¿Nutriscore ayuda a comprar mejores procesados? Leyendo el último estudio

noviembre 5, 2023

2 Views 0

SaveSavedRemoved 0

With everything that Nutriscore is talking about and with the doubts that some of us have regarding its usefulness (see), it is normal that we pay attention to scientific publications about this system. And just a few days ago we met a new one, «The Impact of the Nutri-Score Front-Pack Nutrition Label on Purchase Intentions for Processed and Unprocessed Foods: Post Hoc Analysis of Three Randomized Controlled Trials» (2021).

First of all, it is worth clarifying a couple of issues. The first is that, as mentioned in the title, this work is not «original», it is a reanalysis of data from a previous study, this one in particular. And the second, which can also be deduced from the title, is that it is not based on making real purchases. In the included trials, purchase simulations were carried out using a computer program. Obviously, this second circumstance is an important limiting factor, because we do not have solid evidence that people will behave in the same way in a computer simulation as in a real purchase. And much less in the medium long term.

After these points, let’s get to the point.

The first thing to highlight about this latest research is that it aimed to study whether the use of the Nutriscore label affects the purchase of products that do not carry Nutriscore, that is, fresh and unpackaged products. And their final conclusions are favorable, as can be seen in the translation of the authors’ summary:

«These findings provide new insights into the positive effect of Nutriscore, which appears to decrease the purchase of processed products, resulting in a higher proportion of unprocessed and unpackaged foods, in line with public health recommendations.»

But I am not going to focus on this result. I prefer to delve into other data related to the purpose for which Nutriscore was created: to help us choose «better» packaged products through its letter and color evaluation system. Because it is worth remembering that Nutriscore only applies to processed and packaged products and that this label was created precisely to provide us with information that encourages us to buy (and consume) more products qualified with level A and B (healthier according to their algorithm) and less. with level D and E (less healthy). It is assumed that it is its added value compared to other labels, such as the alert type (high in salt, high in sugar, etc.), which only provide negative evaluations.

Well, the truth is that in the original study The basis of this reanalysis did not come to very encouraging conclusions regarding the role of Nutriscore in helping us choose better packaged products. This is what its authors said then:

«…there were no significant differences between the Nutri-Score and unlabeled groups»

But what is interesting about the new reanalysis is that it provides complementary data from the original study that was not accessible at that time and that helps us now evaluate the effectiveness of Nutriscore in this regard in more detail. It is enough to take the data from the table 2 and do some simple calculations to arrive at some especially interesting results: know the percentage of each level (A, B, C, D and E), both when using the Nutriscore label and without a label of any kind, regarding the quantity of packaged products. This last condition is especially important, calculating the percentage with respect to the packaging, instead of with respect to the total purchase (which is what the authors do). This way we can know if Nutriscore, in addition to buying more unpackaged products, has helped us buy «better» packaged products. Which is, I insist, its «original function.»

Well, after doing the numbers I have graphically represented the result. Bars with solid lines show the percentage of packaged products purchased that would fit into each Nutriscore level when no FOPL tag is used, regarding the total number of packaged products. And the bars with dashed lines show the same thing, but in this case when using the Nutriscore label:

It can be seen that the percentage of A and B – the products classified as healthier – has been reduced, just the opposite of what would be desirable (and expected by Nutriscore defenders). On the contrary, the percentage of D has increased, which is positive, but that of E – the least healthy group – has practically not changed. Finally, the C percentage – an «intermediate» level that is not known to be healthy or not but that I personally consider is nothing to write home about – has increased significantly.

In short, according to this data it seems that Nutriscore has not served to buy better packaged or processed products. The percentage distribution with a dashed line (with Nutriscore) is not significantly better than the percentage distribution with a solid line (without labels).

In other words, this study confirms one of the issues that we consider most critical and that we commented on in this previous post: Fresh foods – which are precisely those that do not carry the Nutriscore label – are usually included in the evaluation of their effectiveness, interfering in a way very important in this evaluation, since they provide benefits and nutrients that Nutriscore considers positive. That is, fresh products act as a confounding variable. But what we really need to know is whether Nutriscore is effective and provides benefits considering only its effects on products that carry this label. In this study we have data to know, doing the calculations to know the percentage of each category only in packaged products. And by doing so, the usefulness and benefits of Nutriscore in this regard will disappear. That is to say, it only seems to be effective in reducing (modestly) the percentage of processed food, in general, without providing any nuance regarding its different processed categories. But then, why so many letters, colors and algorithms? Wouldn’t it be much easier (and possibly more effective) to use «alert» type stamps?

Regarding the influence on unpackaged products

Let’s return to the main objective of the article, to analyze the influence of the Nutriscore label on the purchase of unpackaged products. As I have already mentioned, this result is positive, as shown in the following table included:

I have framed in green, at the bottom of the table, the result that has given rise to the main conclusions highlighted by the authors: the percentage of non-packaged products purchased has increased by almost 6%. In principle it is a good result, although quite modest.

But it is also worth clarifying that if instead of the percentage of products we look at the number, We will see that, paradoxically, the number of unpackaged products has actually been reduced with Nutriscore. About 2%, as framed in red. But since the number of products has been reduced in all cases, the overall balance in percentage has been positive.

In any case, there is something more striking (and not very positive) about the specific effect that Nutriscore has had on some food groups. We can see the data in table 4from which I have extracted a part, with the differences (in percentages) between not using labels and using Nutriscore:

I have marked in red the differences that can be considered negative (because the proportion of products has changed for the worse when using Nutriscore labeling) and there are a few. Certainly, the rest of the results are mostly positive, but there are some of the framed ones that are quite worrying: Reduction in the purchase of vegetables, legumes and fresh fish. And increase in fresh and processed meat, as well as juices and sugary drinks. None of these dietary changes can be considered positive, quite the opposite.

Conclusions

In short, in this study, on the one hand, it seems that Nutriscore helps to slightly increase the purchase of unpackaged or fresh products. Also to reduce some not recommended and increase some healthy ones. But on the other hand, it also seems to cause other undesirable dietary changes.. Furthermore, it can be deduced from the data included that Nutriscore has not been effective in helping people purchase better packaged products, which It should be its primary function.

Therefore, I think that this study adds little in favor of Nutriscore. YesYour design – a computer simulation – is unrealistic; and its results – diverse and heterogeneous – are not at all conclusive regarding its real usefulness or possible benefits.