To the best of our knowledge, this is the largest study to date to investigate the effect of a front-of-pack nutrition label on the development of healthier food products. Our data showed that most products carrying the logo as a result of reformulation and of new product development were soups and snacks, respectively. Sodium was the nutrient reformulated in the most products groups, namely in processed meats, sandwiches, soups and sandwich fillings. Dietary fiber was significantly higher in most newly developed Choices product groups when compared with reference products, namely in fruit juices, processed meats, dairy products, sandwiches and soups.
The finding that sodium is an important nutrient for reformulation is in agreement with a study from New Zealand that showed that the Pick the Tick logo effectively reduced the sodium content in a relatively small sample of food products [
12]. In addition to a reduction in sodium, our study showed that the Choices logo also led to an improvement in the other nutrients with defined Choices criteria. For example, SAFA and added sugar were significantly decreased in both reformulated and newly developed dairy products. Dietary fiber was increased, also in product groups for which no fiber criteria were defined, such as processed meats and dairy products, possibly due to technological reasons. Newly developed Choices sandwiches, however, had a significantly higher sugar content than reference sandwiches, possibly to compensate for changes in other nutrients; this change deserves attention from a health perspective. Further, we only found significant reductions in the caloric content of dairy products and sandwich fillings after reformulation, and reductions in the caloric content of some newly developed snacks compared with reference snacks. The lack of major reductions in energy density is somewhat disappointing because a high intake of energy-dense food products is one of the major contributors to the prevalence of obesity [
1]. Nevertheless, even small changes in calories can have a far-reaching public health impact. Roodenburg and colleagues showed a potential reduction in nutrient intakes, including calories, with the consumption of a diet complying with the Choices criteria, indicating their potential impact on energy balance [
14]. This study is further discussed below.
Most newly developed Choices products were found in the category of snacks. Although the consumption of a limited number of snacks is promoted in the Netherlands, around 30% of a person's daily energy intake comes from food consumption between meals, and the greater part of that amount is snacks [
15]. Our study showed that Choices snacks generally have a lower caloric content than regular snacks. Other nutrients were found to be changed in positive directions as well, such as decreased levels of SAFA in ice creams (milk-based) and decreased levels of sodium in licorice. This stresses the importance of further encouraging food manufacturers to develop healthier snacks. It has been debated whether it is justifiable to assign a health logo to snacks since the logo could stimulate snack consumption, which could constitute a negative side effect of the logo. Steenhuis and colleagues, however, showed that the use of the Choices logo had no negative side effects on the consumption of a chocolate mousse cake among females in a university community when they compared a cake with the logo to the same cake without it [
16]. Nevertheless, it is of interest to note that the chocolate mousse cake was not perceived as healthy in that study. Other research indicates that the perception that a snack food was healthy did increase the actual intake of the food [
17].
Our study does have some limitations. First, one could question whether our data can be considered a representative sample of the total number of Choices logo products available on the market. By collecting data from food manufacturers representing different types of industries, including multinationals, medium and small enterprises, retailers and caterers, we tried to create a sample that was as representative as possible and we did collect data from all product groups. Nevertheless, we did not collect enough data to be able to analyze all product groups, such as breads for example. Future research should try to include data on the product reformulation of breads because this category is the major source of sodium intake in the Netherlands and, therefore, is regarded as an important product for reformulation [
18]. Secondly, it should be noted that some nutrients in quite a few product groups had a large standard deviation, due to the large variety of products within those product groups. Thirdly, the reference values for the newly developed Choices products could have been selected differently. It is possible that the food manufacturers developed new Choices products (for example, mango yogurt) based on existing non-Choices products (for example, strawberry yogurt) which were only slightly different from the Choices guidelines (for example, less sugar was added to the mango yogurt than to the strawberry yogurt, making the mango yogurt compliant with the Choices criteria). It could be useful for future food reformulation studies to ask food manufacturers more extensive questions about the composition development of newly introduced products. In this way, more valid reference values could be obtained (in this example the strawberry yogurt would have been the reference product for the newly developed Choices mango yogurt).
Finally, we collected data on a voluntary basis and all nutrient composition data were self-reported by the food manufacturers. The response rate was quite low and it is possible that only motivated food manufacturers participated in our research, especially those manufacturers that had significantly improved their products. Unfortunately, no data were collected about how many unhealthy products, or those not meeting the Choices criteria were introduced during the same time frame, to be able to evaluate the overall picture of the food supply. Nevertheless, the finding that motivated food manufacturers improved their products can be considered a positive starting point for the improvement of the availability of healthy products for consumers. It would be interesting for further research to explore why some food manufacturers are motivated to improve their products and others are not, and which aspects of company policies play a role in these decisions.
Despite these limitations, this is the largest study to date to explore the impact of a front-of-pack nutrition label on the development of healthier food products. Whether all significant changes can be considered nutritionally relevant remains to be determined. No consumption data and sales data were collected for this study. Consequently, we are only able to relate our findings to individual product groups and cannot make statements about the actual impact of the Choices logo on a population's health outcomes. Nevertheless, consuming a Choices-compliant diet could
potentially lead to substantial improvements in nutrient intake, as reported by Roodenburg and colleagues [
14]. In this study, the researchers combined food composition data and food consumption data and calculated the usual nutrient intake distributions in the Dutch population of young adults. Additionally, food products not complying with the Choices criteria were replaced by products that did comply. As a result, nutrient intakes for energy, total fat, SAFA, TFA, sodium, and total sugar decreased and fiber intake increased (these are the nutrients included in the Choices criteria). Additionally, positive changes were found for protein, total carbohydrate, PUFAs, MUFAs, calcium, iron and folic acid (nutrients not included in the Choices criteria). The challenge now is how to investigate the
actual effect of the Choices logo by combining reformulation data with intake data and sales data. Consequently, possible health gains can be estimated, such as the prevalence of cardiovascular disease, life expectancy and health care costs. For example, in the United States, the Coronary Heart Disease Policy Model has been used to estimate the cost-effectiveness of a population-wide dietary salt reduction [
19]. In future studies, using such a model could be helpful in estimating the impact of a front-of-pack nutrition logo on a population's health outcomes.
1 Ingrid HM Steenhuis,1 Annet JC Roodenburg,1 Johannes Brug,2 and Jacob C Seidell1,2
This article has been 
