Impacts of sugar and sweetener warning labels on substitution between sugar- and non-sugar-sweetened beverages in a non-hypothetical selection task

Abstract

Background

Front-of-pack (FOP) warning labels have demonstrated effectiveness for reducing sugar-sweetened beverage (SSB) consumption and switching to water. However, an unintended consequence is that they may also increase switching to non-sugar-sweetened beverages (NSSBs). A non-hypothetical experimental study examined the effectiveness of combining sugar and sweetener FOP warning labels to reduce sugary drink consumption and prevent NSSB substitution. The study also examined potential integration with Australia and New Zealand’s existing Health Star Rating (HSR) system to determine suitability for local context and other jurisdictions with interpretive labelling schemes already in place.

Methods

Participants (N = 414) accessed an online convenience store app via an on-campus laptop to select one drink from an array of 10 beverages, on three occasions. Drink options included a variety of SSBs, 100% fruit juice, NSSBs, and water. Following an escalating exposure procedure, drinks were presented (1) without any additional labelling, (2) with warning labels added to sugary drinks or to both sugary drinks and NSSBs (according to allocated condition), then (3) with HSR icons added to all drinks. Participants were informed they would receive a complementary drink, based on their selections, following the completion of a brief questionnaire.

Results

Baseline results indicated that SSBs and water were the most and least popular drink choices, respectively. Placing FOP warning labels on sugary drinks decreased selection of SSBs and increased NSSB and water choices. Water became the most popular individual drink choice in response to warnings on sugary drinks. Placing FOP warning labels on both sugary drinks and NSSBs successfully avoided an increase in NSSB choices, whilst also increasing water selections, but did not significantly decrease selection of SSBs until HSR ratings were added. The incorporation of HSR icons consolidated warning label effects on NSSB and water selection across both conditions.

Conclusions

Results demonstrate the potential of FOP sugar warning labels for addressing beverage consumption behaviours. The incorporation of sweetener warning labels may successfully avoid substitution towards NSSBs, whilst still promoting water choices, but may also dilute the impact of the sugary drinks warning labels. Warning labels were complementary to existing interpretive FOP labels.

Background

Globally, consumption of sugary drinks is high. In 2018, sugar-sweetened beverages (SSBs) were consumed at an average rate of 2.7 serves per week, with this increasing to 7.8 serves across Latin America and the Caribbean [1]. In Australia, 6.4% of the population reported consuming SSBs daily in 2020–2021, at an average rate of 2.9 cups per day, with young adults (18–24 years) being the highest consumers. Overconsumption of sugary drinks significantly contributes to the risk of developing obesity and related chronic health conditions including type II diabetes, cardiovascular disease, tooth decay, and periodontal disease [2,3,4,5]. With sugary drink consumption being identified as a leading modifiable risk factor for the development of many health impacts, effective interventions for reducing excess consumption of sugary drinks are urgently needed.

Front-of-pack (FOP) warning labels offer a promising approach to discourage the consumption of sugary drinks, and over 30 countries have already implemented some kind of labelling system [6, 7]. Chile, being the first country to mandate FOP nutrient warning labels, demonstrated the success of this approach [8], and a recent systematic review and meta-analysis demonstrated the effectiveness of SSB warning labels for discouraging the selection of such drinks [9]. To date, warning label schemes have largely involved communicating risks associated with the consumption of labelled products, with the most common approach involving denoting products as “high in” a particular nutrient such as sugar [6]. Chile implemented this approach in 2016, and several countries have followed suit with similar mandatory labelling schemes, including Peru, Mexico, Uruguay, Argentina, Venezuela, Brazil, and Canada [10]. In addition to “high in” labels, other label types, such as health warnings and labels denoting teaspoons of sugar, have also shown potential. Whilst research directly comparing the effectiveness of different label types is limited, a recent evaluation of 27 label types indicated that labels providing concrete, factual, and easy-to-interpret information, such as teaspoons of sugar, provide benefit for consumers beyond the “high in” labels [11]. Comparisons of the effectiveness of label types are still needed, especially in non-hypothetical contexts, to ascertain actual behaviour change in response to the intervention.

Policies targeting sugary drink consumption are designed to reduce excess consumption. An inevitable consequence is that they also generate switching behaviours to other beverages. Research on FOP warning labels on SSBs has demonstrated potential switching behaviours, including increased consumption of non-sugar-sweetened beverages (NSSBs). A recent study assessing the effectiveness of a range of SSB warning labels, in a real-world setting, found an increase in NSSB choices in conjunction with the decrease in SSB and increase in water choices [12]. Countries that have implemented SSB policies have also noted a corresponding increase in NSSB purchases [13, 14]. Switching from SSBs to NSSBs, rather than water, may also result in increased risk of long-term health effects. There is increasing evidence linking NSSB consumption to health effects, with a World Health Organization (WHO) review on non-sugar sweetener consumption concluding there is some evidence from prospective cohort studies that greater consumption may be associated with increased body weight and increased risk of type 2 diabetes, cardiovascular disease, and all-cause mortality [15]. The review also demonstrated no significant reductions in body weight or BMI when non-sugar sweeteners were used as a direct replacement for sugars in randomised controlled trials (with these results being primarily based on studies where SSB consumption was replaced with NSSB consumption). The WHO caution against including non-sugar sweeteners in the diet [16], as such, it is important to find effective methods for reducing SSB consumption that minimise substitution to NSSBs and promote water. One potential approach would be to also apply labels to NSSBs; informing consumers that the product contains sweeteners. Evidence from taxing policies indicates that incorporating sweeteners into sugar policies may be an effective means of avoiding substitution towards NSSBs. For example, evaluations of real-world sugar taxes demonstrate high rates of substitution towards NSSBs when NSSBs are excluded from the tax [13, 14], but limited substitution towards NSSBs when they are incorporated [17, 18]. This indicates the potential for a warning label policy incorporating both sugar and sweeteners to similarly avoid the substitution towards NSSBs that is seen when warning labels are placed only on sugary drinks. Further, placing warning labels on NSSBs may confer other benefits for consumers as research indicates that some consumers are concerned about sweetener intake and, perhaps more importantly, are often unaware of sweetener content in products [19].

In Australia, whilst FOP warning labels are not currently in place, there is a voluntary interpretive Health Star Rating (HSR) labelling scheme, which was introduced in 2014 and revised in 2020 [20]. HSRs range from 0.5 to 5 stars (with higher stars indicating greater nutritional quality). A mandatory warning label system would complement the current voluntary HSR system and provide clarity for consumers, given that uptake has been relatively low and skewed towards products with a higher rating [21, 22]. A recent real-stakes Australian field study demonstrated the potential of the two schemes to coexist, with the addition of HSR icons alongside sugar warning labels reinforcing preferences for waters over SSBs [12]. Further research is needed, however, to ensure that a combined labelling approach (targeting both sugary drinks and NSSBs) would work within the Australian context, alongside the existing HSR icons.

Whilst research is limited, one hypothetical online study showed that non-sugar sweetener-specific warnings may help to reduce substitution towards NSSBs, with participants perceiving NSSBs to be less healthy for children when shown a drink with (versus without) a non-sugar sweetener warning label [23]. Research comparing different types of warning labels in non-hypothetical settings, however, is required. Non-hypothetical studies have greater potential to inform policies regarding substitution effects because they provide greater insight into actual, rather than hypothetical, behaviour change. Evidence from real-world evaluations of taxing policies, along with results of the hypothetical online study [23], demonstrate that the incorporation of NSSBs into taxes can avoid substitution towards NSSBs. However, it is currently unknown whether a similar effect may be achieved from combining sugar and sweetener warnings to discourage SSB consumption whilst avoiding substitution towards NSSBs. The present study sought to address this gap, assessing the potential benefit of incorporating sweetener labels, alongside sugar warning labels, for discouraging SSB choices (and increasing water choices) without simultaneously increasing NSSB choices, in a naturalistic, non-hypothetical environment. The study also sought to determine whether such warning label schemes would be effective alongside existing interpretive FOP labels, namely the existing Australian and New Zealand HSR system, to ensure suitability for implementation in an Australian context. It was hypothesised: (i) that sugar warning labels would decrease the selection of SSBs whilst increasing the selection of water and NSSBs, in line with existing research [12,13,14]; (ii) that the combination of sugar and sweetener warning labels would decrease SSB choices whilst increasing water but not NSSB choices [17, 18]; and (iii) that these effects would be maintained with the addition of the HSR icons [12].

Methods

Design

The study used a randomised, between-group, repeated measures design. Participants were randomly allocated to one of four warning label conditions; two conditions were exposed only to warning labels relevant to sugary drinks (sugary drinks warning labels), displaying (i) sugar content (sugar pictogram [SP]), or (ii) health effects (health text [HT]), and two conditions were exposed to warning labels relevant to both sugary drinks and sweeteners (sugary drink plus NSSB warning labels), displaying (iii) sugar content and sweetener warnings (SP + NSSB), or (iv) health effects and sweetener warning labels (HT + NSSB; see Fig. 1). Participants were exposed to a range of ten cold drinks on three occasions (Rounds 1 to 3), following an escalating exposure procedure, and were asked to select one drink each time. The drinks were presented to participants via a trial online convenience store app, accessed on laptops. The ten beverages were displayed simultaneously, with drink order randomised for each participant, and in each round. Drink options included four SSBs (soft drink, energy drink, sports drink, iced tea), four NSSBs (no-sugar versions of the four SSBs), one 100% fruit juice, and one water (see Table 1). Information about the drinks was progressively added across the three rounds, via warning labels and Health Star Rating (HSR) icons, as shown in Fig. 1. In Round 1, (control) drinks were presented without any warning labels or HSR icons. In Round 2, sugar warning labels were added to SSBs and/or sweetener warning labels to NSSBs, according to allocated condition. In Round 3, HSR icons were added to all drinks, alongside the warning labels from Round 2. Warning labels and HSR icons were displayed on the front of each bottle with enlarged versions of the labels/icons presented alongside each image to ensure they were visible to participants (see Fig. 2 for an example). When participants hovered over a drink in the app, a zoomed-in image of the drink label, which also displayed any warning labels, was shown, and an enlarged image of the whole bottle was displayed when participants selected a drink, along with the option to confirm their drink selection.

Fig. 1

Experimental procedure

Table 1 Descriptive characteristics of drinks included in the study

Fig. 2

Example image from app

Procedure

A booth was set up in a centrally located position on a university campus for a total of seven days throughout the university’s orientation week (3 days; 20–22 February 2023) and subsequent semester weeks (4 days between 28 March and 3 April 2023). Potential participants (16 + years) were recruited from the area around the booth, by trained research assistants, and were invited to participate in a trial of a convenience store app for selling cold beverages. Up to four participants completed the study at one time on laptops at the booth. Participants were provided brief verbal information and instructions on the study and the study information sheet. Participants provided informed consent before proceeding with the task, with brief on-screen instructions provided prior to commencing each round, and subsequently made their three drink selections. Participants then completed follow-up questions via an online survey. Survey completion was confirmed by one of the research assistants, who then checked the on-screen code which indicated participants’ Round 3 drink choice, to determine their complementary drink which was then provided to the participant.

A total of 414 people participated in the study, with most participating during the university’s orientation week (62.3%), and the highest proportion of participants being recruited on day two of orientation week (28.5%). Participation times ranged from 9.30 am to 4.00 pm, with most (85.8%) participating between 11.00 am and 3.00 pm. The mean temperature during data collection periods was 27.5 °C (range: 13.3–37.8).

Measures

The survey was programmed using RedCap survey software [24, 25]. The survey questions were developed from previously published measures [19] and have also been used in a similar previous study on warning labels on sugary drinks [12]. Questions assessed consumption, perceptions, and health beliefs relating to SSBs (referred to as sugary drinks), 100% fruit juice (referred to as unsweetened juice), and water, with new questions added for NSSBs (referred to as artificially sweetened drinks). The study protocol was refined through pilot testing with approximately 40 participants. Demographic items included age (in years) and gender. Typical beverage consumption was assessed by asking participants to indicate “How often [they] normally drink a bottle or can of the following” (daily, weekly, monthly, less than monthly, never) for a range of SSBs (soft drink, energy drink, sports drink, iced tea), NSSBs (no-sugar soft drink, energy drink, sports drink, and iced tea), 100% fruit juice, and water. Perceptions of consumption were assessed by asking participants “[their] opinion, [on] the amount of [sugary drinks/artificially sweetened drinks/unsweetened juice] that [they] drink during an average week…” (just right, too much, not enough). Influences on drink selection were assessed by asking “When making a drink selection, I am normally influenced by (select one): taste, familiarity, thirst, healthiness of the drink, energy boost, time of day, enjoyment, treat, other.” Health beliefs were assessed by asking participants “If an average person like you drank a [sugary drink/artificially sweetened drink/unsweetened juice] every day, would they be likely or unlikely to have health problems later in life?” (response options: very unlikely, somewhat unlikely, neither likely nor unlikely, somewhat likely, very likely), and extent to which “drinking [sugary drinks/artificially sweetened drinks/unsweetened juice] every day would increase [their] risk of weight gain, diabetes, tooth decay, heart disease, cancer, and depression” (not at all, somewhat, moderately, considerably, a great deal). Attention to and perceived effectiveness of labels was assessed through asking whether the participant recalled seeing a (i) HSR icon and (ii) warning label, and to what extent they agreed or disagreed (strongly disagree, disagree, neither, agree, strongly agree) that the icon/label grabbed their attention; they read and looked closely at the icon/label; the icon/label made them think about the health problems caused by sugary drinks; and the icon/label made them think about the health problems caused by artificially sweetened drinks. Participants were also asked to identify which of the warning labels they recalled seeing, from example images of each label type (sugar pictogram, health text, sweetener). The outside temperature for each participant at the time they completed the study was also collected. Ethics approval was obtained from The University of Adelaide Human Research Ethics Committee (approval number: H-2023–013).

Analyses

As the primary aim of this study was to determine the effect of adding sweetener labels to sugar warning labels on participants’ drink choices, the four individual label conditions (SP, HT, SP + NSSB, HT + NSSB) were grouped into two conditions for analyses: a sugary drinks warning label condition (SP and HT), and a sugary drinks plus NSSB warning label condition (SP + NSSB and HT + NSSB). The primary outcome for the main analyses of this study was a change in drink selection across each of the three rounds, assessed according to condition type (sugary drinks vs. sugary drinks plus NSSB warning labels). Drink options were grouped into four drink categories: SSBs (soft drink, energy drink, sports drink, and iced tea), NSSB versions of the SSBs, unsweetened juice (one option only), and water (one option only) for statistical analysis. Comparisons between variables were analysed using chi-square tests, one-way ANOVAs, and repeated samples Cochran’s Q tests, with McNemar tests for pairwise comparisons. Pairwise comparisons were adjusted using the Bonferroni correction and p ≤ 0.05 was used to indicate statistical significance.

A series of supplementary analyses were also conducted (see Additional file 1). To provide insight into any issues with randomisation, participant characteristics, usual consumption, and baseline (Round 1) drink choices were compared across conditions and genders. These analyses indicated gender differences in usual consumption patterns; therefore, analyses were also conducted separately among males and females, with results of note reported in the text, and detailed results available in Additional file 1. Additional analyses also compared key outcomes across the four individual label conditions and examined whether changes in drink selections varied according to participant characteristics, label perceptions, and participants’ attitudes and beliefs.

Results

Sample characteristics and baseline consumption patterns

Participant characteristics are presented in Table 2. There was a greater proportion of males (50.5%) than females (42.8%; 6.8% other/prefer not to say) and the mean age was 21.8 years (range: 16–69). Usual beverage consumption (weekly or more) was highest for bottled water (61.4%), followed by SSBs (38.2%), NSSBs (36.1%), and juice (35.0%), with participants nominating as many beverage types as relevant. For Round 1 drink choices, 42.3% chose an SSB, 32.6% chose an NSSB, 16.4% chose juice, and 9.2% chose bottled water.

Table 2 Participant characteristics, usual drink consumption, Round 1 choice, and average testing temperature, for total sample

Results of comparisons of sample characteristics across conditions are available from Additional file 1: Table S1, and indicate randomisation was successful, and that regular consumption of SSBs and NSSBs was significantly higher among males than females (p < 0.01). With respect to Round 1 drink choices (control; no labels), sports drink and no-sugar energy drink selection was higher among males than females (p < 0.01), whereas iced tea, no-sugar iced tea, and water selection were higher among females than males (p < 0.05).

Changes in drink popularity in response to warning labels (R2) and HSR icons (R3)

Participants’ drink choices are reported by drink category (grouped as all SSBs, Juice, all NSSBs, and water) and by individual drink choice for each round, with comparisons made between conditions (see Table 3). The pattern of drink choices in Round 1 (control; no label) was consistent across conditions; SSBs were the most popular drink category, followed by NSSBs, juice, and then water. With respect to individual drinks, iced tea and juice were the most popular choices in Round 1.

Table 3 Changes in drink popularity across rounds, by experimental condition

Following the addition of the warning labels in Round 2, across both conditions, NSSBs became the most popular drink category, followed by SSBs, water, and juice. In the sugary drinks warning label condition, there was a significant decrease in the selection of SSBs (− 16.9%), and a significant increase in the selection of NSSBs and water (+ 11.7% and + 9.4%, respectively; ps < 0.005). In the sugary drinks plus NSSB warning label condition, the pattern of results was similar, albeit less prominent: there was a slight decrease in selection of SSBs (− 6.0%; not significant), a slight increase in NSSB choices (+ 4.0%; not significant), and a significant increase in water selection (+ 7.0%; p = 0.032). Juice selection decreased slightly across both conditions (− 4.2% and − 5.0%, respectively) though not significantly. With respect to the selection of individual drinks, water became the most popular drink choice in the sugary drinks warning label condition and the second most popular drink choice in the sugary drinks plus NSSB warning label condition (with iced tea remaining the most popular drink choice in this condition).

Overall, the HSR icons presented in Round 3 had little additional effect in terms of change in drink choices across rounds (i.e. R2–R3; see Table 3). The combined effect of warning labels and HSR icons (R1–R3) further consolidated the effect of warning labels on drink selection in the sugary drinks warning label condition (Table 3). Moreover, the sugary drink plus NSSB warning label plus HSR icon condition (R3) corresponded with a significant reduction in SSB selection compared to baseline (R1; − 11.4%, p = 0.003), unlike the warning labels without the HSR icons condition (R2; − 6.0%, p = 0.006).

Individual-level (within participant) changes in drink choices in response to warning labels (R2) and HSRs (R3)

The extent to which individuals changed their drink selection in response to warning labels (i.e. from R1-R2) is presented in Table 4. Overall, a slightly greater proportion of participants changed their drink choices in response to the sugary drinks warning labels (40.8%), than the sugary drinks plus NSSB warning labels (33.3%). Participants who chose an NSSB in Round 1 were the least likely to switch drinks, across both conditions, though less switching occurred in the sugary drinks plus NSSB warning label condition (19.3%) than the sugary drinks warning label condition (27.4%). Of the few participants who did switch from NSSBs in R1 in the sugary drinks warning label condition (n = 17), participants switching from NSSBs were most likely to switch to water. In the sugary drinks plus NSSB warning label condition, however, participants switching from NSSBs (n = 14) were equally likely to switch to water or an SSB.

Table 4 Within-participant changes in drink selection across rounds and subsequent drink choice, by experimental condition

In contrast, switching drink selection was more common among participants who chose an SSB in Round 1, compared to those who had initially chosen an NSSB (n = 74 vs n = 31). Across both conditions, participants who switched from an SSB in Round 1 were most likely to switch to an NSSB in Round 2. The number of individuals switching from SSBs increased even further in Round 3 with the addition of HSR icons (n = 94). In the sugary drinks label condition, the majority switched to NSSBs (69%), followed by water (20%) and juice (11%). Switching to NSSBs was also common in the sugary drinks plus NSSB warning label condition, albeit to a lesser degree (51%), and water selection was greater (33%).

From Round 2 (warning labels) to Round 3 (warning labels and HSRs), approximately one third of participants changed their drink selection, with this finding consistent across conditions. The combined effect of warning labels and HSRs together (R3) in comparison to the control round (no warning labels; R1) was similar to that of warning label effects alone (R1–R2). Patterns of drink choices in Round 3 (warning labels and HSRs applied) were generally similar to Round 2 drink choices when only warning labels were applied.

Attention to labels and perceptions of label effectiveness, and general perceptions, attitudes, and health beliefs towards drinks

Attention to labels and perceptions of label effectiveness are presented in Table 5. Most participants reported noticing a warning label (65.9%), and this was consistent across conditions (sugary drinks: 61.5%; sugary drinks plus NSSBs: 70.6%). The proportion of participants who correctly identified the warning label they had been exposed to (prompted) was significantly higher in the sugary drinks warning label condition (43.2%) than the sugary drinks plus NSSB warning label condition (29.4%; p = 0.005).

Table 5 Attention to and perceived effectiveness of labels, and beverage attitudes/beliefs (total sample and experimental condition)

Among participants who reported seeing the warning labels, high proportions indicated the warning labels grabbed their attention (85.7%) and that they read and looked closely at the warning labels (75.5%), and around half indicated that the warning labels made them think about health problems caused by sugary drinks (55.3%), and artificially sweetened drinks (45.4%). These proportions were generally equivalent across conditions; although, a greater proportion of participants reported they read and looked closely at the warning labels in the sugary drinks plus NSSB warning label condition than the sugary drinks condition. A high proportion (86.7%) of participants recalled seeing the HSR icons (prompted, assessed prior to warning label recognition). Noticing, and perceived effectiveness of HSR icons was similar across conditions (see Table 5).

Health beliefs and attitudes towards SSBs and NSSBs were largely similar across conditions (see Table 5), although participants in the sugary drinks plus NSSB warning label condition were significantly more likely to believe that an average person would be likely to have health problems if they consumed unsweetened juice every day, compared to participants in the sugary drinks condition. Results of supplementary analyses, however, indicated some self-perceptions of consumption and health beliefs were higher among lower SSB consumers, lower NSSB consumers, and females (see Table S2). Perceiving current consumption as ‘just right’ was more likely among low SSB consumers for SSBs, among low NSSB consumers for NSSBs, and among females for both NSSBs and juice. Lower SSB consumers were also more likely to believe that daily SSB consumption would likely lead to health problems and increase their risk of weight gain, diabetes, tooth decay, heart disease, and cancer. Lower NSSB consumers were also more likely to believe that daily NSSB consumption would increase their risk of weight gain, diabetes, and tooth decay, as well as being more likely to believe that daily unsweetened juice consumption would likely lead to health problems. Females were more likely to believe that daily SSB, NSSB, and juice consumption would likely lead to health problems and that drinking NSSBs every day would increase their risk of weight gain and diabetes. Results of further supplementary analyses also indicated there were few significant differences regarding change in drink choice across rounds, according to attention to and perceived effectiveness of warning labels and HSR icons (see Table S3).

Discussion

Existing evidence supports the efficacy of sugar-focussed policy options, including FOP warning labels, for reducing SSB consumption and increasing water intake. However, these policies have also been found to increase consumption of other beverages, such as NSSBs [12,13,14]. Finding ways to implement SSB policies that do not increase consumption of NSSBs, given the increasing evidence of potential health effects of consuming NSSBs [15] is an important practical issue in public health policy. To assess the impact of FOP warning labels on beverage choices in a non-hypothetical environment, the present study investigated the effectiveness of sugar warning labels for influencing consumption behaviours, and the additional impact of adding sweetener warning labels alongside sugar warning labels.

The findings of this study demonstrate the popularity of high-sugar drinks, reinforcing the urgent need for interventions to curb the overconsumption of such products. In the absence of any labelling (Round 1; control), SSBs were the most popular drink type, and iced tea and juice the most popular individual drinks. Adding sugar warning labels to beverages effectively reduced the selection of SSBs, whilst increasing NSSB and water choices, in line with existing research [26, 27]. Notably, water became the most popular individual drink choice in Round 2 following the introduction of warning labels in the sugary drinks warning label condition, underscoring their potential for encouraging healthier beverage choices. However, when looking at drink types, NSSBs (i.e. artificially sweetened soft drink, iced tea, sports drink, energy drink) became the most popular drink type in Round 2, in line with existing evidence on the impact of SSB policies [26, 27]. The participants exposed to both sugary drink and NSSB warning labels in the present study did not display the same NSSB substitution effects in Round 2, but did increase water choices. SSB choices also decreased in this condition, albeit, not significantly or to the same extent as they did in the sugary drinks warning label condition, potentially indicating that the inclusion of sweetener warning labels may have diluted the impact of sugar warning labels on SSB choices. Currently, only three of the countries that have implemented mandatory warning labels have incorporated NSSBs into their policies (Argentina, Mexico, and Colombia) and there are currently no evaluations of their effectiveness. However, one study has examined the potential impact of Mexican sweetener warning labels which focus on children’s consumption of such products, stating “contains sweeteners – not recommended in children” [23]. Participants exposed to the sweetener warning labels had lower perceptions of healthfulness of sweetened drinks for children than when exposed to beverages without the label, and over 60% of participants indicated they used the warning labels at least sometimes when making purchasing decisions, indicating such labels may influence drink choices.

The less marked decrease in SSB choices in the condition with sugary drink and NSSB labels in the present study may also be due to a label ‘saturation’ effect given that, with the exception of water, every drink in the sugary drinks plus NSSB warning label condition had a warning label applied. As such, the only unlabelled option for participants choosing to avoid warning labels was water, which may not be the preference for many participants. Indeed, prior research indicates participants consider many factors when making drink selections and purchases, such as taste, value for money, and brand preferences [28]. In contrast, in the sugary drinks warning label condition, there were more unlabelled options for participants wishing to avoid labelled drinks. The NSSBs also offered direct alternatives to SSBs (e.g. Coke versus No-Sugar Coke) for participants, but similar unlabelled alternatives were not available in the sugary drinks plus NSSB warning label condition. Reinforcing this, there appeared to be more within-participant movement in choices in the sugary drinks warning label condition, such that switching was more common in the sugary drinks warning label condition (40.8%) compared to the sugary drinks plus NSSB warning label condition (36.8%). Participants who chose an SSB in Round 1, were more likely to switch in the sugary drinks warning label condition (46.5%) than those in the sugary drinks plus NSSB warning label condition (36.8%), with most of these participants (in both conditions) switching to an NSSB. Whilst the switch to NSSBs was greater in the absence of NSSB labels, the absence of unlabelled alternative drink options may have limited the impact of the intervention. Additional research is needed to determine the effectiveness of a combined (sugar-plus-sweetener) labelling approach on selection of SSBs and NSSBs in an environment with greater availability and variety of products without labels (e.g. selection of waters or other non-sweetened drinks).

Australia currently has a voluntary Health Star Rating (HSR) labelling system that can be applied to products at the discretion of the industry. The latest review of this system (2023) indicated that HSR uptake had occurred for around 37% of eligible products [29], with evidence suggesting that manufacturers are selective in the products that display stars, with the four-star rating most commonly applied [30]. At the time of this study, only one of the beverages displayed an on-package HSR icon, and thus for the purpose of this study HSRs were added uniformly to all beverages, by the researchers. Despite low actual uptake of the voluntary scheme, we explored the combined impact of warning labels and HSR icons in this study, to ensure FOP warning labels would be well suited to the Australian and New Zealand context specifically, and more generally to other jurisdictions with interpretive labels. HSRs were found to have little additional effect in comparison to warning label effects, but complemented warning label effects. For example, the addition of HSR icons with sugary drink and NSSB warning labels increased effects on SSB choices, resulting in decreased selection of SSBs and increased selection of water in response to the combination of these labels. Further, the addition of HSR icons in the sugary drinks warning label condition increased the effect on water choices, resulting in significant decreases in SSB choices and significant increases in NSSB and water choices in response to these labels. This suggests the addition of the HSR icons clearly rating water as healthier (5 stars) may have helped to shift choices away from SSBs (when combined with sugary drink and NSSB warning labels) and towards water (when combined with sugary drinks warning labels). Overall, results demonstrate that the warning label and HSR icon labelling schemes would complement each other. These findings are consistent with the findings of a previous study that also found sugar warning labels can complement HSR icons in encouraging healthier drink selection [12]. However, this study extends these findings by showing sweetener warning labels can also complement HSR icons.

The majority of participants reported noticing the warning labels. Correct identification of warning labels (prompted) was lower for the sugary drinks plus NSSB warning label condition, compared to the sugary drinks warning label condition. This may be due to the amount of information participants had to recall (two labels, as opposed to one), or lower familiarity with considering the health effects of non-sugar sweeteners. Whilst evidence on the health effects of sweeteners is increasing [31], the health risk associated with non-sugar-sweetened beverage consumption are not as well-known as that of sugary drinks consumption [19], also observed in the results of this study. A higher proportion of participants noticed HSR icons compared to warning labels, but this was likely due to the familiarity of HSR icons among the Australian population, having been voluntarily implemented since 2014 [29].

The present study had many strengths, including its non-hypothetical decision-making, direct comparison of different types of sugar warning labels, and novel testing of the incorporation of sweetener warning labels, as well as demonstrating the combined impact of HSR and warning label schemes specific to the Australian context. However, there were several limitations. The study setting, a university campus was chosen to provide access to young adults, the primary consumers of sugary drinks; however, there are sample limitations and results may not be generalisable to the wider population. Additionally, due to the limitations of the convenience store app and the maximum number of beverages that could be displayed without compromising visibility, the number of drinks was limited. This resulted in a disproportionate amount of SSBs (n = 4) and NSSBs (n = 4) compared to unsweetened juice (n = 1) and water (n = 1), as such there was very limited availability of non-labelled products in the sugary drinks plus NSSB warning label conditions. Whilst these proportions may be somewhat in line with the range of products available in stores, which are heavily dominated by SSBs and NSSBs and have a limited variety of healthier options [32], future research would benefit from including a broader range of beverages. Relatedly, testing with a broader range of beverages would likely provide insight into the potential of any brand effects, although the beverages included in the present study were all chosen as common commercially available products to minimise any impact of brand effects and drink popularity; participants who initially chose an SSB in the present study were also the most likely to change their drink choice, suggesting that these participants were not anymore attached to these popular beverages compared to the other drink categories. Finally, it is possible that given the nature of the study, participants may have inferred study aims and responses may have been at risk of potential social desirability bias, however, this risk was mitigated by the non-hypothetical nature of the study as participants were also aware they would receive one of their chosen beverages upon completion of the study.

Conclusions

This study supports and extends the results of previous online experimental studies focussing on sugary drink warnings to a non-hypothetical experiment that also addresses substitution effects demonstrated in response to sugary drink interventions. The results demonstrate the potential for the advancement of FOP warning labels into policy, demonstrating a reduced selection of SSBs in response to warning labels. Incorporation of sweetener labels may assist in mitigating substitution to NSSBs, but may also dilute the impact of sugary drinks warning labels if there are few non-label drink options. More in-depth research is required to fully evaluate the potential of including warning labels on NSSBs as well as sugary drinks, and the viability of enabling access to a wider variety of other healthier beverage options, besides water, to encourage healthier choices.