Abstract
Background
In Chile, added sugar warning labels prompted product reformulation with nonnutritive sweeteners (NNS). Some countries have since adopted NNS labels to discourage this industry response, but evidence of their effects on consumer behavior is limited.
Objectives
This study aimed to examine the impact of 2 types of NNS labels on Chilean parents’ NNS-containing product identification, product choices for their children, and product perceptions.
Methods
In 2025, 3523 Chilean parents (children age 2–14 y) were randomly assigned to either no NNS label (control), a child-focused NNS warning (“CONTAINS NON-NUTRITIVE SWEETENERS, AVOID CONSUMPTION AMONG CHILDREN”), or an NNS disclosure (“CONTAINS NON-NUTRITIVE SWEETENERS”). Participants viewed sets of 4 products (one without added sweeteners and others with sugar, NNS, or both) and identified NNS-containing products, selected one product for their child, and, afterward, recalled labels seen. Next, participants viewed NNS-containing products and rated labels’ perceived message effectiveness (PME), products’ perceived healthfulness, and intentions to limit children’s NNS intake. Average differential effects (ADEs) are reported in percentage points (pp) or 1 to 5 scales.
Results
Both the child-focused NNS warning and NNS disclosure led to higher correct identification of NNS-containing products (ADE = 67 pp; ADE = 69 pp, both P < 0.001), lower selection of NNS-containing products (ADE range: −22 pp to −35 pp; ADE range: −22, −33 pp, all P < 0.001), and higher selection of products without added sweeteners (ADE range: 20–35 pp, ADE range: 22–31 pp, all P < 0.001) than control, respectively. Neither NNS label increased selection of sugar-sweetened products. NNS label recall was 63% for the child-focused NNS warning, and 84% for the NNS disclosure (P < 0.001). Both the child-focused NNS warning and NNS disclosure led to higher PME (ADE = 1.53; ADE = 0.92, both P < 0.001), lower product healthfulness perceptions (ADE = 1.43; ADE = 0.81, both P < 0.001), and higher intentions to limit children’s NNS intake (ADE = 0.66; ADE = 0.17, both P < 0.001) than control, respectively.
Conclusions
Both NNS labels improved parents’ NNS-containing product identification and encouraged choice of products without added sugar or NNS for their children.
This trial was registered at the clinicaltrials.gov as NCT06842693. https://clinicaltrials.gov/study/NCT06842693
Keywords: nonnutritive sweeteners, nonsugar sweeteners, front-of-package labeling, front-of-pack labeling, warning labels, food labeling
Introduction
Nonnutritive sweeteners (NNS) are food additives widely used as sugar substitutes whose role in diet-related disease prevention is debated. Upon reviewing the available evidence in 2023, the WHO issued a conditional recommendation that NNS should not be used for weight management or noncommunicable disease prevention [1]. Although the WHO recommendation is intended for nondiabetic individuals of any age, consumption of NNS among children is particularly concerning, as sweet taste preferences develop in early childhood [[2], [3], [4]] and children can be at higher risk of exceeding acceptable dietary intake levels of NNS [[5], [6], [7], [8], [9], [10], [11]] because of their lower body weight [12,13].
Front-of-package labeling (FOPL) is an increasingly popular policy intervention gaining traction in several countries as part of a strategy to curb diet-related diseases. NNS use is particularly salient in the context of FOPL policies, in part because of Chile’s experience as the first country to implement warning-style FOPL labels in 2016 [14]. After implementation of its policy—which includes warning labels indicating when products are high in added sugar, sodium, saturated fat, and/or calories—Chile saw marked decreases in the percentage of products that qualify as high in added sugars [15]. However, the use of NNS in packaged products concomitantly rose from 37.9% to 43.6% [16], and the prevalence of Chilean preschool children who consume NNS increased from 77.9% to 92% [17].
Given this potential unintended consequence of FOPL policies (and sugar warning labels specifically), the Pan American Health Organization currently includes NNS in its nutrient profile model as a food additive of concern to be targeted by nutrition policies [18], and some countries have introduced different types of NNS labels as part of their FOPL systems: Colombia has implemented an octagonal label (resembling stop signs and matching the design of nutrient labels) that discloses the presence of NNS (i.e., NNS disclosure label); Mexico and Argentina, alternatively, have opted for longer, rectangular warnings that recommend against NNS consumption for children (i.e., child-focused NNS warning labels). Evidence from Mexico suggests that NNS prevalence in packaged products has decreased since the implementation of its FOPL system, suggesting that its NNS labels have positively influenced the food supply [19]. However, presently, very limited evidence exists on NNS labels’ effects on consumer behavior [20,21], and no studies have compared the effects of different types of NNS labels.
Amid rising concerns over children’s NNS intake, the Chilean Ministry of Health has proposed adding a child-focused NNS warning, emulating the wording and design of the NNS labels used in Mexico and Argentina, to its FOPL system [22]. To inform Chile’s and other countries’ NNS labeling initiatives, this study examined and compared the effects of two different types of NNS warning labels currently in use in different countries on Chilean parents’ ability to identify products containing NNS, product choices for their children, perceptions of NNS-containing products, and intentions to limit their children’s NNS intake.
Methods
Participants
In July and August 2025, we recruited an online convenience sample of Chilean parents through the panel company Netquest for a one-time survey. Participants were eligible if they were ≥18 y old, resided in Chile, and had a child between 2 and 14 years old who had never been diagnosed with diabetes, prediabetes, or insulin resistance. Because children of individuals with lower educational level consume less NNS in Chile compared with those of higher-educated parents [14], we established a quota requiring 50% of participants to have a high school education or less. The University of North Carolina at Chapel Hill Institutional Review Board approved this study (#24-2896). All participants provided electronic informed consent. The study design, measures, hypotheses, and analytic plan were registered before data collection on ClinicalTrials.gov (NCT06842693).
Procedures
Participants completed an online survey using Qualtrics and translated to Spanish by native Chilean Spanish speakers. All tasks were completed within this survey. Using the randomizer function in Qualtrics, participants were randomly assigned to 1 of 3 NNS labeling conditions using a 1:1:1 allocation ratio: control (no NNS label), child-focused NNS warning label, or NNS disclosure label. Regardless of NNS labeling condition, sugar-sweetened products carried an octagonal “HIGH IN SUGAR” label, consistent with Chile’s current FOPL policy. The child-focused NNS warning was a rectangular label stating, “CONTAINS NON-NUTRITIVE SWEETENERS, AVOID CONSUMPTION AMONG CHILDREN,” matching the label adopted in Mexico, Argentina, and under consideration in Chile. The NNS disclosure was an octagonal label stating, “CONTAINS NON-NUTRITIVE SWEETENERS,” matching the label adopted in Colombia.
Both types of NNS labels used the term “edulcorates,” consistent with existing labels in Spanish-speaking countries. This technical term encompasses both artificial and natural types of NNS. In a previous qualitative study in Brazil, participants expressed confusion about the difference between “edulcorante” (which is the same in Portuguese and Spanish) and “corante” (Portuguese for food colors/dyes) [23]. Given that the Spanish term for colors/dyes, “colorante,” is similar, before starting all experimental tasks, we informed participants that the term “edulcorante,” which they might encounter during the study, “refers to sweeteners that are not a type of sugar and that have few or no calories.” Because “edulcorantes” does not have a direct English translation, we use the term “non-nutritive sweeteners” for the purposes of this article.
Stimuli
The experiment included tasks using images of fictitious food products to avoid the influence of existing brand preferences. Because of limited screen space, the labels on the product packages were small and difficult to read, so we also displayed enlarged labels in call-out boxes next to each product, consistent with previous studies [[24], [25], [26], [27]] and with practices adopted by Chilean online grocery stores.
Selection tasks
In random order, participants viewed 3 sets of products (fruit drinks, yogurts, and breakfast cereals), each consisting of 4 products, and completed selection tasks for each set (described under Measures). In each set, one product did not have any added sweeteners, one contained only added sugars, one contained only NNS, and one contained both added sugars and NNS. Within each set, all products had the same flavor and similar package design. The NNS-sweetened product in each set stated, “no added sugars” on the package, and the NNS-plus-sugar-sweetened product stated, “with 50% less calories,” reflecting common marketing claims found on these types of products (Figure 1).
FIGURE 1.
Example set of products used in the choice experiment by label type. NNS, non-nutritive sweeteners.
Perceptions and intentions assessments
Participants viewed a chocolate-flavored milk and a cereal bar, which were displayed one at a time in random order, and reported their perceptions of the labels on each product (described under Measures). Both products shown were NNS-sweetened and stated “no added sugars” on the package. In the control condition, products carried a neutral barcode label, consistent with prior studies [24,[28], [29], [30]]. In the NNS label conditions, participants viewed products with NNS labels identical to the labels viewed in the first task (Figure 2).
FIGURE 2.
Example product used in the perceptions and intentions assessments.
Measures
Selection tasks
Within each product set, participants were asked to choose one product to purchase for their child and, next, to select all products that contained NNS. After completing these measures for all product sets, participants were asked whether they recalled seeing FOPLs on the products during the tasks. Those who replied “yes” were asked what the labels that they recalled seeing were about. Response options were not exclusive and included sugar, NNS, calories, sodium, saturated fats, caffeine, and trans fats. Calories, sodium, and saturated fat labels were not displayed during the experiment but are part of Chile’s existing FOPL system, whereas trans fats (which are allowed in Chile in very small amounts [31]) and caffeine labels were neither displayed during the experiment nor are part of Chile’s existing FOPL system.
Perceptions and intentions assessments
For each product viewed, participants rated the labels’ perceived message effectiveness (PME) through 3 items: 1) how much the label discouraged participants from wanting to buy products with NNS for their child, 2) how much the label made participants concerned about the health effects for their child of consuming products with NNS, and 3) how much the label made buying products with NNS for their child seem unpleasant. PME has been shown to predict behavior change in the context of warning label exposure [32,33]. Responses were assessed on a 5-point scale from “not at all” to “very much.” Next, participants rated products’ perceived healthfulness (i.e., how good or bad for their child’s health it would be to consume the product every day). The 5-point response scale ranged from “very bad” to “very good.” Participants also rated products’ relative healthfulness (i.e., how healthful participants believed the product to be compared with a version of the same product sweetened with sugar). The 5-point response scale ranged from “much less healthy” to “much healthier.” Lastly, participants were asked to what extent they intended to limit their child’s NNS consumption over the next week on a 5-point scale ranging from “not at all” to “completely.” Participants also answered questions about their prior experiences with NNS and sociodemographic characteristics.
Analyses
Power
We powered the study to detect differences in the selection of sugar-sweetened products between the control condition and each of the NNS label conditions because, according to our calculations, this was the primary outcome that required the largest sample size. Assuming that a share of consumers would prefer an unlabeled product over a comparable labeled product but would be indifferent between two labeled products, the higher the preference for unlabeled products, the larger the sample size required. To be conservative, we assumed that 90% of participants would prefer unlabeled products, in which case a sample of 3,306 participants provided 80% power to detect differences between the control condition and each NNS label condition using two-sided two-sample proportions tests with a critical α of 0.05 (which best approximate the mixed-effects models used in the data analysis).
Selection task
Outcomes were dichotomous (i.e., selected product or not). Three outcomes were based on participants’ choice of product for their child: 1) choice of one of the NNS-sweetened products, 2) choice of the product without added sweeteners, and 3) choice of one of the sugar-sweetened products. In turn, correct identification of NNS-containing products corresponded to the selection of both NNS-containing products and no others. We used logistic mixed-effects regressions to analyze the effects of the labels on each of these outcomes, treating the intercept as random to account for repeated measures within participants. Models regressed each outcome on indicator variables for labeling conditions, product categories (i.e., fruit drink, yogurt, or breakfast cereal), and the interactions between labeling conditions and product categories. Post–model estimation, we conducted Wald tests to assess the joint statistical significance of the interaction terms. For the correct NNS identification outcome, interactions were not jointly significant, so we did not retain them in the final model and conducted a second Wald test to assess the joint significance of differences in the predicted probabilities of the outcome across product categories. Because this test also did not yield significant results, we report average differential effects (ADEs) on this outcome (i.e., differences in the predicted probability of correct NNS identification between labeling conditions) across product categories. Alternatively, for choice of NNS-sweetened product, choice of product without added sweeteners, and choice of sugar-sweetened product, the interaction terms were jointly significant and thus retained in the final models. For these outcomes, we report ADEs separately for each product category. Alternatively, to analyze how well participants recalled seeing the labels during the selection tasks, which was measured only once, we used logistic models, regressing each response option on labeling conditions.
Perceptions and intentions assessments
We verified that Cronbach’s α was sufficient (i.e., >0.7) and calculated the mean of participants’ scores across the three PME items for each product. Next, we used linear mixed-effects regression models to analyze the effects of the labels on PME, perceived product healthfulness, and relative product healthfulness, treating the intercept as random to account for repeated measures within participants. Models regressed each outcome on indicator variables for labeling conditions, product categories (i.e., chocolate milk or cereal bar), and the interactions between labeling conditions and product categories. Post–model estimation, we conducted Wald tests to assess the joint statistical significance of the interaction terms. None of the models presented jointly significant interactions; interaction terms were thus not retained in the final models. We report ADEs (i.e., differences in predicted means between labeling conditions) across product categories. Alternatively, to analyze the impacts of labeling condition on participants’ intentions to limit their child’s NNS consumption, which was measured only once after all product exposures, we used linear models, regressing the outcome on labeling conditions.
Moderation analyses
We examined whether the effects of the labels on select outcomes from the choice experiment (i.e., correct NNS identification, selection of NNS-sweetened product, and selection of product without added sweeteners) differed by participants’ gender, educational level, and self-reported use of tabletop NNS. We used the latter as a proxy for participants’ intentional use of NNS, assuming that those who voluntarily add NNS to foods or beverages demonstrate more deliberate intentions to use NNS than those who only consume NNS incidentally through ready-to-eat or ready-to-drink products, which they may not be aware contain NNS. We dichotomized moderator variables to maximize power and simplify interpretations. We used mixed-effects regression models, regressing outcomes on indicator variables for the labeling conditions, the moderator (one moderator per model), and their interactions. Because of significant interactions between labeling conditions and product categories in the main models for the selection of NNS-sweetened product and selection of unsweetened product, we fit separate mixed-effects models for each product category (i.e., fruit drink, yogurt, and breakfast cereal) for the moderation analysis of these outcomes.
Participants who completed the survey implausibly fast (i.e., in <1/3 of the median completion time), who completed <90% of the survey, or who had reCAPTCHA (i.e., Completely Automated Public Turing test to tell Computers and Humans Apart) scores of <0.5 were excluded from analysis [25]. Otherwise, analyses included all participants according to the trial arm to which they were randomly assigned. We used complete case analysis, resulting in the exclusion of three participants from some primary outcome analyses. Analyses were conducted using Stata/MP version 19.5 (StataCorp LLC) with a two-sided critical α of 0.05.
We made the following deviations from the registered analytic plan: 1) we did not register our plan to exclude participants with reCAPTCHA scores of <0.5 but decided to exclude such participants per advice from the Qualtrics platform; 2) we registered our plan to only report label noticing and label recall descriptively but decided to conduct statistical testing on label recall post hoc given its relevance to the study’s objectives; 3) we registered our plan to use mixed-effects models for all outcomes but used models with only fixed parameters to analyze effects on label recall and intentions to limit one’s child’s NNS intake because these outcomes did not include repeated measures, and 4) we did not plan to examine moderation by NNS tabletop use but included this analysis post hoc to better understand whether the effects of the NNS labels may differ among individuals who use NNS intentionally.
Results
The final analytic sample included 3,523 participants (Figure 3). Participants’ mean age was 39.6 years and 70% identified as women. Approximately 44% had an educational level of high school or less, 11% identified as indigenous, and 59% considered themselves overweight or very overweight. The mean age of participants’ children was 8.9 years. Approximately 19% of participants considered their child overweight or very overweight, and 27% reported having been previously told by a health care provider that their child needed to lose weight (Table 1).
FIGURE 3.
Participant flow diagram. NNS, nonnutritive sweeteners. reCAPTCHA, Completely Automated Public Turing test to tell Computers and Humans Apart.
TABLE 1.
Participant characteristics (n = 3523)
| Characteristics | n | % |
|---|---|---|
| Gender | ||
| Man | 1,050 | 30 |
| Woman | 2,470 | 70 |
| Other | 3 | <1 |
| Education (n = 3509) | ||
| None, kindergarten, or basic education | 107 | 3 |
| High school | 1,435 | 41 |
| Higher-level technician (1–3 y) | 836 | 24 |
| Professional (>4 y) | 919 | 26 |
| Postgraduate | 212 | 6 |
| Ethnicity (n = 3,506) | ||
| Indigenous | 374 | 11 |
| Region (n = 3,496) | ||
| Metropolitana | 1,774 | 51 |
| Valparaíso | 391 | 11 |
| Biobío | 246 | 7 |
| North (Antofagasta, Atacama, Arica y Parinacota, Coquimbo, Tarapacá) | 271 | 8 |
| Central-West (O’Higgings, Ñuble, Maule) | 365 | 10 |
| South (Aysén, La Araucanía, Los Lagos, Los Ríos, Magallanes) | 449 | 13 |
| Nationality (n = 3,513) | ||
| Chilean only | 3,158 | 90 |
| Chilean and another | 97 | 3 |
| Other | 258 | 7 |
| Perceived weight (n = 3,500) | ||
| Very underweight | 14 | <1 |
| Underweight | 70 | 2 |
| Approximately the right weight | 1,335 | 38 |
| Overweight | 1,833 | 52 |
| Very overweight | 248 | 7 |
| Ever diagnosed with1 (n = 3,506) | ||
| Diabetes | 575 | 16 |
| Hypertension | 545 | 16 |
| High cholesterol | 675 | 19 |
| Cardiovascular disease | 60 | 2 |
| None | 2,114 | 60 |
| Perceived income adequacy2 (n = 3,433) | ||
| Somewhat or very difficult | 2,454 | 71 |
| Neither difficult nor easy | 725 | 21 |
| Somewhat or very easy | 254 | 7 |
| Frequency of FOPL use (n = 3,432) | ||
| Never or rarely | 690 | 20 |
| Sometimes | 1,075 | 31 |
| Often or all the time | 1,667 | 49 |
| Sources of NNS intake in the last 7 days1 (n = 3,513) | ||
| Tabletop NNS | 1,821 | 52 |
| Sugar-free soda | 1,799 | 51 |
| Sugar-free fruit drinks | 1,525 | 43 |
| Sugar-free flavored yogurts | 1,449 | 41 |
| Sugar-free flavored water | 1,230 | 35 |
| Sugar-free flavored milk | 1,138 | 32 |
| Sugar-free sports drinks | 347 | 10 |
| Sugar-free energy drinks | 307 | 9 |
| None | 297 | 8 |
| NNS recommendation from health care provider (n = 3,517) | ||
| Recommendation to consume NNS | 779 | 22 |
| Recommendation to avoid NNS | 876 | 25 |
| No recommendation | 1,862 | 53 |
| NNS-related effort in the last year | ||
| Made effort to consume more NNS | 314 | 9 |
| Made effort to consume less NNS | 2,049 | 58 |
| Did not make any NNS-related effort | 1,148 | 33 |
| Best term to describe NNS | ||
| Artificial sweeteners | 1,289 | 37 |
| Sweeteners | 936 | 27 |
| “Edulcorantes” | 447 | 13 |
| Sugar substitutes | 378 | 11 |
| Non-nutritive sweeteners | 169 | 5 |
| Non-caloric sweeteners | 154 | 4 |
| Diet sweeteners | 113 | 3 |
| Other | 33 | 1 |
| Child's sex | ||
| Male | 1,779 | 50 |
| Female | 1,744 | 50 |
| Parent’s perception of child's weight (n = 3,503) | ||
| Very underweight | 16 | <1 |
| Underweight | 188 | 5 |
| Approximately the right weight | 2,646 | 76 |
| Overweight | 625 | 18 |
| Very overweight | 28 | 1 |
| Health care provider says child needs to lose weight (n = 3,503) | 931 | 27 |
| Mean | SD | |
| Age | 39.6 | 7.8 |
| Child age | 8.9 | 3.6 |
| Number of people in household (n = 3,432) | 4.1 | 1.4 |
Abbreviations: FOPL, front-of-package label; NNS, nonnutritive sweeteners.
Sample size is provided for specific variables when data were not available for the entire analytic sample for that variable.
Nonexclusive options.
Measured as perceived difficulty of making ends meet without going into debt.
Nearly all participants reported consuming common sources of NNS (including tabletop NNS and sugar-free products) in the past 7 days. Similar percentages of participants reported having previously received recommendations from health care providers to avoid (25%) and to consume (22%) NNS—the latter being common in Chile as a recommended method for losing weight and/or reducing sugar intake. Approximately 58% of participants reported trying to reduce their NNS intake in the past year. Participants’ most commonly preferred term for NNS was “artificial sweeteners” (37%), followed by “sweeteners” (27%), and “edulcorantes” (13%; Table 1).
Selection tasks
Compared with participants in the control condition, those in the child-focused NNS warning condition (ADE = 67 pp) and in the NNS disclosure condition (ADE = 69 pp) were both much more likely to correctly identify both NNS-containing products (both P < 0.001). Participants in the NNS disclosure condition were also slightly more likely to correctly identify products with NNS than participants in the child-focused NNS warning condition (P = 0.002; Figure 4, Supplemental Figure 1, Supplemental Table 1).
FIGURE 4.
Predicted probability of correctly identifying both nonnutritive sweeteners (NNS)-containing products (n = 3,523) and of selecting one of the NNS-sweetened products, the unsweetened product, or one of the sugar-sweetened products to purchase for their child (n = 3,520), by label type. ∗Statistically significant difference at the 95% confidence level. Predicted probabilities refer to predicted margin postestimation. There were no interactions between label type and product category for correct identification of NNS-containing products, so results are averaged across products. However, there were interactions between label type and product category for all selection outcomes,so results are presented by product category.
Compared with participants in the control condition, those in the child-focused NNS warning condition (ADE range: −22 to −35 pp) and in the NNS disclosure condition (ADE range: −22 to −33 pp) were both less likely to choose products containing NNS for their child from all three product categories (all P < 0.001). There was only a significant difference on this outcome between the two NNS labels for one product category: participants in the child-focused NNS warning condition were slightly more likely to choose a fruit drink containing NNS for their child than participants in the NNS disclosure condition (ADE = −4 pp, P = 0.01; Figure 4, Supplemental Figure 2, Supplemental Table 1).
Compared with participants in the control condition, those in the child-focused NNS warning condition (ADE range: 20–35 pp) and in the NNS disclosure condition (ADE range: 22–31 pp) were both more likely to choose the product without added sweeteners for their child from all three product categories (all P < 0.001). There were no significant differences on this outcome between the two NNS labels for any of the product categories (all P > 0.05; Figure 4, Supplemental Figure 2, Supplemental Table 1).
For choice of sugar-sweetened product, there was more variation across product categories. For fruit drink, there were no significant differences between any of the labeling conditions (all P > 0.05). For yogurt, compared with participants in the control condition, those in the NNS disclosure condition were slightly less likely to choose a product with added sugar for their child (ADE = −3 pp, P = 0.02); whereas for breakfast cereal, compared with participants in the control condition, participants in the child-focused NNS warning condition were less likely to choose a product with added sugar for their child (ADE = −5 pp, P = 0.005). Importantly, regardless of product category, exposure to either type of NNS label did not increase participants’ likelihood of choosing a sugar-sweetened product for their child compared with the control condition (all P > 0.05; Figure 4, Supplemental Figure 2, Supplemental Table 1).
After seeing all products, among participants who reported noticing any FOPLs, those in the NNS disclosure condition were the most likely to recall seeing labels about NNS (ADE compared with control = 77%, ADE compared with child-focused NNS warning = 21%, both P < 0.001), followed by those in the child-focused NNS warning condition (ADE compared with control = 56 pp, P < 0.001) and, lastly, by those in the control condition. There were no significant differences between the likelihood of recalling seeing labels about sugar between participants in the control condition and those in the NNS disclosure condition (P > 0.05), but both were more likely to recall labels about sugar than participants in the child-focused NNS warning condition (ADE compared with control = −5 pp, ADE compared with NNS disclosure = −4 pp, both P < 0.001). There were no significant differences between the likelihood of recalling seeing labels about any other nutrients or ingredients between any of the labeling conditions (all P > 0.05; Figure 5).
FIGURE 5.
Predicted probability of recalling seeing front-of-package label about each nutrient/ingredient after task 1, by label type (n = 2,937). ∗Statistically significant difference at the 95% confidence level. Predicted probabilities refer to predicted margin post-estimation. Products in the experimental task only carried sugar labels (all arms) and non-nutritive sweeteners (NNS) labels (child-focused NNS warning and NNS disclosure arms). The other nutrients/ingredients were provided as response options for comparison.
Perceptions and intentions assessments
Compared with the control label, participants perceived both the child-focused NNS warning (ADE = 1.53) and the NNS disclosure (ADE = 0.92) as more effective (both P < 0.001). Participants also perceived the child-focused NNS warning as more effective than the NNS disclosure (ADE = 0.61, P < 0.001; Figure 6, Supplemental Table 2).
FIGURE 6.
Predicted mean perceived message effectiveness (n = 3,503), perceived product healthfulness (n = 3,516), and intentions to limit child’s nonnutritive sweetener intake (n = 3,514) by label type. ∗Statistically significant difference at the 95% confidence level. Predicted means refer to predicted margin postestimation. Perceived message effectiveness (PME) and perceived product healthfulness were measured repeatedly with two products. There were no interactions between label type and product in the analytical models. Outcomes were measured on a scale between 1 (least) and 5 (most). NNS, nonnutritive sweeteners
Compared with participants in the control condition, those in the child-focused NNS warning condition (ADE = −1.43) and in the NNS disclosure condition (ADE= −0.81) both perceived products as less healthful (P < 0.001). Participants in the child-focused NNS warning condition also perceived products as less healthful than those in the NNS disclosure condition (ADE = −0.62, P < 0.001; Figure 6, Supplemental Table 2).
Compared with participants in the control condition, those in the child-focused NNS warning condition (ADE = −0.71) and in the NNS disclosure condition (ADE = −0.26) both perceived products (which were NNS-sweetened) as less healthful relative to a version of the same product sweetened with sugar (both P < 0.001). Participants in the child-focused NNS warning condition also perceived products as less healthful relative to a version of the same product sweetened with sugar than those in the NNS disclosure condition (ADE = −0.44, P < 0.001; Supplemental Figure 3, Supplemental Table 2).
Compared with participants in the control condition, those in the child-focused NNS warning condition (ADE = 0.66) and in the NNS disclosure condition (ADE = 0.17) both reported higher intentions to limit their child’s NNS intake (both P < 0.001). Participants in the child-focused NNS warning condition also reported higher intentions to limit their child’s NNS intake than those in the NNS disclosure condition (ADE = 0.49, P < 0.001; Figure 6, Supplemental Table 2).
Moderation analyses
Neither gender nor educational level moderated the effects of either type of NNS label (compared with control) on correct identification of NNS-containing products, choice of NNS-sweetened product, or choice of product without added sweeteners (all P > 0.05 for interaction terms, Supplemental Table 3).
NNS tabletop use did not moderate the effects of the labels on identification of NNS-containing products. In contrast, NNS tabletop use moderated the effects of the NNS disclosure (but not of the child-focused NNS warning) on choice of NNS-sweetened product and on choice of product without added sweeteners for one product category: breakfast cereals. This moderation was such that the effects of the NNS disclosure on these outcomes (compared with control) were smaller among those who reported using tabletop NNS in the past 7 days (respectively: ADE = −0.30 and ADE = 0.27) compared with those who did not report using it (respectively ADE = −0.37 and ADE = 0.37; P = 0.01 for both interaction terms; Supplemental Tables 4 and 5).
Discussion
In this online randomized controlled trial with Chilean parents, a child-focused NNS warning label and an NNS disclosure label both increased participants’ ability to identify products containing NNS compared with a control without NNS labels, with the NNS disclosure exhibiting a slightly larger effect than the child-focused NNS warning. Both labels also led to reductions in participants’ likelihood of choosing NNS-sweetened products for their children, with the child-focused NNS warning exhibiting a slightly larger effect than the NNS disclosure for one product category. Both labels increased participants’ likelihood of choosing products without any added sweeteners for their children, with no differences between the two NNS labels. Neither label led to unintended increases in participants’ choice of sugar-sweetened products for their children. Finally, both labels were perceived as more effective than control labels at discouraging parents from purchasing products with NNS for their children, decreased products’ perceived healthfulness, and increased parents’ intention to limit their children’s NNS intake, with the child-focused NNS warning exhibiting slightly larger effects on these outcomes than the NNS disclosure label. Thus, overall, both NNS labels exhibited their intended effects.
In 2024, a panel of experts recommended NNS warning labels as a potential strategy to address Chilean children’s high NNS intake [34]. Our findings support the hypothesis that such labels can help parents not only identify products containing NNS more easily but also reduce selection of NNS-sweetened products for their children without increasing selection of sugar-sweetened products. In our study, NNS labels led parents to choose products without added sweeteners, thus achieving their intended goal. It is noteworthy that this study consistently offered parents options without any added sweeteners (either sugar or NNS) that were still close substitutes for the sugar-sweetened and NNS-sweetened products offered. Such close substitutes may not always be available in real-world conditions, because the absence of added sweeteners can change products’ flavor profile. However, under a scenario where both added sugar and NNS labels are mandatory, manufacturers have a greater incentive to develop added sweetener-free products, so their availability may increase. In fact, evidence from Mexico suggests that use of sugar and of NNS both decreased for product categories similar to those used in this study, suggesting that reformulation to avoid both added sugar and NNS labels is feasible provided the right policy incentives [19].
When comparing between the two types of NNS labels examined, we found that the NNS disclosure label had a slightly larger effect on participants’ ability to identify products containing NNS and was more readily recalled compared with the child-focused NNS warning label. These results are likely explained by the NNS disclosure label’s shorter message and simpler design, which is more eye-catching and familiar to Chileans—thus possibly drawing attention more easily and requiring less cognitive processing [35]. In contrast, when participants were asked to focus on the label, the child-focused NNS warning was perceived as more effective at discouraging NNS consumption, led to a greater reduction in the perceived healthfulness of the products to which they were applied, and increased participants’ intentions to limit their children’s NNS intake the most. This may be due to the statement that the product was not recommended for children, which, in prior qualitative studies in Brazil and the United States, parents considered particularly compelling [23,36]. However, the visual design of the child-focused NNS warning may limit its effectiveness in more realistic decision-making contexts. Policymakers considering this label might therefore consider increasing its size and/or adding interpretive, attention-grabbing visual elements, such as icons, to draw more attention to it. On a separate note, if policymakers’ aim is to discourage NNS consumption not only among children but also across the broader population—as recommended by the WHO—NNS disclosure labels should, in principle, better align with their goal. However, this hypothesis was not tested in the present study and requires additional research for empirical verification.
Parents’ educational level did not influence the effects of the NNS labels tested in this study. This is noteworthy given that NNS consumption among Chilean children differs based on mothers’ education [37] and that the NNS labels in this study used a somewhat complex technical term (“edulcorantes”) to refer to NNS, which may be more familiar to individuals with higher literacy levels. Providing participants with an explanation for this term before beginning the experiment may have attenuated differential effects by participants’ educational level—and although we cannot determine if this was the case, if so, it would demonstrate the importance of public education about the label’s content to minimize the potential for inequitable effects. Alternatively, previous evidence suggests that nutrition labels’ salience and interpretive elements (e.g., shape, color, statement) tend to drive label effectiveness, such that well-designed labels can be effective even among individuals with lower nutrition literacy [38]. Therefore, another possibility is that participants’ precise understanding of the terminology used may have been less important than the NNS labels’ interpretive features.
This study’s strengths include its experimental design allowing for causal inference, selection tasks providing participants with several product options developed by a professional designer to mitigate the influence of existing product and brand preferences (but following label application guidelines to enhance realism), a large representation of parents with low educational level, and the use of more than one type of NNS label for comparison purposes. However, this study also has limitations. As an online study with fictitious products, we cannot establish how generalizable results would be to real-world shopping contexts, where product preferences are already established and substitution options may be more limited. We also used only a few product categories and thus cannot generalize results to other types of products. However, it is worth noting that we used product categories that have undergone considerable reformulation efforts in Chile and are some of the main sources of NNS intake among children [16]. Finally, we used a convenience sample, which limits the generalizability of our findings—yet previous analyses suggest that online convenience and representative samples tend to produce experimental results similar in direction [39,40].
Conclusion
NNS warning labels greatly improved parents’ ability to identify NNS-containing products and led to healthier product choices for their children, showing promise for helping to curb NNS intake in this vulnerable age group.
Author contributions
The authors’ responsibilities were as follows—ADC, NR, LST, and ACS: designed the study; ADC: performed the data analyses, drafted the manuscript; NR, LST, and ACS: reviewed and contributed; and all authors have approved the final manuscript.
Data availability
The data set supporting the conclusions of this article is available in the Open Science Framework repository, at https://osf.io/5yj4d/overview?view_only=24b7ad7701224ed786311711bdb04d02.
Declaration of Generative AI and AI-assisted technologies in the writing process
The authors declare that no generative AI or AI-assisted technologies were used in the writing of this manuscript.
Funding
This research was funded by Bloomberg Philanthropies (grant #2019-71181). General support to this study was provided by the NIH grant to the Carolina Population Center (grant P2C HD050924 and T32 HD007168). NR is supported by the ANID/Subvención a la Instalación en la Academia Grant #85240064. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders. The funders had no role in the design, execution, interpretation, or writing of the study.
Conflict of interest
The authors report no conflicts of interest.
Acknowledgments
We thank Emily Busey for assistance with graphics development, Maxime Bercholz for assistance with power calculations, and Yelyzaveta Minaieva for assistance with manuscript proofing.
Footnotes
Supplementary data to this article can be found online at https://doi.org/10.1016/j.cdnut.2026.109463.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
References
- 1.World Health Organization . World Health Organization; Geneva, Switzerland: 2023. Use of Non-Sugar Sweeteners: WHO Guideline; p. 1. [Google Scholar]
- 2.Mennella J.A., Bobowski N.K., Reed D.R. The development of sweet taste: from biology to hedonics. Rev. Endocr. Metab. Disord. 2016;17(2):171–178. doi: 10.1007/s11154-016-9360-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Swithers S.E. Artificial sweeteners are not the answer to childhood obesity. Appetite. 2015;93:85–90. doi: 10.1016/j.appet.2015.03.027. [DOI] [PubMed] [Google Scholar]
- 4.Sylvetsky A., Rother K.I., Brown R. Artificial sweetener use among children: epidemiology, recommendations, metabolic outcomes, and future directions. Pediatr. Clin. North Am. 2011;58(6):1467–1480. doi: 10.1016/j.pcl.2011.09.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.World Health Organization Expert Committee on Food Additives (WHO/JECFA) Aspartame [Internet] https://apps.who.int/food-additives-contaminants-jecfa-database/Home/Chemical/62?utm_source=chatgpt.com [July 10, 2026; date cited]. Available from:
- 6.World Health Organization Expert Committee on Food Additives (WHO/JECFA) Acesulfame Potassium [Internet] https://apps.who.int/food-additives-contaminants-jecfa-database/Home/Chemical/926 [July 10, 2026; date cited]. Available from:
- 7.World Health Organization Expert Committee on Food Additives (WHO/JECFA) Saccharin [Internet] https://apps.who.int/food-additives-contaminants-jecfa-database/Home/Chemical/3164?utm_source=chatgpt.com [July 10, 2026; date cited]. Available from:
- 8.World Health Organization Expert Committee on Food Additives (WHO/JECFA) Sucralose [Internet] https://apps.who.int/food-additives-contaminants-jecfa-database/Home/Chemical/2340 [July 10, 2026; date cited]. Available from:
- 9.World Health Organization Expert Committee on Food Additives (WHO/JECFA) Steviol Glycosides [Internet] https://apps.who.int/food-additives-contaminants-jecfa-database/Home/Chemical/267?utm_source=chatgpt.com [July 10, 2026; date cited]. Available from:
- 10.World Health Organization Expert Committee on Food Additives (WHO/JECFA) Advantame [Internet] https://apps.who.int/food-additives-contaminants-jecfa-database/Home/Chemical/6181?utm_source=chatgpt.com [July 10, 2026; date cited]. Available from:
- 11.World Health Organization Expert Committee on Food Additives (WHO/JECFA) Neotame [Internet] https://apps.who.int/food-additives-contaminants-jecfa-database/Home/Chemical/5107 [July 10, 2026; date cited]. Available from:
- 12.Baker-Smith C.M., de Ferranti S.D., Cochran W.J., Committee on Nutrition, Section on Gastroenterology, Hepatology, and Nutrition, Abrams S.A., Fuchs G.J., III The use of nonnutritive sweeteners in children. Pediatrics. 2019;144(5) doi: 10.1542/peds.2019-2765. [DOI] [PubMed] [Google Scholar]
- 13.Campos M.J., L, Silva J.G., Pereira A.M.P.T., Pena A. Non-sugar sweeteners and children: the current picture and controversies. Front. Nutr. 2025;12 doi: 10.3389/fnut.2025.1676373. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Corvalán C., Reyes M., Garmendia M.L., Uauy R. Structural responses to the obesity and non-communicable diseases epidemic: the Chilean law of food labeling and advertising. Obes. Rev. 2013;14(S2):79–87. doi: 10.1111/obr.12099. [DOI] [PubMed] [Google Scholar]
- 15.Reyes M., Taillie L.S., Popkin B., Kanter R., Vandevijvere S., Corvalán C. Changes in the amount of nutrient of packaged foods and beverages after the initial implementation of the Chilean Law of Food Labelling and Advertising: a nonexperimental prospective study. PLoS Med. 2020;17(7) doi: 10.1371/journal.pmed.1003220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zancheta Ricardo C., Corvalán C., Smith Taillie L., Quitral V., Reyes M. Changes in the use of non-nutritive sweeteners in the Chilean food and beverage supply after the implementation of the Food Labeling and Advertising Law. Front. Nutr. 2021;8 doi: 10.3389/fnut.2021.773450. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Rebolledo N., Reyes M., Popkin B.M., Adair L., Avery C.L., Corvalán C., et al. Changes in nonnutritive sweetener intake in a cohort of preschoolers after the implementation of Chile’s Law of Food Labelling and Advertising, Pediatr. Obes. 2022;17(7) doi: 10.1111/ijpo.12895. [DOI] [PubMed] [Google Scholar]
- 18.Pan American Health Organization . Pan American Health Organization [Internet]; 2016. PAHO Nutrient Profile Model.https://www.paho.org/en/nutrient-profile-model 2016. Available from: [Google Scholar]
- 19.Salgado J.C., Pedraza L.S., Contreras-Manzano A., Aburto T.C., Tolentino-Mayo L., Barquera S. Product reformulation in non-alcoholic beverages and foods after the implementation of front-of-pack warning labels in Mexico. PLoS Med. 2025;22(3) doi: 10.1371/journal.pmed.1004533. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Arellano-Gómez L.P., Jáuregui A., Nieto C., Contreras-Manzano A., Quevedo K.L., White C.M., et al. Effects of front-of-package caffeine and sweetener disclaimers in Mexico: cross-sectional results from the 2020 International Food Policy Study. Public Health Nutr. 2023;26(12):3278–3290. doi: 10.1017/S1368980023002100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Munguía A., Sagaceta-Mejía J., Cruz-Casarrubias C., Durán R., Barquera S., Tolentino-Mayo L. Identification and understanding of precautionary legends in the Mexican warning labeling for sweetened beverages. Salud Pública México. 2025;67(2):153–162. doi: 10.21149/15879. [DOI] [PubMed] [Google Scholar]
- 22.Consulta pública para la modificación del reglamento sanitario de los alimentos, decreto supremo n° 977/96 del ministerio de salud introduce nuevo artículo 120 TER. 2024. https://www.minsal.cl/wp-content/uploads/2021/11/Texto-Consulta-Publica-Art-120-ter.pdf [Internet]. Available from: [Google Scholar]
- 23.Grilo M.F., Nunes B.S., Eberle M.F., Vallone N., Nieto C., Cólon-Ramos U., et al. Perceptions of non-sugar sweeteners and front-of-package labels among parents of preschool and school-aged children in Brazil. Public Health Nutr. 2025;28(1) doi: 10.1017/S1368980025101146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.D’Angelo Campos A., Ng S.W., Duran A.C., Khandpur N., Taillie L.S., Christon F.O., et al. “Warning: ultra-processed”: an online experiment examining the impact of ultra-processed warning labels on consumers’ product perceptions and behavioral intentions. Int. J. Behav. Nutr. Phys. Act. 2024;21(1):115. doi: 10.1186/s12966-024-01664-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.D’Angelo Campos A., Grummon A.H., Ng S.W., Puhl R.M., Golden S.D., Hall M.G. Front-of-package food labels and perceived weight stigmatization: a randomized clinical trial. JAMA Netw. Open. 2025;8(6) doi: 10.1001/jamanetworkopen.2025.16821. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Grummon A.H., Hall M.G., Taillie L.S., Brewer N.T. How should sugar-sweetened beverage health warnings be designed? A randomized experiment. Prev. Med. 2019;121:158–166. doi: 10.1016/j.ypmed.2019.02.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Grummon A.H., Gibson L.A., Musicus A.A., Stephens-Shields A.J., Hua S.V., Roberto C.A. Effects of 4 interpretive front-of-package labeling systems on hypothetical beverage and snack selections: a randomized clinical trial, JAMA Netw. Open. 2023;6(9) doi: 10.1001/jamanetworkopen.2023.33515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hall M.G., Grummon A.H., Higgins I.C.A., Lazard A.J., Prestemon C.E., Avendaño-Galdamez M.I., et al. The impact of pictorial health warnings on purchases of sugary drinks for children: a randomized controlled trial. PLoS Med. 2022;19(2) doi: 10.1371/journal.pmed.1003885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Grummon A.H., Taillie L.S., Golden S.D., Hall M.G., Ranney L.M., Brewer N.T. Sugar-sweetened beverage health warnings and purchases: a randomized controlled trial. Am. J. Prev. Med. 2019;57(5):601–610. doi: 10.1016/j.amepre.2019.06.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Taillie L.S., Higgins I.C.A., Lazard A.J., Miles D.R., Blitstein J.L., Hall M.G. Do sugar warning labels influence parents’ selection of a labeled snack for their children? A randomized trial in a virtual convenience store. Appetite. 2022;175 doi: 10.1016/j.appet.2022.106059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Ministerio de Salud de Chile . Reglamento Sanitario de los Alimentos. Decreto N° 977/96. 2016. https://www.dinta.cl/wp-content/uploads/2016/11/DECRETO_977_96_Actualizado-Marzo-2016_.pdf [Internet]. Available from: [Google Scholar]
- 32.Noar S.M., Barker J., Bell T., Yzer M. Does perceived message effectiveness predict the actual effectiveness of tobacco education messages? A systematic review and meta-analysis. Health Commun. 2020;35(2):148–157. doi: 10.1080/10410236.2018.1547675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Baig S.A., Noar S.M., Gottfredson N.C., Boynton M.H., Ribisl K.M., Brewer N.T. UNC perceived message effectiveness: validation of a brief scale. Ann. Behav. Med. 2019;53(8):732–742. doi: 10.1093/abm/kay080. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Reyes M., Pino C., Ortega A., Pemjean I., Corvalán C., Garmendia M.L. Potential actions for preventing high consumption of non-nutritive sweeteners among Chilean children and adolescents: recommendations from a panel of relevant actors. Public Health Nutr. 2024;27(1) doi: 10.1017/S1368980024001745. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Ribeiro M., de Morais Sato D., Rojas C.F.U., Spinillo C.G., Mais L.A., Borges C.A., et al. Experts contributions to the development of a non-sugar sweeteners warning label for Brazilian food products. PLoS One. 2025;20(9) doi: 10.1371/journal.pone.0331302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Vallone N., Taillie L.S., Krieger J.W., Grilo M.F., Patel P., Diaz-Guzman G., et al. Perceptions of nonsugar sweeteners and nonsugar sweetener front-of-package labels among parents in the United States: a qualitative study. J. Acad. Nutr. Diet. 2025;126(1) doi: 10.1016/j.jand.2025.09.006. [DOI] [PubMed] [Google Scholar]
- 37.Venegas Hargous C., Reyes M., Smith Taillie L., González C.G., Corvalán C. Consumption of non-nutritive sweeteners by pre-schoolers of the food and environment Chilean cohort (FECHIC) before the implementation of the Chilean food labelling and advertising law. Nutr. J. 2020;19(1):69. doi: 10.1186/s12937-020-00583-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Roberto C.A., Ng S.W., Ganderats-Fuentes M., Hammond D., Barquera S., Jauregui A., et al. The influence of front-of-package nutrition labeling on consumer behavior and product reformulation. Annu. Rev. Nutr. 2021;41(1):529–550. doi: 10.1146/annurev-nutr-111120-094932. [DOI] [PubMed] [Google Scholar]
- 39.Coppock A., Leeper T.J., Mullinix K.J. Generalizability of heterogeneous treatment effect estimates across samples. Proc. Natl. Acad. Sci. U.S.A. 2018;115(49):12441–12446. doi: 10.1073/pnas.1808083115. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Jeong M., Zhang D., Morgan J.C., Ross J.C., Osman A., Boynton M.H., et al. Similarities and differences in tobacco control research findings from convenience and probability samples. Ann. Behav. Med. 2019;53(5):476–485. doi: 10.1093/abm/kay059. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data set supporting the conclusions of this article is available in the Open Science Framework repository, at https://osf.io/5yj4d/overview?view_only=24b7ad7701224ed786311711bdb04d02.