cancer, bowel cancer, coronary heart disease and obesity. Large prospective cohort studies were commenced to test the hypothesis but the results were overwhelmingly negative in relation to all of these conditions.

Image: Walter Willett: source

Here is how Walter Willett, professor of nutrition and medicine at the Harvard School of Public Health, summed up his presentation on fat and health at a conference in Boston at the end of 2010: In conclusion, let me reiterate that diets with a lower percentage of energy from fat do not reduce the risk of heart disease, diabetes, cancer, or adiposity. For some people, a reduction in total fat can actually be harmful if healthful fats are preferentially reduced. We should just clean up our dietary guidelines and remove any reference to percentage of energy from fat …

Think about this. If the fat content of the whole diet is not related to any outcome of public health significance, what possible purpose could be served by highlighting the fat content of an individual food?

What about saturated fat?

Saturated fat is a useful criterion for assessing fat-rich foods such as vegetable oils, margarines, salad dressings, dairy foods and meats. But it is totally inappropriate for assessing carbohydrate-rich foods based on, say, potato, cereal or fruits. Saturated fat and carbohydrate confer the same risk for coronary heart disease. Giving a green light on saturated fat to a carbohydrate-rich food would not only be wrong-headed scientifically it would send a misleading message to the general public.

Heart disease risk falls when saturated fat is replaced by unsaturated fats. This is the exchange that food labelling needs to encourage, not just lowering saturated fat.


In an earlier post we noted that health authorities in both the United States and Europe have been unable to set an upper limit for sugar consumption for the prevention of any chronic disease, including obesity, so we won’t go over this old ground again. But the situation is analogous to that of total fat: if the sugar content of the whole diet has negligible effect on health, what is the logic in focussing on the sugar content of an individual food?

The muesli example

A good example of how focussing on fat and sugar can send misleading and confusing messages was a review of 159 mueslis in the Australian Consumers Association’s Choice magazine published last year. Much of the information presented in the article is factual, but the accompanying press release was titled ‘Spoonful of muesli may be mouthful of sugar and fat’. The media loved it – it was one of those ‘you think it’s good for you but it’s really not’ stories. But it was very misleading.

The article noted that much of the fat in the mueslis actually came from nuts and much of the sugar came from dried fruit. So in this case the fat and sugar in the mueslis were markers of nutritious ingredients – nuts and fruits. So what was all the fuss about? How would Choice magazine advise a muesli manufacturer to improve the nutritional quality of its products? Cut back on the nuts? Cut back on the fruits?

Sugar and total fat are simply not useful measures of the healthiness of foods and diets. Highlighting their content in foods does not enable consumers to make healthier choices.

Would traffic lights really make decision making easier?

While you are looking at the Choice muesli article run you eye down the long table of mueslis, all with traffic light-type symbols for total fat, saturated fat, sugar and sodium. The first thing you notice is that none of the products gets four green lights. And none of them gets four red lights, so there is no clear endorsement or rejection of any of the mueslis. In fact, the majority of products have a combination of red, amber and green lights – the ultimate mixed message.

How is this supposed to help consumers make a healthy choice? If you were in your car at the traffic lights and the red, amber and green lights all shone at once, what would you do?

Image: source


12 thoughts on “Why traffic light labelling of foods won’t work

    • Hello Jenny. Thanks for highlighting this report. To me there is a mixture of good and bad in this model. On the positive side there is a focus on protein and vegetables, which (together) is a surrogate for nutrient density – all about meeting essential nutrient requirements.
      However, the nutrients of concern are essentially the same as those I discussed in my post. The UK model uses energy rather than total fat, but energy density is largely driven by fat (and water) content. This model basically says ‘fat is bad for you’ and that is just wrong in science. The idea that eating high energy foods makes people fat is a myth (see my earlier post on this issue).
      Using saturated fat alone i.e. not considering its replacement macronutrient, also discriminates wrongly against fat. Admittedly, this model was put together before this aspect of the science became clear in 2009.
      If you were to put any unsaturated vegetable oil through this model e.g. canola, sunflower or olive oil, it would generate negative points for energy and saturated fat. However, the use of unsaturated oils in place of saturated fats is one of the few dietary strategies that has been tested in randomised controlled trials with clinical end points, and the results were beneficial. How can a good model discriminate against these oils? Can I quote Willett again? … a reduction in total fat can actually be harmful if healthful fats are preferentially reduced.
      The use of ‘non-milk extrinsic sugars’ as a criterion stretches credulity. Is there a scientific study anywhere in the world that links this factor with any health outcome? Nutrient dilution is the only potential concern with sugar-rich foods and beverages and this should be captured with nutrient density measures.
      This model is locked into old hypotheses and belief systems that just don’t stand up to scientific scrutiny today.
      Regards, Bill

  1. Hi Bill – I agree there are inherent problems in a simple traffic light approach. Have you given any consideration to the Healthy Eating System developed by Sanitarium? It tested well in terms of understanding and appropriate interpretation, and gives a much richer picture of the food, discriminating well between foods according to modelling. There’s more scope for development, which is why Sanitarium released it freely for review. You can find the report here: http://www.sanitarium.com.au/~/media/sanitarium/about-us/traffic-light-report.ashx
    [Declaration: I worked on the HES and consult for Sanitarium.]

    • Hi Liz. Firstly, I think Sanitarium needs to be congratulated for supporting the HES project. It has driven the debate along and put some ideas out there. We need plenty of creativity and debate on this issue.
      The idea of an overall interpretation of the nutrition information i.e. ‘eat often’, would certainly be appreciated by consumers. However, apportioning a red colour to the ‘eat sparingly’ foods would be resisted strongly by sections of the food industry. No company wants a red flag on its products, full stop. Personally, I think we need to be a little pragmatic on this issue.
      Didn’t Sanitarium produce another front-of-pack labelling system which was a continuum, with more or less stars per product? A system that reflects the shades of grey is attractive as it gets away from the idea of goods and bad foods, which generates a lot of hostility.
      The criteria for any model are critical. In relation to the HES criteria, there is a whole food criterion (fruit, veg, grains, nuts, seeds) as well as a fibre criterion, which could be criticised for considering similar qualities twice. Saturated fat (in isolation) doesn’t work as a criterion. People really need to get their head around this – the science in relation to saturated fat has changed.
      And I struggle with ‘added’ sugar. For a start, you can’t analyse for added sugar, so it’s not practical. More importantly, if sugar is bad for us we firstly need to see the scientific argument. Then we need to see sugar criteria applied universally. For too long nutritionists have been saying sugar is bad, but not sugar in fruits. And fat is bad, but not fat in nuts. If criteria can’t be applied objectively across all relevant foods then there is something wrong with the criteria.
      Regards, Bill

  2. Apologies, I unintentionally posted this previously under a pseudonym.

    I don’t think anyone is arguing that TL labeling is *the* solution. Consequently, I think your definition of “work” is unrealistic and unhelpful.

    The truth is we don’t know the heuristic processes people use to choose food. Your rebuttal assumes one – rather simplistic – heuristic model. In my experience the public are not as simple as we too often assume them to be.

    It’s easy to build straw men, it is much harder to suggest practical alternative solutions. Perhaps traffic light labeling of foods won’t work simply because we can’t agree on our measures of success.

    • Hi Steve. One thing I never do is assume that the general public is simple.
      Can I put it too you that the reason nutritionists can’t agree on our measures of success is that many of our measures are not evidence-based and therefore are open to challenge? Regards, Bill

  3. I think you are trying to over complicate things.

    This is for the 90% of the population that don’t read food labels or really understand nutrition.

    I don’t think traffic lights are perfect, but I think the concept is right.

    • Hello Geoff
      At the heart of your comment is a belief that if the general public could only understand how much fat and sugar was in their food they would become healthier and not gain weight. I’m sorry. These parameters are just not linked with health and body weight.
      What is the point of better communicating the wrong information?
      Regards, Bill

  4. Total fat in particular is a problem for this style of labelling.

    However, a good case can be made for sodium.

    Milligram amounts on labels are always going to be hard to understand. %DI based on a target which is not consistent with chronic disease (2300mg) is worse.

    Manufacturers need a greater incentive to bring low sodium products to market. A green light can give a marketing advantage to this.

    Until then, will we keep hearing people say they don’t each much salt, while consuming cornflakes, vegemite and corned beef?

Leave a Reply

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


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>