Low- and no-calorie sweetener intakes from beverages – an up-to-date assessment in four regions: Brazil, Canada, Mexico and the United States

Abstract

The current assessment estimated exposure to four low- and no-calorie sweeteners (LNCS) (aspartame, acesulfame potassium (AceK), steviol glycosides and sucralose) from beverages in Brazil, Canada, Mexico and the United States, using up-to-date nationally representative consumption data and industry reported-use level information. Two modelling scenarios were applied – the probabilistic model was guided by reported use level data, with estimated intake for an individual leveraging market-weighted average use level of a particular LNCS in any given LNCS-sweetened beverage type, while the distributional (brand-loyal) model assumed consumer behaviour-led patterns, namely that an individual will be brand loyal to a pre-determined beverage type. Consumer-only and general population intake estimates were derived for the overall population and individual age categories, and compared to the respective acceptable daily intake (ADI) as established by the Joint FAO/WHO Expert Committee on Food Additives (JECFA) for each LNCS. The mean, 90th percentile and 95th percentile intake estimates were substantially lower than the ADI in both modelling scenarios, regardless of the population group or market. In the probabilistic model, the highest consumer-only intake was observed for AceK in Brazilian adolescents (95th percentile, 12.4% of the ADI), while the highest 95th percentile intakes in the distributional model were observed for sucralose in Canadian adults at 20.9% of the ADI. This study provides the latest insights into current intakes of LNCS from water-based non-alcoholic LNCS-sweetened beverages in these regions, aligning well with those published elsewhere.

Keywords:

• Low- and no-calorie sweeteners (LNCS)
• food additive
• exposure
• probabilistic modelling
• distributional modelling
• acesulfame potassium (AceK)
• aspartame
• steviol glycosides
• sucralose

Introduction

A 2018 review of global intakes of low- and no-calorie sweeteners (LNCS) concluded that the available data raised no concerns with respect to the exceedance of the acceptable daily intake (ADIs) among the general population globally (Martyn et al. 2018). Regular monitoring of intakes for any dietary substance is critical to track patterns and trends in consumption over time. This is particularly pertinent for LNCS which serve the role of sugar substitutes (or non-sugar sweeteners), as competent authorities update their respective population-based guidelines to advising limiting added and/or free sugar intakes (WHO 2015; Health Canada 2019; CDC 2021; EFSA 2022; Martyn et al. 2022). In support of national and global sugar reduction goals, the beverage industry has embarked on commitments to reduce sugar in consumers’ diets (from beverages) by offering reduced, low- and zero-sugar versions of products in addition to smaller packages for the sugar version, and making sure these options are widely available and marketed effectively (Conference Board of Canada 2020; ABA 2022; KPMG 2022; UNESDA 2022).

Sweetness preference remains a complex topic, highly variable depending on age, region, culture, among others (Venditti et al. 2020). As such, examining the latest trends in exposure allows an assessment over time as to the impact (if any) of government- and industry-led changes to the diet of the overall population. Two core elements to preparing a robust exposure assessment for dietary substances in foods are up-to-date consumption data, and current use level patterns for the dietary substance (FAO/WHO 2020). Other factors such as brand loyalty, selected food groups, and consumer cohorts may vary depending on the purpose of the assessment, the ingredient investigated and its use in the food supply. Exposure assessors seek to obtain representative estimates of intake to ensure it is fit-for-purpose in the context of risk assessment.

Several publications have investigated LNCS intakes in various regions (Barraj, Scrafford, et al. 2021; Tran et al. 2021). In brief, Tran et al. (2021) applied a conservative tiered exposure assessment approach to estimate hypothetical maximum LNCS intakes from beverages globally and the European Union (EU), leveraging food consumption data from the United States (U.S.) 2013–2016 National Health and Nutrition Examination Survey (NHANES) and the United Kingdom (U.K) 2008–2017 National Diet and Nutrition Survey (NDNS) as representative markets, respectively. In the refined exposure assessment models, all LNCS intake estimates were below the ADI. Separately, Barraj, Scrafford, et al. (2021) determined the exposure to LNCS in Brazil using nationally representative data collected in 2008–2009 from foods and beverages containing LNCS, by applying a conservative brand loyal scenario as well as a scenario to consider estimated intakes for the general population. In both models, all LNCS intake estimates were below the ADI. These findings also align with Martyn et al. (2022), wherein LNCS intakes were estimated in Brazil by applying an ‘added sugar substitutional modelling’ approach (based on sucrose sweetness equivalence) to nationally representative food consumption data (2008–2009) for two scenarios considering replacement of total added sugar content, and the replacement of 50% of the added sugar content of foods and beverages. Both models predicted that the ADI would not be exceeded for AceK, aspartame, saccharin, steviol glycosides and sucralose, although high-level estimates for cyclamate reached or exceeded the ADI depending on the model applied (Martyn et al. 2022).

The current analysis further complements these publications by focusing on markets with typically higher beverage consumption, using the most recently released nationally representative food consumption data for each market. The aim of the current analysis was to estimate the exposure to four LNCS (AceK, aspartame, steviol glycosides, and sucralose) from water-based flavoured non-alcoholic beverages using up-to-date national survey consumption data (2015–2019) and representative beverage consumption habits (e.g. considering ‘brand loyal’ consumer behaviours), combined with industry reported use levels (2019) among the higher consuming beverage markets – North America (U.S. and Canada) and Latin America (LATAM) (Mexico and Brazil) (Popkin and Hawkes 2016).

Materials and methods

Food consumption data

Brazil

Food and beverage consumption data used in the assessment were based on the National Food Survey (Inquérito Nacional de Alimentação [INA]) component of the Pesquisa de Orcamenos Familiares (POF; Household Budget Survey 2017–2018) conducted by The Brazilian Institute of Geography and Statistics (Instituto Brasileiro de Geografia e Estatística) (IBGE 2018). A subsample of households (25%; n = 20,112) were randomly selected to provide food consumption data. Information on food consumption was collected from individuals over 10 years of age (n = 46,164) via 24-hour (24-h) dietary recalls on 2 non-consecutive days, recording all food and beverages consumed inside and outside of the home. Of note, no consumption data were collected for individuals <10 years of age. Further details on the data collection are available from IBGE (2020a). The consumption data were collated into approximately 1593 food and drink codes from the 2017–2018 POF database, including methods of preparation and pre-defined portion sizes (IBGE 2020b). Additional details on the data collection are available in Nogueira Bezerra et al. (2021).

Canada

Statistics Canada collates nationally representative food-consumption data as part of the Canadian Community Health Survey – Nutrition (CCHS) for individuals aged 1 year and older (Statistics Canada 2018, 2022). Data from the CCHS 2015 Public Use Microdata File (PUMF) are applied in the current assessment (Statistics Canada 2022). As anthropometric data were not collected in children aged less than 2 years, this population group was excluded from the current analysis. A 24-h dietary recall was used to assess an individual’s food and beverage intake, and data for individuals who completed two 24-h dietary recalls were included in the current assessment (n = 5601). Survey weights were applied by Statistics Canada to ensure all analyses were nationally representative (Health Canada 2017). Additional details on the data collection are available from Health Canada (2017).

Mexico

Food-consumption data from the Instituto Nacional de Salud Pública (INSP) Encuesta Nacional de Salud y Nutrición 2018–2019 (ENSANUT), were utilised for the current assessment. This was a nationally representative household survey, with a multistage stratified sample design, representative of urban–rural strata and all states. Dietary information was obtained for a subsample of the population using two different methods: a 24-h dietary recall and a semi-quantitative food-frequency questionnaire (FFQ). The FFQ was based on a validated questionnaire consisting of a closed list of foods (n = 140) and the data were made publicly available. For each food and beverage item, participants recorded the number of days the item was consumed per week, the number of times per day, number of portions and portion size consumed during the 7 days prior to the interview. The available individual FFQ consumption data, by population group (2–11 years, n = 6201; 12+ years, n = 22,807), for each reported food type, averaged across 7 days, are applied in the current assessment. Further details are available in Romero-Martínez et al. (2019).

United States

Food consumption data were obtained from the U.S. National Center for Health Statistics’ (NCHS) NHANES 2017–2018 survey (CDC 2022a, 2022b; USDA 2022). The NHANES is a continuous programme released in 2-year cycles, and includes nationally representative food consumption data for the U.S. population. The dietary component is comprised of two 24-h dietary recalls issued on non-consecutive days. The first dietary recall is collected via in-person interviews with trained individuals; the second is collected by telephone 3–10 days after the first recall. Detailed information on each food or beverage item consumed (including the description, amount, and nutrient content) is collected for each individual surveyed and collated into food files. Consumption data for individuals who completed both dietary recall days (n = 6700) were included in the current assessment. Further details are available on the Centers for Disease Control and Prevention’s website (CDC 2022a).

LNCS use level data

The LNCS use levels were obtained from a survey conducted between July and December 2019, facilitated through the International Council of Beverages Associations (ICBA) membership. LNCS beverage use levels (for each of the four LNCS) were gleaned from the relevant markets – Brazil, Canada, Mexico, and U.S. Multi-national beverage companies (including, among others, Keurig Dr Pepper, Pepsi Co, The Coca-Cola Company) and national and also regional beverage associations such as the American Beverage Association (ABA), Associacao Brasileira das Industrias de Refrigerantes e de Bebidas nao Alcoolicas (ABIR), Canadian Beverage Association (CBA), among others, participated in the surveys. Further details are provided by Tran et al. (2021).

The Beverage Marketing Corporation (2019) frequently collects beverage market share information in the U.S. and elsewhere. Similar to Tran et al. (2021), LNCS-sweetened water-based flavoured non-alcoholic beverages were stratified according to conventional LNCS-sweetened beverage types, namely: carbonated soft drinks (CSDs), flavoured waters, juice drinks, energy drinks, sports drinks, ready-to-drink (RTD) teas and RTD coffees. Volume data (in millions of gallons) of the top contributing brands for each of these beverage types were gathered for 2018. Ingredient labels (for each brand that contributed to the top 80% of the total market volume for the respective beverage type) were reviewed to confirm LNCS presence when company data were missing. Table 1 presents the percentage of the total volume of LNCS-sweetened products for a given beverage type that was determined to contain the specific LNCS in question.

Table 1. Percent presence, weighted average and highest reported use levels of LNCS (AceK, Aspartame, Steviol Glycosides, Sucralose) in four jurisdictions by Codex GSFA Category for each LNCS-sweetened beverage type, as utilised in exposure assessment models.

	AceK
LNCS beverage types (Codex GSFA Category)^a	Brazil			Canada			Mexico			United States
	Percent presence (%)^b	Avg.^c,d	Max.^d	Percentage of presence (%)^b	Avg.^c,d	Max.^d	Percentage of presence (%)^b	Avg.^c,d	Max.^d	Percentage of presence (%)^b	Avg.^c,d	Max.^d
CSDs (14.1.4)	83	123	160	98	85	141	99	144	275	88	106	184
Fruit juice-based drinks, including concentrates/mixes (14.1.4)	0	0	0	24	25	103	100	92	92	57	57	169
Sports drinks (14.1.4)	0	0	0	100	36	36	100	26	26	97	33	49
Energy drinks (14.1.4)	62	45	80	87	96	227	100	59	80	95	167	227
RTD coffee (14.1.4)	n.a.	n.a.	n.a.	100	84	84	7	6	84	0	0	0
RTD tea (14.1.4)	59	55	154	99	133	154	31	15	51	89	67	144
Flavoured water drinks (14.1.4)	100	81	81	89	46	58	7	4	54	60	33	58
Powdered fruit drinks (14.1.4.3)	100	69	69	100	52	52	100	177	208	100	52	58
Fruit nectar (14.1.3.1)	78	31	40	100	129	160	0	0	0	n.a.	n.a.	n.a.
	Aspartame
CSDs (14.1.4)	82	104	347	99	325	517	98	252	579	96	285	567
Fruit juice-based drinks, including concentrates/mixes (14.1.4)	0	0	0	25	58	237	0	0	0	34	86	526
Sports drinks (14.1.4)	0	0	0	0	0	0	0	0	0	0	0	0
Energy drinks (14.1.4)	0	0	0	14	28	200	28	56	200	25	51	200
RTD coffee (14.1.4)	n.a.	n.a.	n.a.	0	0	0	0	0	0	0	0	0
RTD tea (14.1.4)	21	44	210	1	7	561	13	34	561	62	168	561
Flavoured water drinks (14.1.4)	100	315	315	89	62	70	0	0	0	19	12	70
Powdered fruit drinks (14.1.4.3)	100	417	417	100	254	254	100	322	465	56	142	254
Fruit nectar (14.1.3.1)	0	0	0	0	0	0	0	0	0	n.a.	n.a.	n.a.
	Steviol Glycosides
CSDs (14.1.4)	0	0	0	0	0	0	1	<1	25	0	0	0
Fruit juice-based drinks, including concentrates/mixes (14.1.4)	0	0	0	0	0	0	0	0	0	22	4	17
Sports drinks (14.1.4)	0	0	0	0	0	0	0	0	0	0	0	0
Energy drinks (14.1.4)	0	0	0	0	0	0	0	0	0	0	0	0
RTD coffee (14.1.4)	n.a.	n.a.	n.a.	0	0	0	0	0	0	0	0	0
RTD tea (14.1.4)	0	0	0	4	<1	10	89	29	57	32	9	33
Flavoured water drinks (14.1.4)	0	0	0	100	77	82	0	0	0	67	59	115
Powdered fruit drinks (14.1.4.3)	0	0	0	0	0	0	2	1	28	0	0	0
Fruit nectar (14.1.3.1)	0	0	0	0	0	0	0	0	0	n.a.	n.a.	n.a.
	Sucralose
CSDs (14.1.4)	1	2	179	3	5	242	6	8	209	64	93	219
Fruit juice-based drinks, including concentrates/mixes (14.1.4)	68	273	400	28	42	146	100	188	188	71	102	176
Sports drinks (14.1.4)	100	250	250	93	175	188	100	88	88	98	64	88
Energy drinks (14.1.4)	100	141	207	83	168	250	96	147	230	66	161	279
RTD coffee (14.1.4)	n.a.	n.a.	n.a.	100	97	97	100	81	97	0	0	0
RTD tea (14.1.4)	100	103	150	99	105	128	50	38	200	91	122	208
Flavoured water drinks (14.1.4)	0	0	0	89	86	96	100	152	152	70	65	105
Powdered fruit drinks (14.1.4.3)	0	0	0	81	119	147	2	2	76	33	69	216
Fruit nectar (14.1.3.1)	67	81	120	100	101	144	94	160	170	n.a.	n.a.	n.a.

AceK: Acesulfame potassium; Avg: average; CSDs: carbonated soft drinks; GSFA: General Standard for Food Additives; LNCS: low- and no- calorie sweeteners; Max.: maximum; RTD: ready-to-drink; n.a.: not available.

^aBeverage types as defined by industry.

^bPercentage market volume penetration by a specific LNCS within a given LNCS-sweetened beverage type based on presence (either according to labels or manufacturers’ confirmation of use). These values have been weighted according to the market volume share of the available brands in the specific jurisdiction.

^cMarket volume-weighted average use level of LNCS in LNCS-sweetened beverage type.

^dmg kg⁻¹.

Average use levels for each LNCS in a given beverage type (for a given market) were calculated based on the market volume distribution across beverage brands within that beverage type (Beverage Marketing Corporation 2019). The distribution across the top 80% of the market volume for a given beverage type in a given market was assumed to be the same for the bottom 20% for which data were lacking. These market average values were incorporated into the brand-loyal model and applied to the non-brand-loyal beverage types (see below). In cases where market share data were not available for 2018, data from 2017 (where available) were used.

Any beverage included in the market survey that did not have a corresponding LNCS use level assigned to it was excluded, except in cases where use levels were noted for the same brand in any of the other three markets evaluated. It was assumed that the use levels for steviol glycoside were relevant for the whole compound, as the steviol conversion factors range from 1.51-4.51 (average ∼3) depending on the molecular weight of the bonded glycoside. Therefore, in the current assessment to estimate steviol equivalent, the use levels provided for steviol glycosides were divided by 3 (U.S. Food and Drug Administration (U.S. FDA) 2018; Tran et al. 2021). The market-weighted average use levels determined in the current assessment (see Table 1) were comparable to the average use levels reported by Tran et al. (2021) confirming concordance between the calculated typical use levels.

Exposure assessment methodology

Food code selection and categorisation

The industry-defined categorisation of LNCS-sweetened water-based flavoured non-alcoholic beverages aligns with the Codex General Standard for Food Additives (GSFA) Food Category 14.1.4. and was followed for all four markets (CAC 2019). Regular calorically sweetened CSDs were typically excluded as these would not typically use LNCS (exceptions noted elsewhere). The relevant food codes were identified from each food consumption database and categorised according to the industry-reported beverage types (see Table 1).

Due to limited food codes for LNCS-sweetened beverages in the Brazilian POF and Mexican ENSANUT databases, the calorically sweetened beverage food codes were included as surrogates for their LNCS-sweetened counterparts to adequately assess a plausible pattern of consumption (Table S1). For example, regular calorically sweetened CSDs were included alongside LNCS-sweetened soft drinks to account for the low proportion of LNCS-sweetened soft drink consumers (Brazil at 1.0%, Mexico at 5.3%). Although regular calorically sweetened CSDs do not contain LNCS, the overall pattern and volume of consumption were assumed for LNCS-sweetened CSDs consumers, including these food codes to increase the sample size for these datasets. The same approach was applied to other relevant beverage types in both these markets.

In some instances, as food codes pertaining to certain beverage types were not available in some surveys (depicted as ‘n.a.’ in Table 2), anticipated contribution to LNCS consumption from these beverage types could not be evaluated. For powdered juice drinks, product-specific reconstitution factors were applied to calculate the final concentration.

Table 2. Percent consumers^a of individual LNCS-sweetened water-based flavoured drinks by population group in four regions.

	n	Percent consumers (%)^b
		LNCS- CSDs	LNCS- flavoured water drinks	LNCS-fruit juice-based drinks	LNCS- powdered fruit drinks	LNCS- nectars	LNCS- energy drinks	LNCS- sports drinks	LNCS- RTD tea	LNCS- RTD coffee	Any LNCS- sweetened water-based flavoured drink
CCHS-Nutrition (2015)
Toddlers (2–3 years)	137	0.1	0.6	0.0	0.0	0.0	0.0	0.0	0.0	n.a.	0.7
Older children (3–9 years)	762	3.1	0.8	0.5	0.1	0.0	0.0	0.1	0.0	n.a.	4.3
Adolescents (10–17 years)	1,426	4.6	1.2	0.9	1.3	0.0	0.0	0.0	0.0	n.a.	7.8
Adults (18–64 years)	2,084	10.6	1.1	0.6	0.5	0.0	0.5	0.1	0.0	n.a.	12.4
Elderly adults (65–74 years)	635	12.0	0.6	2.0	0.2	0.0	0.0	0.1	0.0	n.a.	14.4
Very elderly (75+ years)	557	5.5	0.5	0.4	0.0	0.0	0.0	0.0	0.0	n.a.	6.4
Total population (2+ years)	5,601	8.5	0.9	0.8	0.5	0.0	0.3	0.1	0.0	n.a.	10.4
Brazil, POF 2017–2018^c,d
Adolescents (10–17 years)	6,793	33.2	n.a.	5.2	2.9	0.3	0.1	n.a.	<0.1	n.a.	36.8
Adults (18–64 years)	33,666	25.4	n.a.	4.8	1.9	0.2	0.2	n.a.	<0.1	n.a.	29.3
Elderly adults (65–74 years)	3,451	15.5	n.a.	3.7	1.3	0.3	0.0	n.a.	0.0	n.a.	18.9
Very elderly (75+ years)	2,254	12.1	n.a.	4.5	0.9	0.5	0.0	n.a.	0.0	n.a.	16.8
Total population (10+ years)	46,164	25.2	n.a.	4.8	1.9	0.2	0.1	n.a.	<0.1	n.a.	29.0
Mexican ENSANUT (2018)^d
Children (2–11 years)	6,201	72.5	2.7	2.1	n.a.	31.2	n.a.	n.a.	18.8	n.a.	84.1
Adolescents (12–17 years)	4,351	79.7	2.8	2.7	n.a.	27.5	n.a.	n.a.	17.0	n.a.	87.3
Adults (18–64 years)	15,511	76.2	3.1	3.5	n.a.	14.1	n.a.	n.a.	25.4	n.a.	86.3
Elderly adults (65–74 years)	1,736	59.2	2.2	4.8	n.a.	8.0	n.a.	n.a.	32.2	n.a.	76.9
Very elderly (75+ years)	1,209	54.4	2.2	4.7	n.a.	10.3	n.a.	n.a.	32.6	n.a.	76.4
Total population (12+ years)	22,807	74.4	2.9	3.5	n.a.	16.1	n.a.	n.a.	24.7	n.a.	85.2
U.S. NHANES (2017–18)
Infants and toddlers (0–3 years)	599	0.1	1.8	1.8	1.2	n.a	0.0	0.2	0.4	0.0	5.0
Older children (3–9 years)	752	1.6	1.3	6.2	4.4	n.a	0.0	0.4	0.7	0.1	14.3
Adolescents (10–17 years)	933	4.1	0.7	1.3	2.7	n.a	<0.1	1.3	1.0	0.1	10.4
Adults (18–64 years)	3,253	9.1	2.0	1.6	1.8	n.a	1.2	1.3	2.2	0.4	17.4
Elderly adults (65–74 years)	625	18.2	1.5	3.6	4.0	n.a	0.1	0.5	5.7	1.2	28.0
Very elderly (75+ years)	421	17.9	1.8	1.7	3.2	n.a	0.0	0.9	4.7	0.3	26.7
Total population (0+ years)	6,583	9.0	1.7	2.2	2.4	n.a	0.7	1.1	2.3	0.4	17.5

LNCS: Low- and no-calorie sweeteners; CSDs: carbonated soft drinks; n: number of subjects (unweighted); RTD: ready-to-drink; n.a.: no food code available

^aConsumers were identified as individuals who reported consumption of any of the LNCS-sweetened beverage types during the dietary survey reporting period. For Brazil and Mexico, consumers of regular calorically sweetened beverage types were assumed to be surrogates for their LNCS-sweetened counterparts due to insufficient reporting for the latter.

^bWeighted values.

^cConsumption data only available for individuals over 10 years.

^dFor Brazil and Mexico, regular CSDs were used as surrogates for LNCS-CSDs, to reflect the anticipated pattern of consumption.

Exposure assessment models

For all four markets, both the probabilistic and distributional exposure assessment modelling scenarios were applied, as described below, to determine intake estimates across subpopulations.

Probabilistic model (non-brand loyal)

Food consumption data in the four dietary intake surveys were only available at the food code level, not on an individual brand level. To account for all available use level data, a distribution curve based on the market share volume of individual brands was applied. Therefore, the chance that each reported LNCS use level was encountered by a consumer on a given eating occasion was related to the market volume share of the corresponding brand within the relevant beverage type.

Distributional model (brand loyal)

In this scenario, ‘brand loyal’ consumers are assumed to habitually consume the top-contributing beverage type that contains the LNCS of interest at the highest use level. These consumers are also assumed to consume other beverage types at market-weighted average use levels. The maximum use level, assuming a 100% presence probability, was applied to the ‘brand loyal’ beverage type for each LNCS (e.g. LNCS-sweetened CSDs were the brand loyal beverage type for sucralose). Then, the market-weighted average use levels were applied to all other beverage types. Table 1 captures the market average and highest reported use level for each beverage type as incorporated into this model.

To identify the ‘brand loyal’ beverage type, a preliminary exposure assessment was completed in each market to determine the top contributing beverage type to total intakes for each LNCS after applying the maximum use level to each beverage type (data not shown). LNCS-sweetened CSDs were the top contributor to AceK, aspartame, and sucralose intakes in all markets and to steviol glycosides in Mexico, therefore were selected as the ‘brand loyal’ beverage type for these LNCS. Due to limited survey data for steviol glycosides, the ‘brand loyal’ beverage type varied by market, with LNCS-sweetened flavoured water drinks being the top contributor in the U.S. and Canada, with no data provided for Brazil. In the ‘brand loyal’ model, the maximum use level was applied to the food codes representing the brand loyal beverage type, and the market-weighted average use level was applied to all other beverage types (non-brand loyal).

Data analysis

Statistical analyses and data management were conducted with DaDiet Software (Dazult Ltd. 2018). Individual-level food consumption data were available in each of the four markets and were utilised in both exposure assessment models. The percentage of individual LNCS consumers in the total population (i.e. the proportion of the population that consumed a specific LNCS-sweetened beverage type in the context of any LNCS-sweetened beverage for that given beverage type over the dietary survey period) was determined for each market and for each subpopulation group based on the age brackets defined by the Joint FAO/WHO Expert Committee on Food Additives (JECFA) (see Table 2); toddlers and young children (1–7 years), other children, including adolescents (8–17 years), adults (18–64 years), elderly (65–74 years), very elderly (75+ years) and the total population.

The probabilistic models were conducted iteratively using a Monte Carlo simulation. This approach uses a specific number of iterations to represent the collection of all possible combinations of individual consumption patterns and concentrations – specifically combining the full range of individual beverage consumption amounts, in each survey, with the full range of use levels available (weighted according to the market volume share of the brand) (Boon and Van Klaveren 2003; Boon et al. 2004 as reviewed in EFSA 2012). In brief, every single consumption event from the dietary survey was simulated to contain one of the reported use levels; and repeated to reach a target number of iterations to reduce uncertainty (Brazil, n = 85,018 persons; Canada, n = 11,202 persons; Mexico, n = 29,440 persons; U.S. n = 12,277 persons) from which mean, 90th and 95th percentile estimates were determined across either the general population or the consumer-only population.

The brand loyal models also applied individual-level dietary data, with the specific beverage consumption data multiplied by the appropriate use level data. Individuals consuming the ‘brand-loyal’ beverage type (as mentioned above) had their consumption multiplied by the maximum use levels, while the remaining beverage types consumed were multiplied by the average (market volume-weighted) use levels.

The mean, 90th and 95th percentile estimates were then determined across either the general population or the consumer-only population. The general population exposure refers to the estimated intake of the respective LNCS over all individuals in the dietary survey, regardless of whether they consumed LNCS-containing beverage types or not. For the probabilistic model, this included any individual who was simulated to consume beverages containing the selected LNCS during the iterations, and consequently beverages that were not simulated to contain the selected LNCS (depending on the market) were removed. The consumer-only exposure represents only the individuals who consumed any LNCS-sweetened beverage during the dietary survey period. For the brand loyal model, this included consumers of any LNCS-sweetened beverage where uses were reported – as explained above, 100% use of LNCS was assumed for given LNCS-sweetened beverage types for which the LNCS use was reported (see Table 1). For both models, estimated daily intakes were determined on a body weight (bw) basis (mg kg⁻¹ bw day⁻¹) and as a fraction of the relevant JECFA ADI (% ADI) (JECFA 2019). As the consumer-only analyses provide conservative estimates for higher-level LNCS exposure, these are presented herein, with the general population estimates accessible in the Supplementary Data Files.

Results

LNCS use in water-based flavoured non-alcoholic beverages

The percent of beverages containing AceK, aspartame, steviol glycosides, and sucralose within each type of LNCS-sweetened water-based flavoured drink is summarised by market (see Table 1).

For all markets evaluated, greater than 50% of products in each beverage type were reported to contain AceK, except for fruit juice-based drinks, RTD coffees, RTD teas, and fruit nectars. The LNCS-sweetened beverage types with the highest proportion of products containing AceK were powdered fruit drinks (100%), followed by sports drinks (97–100%; no use reported for Brazil), CSDs (83–99%), fruit nectar (78–100%; no use reported for the U.S. and Mexico), energy drinks (62–100%), fruit-juice based drinks (24–100%; no use reported for Brazil), RTD tea (31–99%), and flavoured water drinks (7–100%). AceK was reportedly used in large a proportion of RTD coffee in Canada (100%), with a lower proportion in Mexico (7%), no use reported for the U.S. and no data reported for Brazil.

The LNCS-sweetened beverage type with the highest proportion of products containing aspartame were CSDs (82–99%), followed by powdered fruit drinks (56–100%) and flavoured water drinks (19–100%). Canada and the U.S. reported aspartame use in fruit juice-based drinks, at 25% and 34%, respectively. Aspartame used in RTD teas ranged from 1% to 62%. Due to a gap in use level data for Mexico and Canada, data for similar products being sold in the U.S. were replicated for these regions. Aspartame was reportedly used in 14% of energy drinks in Canada, and it was assumed a similar proportion of use in the U.S. and Mexico due to brand similarities (having confirmed presence in ingredient statements on randomly selected labels). For energy drinks, the available use level data from Canada was applied to the energy drinks assessment for the U.S. (25%) and Mexico (28%). No use levels were reported for aspartame in sports drinks (confirmed by the absence of aspartame on the ingredient label), these were excluded from the respective assessments. No use of aspartame in RTD coffee or fruit nectar was reported.

Steviol glycosides were not reported to be used in any beverage type in Brazil. Flavoured water drinks were the beverage type with the highest proportion of products containing steviol glycosides in the U.S. and Canada (67–100%; no use reported in Mexico), followed by RTD tea (4–89%). It was reported to be used in 22% of fruit juice-based drink products in the U.S., with no use reported in Canada or Mexico. LNCS-sweetened RTD coffee was limited to a single brand, therefore 100% presence was reported. The reported use was 0% or minimal (<2%) in CSDs, sports drinks, energy drinks, powdered fruit drinks, and fruit nectars.

Sucralose was reportedly used in most beverage types. Sports drinks had the highest proportion of beverage types containing sucralose across all 4 markets investigated (93–100%), followed by flavoured water drinks (70–100%; no use reported in Brazil), RTD tea (50–100%), fruit juice-based drinks (28–100%), energy drinks (66–100%) and CSDs (1–64%). Sucralose was reportedly used in high proportions in powdered fruit drinks in Canada (81%), with lower proportions reported for the U.S. (33%) and Mexico (2%), with no use reported for Brazil. Data on the use of LNCS-sweetened RTD coffee was limited to a single brand in Canada and Mexico, therefore, 100% presence was reported in these regions, with no use reported in the U.S. and no data available for Brazil.

Percentage consumers of any LNCS-sweetened water-based flavoured non-alcoholic beverages

The percentage of consumers, in the total population, for each beverage type is presented in Table 2 by subpopulation group and market. The highest proportion of consumers was identified for LNCS-sweetened CSDs in Mexico and Brazil at 54.4–79.7% and 12.1–33.2%, respectively, compared to Canada and the U.S. at 0.1–12.0% and 0.1–18.2%, respectively. It should be noted that for both Mexico and Brazil, regular calorically sweetened beverages were utilised as surrogates for their LNCS-sweetened counterparts, which may explain the high percentage of consumers in these markets. All other LNCS-sweetened beverage types were reported to be consumed by less than 10% of the population for any market, except RTD teas (17.0–32.6%) and nectars (8.0–31.2%) in the Mexican population.

Dietary exposure to LNCS

Probabilistic model (non-brand loyal)

Given that different survey methodologies were employed in each market, findings across markets were not directly comparable (see Table 3). Overall patterns can be noted, however. The intake estimates for the general population are provided in Table S2. As noted above, to ensure a conservative intake estimate, surrogates were included in the Brazilian and Mexican assessments. For completeness, the results of the analysis for these regions in which surrogates were not used are shown in Table S3. Percent consumers refer to the proportion of a particular subgroup who were identified as consumers of the LNCS-sweetened beverages reported to contain the sweetener of interest.

Table 3. Estimated daily exposure to LNCS (AceK, Aspartame, Steviol Glycosides, Sucralose) from water-based flavoured non-alcoholic beverages by subpopulation group in LNCS-sweetened beverage consumers^a only - probabilistic model.

Population Group	n	AceK							Aspartame							Steviol glycosides							Sucralose
		% cons.	mg kg^−1 bw day^−1			% JECFA ADI (15 mg kg^−1 bw day^−1)			% cons.	mg kg^−1 bw day^−1			% JECFA ADI (40 mg kg^−1 bw day^−1)			% cons.	mg kg^−1 bw day^−1			% JECFA ADI (4 mg kg^−1 bw day^−1)			% cons.	mg kg^−1 bw day^−1			% JECFA ADI (15 mg kg^−1 bw day^−1)
			Mean (SD)	P90	P95	Mean	P90	P95		Mean (SD)	P90	P95	Mean	P90	P95		Mean (SD)	P90	P95	Mean	P90	P95		Mean (SD)	P90	P95	Mean	P90	P95
Brazil, POF 2017–18^b,c
Adolescents (10–17 years)	6,793	35.3%	0.75 (0.62)	1.50	1.86	4.99%	9.99%	12.37%	35.0%	0.78 (0.81)	1.47	2.17	1.94%	3.67%	5.43%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	36.8%	0.18 (0.49)	0.67	1.27	1.20%	4.45%	8.49%
Adults (18–64 years)	33,666	27.0%	0.52 (0.43)	1.03	1.31	3.43%	6.87%	8.73%	26.8%	0.53 (0.55)	1.04	1.49	1.31%	2.61%	3.72%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	29.3%	0.15 (0.39)	0.58	0.97	1.01%	3.89%	6.45%
Elderly (65–74 years)	3,451	16.6%	0.39 (0.28)	0.72	1.02	2.60%	4.82%	6.80%	16.5%	0.43 (0.52)	0.80	1.10	1.08%	1.99%	2.76%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	18.9%	0.17 (0.40)	0.57	1.03	1.10%	3.83%	6.89%
Very elderly (75+ years)	2,254	13.3%	0.42 (0.55)	0.77	1.14	2.81%	5.10%	7.61%	12.8%	0.44 (0.55)	0.94	1.54	1.10%	2.35%	3.85%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	16.8%	0.19 (0.34)	0.62	0.93	1.25%	4.13%	6.23%
Total population (10+ years)	46,164	26.7%	0.55 (0.48)	1.10	1.46	3.67%	7.33%	9.70%	26.5%	0.57 (0.61)	1.16	1.60	1.42%	2.91%	4.01%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	29.0%	0.16 (0.41)	0.60	1.02	1.06%	3.98%	6.81%
Canada, CCHS 2015
Toddlers (24–35 months)	137	0.7%	0.17^e (0.18)	0.26^f	0.43^g	1.11%^e	1.76%^f	2.85%^g	0.7%	0.45^e (0.77)	0.87^f	1.57^g	1.12%^e	2.17%^f	3.91%^g	0.6%	0.15^e (n.a)	0.15^f	0.15^g	3.70%^e	3.70%^f	3.70%^g	0.7%	0.14^e (0.05)	0.15^f	0.16^g	0.96%^e	0.99%^f	1.05%^g
Young children (3–9 years)	762	4.6%	0.29^e (0.20)	0.59^f	0.66^g	1.95%^e	3.95%^f	4.37%^g	4.4%	0.83^e (0.62)	1.43^f	2.20^g	2.07%^e	3.56%^f	5.49%^g	0.8%	0.74^e (0.09)	0.75^f	0.81^g	18.51%^e	18.87%^f	20.15%^g	4.6%	0.25^e (0.37)	0.80^f	0.88^g	1.64%^e	5.33%^f	5.86%^g
Adolescents (10–17 years)	1,426	9.0%	0.29 (0.20)	0.59	0.70^g	1.96%	3.94%	4.68%^g	9.0%	1.04 (0.86)	2.50	2.71^g	2.60%	6.25%	6.76%^g	1.2%	0.45^e (0.27)	0.73^f	0.82^g	11.35%^e	18.32%^f	20.43%^g	9.0%	0.20 (0.27)	0.48	0.82^g	1.33%	3.19%	5.45%^g
Adults (18–64 years)	2,084	12.8%	0.37 (0.32)	0.75	1.10	2.43%	4.99%	7.33%	12.6%	1.32 (1.27)	2.80	4.21	3.29%	7.00%	10.53%	1.1%	0.34 (0.28)	0.53^f	0.75^g	8.43%	13.13%^f	18.82%^g	12.8%	0.10 (0.20)	0.31	0.54	0.66%	2.08%	3.60%
Elderly (65–74 years)	635	16.8%	0.25 (0.22)	0.48	0.72^g	1.67%	3.23%	4.83%^g	16.6%	0.94 (0.87)	1.86	2.77^g	2.35%	4.64%	6.94%^g	0.6%	0.19^e (0.15)	0.24^f	0.38^g	4.87%^e	6.04%^f	9.50%^g	16.8%	0.06 (0.09)	0.17	0.17^g	0.39%	1.14%	1.14%^g
Very elderly (75+ years)	557	8.4%	0.37 (0.36)	1.24^f	1.24^g	2.49%	8.26%^f	8.26%^g	8.4%	1.42 (1.41)	4.75^f	4.75^g	3.54%	11.87%^f	11.87%^g	0.5%	0.17^e (0.10)	0.26^f	0.30^g	4.31%^e	6.48%^f	7.44%^g	8.4%	0.08 (0.09)	0.23^f	0.23^g	0.52%	1.52%^f	1.52%^g
Total population (2+ years)	5,601	11.3%	0.33 (0.30)	0.65	0.87	2.22%	4.34%	5.79%	11.2%	1.20 (1.17)	2.50^f	3.33^g	3.00%	6.25%^f	8.32%^g	0.9%	0.37 (0.29)	0.76^f	0.77^g	9.29%	18.96%^f	19.27%^g	11.3%	0.11 (0.20)	0.38	0.49	0.72%	2.52%	3.27%
Mexico, ENSANUT 2018^d
Children (2–11 years)	6,201	79.1%	0.35 (0.44)	0.78	1.15	2.35%	5.21%	7.66%	76.1%	0.61 (0.76)	1.36	1.97	1.53%	3.39%	4.93%	78.1%	0.02 (0.05)	0.06	0.10	0.42%	1.41%	2.43%	84.5%	0.20 (0.38)	0.53	0.76	1.32%	3.50%	5.05%
Adolescents (12–17 years)	4,351	84.5%	0.24 (0.27)	0.57	0.81	1.61%	3.78%	5.38%	83.8%	0.42 (0.47)	0.98	1.40	1.05%	2.45%	3.50%	83.7%	0.01 (0.02)	0.02	0.04	0.18%	0.53%	1.05%	87.3%	0.08 (0.14)	0.18	0.28	0.51%	1.21%	1.89%
Adults (18–64 years)	15,511	84.2%	0.19 (0.21)	0.45	0.62	1.25%	2.99%	4.11%	83.7%	0.33 (0.37)	0.78	1.08	0.82%	1.95%	2.69%	83.1%	0.01 (0.02)	0.03	0.05	0.25%	0.87%	1.32%	86.3%	0.05 (0.12)	0.13	0.20	0.37%	0.84%	1.35%
Elderly (65–74 years)	1,736	75.3%	0.15 (0.17)	0.36	0.51	1.01%	2.40%	3.39%	73.3%	0.26 (0.31)	0.60	0.89	0.66%	1.5%	2.23%	73.0%	0.02 (0.03)	0.05	0.07	0.43%	1.33%	1.74%	76.8%	0.06 (0.1)	0.15	0.27	0.40%	1.01%	1.77%
Very elderly (75+ years)	1,209	73.6%	0.17 (0.23)	0.40	0.64	1.16%	2.69%	4.27%	70.9%	0.31 (0.41)	0.70	1.12	0.77%	1.75%	2.80%	70.8%	0.02 (0.04)	0.07	0.10	0.52%	1.68%	2.45%	76.3%	0.08 (0.13)	0.18	0.28	0.50%	1.23%	1.90%
Total population (12+ years)	22,807	83.0%	0.20 (0.22)	0.46	0.65	1.30%	3.07%	4.32%	82.2%	0.34 (0.39)	0.80	1.13	0.85%	2.00%	2.81%	81.8%	0.01 (0.02)	0.04	0.06	0.26%	0.91%	1.38%	85.2%	0.06 (0.12)	0.14	0.23	0.40%	0.95%	1.53%
United States, NHANES 2017–18
Infants and Toddlers  (0–35 months)	599	5.1%	0.55^e (0.43)	1.25^f	1.36^g	3.65%^e	8.31%^f	9.09%^g	4.9%	0.97^e (1.15)	3.09^f	3.32^g	2.43%^e	7.72%^f	8.29%^g	4.1%	0.36^e (0.52)	0.77^f	0.89^g	9.12%^e	19.32%^f	22.29%^g	5.1%	0.92^e (0.67)	2.03^f	2.14^g	6.11%^e	13.51%^f	14.28%^g
Young children (3–9 years)	752	14.2%	0.36 (0.34)	0.64	0.94^g	2.41%	4.25%	6.28%^g	13.9%	0.78 (0.93)	1.71	2.52^g	1.95%	4.27%	6.29%^g	8.0%	0.05^e (0.08)	0.22^f	0.22^g	1.37%^e	5.62%^f	5.62%^g	14.2%	0.53 (0.43)	0.89	1.01^g	3.55%	5.95%	6.70%^g
Adolescents (10–17 years)	933	10.2%	0.28 (0.24)	0.49	0.84^g	1.85%	3.29%	5.60%^g	9.5%	0.73 (0.68)	1.32^f	2.25^g	1.82%	3.30%^f	5.62%^g	2.9%	0.07 (0.11)	0.25^f	0.29^g	1.72%	6.18%^f	7.18%^g	10.2%	0.31 (0.23)	0.66	0.74^g	2.10%	4.40%	4.96%^g
Adults (18–64 years)	3,253	17.0%	0.49 (0.68)	1.08	1.50	3.27%	7.23%	10.00%	15.9%	1.24 (1.78)	2.73	4.02	3.10%	6.83%	10.05%	6.3%	0.11 (0.20)	0.29	0.34	2.67%	7.34%	8.58%	17.0%	0.53 (0.67)	1.07	1.47	3.56%	7.16%	9.82%
Elderly (65–74 years)	625	27.6%	0.54 (0.48)	1.09	1.17^g	3.58%	7.24%	7.78%^g	27.0%	1.43 (1.29)	2.79	3.15^g	3.57%	6.98%	7.87%^g	10.9%	0.05 (0.10)	0.19^f	0.20^g	1.34%	4.80%^f	5.11%^g	27.6%	0.59 (0.51)	1.24	1.35^g	3.96%	8.25%	9.00%^g
Very elderly (75+ years)	421	26.1%	0.40 (0.32)	0.79	1.06^g	2.66%	5.26%	7.06%^g	25.2%	1.06 (0.87)	2.12	2.83^g	2.65%	5.29%	7.08%^g	9.2%	0.09 (0.12)	0.25^f	0.32^g	2.34%	6.37%^f	7.92%^g	26.1%	0.45 (0.39)	0.97	1.17^g	3.01%	6.47%	7.80%^g
Total population (0+ years)	6,583	17.2%	0.47 (0.59)	1.08	1.33	3.12%	7.18%	8.85%	16.3%	1.19 (1.55)	2.68	3.56	2.97%	6.69%	8.91%	6.5%	0.10 (0.19)	0.24	0.34	2.39%	5.92%	8.58%	17.2%	0.53 (0.60)	1.09	1.35	3.51%	7.27%	9.00%

LNCS: low- and no-calorie sweeteners; AceK: acesulfame potassium; n: number of subjects in total population (unweighted); % cons.: percentage of consumers (weighted) within specified population group; JECFA: Joint (FAO/WHO) Expert Committee on Food Additives (and Contaminants); ADI: acceptable daily intake; bw: body weight; SD: standard deviation; P90: 90th percentile; P95: 95th percentile; n.a.: data not available.

^aConsumers were identified as individuals who reported consumption of any of the GSFA categories or subcategories during the dietary survey reporting period, which were simulated to contain LNCS during the iterations.

^bConsumption data only available for individuals over 10 years.

^cDue to limited food codes, regular calorically sweetened CSDs, fruit juice, energy drinks, RTD teas and nectars were included as surrogates for their LNCS-sweetened counterparts to assess anticipated patterns of consumption of LNCS-sweetened beverages.

^dDue to limited food codes, regular calorically sweetened CSDs, fruit juice and fruit nectars were included as surrogates for their LNCS-sweetened counterparts to assess anticipated patterns of consumption of LNCS-sweetened beverages.

^eIndicates an intake estimate that may not be statistically reliable, as the sample size does not meet the minimum reporting requirements of n = 30 (Korn and Graubard 1999).

^fIndicates an intake estimate that may not be statistically reliable, as the sample size does not meet the minimum reporting requirements of n = 80 (Kroes et al. 2002).

^gIndicates an intake estimate that may not be statistically reliable, as the sample size does not meet the minimum reporting requirements of n = 130 (Conover 1971).

AceK

Estimated intakes for AceK were substantially lower than the ADI of 15 mg kg⁻¹ bw day⁻¹ across all subpopulations for each market. The percent consumers were highest in Mexico (73.6–84.5%), followed by Brazil (13.3–35.3%), the U.S. (5.1–27.6%) and Canada (0.7–16.8%). The mean and 90th percentile intakes across all markets ranged from 1.01–4.99% and 2.40–9.99% of the ADI, respectively, and the highest intake at the 95th percentile was estimated in Brazilian adolescents (ages 10–17 years) at 12.37% of the ADI. All other 95th percentile intakes, regardless of the market or age group evaluated, were less than 10.0% of the ADI.

Aspartame

A similar trend was observed for aspartame, with Mexico representing the highest percentage of consumers (70.9–83.8%), followed by Brazil (12.8–35.0%), the U.S. (4.9–27.0%) and Canada (0.7–16.6%). Relative to the JECFA ADI of 40 mg kg⁻¹ bw day⁻¹ for aspartame, estimated intakes at the mean and 90th percentile across markets ranged between 0.66–3.57% and 1.50–7.00% of the ADI, respectively. At the 95th percentile, the highest intakes were observed in Canadian adults at 10.53% of the ADI, followed by U.S. adults at 10.05% of the ADI. The data were not statistically reliable for any other age group within Canada or U.S. The estimated 95th percentile intakes were similar in Brazil and Mexico, ranging between 2.76–5.43% and 2.23–4.93% of the ADI, respectively.

Sucralose

The greatest percentage of consumers of sucralose-containing beverages was observed in Mexico (76.3–87.3%), followed by Brazil (16.8–36.8%), the U.S. (5.1–27.6%) and Canada (0.7–16.8%). Estimated mean intakes of sucralose (JECFA ADI of 15 mg kg⁻¹ bw day⁻¹) were relatively low across all markets ranging from 0.37–3.96% of ADI. Across all markets, the 90th percentile intakes ranged from 0.84–8.25% of the ADI, with the highest 95th percentile estimates in U.S. adults at 9.82% of the ADI, with the lowest observed in Mexican adults at 1.35% of the ADI.

Steviol glycosides

A high proportion of consumers of LNCS-sweetened beverages containing steviol glycosides was reported in Mexico (70.8–83.7%), with a much lower percentage of consumers reported in the U.S (2.9–10.9%) and Canada (0.5–1.2%). No user-level data were provided for Brazil and the results obtained for Canada were not statistically reliable due to the low sample size. Estimated intakes of steviol glycosides in Mexico and the U.S. were notably lower than the ADI of 4 mg kg⁻¹ bw day⁻¹ (see Table 3). Estimated intakes of steviol glycosides in Mexico ranged were 0.18–0.52% ADI at the mean, 0.53–1.68% ADI at the 90th percentile and 1.05–2.45% ADI at the 95th percentile. In the U.S., 95th percentile intakes could only be estimated in adults (Mean, 2.67% ADI; 90th percentile, 7.34% ADI; 95th percentile, 8.58% ADI) and the total population (Mean, 2.39% ADI; 90th percentile, 5.92% ADI; 95th percentile, 8.58% ADI).

Distributional model (brand loyal)

The intake estimates of each of the four LNCS in consumers of LNCS-sweetened water-based non-alcoholic beverages determined via distributional ‘brand loyal’ modelling are presented in Table 4. Estimated intakes for the general population (i.e. including non-consumers) are provided in Table S4, with the analyses for Brazil and Mexico (not assuming surrogates) shown in Table S5. Overall, estimated intakes for each of the four LNCS were higher in this model as compared to those from the probabilistic model presented above. The percent consumer patterns are similar to those reported for the probabilistic model (i.e. highest in Mexico, followed by Brazil, the U.S. and Canada).

Table 4. Estimated daily exposure to LNCS (AceK, Aspartame, Steviol Glycosides, Sucralose) from water-based flavoured non-alcoholic beverages by subpopulation group in LNCS-containing beverage consumers^a only – brand loyal model.

Population group	n	AceK							Aspartame							Steviol Glycosides							Sucralose
		% cons.	mg kg^−1 bw day^−1			% JECFA ADI (15 mg kg^−1 bw day^−1)			% cons.	mg kg^−1 bw day^−1			% JECFA ADI (40 mg kg^−1 bw day^−1)			% cons.	mg kg^−1 bw day^−1			% JECFA ADI (4 mg kg^−1 bw day^−1)			% cons.	mg kg^−1 bw day^−1			% JECFA ADI (15 mg kg^−1 bw day^−1)
			Mean (SD)	P90	P95	Mean	P90	P95		Mean (SD)	P90	P95	Mean	P90	P95		Mean (SD)	P90	P95	Mean	P90	P95		Mean (SD)	P90	P95	Mean	P90	P95
Brazil, POF 2017–18^b,c
Adolescents (10–17 years)	6,793	35.3%	0.96 (0.80)	1.92	2.40	6.43%	12.80%	16.00%	35.0%	2.21 (1.78)	4.34	5.48	5.52%	10.84%	13.70%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	36.8%	1.18 (0.93)	2.29	2.93	7.86%	15.23%	19.53%
Adults (18–64 years)	33,666	27.0%	0.66 (0.55)	1.33	1.70	4.42%	8.89%	11.31%	26.8%	1.51 (1.23)	2.99	3.88	3.78%	7.48%	9.70%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	29.3%	0.81 (0.65)	1.58	2.07	5.40%	10.53%	13.77%
Elderly (65–74 years)	3,451	16.6%	0.50 (0.37)	0.94	1.23	3.35%	6.24%	8.21%	16.5%	1.17 (0.87)	2.17	3.06	2.93%	5.42%	7.65%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	18.9%	0.64 (0.46)	1.23	1.58	4.26%	8.18%	10.53%
Very elderly (75+ years)	2,254	13.3%	0.54 (0.72)	1.00	1.50	3.62%	6.67%	9.97%	12.8%	1.27 (1.57)	2.31	3.24	3.18%	5.78%	8.11%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	16.8%	0.66 (0.74)	1.22	1.65	4.38%	8.14%	11.02%
Total population (10+ years)	46,164	26.7%	0.71 (0.62)	1.42	1.88	4.73%	9.48%	12.55%	26.5%	1.62 (1.38)	3.20	4.27	4.05%	8.01%	10.68%	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	29.0%	0.86 (0.72)	1.71	2.24	5.76%	11.39%	14.92%
Canada, CCHS 2015
Toddlers (24–35 months)	137	0.7%	0.23^e (0.32)	0.40^f	0.69^g	1.51%^e	2.67%^f	4.60%^g	0.7%	0.65^e (1.26)	1.34^f	2.47^g	1.64%^e	3.34%^f	6.17%^g	0.6%	0.16^e (n.a.)	0.16^f	0.16^g	3.94%^e	3.94%^f	3.94%^g	0.7%	0.40^e (0.55)	0.70^f	1.19^g	2.66%^e	4.65%^f	7.94%^g
Young children (3–9 years)	762	4.3%	0.38^e (0.25)	0.62^f	0.78^g	2.51%^e	4.12%^f	5.22%^g	4.4%	1.14^e (0.98)	2.13^f	2.82^g	2.84%^e	5.33%^f	7.05%^g	0.8%	0.79^e (0.10)	0.80^f	0.86^g	19.69%^e	20.07%^f	21.44%^g	4.6%	0.66^e (0.45)	1.14^f	1.43^g	4.41%^e	7.59%^f	9.54%^g
Adolescents (10–17 years)	1,426	7.8%	0.37 (0.32)	0.76	1.17^g	2.49%	5.08%	7.81%^g	9.0%	1.15 (1.20)	2.38	4.28^g	2.87%	5.96%	10.70%^g	1.2%	0.48^e (0.28)	0.78^f	0.87^g	12.08%^e	19.49%^f	21.74%^g	9.0%	0.63 (0.52)	1.12	2.00^g	4.22%	7.44%	13.36%^g
Adults (18–64 years)	2,084	12.4%	0.57 (0.53)	1.13	1.83	3.82%	7.55%	12.17%	10.6%	1.81 (1.95)	3.95	4.97	4.53%	9.87%	12.43%	1.1%	0.36 (0.30)	0.56^f	0.80^g	8.97%	13.97%^f	20.02%^g	12.8%	0.98 (0.90)	1.93	3.14	6.54%	12.90%	20.93%
Elderly (65–74 years)	635	14.4%	0.41 (0.39)	0.80	1.20^g	2.71%	5.36%	8.01%^g	16.6%	1.28 (1.44)	2.80	3.88^g	3.20%	7.00%	9.71%^g	0.6%	0.21^e (0.16)	0.26^f	0.40^g	5.18%^e	6.43%^f	10.11%^g	16.8%	0.63 (0.65)	1.31	1.82^g	4.20%	8.74%	12.12%^g
Very elderly (75+ years)	557	7.9%	0.67 (0.68)	2.06^f	2.06^g	4.49%	13.71%^f	13.71%^g	8.4%	1.86 (2.43)	7.55^f	7.55^g	4.65%	18.88%^f	18.88%^g	0.5%	0.18^e (0.11)	0.28^f	0.32^g	4.59%^e	6.90%^f	7.91%^g	8.4%	0.92 (1.10)	3.54^f	3.54^g	6.17%	23.59%^f	23.59%^g
Total population (2+ years)	5,601	10.4%	0.52 (0.50)	1.08	1.35	3.45%	7.17%	9.03%	11.2%	1.71 (1.83)	3.88	4.97	4.29%	9.70%	12.43%	0.9%	0.40 (0.30)	0.81^f	0.82^g	9.89%	20.18%^f	20.50%^g	11.3%	0.86 (0.84)	1.82	2.33	5.71%	12.11%	15.52%
Mexico, ENSANUT 2018^d
Children (2–11 years)	6,201	79.1%	0.66 (0.83)	1.48	2.17	4.40%	9.87%	14.47%	76.10%	1.39 (1.74)	3.12	4.53	3.47%	7.79%	11.33%	78.1%	0.08 (0.08)	0.17	0.24	1.89%	4.17%	5.88%	84.4%	0.64 (0.75)	1.38	1.98	4.24%	9.20%	13.23%
Adolescents (12–17 years)	4,351	84.5%	0.45 (0.51)	1.06	1.51	3.03%	7.08%	10.06%	83.8%	0.96 (1.07)	2.22	3.22	2.39%	5.55%	8.04%	83.7%	0.05 (0.05)	0.11	0.15	1.20%	2.76%	3.73%	87.3%	0.39 (0.41)	0.90	1.22	2.60%	5.97%	8.16%
Adults (18–64 years)	15,511	84.2%	0.35 (0.40)	0.85	1.17	2.33%	5.65%	7.80%	83.7%	0.74 (0.84)	1.79	2.47	1.84%	4.47%	6.16%	83.1%	0.04 (0.04)	0.09	0.12	1.03%	2.29%	3.08%	86.3%	0.30 (0.32)	0.69	0.94	1.98%	4.57%	6.25%
Elderly (65–74 years)	1,736	75.3%	0.28 (0.33)	0.64	0.97	1.84%	4.26%	6.46%	73.3%	0.58 (0.71)	1.35	2.05	1.46%	3.37%	5.13%	73.0%	0.04 (0.04)	0.09	0.12	1.04%	2.26%	3.07%	76.8%	0.25 (0.26)	0.58	0.78	1.64%	3.84%	5.21%
Very elderly (75+ years)	1,209	73.6%	0.32 (0.44)	0.75	1.18	2.11%	4.97%	7.90%	70.9%	0.68 (0.94)	1.58	2.52	1.7%	3.94%	6.3%	70.8%	0.05 (0.05)	0.11	0.14	1.23%	2.76%	3.46%	76.3%	0.29 (0.35)	0.65	0.98	1.90%	4.35%	6.54%
Total population (12+ years)	22,807	83.0%	0.36 (0.42)	0.87	1.22	2.42%	5.80%	8.15%	82.2%	0.77 (0.90)	1.84	2.58	1.91%	4.59%	6.45%	81.8%	0.04 (0.04)	0.10	0.13	1.07%	2.39%	3.23%	85.2%	0.31 (0.34)	0.72	0.99	2.08%	4.78%	6.61%
United States, NHANES 2017–18
Infants and toddlers  (0–35 months)	599	5.1%	0.56^e (0.43)	1.26^f	1.37^g	3.73%^e	8.41%^f	9.14%^g	4.7%	0.91^e (1.10)	2.66^f	3.14^g	2.27%^e	6.65%^f	7.86%^g	4.1%	0.69^e (1.03)	1.52^f	1.75^g	17.35%^e	37.93%^f	43.82%^g	5.1%	0.93^e (0.66)	2.02^f	2.15^g	6.21%^e	13.46%^f	14.33%^g
Young children (3–9 years)	752	14.2%	0.40 (0.44)	0.70	1.14^g	2.69%	4.66%	7.63%^g	13.6%	0.95 (1.38)	1.91	2.99^g	2.38%	4.78%	7.47%^g	8.0%	0.09 (0.16)	0.44^f	0.44^g	2.21%	11.01%^f	11.01%^g	14.2%	0.60 (0.56)	0.99	1.95^g	4.01%	6.62%	12.98%^g
Adolescents (10–17 years)	933	10.2%	0.41 (0.43)	0.86	1.47^g	2.72%	5.73%	9.77%^g	9.2%	1.26 (1.39)	2.65^f	4.51^g	3.15%	6.61%^f	11.28%^g	2.9%	0.13 (0.22)	0.48^f	0.56^g	3.14%	12.10%^f	14.05%^g	10.2%	0.52 (0.49)	1.02	1.74^g	3.49%	6.81%	11.61%^g
Adults (18–64 years)	3,325	17.0%	0.73 (1.08)	1.76	2.62	4.88%	11.71%	17.46%	15.0%	2.21 (3.41)	5.45	8.07	5.53%	13.64%	20.16%	6.3%	0.19 (0.40)	0.57	0.67	4.83%	14.37%	16.73%	17.0%	0.92 (1.28)	2.16	3.11	6.13%	14.37%	20.75%
Elderly (65–74 years)	625	27.6%	0.81 (0.78)	1.60	1.82^g	5.37%	10.69%	12.14%^g	25.9%	2.48 (2.45)	5.09	5.45^g	6.20%	12.73%	13.63%^g	10.9%	0.08 (0.19)	0.36^f	0.40^g	2.06%	8.89%^f	9.98%^g	27.6%	1.02 (0.94)	2.03	2.22^g	6.81%	13.55%	14.81%^g
Very elderly (75+ years)	466	26.1%	0.60 (0.51)	1.37	1.84^g	3.98%	9.15%	12.28%^g	22.7%	1.84 (1.60)	4.22	5.90^g	4.59%	10.56%	14.76%^g	7.8%	0.15 (0.24)	0.49^f	0.62^g	3.87%	12.22%^f	15.50%^g	26.1%	0.77 (0.63)	1.63	2.19^g	5.12%	10.87%	14.60%^g
Total population (0+ years)	6,583	17.2%	0.69 (0.94)	1.52	2.14	4.57%	10.12%	14.26%	15.4%	2.06 (2.94)	4.88	6.95	5.14%	12.21%	17.37%	6.5%	0.17 (0.37)	0.46	0.67	4.23%	11.58%	16.73%	17.5%	0.88 (1.11)	1.94	2.57	5.84%	12.96%	17.11%

LNCS: low- and no-calorie sweeteners; AceK: acesulfame potassium; n: number of subjects in total population (unweighted); % cons.: percentage of consumers (weighted) within specified population group; JECFA: Joint (FAO/WHO) Expert Committee on Food Additives (and Contaminants); ADI: acceptable daily intake; bw: body weight; SD: standard deviation; P90: 90th percentile; P95: 95th percentile; n.a.: data not available.

^aConsumers were identified as individuals who reported consumption of any of the GSFA categories or subcategories during the dietary survey reporting period, which were simulated to contain LNCS during the iterations.

^bConsumption data only available for individuals over 10 years.

^cDue to limited food codes, regular calorically sweetened CSDs, fruit juice, energy drinks, RTD tea and nectars were included as surrogates for their LNCS-sweetened counterparts to assess the anticipated patterns of consumption of LNCS-sweetened beverages.

^dDue to limited food codes, regular calorically sweetened CSDs, fruit juice and fruit nectars were included as surrogates for their LNCS-sweetened counterparts to assess the anticipated patterns of consumption of LNCS-sweetened beverages.

^eIndicates an intake estimate that may not be statistically reliable, as the sample size does not meet the minimum reporting requirements of n = 30 (Korn and Graubard 1999).

^fIndicates an intake estimate that may not be statistically reliable, as the sample size does not meet the minimum reporting requirements of n = 80 (Kroes et al. 2002).

^gIndicates an intake estimate that may not be statistically reliable, as the sample size does not meet the minimum reporting requirements of n = 130 (Conover 1971).

AceK

Estimated intakes for AceK at the mean, 90th and 95th percentile across all markets ranged between 1.84–6.43%, 4.26–12.80%, and 6.46–16.00% of the ADI, respectively. Brazilian adolescents (ages 10–17 years) were the highest consumers of AceK at the 95th percentile, at 16.00% of the ADI.

Aspartame

Estimated intakes for aspartame at the mean, 90th and 95th percentile ranged between 1.46–6.20%, 3.37–13.64%, and 5.13–20.16% of the ADI, respectively. The highest 95th percentile intake was observed in U.S. adults at 8.07 mg kg⁻¹ bw day⁻¹.

Sucralose

Estimated intakes for sucralose at the mean and 90th percentile across all markets ranged between 1.64–7.86% and 3.84–15.23% of the ADI, respectively. The 95th percentile intake estimates ranged from 5.21% ADI in the elderly Mexican population to 20.93% ADI in Canadian adults.

Steviol glycosides

There were no data available for steviol glycosides in Brazil, and the data for Canada were not statistically reliable due to the limited sample size. As for the other regions, estimated intakes for steviol glycosides in Mexico ranged between 1.03–1.89%, 2.26–4.17% and 3.07–5.88% of the ADI at the mean, 90th and 95th percentile, respectively. In the U.S., statistically reliable estimated intakes were observed only in adults (90th percentile, 14.37% ADI; 95th percentile, 16.73% ADI) and the total population (90th percentile, 11.58% ADI; 95th percentile, 16.73% ADI).

Discussion

The current analyses applied two modelling approaches using up-to-date nationally representative food consumption data to estimate intakes of four LNCS from LNCS-sweetened water-based beverages in four priority regions. The probabilistic model was guided by reported use level data, with estimated intake for an individual informed by the pattern of LNCS use in a particular beverage type in a given market. The brand-loyal model was driven by assumed consumer behaviour-led patterns, namely that an individual will be brand loyal to a pre-determined beverage type.

The probabilistic approach is more representative of an average or ‘non-brand loyal’ consumer, who may select any available beverage brand on a given ‘eating occasion’. This model considers the variability in input factors which is influenced by the market volume and industry-reported patterns of use, by applying use level data to each relevant beverage code, with the chance that the reported use level being consumed is directly related to the market-volume share of the corresponding brand within the beverage type. To put this into context, if ‘Participant A’ consumed an LNCS-sweetened CSD during the Brazilian dietary intake survey, the likelihood of the 123 mg/L use level being applied to the consumption volume for that ‘eating occasion’ is 83%; there is a 17% chance that one of the other available use level values will be selected (all values included in Table 1).

The distributional approach, on the other hand, aims to model exposure by ‘brand loyal’ consumers, wherein the individual consumers are assumed to be ‘brand loyal’ to (or habitually consume) a pre-determined beverage type which is assumed to contain the maximum use level for the LNCS. The ‘brand loyal’ beverage type is identified by determining the highest contributor beverage type to the total exposure of the LNCS. It is also assumed that the individual consumes all other ‘non-brand loyal’ beverages (which may contain the LNCS) but at the average (market volume weighted) use levels. The market volume data as applied in the brand loyal scenario does not influence an individual’s overall pattern of LNCS consumption as occurs for the probabilistic model. A key difference between the two models is that 100% presence of the LNCS is assumed for all beverage types in which use has been reported in the ‘brand loyal’ model (i.e. only those beverage types that did not have reported use of the LNCS were removed from the brand-loyal assessment). Despite the differences between the two models, estimated intakes for all four LNCS across all four markets investigated were substantially lower than the respective ADIs (at 21% or less) for all subpopulations even at the highest 95th percentile intake estimates.

In 2018, Martyn et al. compiled an extensive review of the literature on the intakes of LNCS globally, noting data and exposure assessments were limited for most markets/regions, except for the EU, Japan and Korea. Since this time, several studies have been published to contribute to data gaps, with a focus on the Americas (Barraj, Scrafford, et al. 2021; Tran et al. 2021; Martyn et al. 2022). The current research contributes to this body of data. Overall, the findings of this study reinforce those in the literature. That is, regardless of modelling approach executed, exposure to any of the four major sweeteners evaluated remains far below the ADI across subpopulations, even among the highest consuming population (95th percentile).

The complex modelling approaches applied for the EU as summarised by Martyn et al. (2018) also resulted in estimated intakes (% ADI) that are comparable to the current analysis. More recently, Tran et al. (2021) assessed global LNCS exposure by combining U.S. food consumption data as a surrogate for the world with the global maximum reported LNCS use level. As part of its tiered intake assessment approach, Tran et al. (2021) conducted a ‘brand loyal deterministic, Scenario 2’ (BLD-2) modelling analysis – which is comparable to the ‘brand loyal assessment’ herein. Tran et al. (2021) leveraged food consumption data from the 2013–2016 NHANES cycles while the current analysis applied the most up-to-date fooc consumption data from the 2017–2018 NHANES. Similar to the current analysis, the ADI was not exceeded for any of the LNCS investigated by Tran and colleagues for any subpopulation in their refined exposure assessment models. Overall, the estimated intakes observed were largely similar across these two analyses, with a few exceptions. For the U.S., the estimated intakes of AceK, aspartame, and sucralose in this paper were lower in toddlers, likely attributable to the smaller sample size in the current assessment, with statistical reliability often not being achieved due to low percent consumers for this age group (∼5%). Additionally, the estimated intakes of AceK, steviol glycosides and sucralose were lower in adults in the current analysis, likely due to the exclusion of hot teas and coffees. Overall, the patterns noted for LNCS-sweetened beverages in the U.S. and Canada in our current analysis are consistent with the outcomes from the BLD-2 scenario in Tran et al. (2021), further reaffirming the safety of LNCS consumed at current use levels.

Data pertaining to LNCS exposure from nationally representative data in LATAM are limited. Barraj, Scrafford, et al. (2021) examined LNCS exposure in the Brazilian population. The results obtained were comparable to the current analysis, despite some key methodological differences including the food consumption data applied and Barraj, Scrafford, et al. (2021) including both food and beverages in the analysis. This can be explained by our inclusion of calorically sweetened surrogates for those LNCS-sweetened beverage types that had generally low reported consumption in the total population, or where food codes were lacking for the current analysis. Also, the results were comparable likely due to the greater contribution from beverages to LNCS intakes. It is worth noting that many LATAM countries, starting with Ecuador, have implemented a front-of-package (FOP) labelling law since 2014, including Chile in 2016 (Sambra et al. 2020). In Chile, exceeding thresholds established for nutrients to limit (e.g. sugar) would require on-pack placement of the relevant FOP scheme. In some cases, food and beverage companies are reformulating their products to fully or partially replace sugar with LNCS. Although the Chilean nationally representative food consumption data dates to 2010 (well before the Chilean FOP law was implemented), a number of later studies evaluated LNCS intakes in preschoolers, children, and pregnant women prior to and following the implementation using an FFQ (Fuentealba Arévalo et al. 2019; Martínez et al. 2020; Venegas Hargous et al. 2020). They all concluded that none of the estimated LNCS intakes exceeded any of the ADIs for any age group, even for high consumers, as also reported Barraj, Bi, et al. (2021) using national sales data and product labels.

Several studies have reported that the prevalence of LNCS beverage consumption in the U.S. is 2–3 times higher than the consumption of LNCS-sweetened foods (Gardner et al. 2012; Sylvetsky et al. 2017; Malek et al. 2018). Given that the highest intake, at the 95th percentile of intake, was only one-quarter of the ADI for AceK, aspartame, and sucralose by any subpopulation group in the current analyses, the use of these LNCS in beverages should be considered safe based on current patterns of use. Given that the highest intake at the 95th percentile is less than 20% of the ADI in all subpopulations (and is in agreement with Martyn et al. 2022), the use of these LNCS in beverages in Brazil and Mexico should also be considered safe based on current patterns of use.

The inclusion of the most up-to-date, nationally representative food consumption data in each of the four markets, coupled with a quantitative brand-specific weighting scheme – that is, relative to the average use level for the ‘non-brand loyal’ beverage types in the distributional brand-loyal model or market volume-weighted probabilistic scenario is a strength of the current analysis. The latter provides for a more accurate representation of industry- and consumer-practices in these key regions. Furthermore, this was the first analysis of its kind, to the authors’ knowledge, in Canada and Mexico. The findings show the consumption of four key LNCS in water-based flavoured non-alcoholic beverages does not present any safety concern when compared with the JECFA ADI (JECFA 2019).

An uncertainty analysis was conducted in accordance with the criteria outlined in Codex and the European Food Safety Authority (EFSA) (CAC 2003; EFSA 2006) to identify all potential sources of over- or underestimation in the probabilistic and distributional models, with conclusions summarised in Table 5. The application of food consumption data has several recognised limitations – specifically the duration of surveys (typically 1–2 days) can result in an overestimate of chronic dietary exposure (Vin et al. 2013); under- or overreporting of certain foods and beverages is well acknowledged, particularly underreporting of foods that may be perceived to be unhealthy, such as soft drinks (Han and Powell 2013). Importantly, however, even though the Brazilian and Mexican dietary intake surveys were limited by the number of relevant food codes available for the LNCS beverage types (e.g. n = 6 for Brazil, n = 2 for Mexico, compared to n = 59 for U.S. or n = 25 for Canada), this limitation was overcome by assuming the consumption pattern for LNCS-sweetened beverages emulated that for calorically sweetened beverages. However, this may also have then contributed to an overestimation of the number of consumers of the particular beverage, given the limited number of codes available to allocate the beverage type to in the Brazil and Mexican surveys. Furthermore, the distributional ‘brand-loyal’ model is conservative, as it assumes 100% LNCS use across all beverage types, and that all beverages in the ‘brand loyal’ beverage type contain the LNCS at the maximum level. In reality, only a fraction of beverages would contain any given LNCS, and a much smaller fraction of those would contain the LNCS at the maximum level such that 100% market penetration rarely occurs.

Table 5. Qualitative evaluation of the influence of uncertainties on exposure estimates to low- and no-calorie sweeteners from water-based flavoured non-alcoholic beverages in four markets^a.

Source of uncertainty	Direction and magnitude^b
Consumption data
Mis/under-reporting of beverage consumption in dietary surveys	–
Estimating chronic exposure with short-term dietary intake data	+
Limited categories to capture beverage types (Mexico and Brazil)	±
Brand-specific beverage consumption data not available	±
Rounding errors in food consumption data	±
Use-level data
Use-level data and beverage consumption data were collected at two separate time points	±
Use levels applied represented approximately 80% of the market, use data for smaller brands not included	±
Rounding errors, incorrect entries for concentration data	±
Probabilistic model
No consideration for brand loyal consumers	±
Probability of LNCS use levels being encountered for each consumer was based on market share  (by volume) and not actual consumer behaviour due to non-branded food codes	±
Brand loyal model
100% presence assumed for all beverage categories in which LNCS were reported for use,  regardless of whether these reported values accounted for a large proportion of the market	+

^aConducted in accordance with the criteria specified in Codex and EFSA (European Food Safety Authority) (CAC 2003; EFSA 2006).

^b+, ++ or +++ is the uncertainty likely to cause small, medium or large overestimates of exposure, respectively, and −, − − or − − − is the uncertainty likely to cause small, medium or large underestimates of exposure, respectively.

In summary, our robust analysis of current exposures to LNCS from water-based flavoured non-alcoholic LNCS-sweetened beverages within ‘high consuming’ beverage markets continues to support the safety of these LNCS. The 95th percentile intake estimates, even for the brand-loyal scenario, were substantially below each respective JECFA ADI, across all subpopulations.

Abbreviations
ABA	=	American Beverage Association
ABIR	=	Associacao Brasileira das Industrias de Refrigerantes e de Bebidas nao Alcoolicas
AceK	=	Acesulfame potassium
ADI	=	Acceptable Daily Intake
BLD-2	=	Brand Loyal Deterministic: Scenario 2
bw	=	Body weight
CBA	=	Canadian Beverage Association
CCHS	=	Canadian Community Health Survey – Nutrition
CSDs	=	Carbonated Soft Drinks
EFSA	=	European Food Safety Authority
ENSANUT	=	Encuesta Nacional de Salud y Nutrición
FDA	=	U.S. Food and Drug Administration
FFQ	=	Food-Frequency Questionnaire
FOP	=	Front-ofpackage
GSFA	=	General Standard for Food Additives
IBGE	=	Instituto Brasileiro de Geografia e Estatística
ICBA	=	International Council of Beverages Associations
INA	=	Inquérito Nacional de Alimentação
INSP	=	Instituto Nacional de Salud Pública
JECFA	=	Joint FAO/WHO Expert Committee on Food Additives
LATAM	=	Latin America
LNCS	=	Low- and No-Calorie Sweeteners
NCHS	=	National Center for Health Statistics
NDNS	=	National Diet and Nutrition Survey
NHANES	=	National Health and Nutrition Examination Survey
P OF	=	Pesquisa de Orcamenos Familiares
P UMF	=	Public Use Microdata File
RT D	=	ready-to-drink
U.K.	=	United Kingdom
U.S.	=	United States

Disclosure statement

YML, JM, DMM, MD, LSG and ET do not report any conflicts of interest. MMJ is employed by the American Beverage Association.

Additional information

Funding

This work was funded by the American Beverage Association.