Decreasing Trends in Heavy Sugar-Sweetened Beverage Consumption in the United States, 2003 to 2016

Published:September 24, 2020DOI:https://doi.org/10.1016/j.jand.2020.07.012

      Abstract

      Background

      Although previous studies have documented declines in intake from sugar-sweetened beverages (SSB) in the United States, it is important to examine whether heavy SSB intake (≥500 kcal/day) is decreasing in parallel. Examining the intake patterns of heavy SSB consumers is imperative because these individuals face the greatest health risks and thus may benefit the most from targeted policy and programmatic efforts to reduce intake.

      Objective

      To provide the most recent national estimates for trends in heavy SSB intake among children and adults in the United States between 2003-2004 and 2015-2016, to examine whether these trends differ by sociodemographic characteristics, and to describe where SSB are acquired and consumed by the heaviest SSB consumers.

      Design

      Trend analyses of demographic and 24-hour dietary recall data in the 2003-2004 to 2015-2016 National Health and Nutrition Examination Survey.

      Participants/setting

      Participants were 21,783 children (aged 2 to 19 years) and 32,355 adults (aged ≥20 years).

      Main outcome measures

      Heavy SSB intake (≥500 kcal/day).

      Statistical analysis

      Survey-weighted logistic regression was used to estimate the proportion of heavy SSB consumers, overall and by age group, race/ethnicity, sex, and income status (lower income = <130% Federal Poverty Level). Proportions were used to summarize where SSB are most often acquired and consumed.

      Results

      Between 2003-2004 and 2015-2016, the prevalence of heavy SSB intake declined significantly among children (10.9% to 3.3%) and adults (12.7% to 9.1%). For children, these declines were observed across age group, sex, family income status, and most races/ethnicities. For adults, these significant declines were observed among 20- to 39-year olds, most races/ethnicities, and higher-income adults. However, there was a significant increase in heavy SSB intake among adults aged ≥60 years and no significant change among 40- to 59-year olds and non-Mexican Hispanic adults. The majority of energy intake from SSB consumed by heavy SSB drinkers was from products acquired from stores and was consumed at home.

      Conclusions

      Heavy SSB intake is declining, but attention must be paid to certain subgroups with high intake for whom trends are not decreasing, particularly 40- to 59-year olds and non-Mexican Hispanic adults.

      Keywords

      The Continuing Professional Education (CPE) quiz for this article is available for free to Academy members through the MyCDRGo app (available for iOS and Android devices) and through www.jandonline.org (click on “CPE” in the menu and then “Academy Journal CPE Articles”). Log in with your Academy of Nutrition and Dietetics or Commission on Dietetic Registration username and password, click “Journal Article Quiz” on the next page, then click the “Additional Journal CPE quizzes” button to view a list of available quizzes. Non-members may take CPE quizzes by sending a request to [email protected] There is a fee of $45 per quiz (includes quiz and copy of article) for non-member Journal CPE. CPE quizzes are valid for 3 years after the issue date in which the articles are published.
      Research Question: What are the trends in heavy sugar-sweetened beverage intake (≥500 kcal/day) among children and adults in the United States between 2003-2004 and 2015-2016?
      Key Findings: Heavy sugar-sweetened beverage intake has declined in the US population overall, but attention must be paid to certain subgroups with high intake for whom trends are not decreasing, particularly 40- to 59-year olds and non-Mexican Hispanic adults.
      Although sugar-sweetened beverages (SSB) are widely consumed in the United States, research suggests that intake is declining.
      • Mesirow M.S.
      • Welsh J.A.
      Changing beverage consumption patterns have resulted in fewer liquid calories in the diets of US children: National Health and Nutrition Examination Survey 2001-2010.
      • Welsh J.A.
      • Sharma A.J.
      • Grellinger L.
      • Vos M.B.
      Consumption of added sugars is decreasing in the United States.
      • Bleich S.N.
      • Vercammen K.A.
      • Koma J.W.
      • Li Z.
      Trends in beverage consumption among children and adults, 2003-2014.
      Between 2003-2004 and 2013-2014, the proportion of the population consuming at least one SSB on a typical day fell from 80% to 61% among children and from 62% to 50% among adults.
      • Bleich S.N.
      • Vercammen K.A.
      • Koma J.W.
      • Li Z.
      Trends in beverage consumption among children and adults, 2003-2014.
      Given the link between SSB intake and increased risk of a wide range of adverse outcomes such as weight gain, type 2 diabetes, and mortality,
      • Malik V.S.
      • Li Y.
      • Pan A.
      • et al.
      Long-term consumption of sugar-sweetened and artificially sweetened beverages and risk of mortality in US adults.
      • Malik V.S.
      • Popkin B.M.
      • Bray G.A.
      • Després J.P.
      • Willett W.C.
      • Hu F.B.
      Sugar-sweetened beverages and risk of metabolic syndrome and type 2 diabetes: A meta-analysis.
      • Malik V.S.
      • Schulze M.B.
      • Hu F.B.
      Intake of sugar-sweetened beverages and weight gain: A systematic review.
      the recent declines in SSB intake signal promising progress. These declines in the average intake of SSB over time may be driven by a shift in the population distribution of consumption or by reductions in intake among the heaviest consumers. Thus, it is critically important to examine whether SSB intake is also declining amongst the heaviest consumers. Examining the intake patterns of heavy SSB consumers is imperative because these individuals face the greatest health risks and thus may benefit the most from targeted policy and programmatic efforts to reduce intake.
      A few prior studies have examined trends in heavy SSB intake over time.
      • Han E.
      • Powell L.M.
      Consumption patterns of sugar-sweetened beverages in the United States.
      ,
      • Mendez M.A.
      • Miles D.R.
      • Poti J.M.
      • Sotres-Alvarez D.
      • Popkin B.M.
      Persistent disparities over time in the distribution of sugar-sweetened beverage intake among children in the United States.
      One study by Han and Powell
      • Han E.
      • Powell L.M.
      Consumption patterns of sugar-sweetened beverages in the United States.
      examined trends in heavy SSB intake (defined as ≥500 kcal per day from SSB) among SSB consumers between 1999-2000 and 2007-2008 and found that heavy SSB intake increased from 4% to 5% among children, decreased from 22% to 16% among adolescents, and decreased from 29% to 20% among young adults. A more recent study by Mendez and colleagues
      • Mendez M.A.
      • Miles D.R.
      • Poti J.M.
      • Sotres-Alvarez D.
      • Popkin B.M.
      Persistent disparities over time in the distribution of sugar-sweetened beverage intake among children in the United States.
      examined changes in the distribution of SSB intake among children between 2003-2004 and 2013-2014 and found that intake at the 90th percentile of SSB intake declined, but disparities in heavy SSB intake persisted over time. For example, higher income was associated with lower SSB intakes at the 90th percentile for non-Hispanic White, but not non-Hispanic Black children.
      • Mendez M.A.
      • Miles D.R.
      • Poti J.M.
      • Sotres-Alvarez D.
      • Popkin B.M.
      Persistent disparities over time in the distribution of sugar-sweetened beverage intake among children in the United States.
      Our study adds to the existing literature by extending prior estimates of heavy SSB intake using the most recent data from the National Health and Nutrition Examination Survey (NHANES). First, this study makes an important contribution by updating trends among adults because Han and Powell’s
      • Han E.
      • Powell L.M.
      Consumption patterns of sugar-sweetened beverages in the United States.
      estimates end in 2007-2008, leaving an 8-year gap in surveillance to the most recent 2015-2016 NHANES data. This is a particularly important gap in the literature in light of evidence that young adults have higher per capita energy intake from SSB than any other age group, suggesting they are an important population to monitor.
      • Bleich S.N.
      • Vercammen K.A.
      • Koma J.W.
      • Li Z.
      Trends in beverage consumption among children and adults, 2003-2014.
      Second, whereas Mendez and colleagues’
      • Mendez M.A.
      • Miles D.R.
      • Poti J.M.
      • Sotres-Alvarez D.
      • Popkin B.M.
      Persistent disparities over time in the distribution of sugar-sweetened beverage intake among children in the United States.
      time trend estimates among children extend to 2013-2014, we still believe that updating estimates with the most recent data is important given the number of SSB reduction policies that have passed since 2014. For example, beginning in 2014, several local and tribal governments, such as the city of Berkeley, CA, and the Navajo Nation, passed SSB excise taxes.
      Healthy Food America
      Sugary beverages: Map and chart the movement.
      In addition, since 2015, more than a dozen municipalities have passed policies requiring restaurants to serve only healthy beverages instead of SSB with children’s meals, and many of the leading chains have voluntarily replaced soda (ie, sweetened carbonated beverage) in kids’ meals with milk and 100% juice.
      Healthy Food America
      Sugary beverages: Map and chart the movement.
      ,
      California Legislative Information. Senate Bill No. 1192: Children’s meals.
      Although these local policies may not affect national intake levels, the increasing frequency of beverage taxes and healthy default beverage laws is indicative of growing recognition of the health harms of SSB over the past several years. This increased awareness is reflected in declining SSB sales,
      • Taylor K.
      People are drinking less Pepsi and Coke than ever—and it reveals the power of the ‘biggest marketing trick of the century.’.
      yet trends in intake after 2014 have not yet been published.
      In addition to updating time trends in heavy SSB intake among children and adults, this study also contributes to the literature by documenting where SSB are most frequently acquired (eg, restaurants and stores) and consumed (ie, at home or away from home) by heavy SSB consumers. A previous study using 2005-2008 NHANES data found that about half of the total energy intake from SSB is consumed at home, with products purchased in stores accounting for the vast majority of this energy intake.
      • Ogden C.L.
      • Kit B.K.
      • Carroll M.D.
      • Park S.
      Consumption of sugar drinks in the United States, 2005-2008.
      However, these estimates are now dated and were not specific to heavy SSB drinkers. Understanding where heavy SSB drinkers are most likely to acquire and consume SSB could help target research, policy, and advocacy efforts to curb excessive SSB intake in the United States.
      The objectives of this study were to examine trends in heavy SSB intake among children and adults between 2003-2004 and 2015-2016; examine whether there are differences in these trends by age group, sex, race/ethnicity, and income; and describe where SSB are acquired and consumed by the heaviest SSB consumers. We hypothesize that there will be declines in heavy SSB intake, but that these declines will not be observed to as great an extent among groups who are disproportionately exposed to SSB marketing (ie, racial/ethnic minorities and low-income populations). We hypothesize that most energy intake from SSB will be from products purchased in stores and consumed at home.

      Materials and Methods

      Data and Study Population

      This trend analysis used data from seven survey cycles (2003-2004 to 2015-2016) of the NHANES, a repeated cross-sectional study released every 2 years and designed to be representative of the US noninstitutionalized population. A complete description of NHANES is available online.
      National Center for Health Statistics
      NHANES survey methods and analytic guidelines.
      The study sample consisted of individuals aged ≥2 years with complete data on all covariates. Because this study analyzed de-identified publicly available data, it does not constitute human subjects research and institutional review board approval was not required.

      Measures

      SSB Intake

      SSB intake was assessed using a 24-hour diet recall. Survey respondents reported all food and beverages consumed in the previous 24-hour period, specifying the type, quantity, source, and location of each intake occasion. Responses for children aged 2 to 5 years were provided by a caretaker, responses for participants aged 6 to 8 years were provided by a caretaker and assisted by the child, responses for participants aged 9 to 11 years were provided by the child and assisted by a caretaker, and participants aged 12 years and older responded independently. All reported food and beverage items were systematically coded using the US Department of Agriculture Food and Nutrient Database for Dietary Studies and the Food Patterns Equivalents Database to obtain energy and added sugar information.
      SSB were defined as any nondairy or nondairy alternative beverage with >0 g added sugar. The beverage coding strategy used in this analysis updates a version used in previous SSB trends papers.
      • Bleich S.N.
      • Vercammen K.A.
      • Koma J.W.
      • Li Z.
      Trends in beverage consumption among children and adults, 2003-2014.
      ,
      • Wang Y.C.
      • Bleich S.N.
      • Gortmaker S.L.
      Increasing caloric contribution from sugar-sweetened beverages and 100% fruit juices among US children and adolescents, 1988-2004.
      In an effort to make identification of SSB more objective, our updated beverage coding strategy now uses added sugar quantity to classify beverages, whereas the previous coding scheme utilized beverage descriptions to identify whether a beverage was sweetened or not.
      • Bleich S.N.
      • Vercammen K.A.
      • Koma J.W.
      • Li Z.
      Trends in beverage consumption among children and adults, 2003-2014.
      In addition, dairy and dairy alternatives are no longer categorized as SSB to be consistent with definitions used in many policies aimed at reducing SSB intake (although we allowed dairy and/or dairy alternatives to be included in nutrient totals in the case that they were a minor addition to another beverage such as sweetened coffee or tea).
      Consistent with a previous study,
      • Han E.
      • Powell L.M.
      Consumption patterns of sugar-sweetened beverages in the United States.
      an individual was considered to be a heavy SSB drinker in the case that s/he reported consuming ≥500 kcal per day from SSB. This quantity is comparable with definitions used by other studies
      • Park S.
      • Blanck H.M.
      • Sherry B.
      • Brener N.
      • O'Toole T.
      Factors associated with sugar-sweetened beverage intake among United States high school students.
      ,
      • White A.H.
      • James S.A.
      • Paulson S.W.
      • Beebe L.A.
      Sugar sweetened beverage consumption among adults with children in the home.
      and is equivalent to consuming about 3.5 cans of regular soda (assuming 12 oz and about 150 kcal per can) per day. Because of the within-person variation in daily SSB intake, the distribution of intake from a single 24-hour recall is wider than the distribution of true usual (mean daily) intake.
      • Willett W.
      Nutritional Epidemiology.
      This means that estimates of the proportion of individuals consuming ≥500 kcal/day from SSB using a single 24-hour recall will likely be overestimated. The National Cancer Institute (NCI) has developed a method to estimate usual intake from two 24-hour recalls and thus more validly estimate the proportion consuming ≥500 kcal/day
      • Tooze J.A.
      • Midthune D.
      • Dodd K.W.
      • et al.
      A new statistical method for estimating the usual intake of episodically consumed foods with application to their distribution.
      ; however, this method has limited analytic flexibility to provide covariate-adjusted time trend estimates. Thus, our main results are reported from a single 24-hour recall. Sensitivity analyses indicate that unadjusted estimates from the NCI Method are lower than those from a single 24-hour recall, although the overall decreasing trend in heavy SSB intake is evident in both methods. This is consistent with findings by Mendez and colleagues
      • Mendez M.A.
      • Miles D.R.
      • Poti J.M.
      • Sotres-Alvarez D.
      • Popkin B.M.
      Persistent disparities over time in the distribution of sugar-sweetened beverage intake among children in the United States.
      who state that their trends in heavy SSB intake estimated from the NCI method are not significantly different from studies that used a single 24-hour recall.
      Although our primary analyses utilized a 500 kcal/day cutoff for all participants to enable comparisons across age groups and with previous studies that have used the same measurement definition,
      • Han E.
      • Powell L.M.
      Consumption patterns of sugar-sweetened beverages in the United States.
      we acknowledge that individuals vary in their daily energy requirements. Thus, we conducted sensitivity analyses wherein heavy SSB consumption was alternatively defined as consuming ≥25% of daily energy intake from SSB.

      SSB Subtypes, Source, and Location of Intake

      SSB were subcategorized into soda, fruit drinks, energy/sports drinks, low-calorie SSB, and other SSB (see Figure 1, available at www.jandonline.org, for SSB subtype coding scheme). For each food and beverage, the NHANES includes information on whether the eating occasion occurred at or away from home as well as where the food/beverage was acquired (ie, food source). In line with a previous study,
      • Barrera C.M.
      • Moore L.V.
      • Perrine C.G.
      • Hamner H.C.
      Number of eating occasions and source of foods and drinks among young children in the United States: NHANES, 2009–2014.
      we categorized the different food source options into four mutually exclusive categories: stores (grocery, supermarket, and convenience stores), restaurants (restaurants with waiter/waitress, restaurants with fast food/pizza, bar/tavern/lounge, street vendor, sport, recreation, or entertainment facility), child or adult care (cafeteria in kindergarten through grade 12 school, child/adult care center, or home), or other source (soup kitchen or food pantry, Meals on Wheels, community food program, fundraiser, mail-order purchase, grown or caught by individual, vending machine, common coffee pot or snack tray, residential dining facility, a gift, other). Because of the small size of the child or adult care category, it was later collapsed together with the other source category.

      Covariates

      To adjust for potential demographic shifts over time, analyses included the following covariates: age group (2 to 5 years, 6 to 11 years, 12 to 19 years, 20 to 39 years, 40 to 59 years, or >60 years), sex (male or female), race/ethnicity (non-Hispanic White, non-Hispanic Black, Mexican American, non-Mexican Hispanic, other race/ethnicity), and family income (lower income or higher income). Other race/ethnicity included individuals reporting a race other than White or Black or individuals reporting multiracial identity. Lower income was defined as <130% of the Federal Poverty Level, whereas higher income was defined as ≥130% of the Federal Poverty Level).

      Statistical Analysis

      All analyses were weighted to account for the multistage, clustered probability sampling of the NHANES. Logistic regression was used to estimate the proportion of heavy SSB drinkers on a typical day for each survey year. For these models, the primary outcome was the log odds of being a heavy SSB drinker and the covariates were survey year, age group, sex, race/ethnicity, and income status. Analyses for adults and children were conducted separately. To obtain trend estimates within subgroups, separate models were fitted within each subgroup, adjusting for all other covariates (eg, model was fit among non-Hispanic White children, adjusting for survey year, sex, age category, and income). Postregression, Stata’s margins command

      Stata [computer program]. Version 16.0. College Station, TX: StataCorp; 2019.

      was used to estimate the predicted proportion of heavy SSB consumers for each survey year, when all other covariates were set to their mean values. To statistically compare the proportion of heavy SSB drinkers across groups, we fitted a logistic regression model using only 2015-2016 data (separately for children and adults) and included terms for each subgroup.
      We also conducted linear regressions to estimate the per capita energy intake from each SSB subtype (soda, fruit drinks, energy/sports drinks, low-calorie SSB, and other SSB) among heavy SSB consumers. These models were restricted to heavy SSB consumers; the primary outcome was energy intake from each SSB subtype (eg, energy intake from soda) and the covariates were the same as above. Analyses for adults and children were conducted separately. Stata’s margins command

      Stata [computer program]. Version 16.0. College Station, TX: StataCorp; 2019.

      was used to estimate the predicted per capita energy intake from each SSB subtype for each survey year, when all other covariates were set to their mean values. To account for heteroscedasticity of observations derived from the complex sampling survey, all regressions were weighted by survey sampling weights.
      To analyze the significance in trends over time, models were fit with a continuous survey year term (instead of a categorical survey year term). To assess potential nonlinearity in trends over time,
      • Ingram D.D.
      • Malec D.J.
      • Makuc D.M.
      • et al.
      National Center for Health Statistics guidelines for analysis of trends: Data evaluation and methods research.
      quadratic and cubic year terms were also included as covariates, and we performed a joint Wald test of the quadratic and cubic terms. In the case that it was significant, we reported the results from this model. In the case that it was not, we concluded there was no evidence of nonlinearity, and a model including only a linear term was fitted and the results from this model were reported.
      Descriptive statistics (ie, means and proportions) were used to summarize patterns in current (2015-2016) SSB intake. Means were used to summarize energy intake and added sugar from SSB consumed by heavy SSB consumers. Among heavy SSB consumers, the proportion of energy intake from SSB consumed at home was calculated as energy intake from SSB consumed at home divided by total energy intake from SSB, whereas the proportion of energy intake from SSB consumed away from home was calculated as energy intake from SSB consumed away from home divided by the total energy intake from SSB. Similarly, the proportion of energy intake from SSB acquired from each source (store, restaurant, child or adult care, and other) was calculated as the energy intake from SSB acquired from each source divided by total energy from SSB.
      All analyses were conducted in 2019 and 2020 using Stata, version 16.0.

      Stata [computer program]. Version 16.0. College Station, TX: StataCorp; 2019.

      Results

      The total analytic sample included 21,783 children and 32,355 adults. Table 1 reports unweighted sample sizes and proportions by age group, sex, race/ethnicity, and income.
      Table 1Unweighted sample sizes and proportions of participants in analytic sample
      The analytic sample consisted of 54,138 participants in NHANES 2003-2016 with complete data on all covariates (age, sex, race/ethnicity, income) and a valid first 24-hour dietary recall.
      by age group, sex, race/ethnicity, and income status, National Health and Nutrition Examination Survey (NHANES) (2003-2016)
      CharacteristicAnalytic sample
      n (%)
      Children (y)21,783 (40)
      2-55,178 (24)
      6-117,010 (32)
      12-199,595 (44)
      Adults (y)32,355 (60)
      20-3911,315 (35)
      40-5910,411 (32)
      ≥6010,629 (33)
      Sex
      Male26,702 (49)
      Female27,436 (51)
      Race/ethnicity
      Non-Hispanic White21,271 (39)
      Non-Hispanic Black12,728 (24)
      Mexican American10,796 (20)
      Non-Mexican Hispanic4,531 (8)
      Other race/ethnicity4,812 (9)
      Income
      Lower income defined as family income <130% of the Federal Poverty Level. Higher income defined as family income ≥130% of the Federal Poverty Level.
      Lower19,956 (37)
      Higher34,182 (63)
      a The analytic sample consisted of 54,138 participants in NHANES 2003-2016 with complete data on all covariates (age, sex, race/ethnicity, income) and a valid first 24-hour dietary recall.
      b Lower income defined as family income <130% of the Federal Poverty Level. Higher income defined as family income ≥130% of the Federal Poverty Level.

      Trends in Heavy SSB Intake between 2003-2004 and 2015-2016

      The prevalence of heavy SSB intake declined significantly between 2003-2004 and 2015-2016 among children (10.9% to 3.3%; P for trend < 0.001) and adults (12.7% to 9.1%; P for trend = 0.001) (Figure 2).
      Figure thumbnail gr1ab
      Figure 2Trends in prevalence of heavy sugar-sweetened beverage (SSB) intake (≥500 kcal/day) between 2003-2004 and 2015-2016 among National Health and Nutrition Examination Survey (NHANES) participants (A) for children and adults separately, (B) for children by age group, and (C) for adults by age group. To obtain trend estimates, separate models were fitted among children and adults, adjusting for survey year, sex, race/ethnicity, and income status. Negative predicted values were truncated at 0. (A) The proportion of heavy SSB drinkers (≥500 kcal/day) declined significantly among both children and adults in NHANES between 2003-2004 and 2015-2016 (P for trend for children < 0.001, P for trend for adults = 0.001). (B) Among children in NHANES, the proportion of heavy SSB drinkers declined significantly between 2003-2004 and 2015-2016 across all age groups (P for trend for all < 0.001), with 12 to 19-year olds maintaining the highest prevalence of heavy SSB intake across all survey years. (C) Among adults in NHANES, the proportion of heavy SSB drinkers decreased significantly between 2003-2004 and 2015-2016 among 20-39-year olds (P for trend < 0.001), remained relatively constant among 40- to 59-year olds (P for trend = 0.767), and increased significantly among ≥60-year olds (P for trend = 0.007).
      Figure thumbnail gr1c
      Figure 2Trends in prevalence of heavy sugar-sweetened beverage (SSB) intake (≥500 kcal/day) between 2003-2004 and 2015-2016 among National Health and Nutrition Examination Survey (NHANES) participants (A) for children and adults separately, (B) for children by age group, and (C) for adults by age group. To obtain trend estimates, separate models were fitted among children and adults, adjusting for survey year, sex, race/ethnicity, and income status. Negative predicted values were truncated at 0. (A) The proportion of heavy SSB drinkers (≥500 kcal/day) declined significantly among both children and adults in NHANES between 2003-2004 and 2015-2016 (P for trend for children < 0.001, P for trend for adults = 0.001). (B) Among children in NHANES, the proportion of heavy SSB drinkers declined significantly between 2003-2004 and 2015-2016 across all age groups (P for trend for all < 0.001), with 12 to 19-year olds maintaining the highest prevalence of heavy SSB intake across all survey years. (C) Among adults in NHANES, the proportion of heavy SSB drinkers decreased significantly between 2003-2004 and 2015-2016 among 20-39-year olds (P for trend < 0.001), remained relatively constant among 40- to 59-year olds (P for trend = 0.767), and increased significantly among ≥60-year olds (P for trend = 0.007).
      Among children, the proportion of heavy SSB intake declined significantly across all age groups, with 12- to 19-year olds remaining the highest heavy SSB consumers across all survey years. Heavy SSB intake also declined significantly among non-Hispanic White (12.1% to 3.7%; P for trend < 0.001), non-Hispanic Black (10.9% to 3.3%; P for trend < 0.001), Mexican American children (10.7% to 2.5%; P for trend < 0.001), and non-Mexican Hispanic children (8.0% to 2.8%; P for trend = 0.028) (Table 2). For children of other race/ethnicities, there was a decline in heavy SSB intake, but it was not statistically significant (3.4% to 1.8%; P for trend = 0.194). For this group, proportions were usually substantially below the average for children in all survey years. With respect to income, there was a significant decline in heavy SSB intake among both lower-income (9.6% to 3.3%; P for trend < 0.001) and higher-income children (11.5% to 3.3%; P for trend < 0.001). When stratified by sex, the proportion for heavy SSB intake decreased significantly for both female (7.1% to 2.8%; P for trend < 0.001) and male children (14.5% to 3.7%; P for trend < 0.001), although male respondents had higher levels of heavy SSB intake across all years compared with female respondents.
      Table 2Trends in prevalence of heavy sugar-sweetened beverage (SSB) intake (≥500 kcal/day) between 2003-2004 and 2015-2016 for children and adults participating in National Health and Nutrition Examination Survey, by race/ethnicity, income status, and sex
      Characteristic
      To obtain trend estimates, separate models were fitted within each subgroup, adjusting for all other covariates (eg, model was fit among non-Hispanic White children, adjusting for survey year, sex, age category, and income). Negative predicted values were truncated at 0.
      Study YearP value for linear trend
      2003-20042005-20062007-20082009-20102011-20122013-20142015-2016
      % (95% CI)
      Childrenn = 3,654n = 3,851n = 2,878n = 3,000n = 2,907n = 2,825n = 2,668
      Race/ethnicity
      Non-Hispanic White12.1 (9.4- 14.8)11.1 (8.8- 13.4)8.7 (5.6- 11.9)6.7 (5.0- 8.5)6.9 (4.9- 9.0)7.2 (5.2- 9.1)3.7 (2.5- 4.9)< 0.001
      Non-Hispanic Black10.9 (8.4- 13.5)8.8 (6.9- 10.7)5.4 (3.5- 7.3)6.2 (4.4- 8.0)5.6 (4.4- 6.8)4.2 (1.2- 7.2)3.3 (0.9- 5.6)< 0.001
      Mexican American10.7 (8.4-13.0)6.7 (4.8- 8.5)4.4 (1.6- 7.1)7.5 (4.7- 10.3)3.7 (1.3- 6.1)2.3 (0.9- 3.7)2.5 (1.1- 3.9)< 0.001
      Non-Mexican Hispanic8.0 (2.0- 14.0)6.4 (2.4- 10.4)9.5 (6.6- 12.5)4.8 (2.6- 7.0)5 (1.8-8.1)6.8 (3.6- 10.0)2.8 (0.6- 5.1)0.028
      Other race/ethnicity3.4 (0.0-7.7)3.4 (0.9- 5.9)2.2 (0.5- 3.8)4.5 (1.0- 8.1)1.9 (0.0-4.0)1.9 (0.2- 3.6)1.8 (0.0- 3.5)0.194
      Income
      Lower9.6 (7.6- 11.5)9.2 (6.8- 11.5)8.6 (5.3- 11.9)7.4 (5.2- 9.6)7.7 (5.0- 10.3)5.9 (3.5- 8.3)3.3 (1.7- 4.9)< 0.001
      Higher11.5 (8.8-14.1)9.3 (7.4- 11.2)6.6 (4.9- 8.3)6.0 (4.8- 7.2)4.6 (3.3- 5.8)5.4 (3.9- 6.9)3.3 (2.4- 4.1)< 0.001
      Sex
      Female7.1 (5.5-8.8)5.6 (3.7- 7.5)5.4 (2.8- 8.0)3.6 (2.4- 4.8)4.1 (2.3- 5.9)3.2 (1.5- 5.0)2.8 (1.8- 3.7)< 0.001
      Male14.5 (11.8-17.1)12.8 (10.7- 15.0)9.2 (7.1- 11.3)9.2 (8- 10.4)7.4 (5.6- 9.1)7.7 (6.0- 9.5)3.7 (2.8- 4.6)< 0.001
      Adultsn = 4,211n = 4,325n = 4,934n = 5,228n = 4,434n = 4,686n = 4,537
      Race/ethnicity
      Non-Hispanic White12.1 (9.8- 14.4)9.7 (8.1- 11.3)10.7 (7.0- 14.3)8.5 (7.0- 10.0)8.6 (7.5- 9.6)8.7 (6.9- 10.4)9.9 (8.1- 11.7)0.036
      Non-Hispanic Black17.6 (13.5- 21.8)15.2 (11.6- 18.8)12.6 (10.3- 14.9)11.4 (9.3- 13.5)12.7 (10.6- 14.8)12.0 (9.6- 14.3)9.1 (6.1- 12.1)< 0.001
      Mexican American15.8 (12.0- 19.7)14.6 (11.7- 17.6)9.8 (6.5- 13.2)11.5 (9.4- 13.6)11.1 (7.4- 14.7)13.5 (11.3- 15.7)8.4 (5.9- 10.9)0.009
      Non-Mexican Hispanic7.5 (0.4- 14.7)9.7 (5.4- 13.9)11.6 (8.5- 14.6)11.9 (8.8- 15.0)9.8 (5.5- 14.0)10.4 (6.1- 14.6)9.1 (6.8- 11.3)0.969
      Other race/ethnicity8.3 (2.2- 14.5)7.7 (3.1- 12.4)3.5 (1.6- 5.3)4.6 (1.5- 7.6)9.0 (4.9- 13.0)9.3 (4.5- 14.2)3.9 (2.0- 5.9)Nonlinear
      Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0398).
      Income
      Lower-income defined as family income <130% of the Federal Poverty Level. Higher-income is defined as family income ≥130% of the Federal Poverty Level.
      Lower16.7 (13.3- 20.0)16.0 (11.8- 20.2)14.8 (10.3- 19.4)13.9 (12.2- 15.6)13.7 (11.5- 15.8)16.6 (13.3- 19.9)11.7 (9.2- 14.1)0.122
      Higher11.5 (9.4- 13.6)9.1 (7.8- 10.4)9.2 (6.9- 11.5)7.6 (6.3- 8.8)8.1 (6.8- 9.5)7.5 (6.0- 8.9)8.3 (7.3- 9.4)0.001
      Sex
      Female7.8 (6.1-9.5)6.3 (5.0- 7.6)7.0 (4.8- 9.3)6.8 (5.7- 7.9)13.9 (12.2- 15.6)6.5 (4.8- 8.3)5.9 (4.5- 7.3)0.174
      Male18.1 (15.6- 20.6)15.2 (12.4-17.9)14.1 (10.7- 17.6)11.2 (9.4- 13.1)12.4 (11.1- 13.8)13.0 (10.9- 15.1)12.5 (10.7- 14.4)Nonlinear
      Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0314).
      a To obtain trend estimates, separate models were fitted within each subgroup, adjusting for all other covariates (eg, model was fit among non-Hispanic White children, adjusting for survey year, sex, age category, and income). Negative predicted values were truncated at 0.
      b Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0398).
      c Lower-income defined as family income <130% of the Federal Poverty Level. Higher-income is defined as family income ≥130% of the Federal Poverty Level.
      d Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0314).
      Among adults, trends in heavy SSB intake were not consistent across age groups. The proportion of heavy SSB consumers declined significantly among adults aged 20 to 39 years (20.8% to 9.9%; P for trend < 0.001), remained relatively constant among 40- to 59-year olds (12.0% to 11.5%; P for trend = 0.767), and increased significantly among ≥60-year olds (1.8% to 4.0%; P for trend = 0.007). With respect to race/ethnicity, there was a significant decline in heavy SSB intake among non-Hispanic Whites (12.1% to 9.9%; P for trend = 0.036), non-Hispanic Blacks (17.6% to 9.1%; P for trend < 0.001), and Mexican Americans (15.8% to 8.4%; P for trend = 0.009). The proportion of heavy SSB consumers remained relatively constant among non-Mexican Hispanic adults (7.5% to 9.1%; P for trend = 0.969). For adults of other race/ethnicities, there was evidence of a nonlinear trend in heavy SSB intake over time, although this group generally had lower proportions of heavy SSB consumers than the adult population as a whole. When stratified by income, heavy SSB intake declined significantly among higher-income adults (11.5% to 8.3%; P = 0.001). The proportion of heavy SSB drinkers declined among lower-income adults, but not significantly (16.7% to 11.7%; P = 0.122). When stratified by sex, the proportion for heavy SSB intake declined for women, but not significantly (7.8% to 5.9%; P for trend =0.174). For male responders, there was evidence of a nonlinear trend in heavy SSB intake over time; male respondents had higher levels of heavy SSB intake across all years compared with female respondents. Trends stratified by both sex and other sociodemographic characteristics are available in Table 3 (available at jandonline.org)
      In sensitivity analyses wherein heavy SSB intake was defined as consuming ≥25% of total energy intake from SSB, there was a significant increase in the proportion of heavy SSB drinkers over time among 40- to 59-year olds, whereas there was evidence of a nonlinear trend in heavy SSB intake over time among 2- to 5-year olds. Trends within other age groups remained similar.
      Table 4 reports differences in the prevalence of heavy SSB intake across subgroups in 2015-2016. Among children, 12- to 19-year olds had a significantly higher prevalence of heavy SSB intake compared with 2- to 5-year olds. Among adults, men had a significantly higher prevalence of SSB intake compared with women, ≥60-year olds had a significantly lower prevalence compared to 20- to 39-year olds, adults of other race/ethnicity had a significantly lower prevalence compared to non-Hispanic White adults, and higher-income adults had a significantly lower prevalence compared with lower-income adults.
      Table 4Differences in prevalence of heavy sugar-sweetened beverage (SSB) intake (≥500 kcal/day) across sociodemographic groups among National Health and Nutrition Examination Survey participants, 2015-2016
      Demographic characteristic
      Estimates reported here are slightly different than the estimates reported in Table 1 for 2015-2016. This is because different models were used to estimate these results, with the Table 1 estimates coming from a model incorporating all years of data and thus borrowing data across years to improve the fit, whereas the estimates for this table come from a model including only 2015-2016 data. Negative predicted values were truncated at 0.
      Proportion of heavy SSB drinkers
      % (95% CI)
      Children
      Sex
      Female2.6 (1.6-3.7)
      Male3.6 (2.6-4.5)
      Age category (y)
      2-50.4 (0-1.0)
      6-111.6 (0.8-2.5)
      12-195.5 (4.3-6.7)
      Indicates statistically significant difference (P < 0.05) in proportion of heavy SSB drinkers compared with reference group. The reference group for age category was 2 to 5 years for children and 20 to 39 years for adults. The reference group for race/ethnicity was non-Hispanic White for both children and adults.
      Race/ethnicity
      Non-Hispanic White3.5 (2.0-5.0)
      Non-Hispanic Black3.2 (0.8-5.7)
      Mexican American2.5 (0.8-4.2)
      Non-Mexican Hispanic3.0 (0.3-5.6)
      Other race/ethnicity1.9 (0-3.7)
      Income
      Lower3.3 (1.4-5.2)
      Higher3.0 (2.1-3.9)
      Adults
      Sex
      Female5.6 (4.2-6.9)
      Male12.1 (10.2-14)
      Indicates statistically significant difference (P < 0.05) in proportion of heavy SSB drinkers compared with reference group. The reference group for age category was 2 to 5 years for children and 20 to 39 years for adults. The reference group for race/ethnicity was non-Hispanic White for both children and adults.
      Age category (y)
      20-399.6 (7.5-11.8)
      40-5911.4 (9.4-13.5)
      ≥603.9 (2.4-5.3)
      Indicates statistically significant difference (P < 0.05) in proportion of heavy SSB drinkers compared with reference group. The reference group for age category was 2 to 5 years for children and 20 to 39 years for adults. The reference group for race/ethnicity was non-Hispanic White for both children and adults.
      Race/ethnicity
      Non-Hispanic White9.9 (8-11.9)
      Non-Hispanic Black8.4 (5.6-11.2)
      Mexican American6.7 (4.2-9.3)
      Non-Mexican Hispanic7.8 (5.1-10.4)
      Other race/ethnicity3.6 (1.7-5.5)
      Indicates statistically significant difference (P < 0.05) in proportion of heavy SSB drinkers compared with reference group. The reference group for age category was 2 to 5 years for children and 20 to 39 years for adults. The reference group for race/ethnicity was non-Hispanic White for both children and adults.
      Income
      Lower11.8 (8.9-14.6)
      Higher8.0 (6.9-9.0)
      Indicates statistically significant difference (P < 0.05) in proportion of heavy SSB drinkers compared with reference group. The reference group for age category was 2 to 5 years for children and 20 to 39 years for adults. The reference group for race/ethnicity was non-Hispanic White for both children and adults.
      a Estimates reported here are slightly different than the estimates reported in Table 1 for 2015-2016. This is because different models were used to estimate these results, with the Table 1 estimates coming from a model incorporating all years of data and thus borrowing data across years to improve the fit, whereas the estimates for this table come from a model including only 2015-2016 data. Negative predicted values were truncated at 0.
      b Indicates statistically significant difference (P < 0.05) in proportion of heavy SSB drinkers compared with reference group. The reference group for age category was 2 to 5 years for children and 20 to 39 years for adults. The reference group for race/ethnicity was non-Hispanic White for both children and adults.
      There were no significant changes over time in the overall per capita energy intake from SSB among heavy SSB drinkers for children (735 kcal to 701 kcal; P for trend = 0.788) or adults (772 kcal to 796 kcal; P for trend = 0.102) (Figure 3 and see also Table 5, available at www.jandonline.org). For children, per capita energy intake of soda and fruit drinks decreased significantly between 2003-2004 and 2015-2016 (soda: 444 kcal to 303 kcal; P for trend = 0.001; fruit drinks: 202 kcal to 101 kcal; P for trend < 0.001), whereas per capita energy intake of other SSB increased significantly 64 kcal to 254 kcal; P for trend < 0.001). Low-calorie SSB contributed minimal amounts of energy in all survey years, changing from 0 kcal in 2003-2004 to 3 kcal in 2015-2016. Per capita energy intake of energy/sports drinks by heavy SSB consumers did not change significantly (26 kcal to 40 kcal; P for trend = 0.199).
      Figure thumbnail gr2a
      Figure 3Per capita intake of sugar-sweetened beverage (SSB) calories among heavy SSB drinkers (≥500 kcal/day) between 2003-2004 and 2015-2016 for children and adults participating in National Health and Nutrition Examination Survey. There were no significant changes between 2003-2004 and 2015-2016 in the overall per capita energy intake from SSBs among heavy SSB drinkers for children (735 kcal to 701 kcal; P for trend = 0.788) or adults (772 kcal to 796 kcal; P for trend = 0.102). (A) For children, per capita energy intake of soda (ie, sweetened carbonated beverage) and fruit drinks decreased significantly between 2003-2004 and 2015-2016, whereas per capita energy intake of other SSBs increased significantly. Low-calorie SSBs contributed minimal amounts of energy in all survey years, whereas per capita energy intake of energy/sports drinks by heavy SSB consumers did not change significantly. (B) For adults, per capita energy intake of soda decreased significantly between 2003-2004 and 2015-2016, whereas per capita energy intake of energy/sports drinks and other SSBs increased significantly. There was evidence of a significant nonlinear trend over time for per capita energy intake of fruit drinks. Low-calorie SSBs contributed nominal amounts of energy intake in all survey years. Negative predicted values were truncated at 0.
      Figure thumbnail gr2b
      Figure 3Per capita intake of sugar-sweetened beverage (SSB) calories among heavy SSB drinkers (≥500 kcal/day) between 2003-2004 and 2015-2016 for children and adults participating in National Health and Nutrition Examination Survey. There were no significant changes between 2003-2004 and 2015-2016 in the overall per capita energy intake from SSBs among heavy SSB drinkers for children (735 kcal to 701 kcal; P for trend = 0.788) or adults (772 kcal to 796 kcal; P for trend = 0.102). (A) For children, per capita energy intake of soda (ie, sweetened carbonated beverage) and fruit drinks decreased significantly between 2003-2004 and 2015-2016, whereas per capita energy intake of other SSBs increased significantly. Low-calorie SSBs contributed minimal amounts of energy in all survey years, whereas per capita energy intake of energy/sports drinks by heavy SSB consumers did not change significantly. (B) For adults, per capita energy intake of soda decreased significantly between 2003-2004 and 2015-2016, whereas per capita energy intake of energy/sports drinks and other SSBs increased significantly. There was evidence of a significant nonlinear trend over time for per capita energy intake of fruit drinks. Low-calorie SSBs contributed nominal amounts of energy intake in all survey years. Negative predicted values were truncated at 0.
      For adults, per capita energy intake of soda decreased significantly between 2003-2004 and 2015-2016 (483 kcal to 364 kcal; P for trend < 0.001), whereas per capita energy intake of energy/sports drinks and other SSB increased significantly (energy/sports drinks: 13 kcal to 67 kcal; P for trend < 0.001; other SSB: 146 kcal to 327 kcal; P < 0.001). There was evidence of a significant nonlinear trend over time for per capita energy intake of fruit drinks (119 kcal to 35 kcal), which were relatively constant from 2003-2004 to 2011-2012 and then declined substantially in the 2013-2014 and 2015-2016 surveys. Low-calorie SSB contributed nominal amounts of energy intake in all survey years, changing from 11 kcal in 2003-2004 to 3 kcal in 2015-2016.

      Patterns in Current Heavy SSB Intake

      For both children and adults, most energy intake from SSB consumed by heavy SSB drinkers was acquired from stores (64% for children, 74% for adults), followed by restaurants (22% for children, 17% for adults), and other locations (14% for children, 9% for adults). About half of all energy intake from SSB was consumed at home compared to away from home (46% for children, 58% for adults).

      Discussion

      The overall prevalence of heavy SSB intake declined significantly between 2003-2004 and 2015-2016 for both children and adults. For children, these declines were relatively consistent across age group, race/ethnicity, sex, and income status. For adults, the story was less clear. Whereas heavy SSB intake declined among 20- to 39-year olds and most race/ethnicities, there was no significant change in heavy SSB intake among 40- to 59-year olds and non-Mexican Hispanic adults, and an increase in intake among older adults (≥60 years). In the most recent years of the data (2015-2016), non-Hispanic White adults, male adults, and adults aged 20 to 59 years had the highest overall levels of heavy SSB intake. Overall, the findings of this study suggest that although there have been promising declines in heavy SSB intake, attention must be paid to certain subgroups with high intake for whom trends are not decreasing, particularly 40- to 59-year olds and non-Mexican Hispanic adults.
      The trends in heavy SSB intake reported in this study are similar to previously documented declines in the proportion of the total population consuming SSB,
      • Bleich S.N.
      • Vercammen K.A.
      • Koma J.W.
      • Li Z.
      Trends in beverage consumption among children and adults, 2003-2014.
      suggesting that heavy SSB intake appears to be dropping in parallel to mean SSB intake. Moreover, our findings confirm the declining trend in heavy SSB intake among adolescents and young adults first documented by Han and Powell
      • Han E.
      • Powell L.M.
      Consumption patterns of sugar-sweetened beverages in the United States.
      using data from 1999-2008. However, the trend among children appears to have changed over time: from 1999-2008, heavy SSB intake among children was rising, but our more recent data show a promising decline. There are many possible reasons why heavy SSB intake has declined over the past decade. Unfortunately, evaluations of SSB reduction strategies rarely examine effects among heavy SSB drinkers alone, so inferences regarding which strategies have been most effective for decreasing heavy SSB intake are limited.
      More recently, nutrition interventions have shifted toward policy, systems, and environment strategies that aim to make the healthy choice, the easy choice and rely less on individual behavior change. One example is SSB excise taxes
      • Falbe J.
      • Thompson H.R.
      • Becker C.M.
      • Rojas N.
      • McCulloch C.E.
      • Madsen K.A.
      Impact of the Berkeley excise tax on sugar-sweetened beverage consumption.
      • Colchero M.A.
      • Salgado J.C.
      • Unar-Munguía M.
      • Molina M.
      • Ng S.
      • Rivera-Dommarco J.A.
      Changes in prices after an excise tax to sweetened sugar beverages was implemented in Mexico: Evidence from urban areas.
      • Colchero M.
      • Guerrero-López C.M.
      • Molina M.
      • Rivera J.A.
      Beverages sales in Mexico before and after implementation of a sugar sweetened beverage tax.
      • Colchero M.A.
      • Molina M.
      • Guerrero-López C.M.
      After Mexico implemented a tax, purchases of sugar-sweetened beverages decreased and water increased: Difference by place of residence, household composition, and income level.
      that are currently implemented in seven US cities and the Navajo Nation and appear to significantly reduce purchases and intake of SSB.
      • Lee M.M.
      • Falbe J.
      • Schillinger D.
      • Basu S.
      • McCulloch C.E.
      • Madsen K.A.
      Sugar-sweetened beverage consumption 3 years after the Berkeley, California, sugar-sweetened beverage tax.
      • Roberto C.A.
      • Lawman H.G.
      • LeVasseur M.T.
      • et al.
      Association of a beverage tax on sugar-sweetened and artificially sweetened beverages with changes in beverage prices and sales at chain retailers in a large urban setting.
      Healthy Food America
      Taxing sugary drinks.
      Another example is healthy kids’ meals policies, which have been passed by many US states and cities and require restaurants to only offer healthy drinks (eg, 100% juice, milk, or water) with children’s restaurant meals instead of SSB.
      California Legislative Information. Senate Bill No. 1192: Children’s meals.
      These and other SSB reduction policies have attracted wide media coverage of the role of SSB in driving obesity and other negative health outcomes.
      • Koplan J.P.
      • Liverman C.T.
      • Kraak V.I.
      Preventing childhood obesity: health in the balance: Executive summary.
      Thus, these policies and the awareness they have generated may be driving some of the declines seen in recent years of the data.
      Our results generally suggest that as people age, the trend in SSB intake flattens (or even increases, as among the oldest adults). This may be attributable to shifts in the cohort of people comprising each age group over time. For example, many individuals who would have been in the 40- to 59-year age group during 2003-2004 would be in the ≥60-year age group during 2015-2016. This generation of older adults would have grown up during the 1960s to1990s, a time when the food environment was becoming increasingly obesogenic (ie, greater availability of ultraprocessed foods and beverages and increased marketing to children and adolescents). Compared with their predecessors, who would have come of age during the 1920s to 1960s, this generation of older adults may be more likely to have developed heavy SSB intake habits, which could explain the increasing trend among this age group over time. Similarly, the number of federal, state, and local policies and campaigns aimed at reducing SSB intake has increased since the early 2000s, particularly in venues serving children and adolescents, such as schools and early child education and care. This may explain lower and declining heavy SSB intake levels seen in younger generations of children, adolescents, and young adults. In other words, differences in trends across age groups may reflect the changing cohort of people in each age group and their exposure to predominant societal norms and health promotion efforts during childhood and adolescence.
      Although our results suggest that heavy SSB intake is declining overall, there is still a need for further efforts to reduce excessive SSB intake in the United States. This study highlights several important elements that should be incorporated into future efforts. First, our study’s finding that per capita energy intake of soda has declined while intake of other SSB has increased among heavy SSB drinkers for both children and adults indicates the growing popularity of nontraditional SSB. These results are consistent with previous research documenting the rising number of beverages available to consumers at chain restaurants, with much of this increase in beverage offerings driven by nontraditional SSB like sweetened coffees, teas, and blended dairy-based beverages (although primarily dairy-based beverages were not included in our definition of SSB).
      • Frelier J.M.
      • Moran A.J.
      • Vercammen K.A.
      • Jarlenski M.P.
      • Bleich S.N.
      Trends in calories and nutrients of beverages in US chain restaurants, 2012-2017.
      The growing popularity of nontraditional SSB may be due in part to consumer’s perceptions that these beverages are healthier alternatives to traditional SSB like soda, a notion that may be driven by marketing of nontraditional SSB using nutrition-related health claims.
      • Harris J.
      • Schwartz M.B.
      • LoDolce M.
      • et al.
      Sugary Drink FACTS 2014: Some Progress but Much Room for Improvement in Marketing to Youth.
      Overall, this suggests that that SSB reduction strategies must incorporate a greater awareness of the types of SSB being consumed and should include wide SSB coverage to ensure success.
      Next, consistent with a past study,
      • Ogden C.L.
      • Kit B.K.
      • Carroll M.D.
      • Park S.
      Consumption of sugar drinks in the United States, 2005-2008.
      we found that stores and restaurants were the most common source for SSB among both children and adults. This suggests that future efforts must continue to focus on these settings. In addition to SSB taxes discussed above, another store-based SSB reduction strategy could be restricting SSB from purchase with Supplemental Nutrition Assistance Program (SNAP) benefits.
      • Cuffey J.
      • Beatty T.K.
      • Harnack L.
      The potential impact of Supplemental Nutrition Assistance Program (SNAP) restrictions on expenditures: A systematic review.
      Evidence from simulation studies indicates that restricting SSB purchases in SNAP could significantly reduce SSB intake among SNAP participants by a daily average of 24 kcal/person.
      • Basu S.
      • Seligman H.K.
      • Gardner C.
      • Bhattacharya J.
      Ending SNAP subsidies for sugar-sweetened beverages could reduce obesity and type 2 diabetes.
      Moreover, given the reach of SNAP (one in seven Americans participate) and the fact that more than $4 billion SNAP dollars are estimated to be spent on SSB annually,
      US Department of Agriculture, Food and Nutrition Service
      Characteristics of Supplemental Nutrition Assistance Program households: Fiscal year 2017.
      ,
      • Franckle R.L.
      • Moran A.
      • Hou T.
      • et al.
      Transactions at a Northeastern supermarket chain: Differences by Supplemental Nutrition Assistance Program use.
      this approach could have a substantial population-level influence. However, these benefits should be weighed in light of the equity and ethical concerns that come with restricting SNAP benefits.
      • Barnhill A.
      Impact and ethics of excluding sweetened beverages from the SNAP program.
      In addition to healthy kids’ meals policies discussed above, another restaurant-based SSB reduction strategy is to reduce the portion size of SSB either by making default serving sizes smaller or reducing the availability of larger portion sizes. For example, New York City’s Sugary Drinks Portion Cap Rule, which was passed in 2013 and later repealed, prohibited the sale of sugary drinks larger than 16 fl oz in restaurants and similar settings. When implemented widely, these types of strategies could be used to continue and accelerate the declines in heavy SSB intake observed in this study.
      Our study was also able to examine the extent to which declining trends in heavy SSB intake are occurring among people who are disproportionately exposed to SSB marketing (ie, racial/ethnic minorities and low-income populations). Consistent with a previous study,
      • Mendez M.A.
      • Miles D.R.
      • Poti J.M.
      • Sotres-Alvarez D.
      • Popkin B.M.
      Persistent disparities over time in the distribution of sugar-sweetened beverage intake among children in the United States.
      declines in heavy SSB intake were similar across racial/ethnic groups among children. Among adults, heavy SSB intake declined among non-Hispanic Blacks, non-Hispanic Whites, and Mexican Americans, with the largest percentage point decline observed among non-Hispanic Blacks. With respect to income, significant declines were observed for both lower- and higher-income groups among children; among adults, significant declines were only observed among higher-income adults. Taken together, the findings for children and adults may have important implications for reducing health inequities, given the strong associations between SSB intake and negative health outcomes,
      • Malik V.S.
      • Li Y.
      • Pan A.
      • et al.
      Long-term consumption of sugar-sweetened and artificially sweetened beverages and risk of mortality in US adults.
      • Malik V.S.
      • Popkin B.M.
      • Bray G.A.
      • Després J.P.
      • Willett W.C.
      • Hu F.B.
      Sugar-sweetened beverages and risk of metabolic syndrome and type 2 diabetes: A meta-analysis.
      • Malik V.S.
      • Schulze M.B.
      • Hu F.B.
      Intake of sugar-sweetened beverages and weight gain: A systematic review.
      as well as the inordinate burden of diet-related diseases among low-income individuals and racial/ethnic minorities.
      • Kurian A.K.
      • Cardarelli K.M.
      Racial and ethnic differences in cardiovascular disease risk factors: A systematic review.
      ,
      • Ogden C.L.
      • Carroll M.D.
      • Fryar C.D.
      • Flegal K.M.
      Prevalence of obesity among adults and youth: United States, 2011–2014.
      Although most of the results are promising, the prevalence of heavy SSB intake remained relatively constant among non-Mexican Hispanic adults, suggesting that more targeted efforts may be needed within this population to reduce heavy intake.
      This study has a number of limitations. First, because NHANES is a cross-sectional study, making causal inferences about the relationship between sociodemographic characteristics and trends in heavy intake of SSB limited. Moreover, because NHANES is national data released every 2 years, results from this study cannot be definitively tied to any single policy at the local or state level (eg, SSB taxes). Rather, our findings are suggestive that SSB reduction strategies are generally working but cannot identify a single strategy that has led to the observed declines. Second, as discussed in the Methods, the use of a single 24-hour dietary means that the proportion of heavy SSB drinkers is overestimated.
      • Willett W.
      Nutritional Epidemiology.
      However, comparison of our results to the NCI Method, which estimates usual intake, indicates that the overall trends identified by the primary analysis hold for the NCI method as well. Next, because the 24-hour dietary recall for children younger than age 12 years was completed or assisted by primary caregivers, SSB intake may be underestimated in the case that children consume beverages without their caregiver’s knowledge (eg, at daycare settings). In particular, this may mean that the proportion of beverages reported to be consumed outside the home and/or acquired by children from school or daycare settings may be underestimated. Despite these limitations, this study has a number of strengths, including using the most recently available nationally representative data, reporting on both children and adults, and using a systematic beverage coding scheme.

      Conclusions

      Heavy SSB intake has declined in the US population overall, but attention must be paid to certain subgroups with high intake for whom trends are not decreasing, particularly 40- to 59-year olds and non-Mexican Hispanic adults.

      Supplementary Materials

      Figure 1Coding Scheme to categorize beverages reported by National Health and Nutrition Examination Survey (NHANES) participants in 24-hour dietary recall into sugar-sweetened beverage (SSB) subtypes for analysis.
      SSB categoryDefinition
      SodaCarbonated beverage with added sugar; not identified as diet or low-calorie.
      Fruit drinksFruit drinks, fruit juice, and fruit nectars with added sugar; not identified as diet or low-calorie; does not include 100% fruit juices.
      Sports/energy drinksEnergy drinks, sports drinks, and thirst quenchers; not identified as diet or low-calorie.
      Low-calorie SSBAny beverage listed as having added sugar in FPED
      FPED = Food Patterns Equivalents Database.
      that is additionally identified as low-calorie through terminology such as “low-calorie,” “reduced calorie,” or “light”; drinks labeled as “diet” but with >5 kcal were categorized as low-calorie.
      Other SSBBeverage listed as having added sugar that is not categorized as soda, fruit drinks and punches, sports and energy drinks, or low-calorie SSBs; “other” beverage categories include sweetened coffees and teas, sweetened nonalcoholic drinks (eg, nonalcoholic malt beverage), and sweetened waters.

      Other SSBs also include beverage combinations with ≥2 nonwater beverages or any combination of at least one beverage with food (eg, hot chocolate with marshmallow).
      a FPED = Food Patterns Equivalents Database.
      Table 3Trends in prevalence of heavy sugar-sweetened beverage (SSB) intake (≥500 kcal/day) between 2003-2004 and 2015-2016 for children and adults participating in National Health and Nutrition Examination Survey, stratified by sex and other sociodemographic characteristics
      Sociodemographic characteristic
      To obtain trend estimates, separate models were fitted among each subgroup, adjusting for all other covariates (eg, model was fit among non-Hispanic White girls, adjusting for survey year, age category, and income). Results are not reported for children and adults of other race/ethnicity due to small sample sizes. Negative predicted values were truncated at 0. NA indicates non-estimable due to small sample sizes.
      Survey YearP value for linear trend
      2003-20042005-20062007-20082009-20102011-20122013-20142015-2016
      % (95% CI)
      Children
      Male
      Age group (y)
      2-54.7 (1.5-7.9)0.9 (0.2- 1.5)2 (0-4.8)1.2 (0.2-2.2)1.5 (0-3.3)0.6 (0-1.6)NA0.003
      6-1111.1 (4.6-17.7)5.3 (1.1-9.5)7.3 (4.3-10.3)2.6 (0.7-4.5)4.6 (1.9-7.3)4.6 (2.4- 6.8)2.1 (1.1-3.1)0.001
      12-1921.8 (17.4-26.1)23.8 (20.2-27.5)14.1 (9.4-18.8)18.3 (14.9-21.6)12.2 (8.7-15.6)13.5 (8.9-18)6.7 (4.4-9)< 0.001
      Race/ethnicity
      Non-Hispanic White16.6 (13.1-20.1)15.3 (11.9-18.7)10.5 (7.5-13.5)9.7 (7.7-11.7)8.2 (6-10.4)10.1 (7.6-12.6)4.8 (2.6-6.9)< 0.001
      Non-Hispanic Black13.9 (10.2-17.6)12.4 (9.5-15.3)6.8 (2.8-10.8)7.5 (5-10)6.6 (4.8-8.4)4.4 (0-8.9)2.2 (1.0- 3.4)< 0.001
      Mexican American16.0 (12.4-19.6)7.6 (4.9-10.3)5.8 (1.5-10)11 (5.8-16.3)5.8 (0.9-10.7)2.9 (0.9-4.9)2.6 (0.9-4.3)< 0.001
      Non-Mexican Hispanic4.5 (0-9.4)8 (0.1-15.9)13.5 (7.3-19.7)8.4 (3.8-13.1)7 (1.2-12.9)9.4 (5-13.8)2.6 (0-5.1)Nonlinear
      Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0357).
      Income
      Lower12.8 (10.4-15.3)12.2 (9.5-15)9.7 (6-13.4)11.7 (8.1-15.2)9.6 (5.4-13.7)6.6 (3.8-9.4)2.8 (1.1-4.4)Nonlinear
      Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0206).
      Higher15.2 (11.4-19)13.1 (10.4-15.7)8.8 (6.1-11.5)8.3 (6.5-10.2)5.8 (3.6-8)8.4 (6-10.7)4.1 (2.9-5.4)< 0.001
      Female
      Age group (y)
      2-53.2 (0-6.5)1.2 (0.1-2.2)2.1 (0.6-3.7)0.5 (0.1-0.9)0.3 (0-0.8)NA0.7 (0-1.9)0.014
      6-113.2 (1.8-4.6)2.3 (0.4-4.3)3.9 (1.5-6.3)1.4 (0.3-2.5)1.1 (0.3-2)1.1 (0.2-2)1.2 (0.2-2.3)0.001
      12-1911.8 (8.2-15.5)10 (6.8-13.2)8.1 (3.8-12.4)6.7 (4.1-9.3)8.1 (4.6-11.7)6.4 (3.1-9.8)4.9 (3.3-6.4)0.001
      Race/ethnicity
      Non-Hispanic White7.3 (4.8-9.9)6.5 (3.6-9.5)6.8 (2.8-10.9)3.6 (1.4-5.8)5.5 (2.1-8.9)3.9 (1.3-6.6)2.5 (1.6-3.4)0.001
      Non-Hispanic Black8 (5.7-10.2)5.4 (2.9-7.9)4 (2.4-5.6)4.9 (1.9-8)4.6 (2.7-6.5)3.9 (0-8.2)4.5 (0.6-8.4)0.173
      Mexican American5.2 (3.6-6.9)5.8 (3.3-8.2)2.9 (0-6.4)3.7 (1.3-6)1.3 (0.1-2.6)1.7 (0.1-3.2)2.3 (0.6-4.1)0.002
      Non-Mexican Hispanic13.3 (3.3-23.3)4.8 (2.1-7.5)5.5 (1.7-9.3)1 (0-2)2.7 (0.1-5.4)3.8 (0-7.8)3.1 (0-6.7)0.061
      Income
      Lower6.4 (3.7-9)6.3 (3.7-9)7.7 (2.6-12.7)3.9 (2.7-5)5.8 (3.1-8.5)5.2 (1.5-8.9)3.7 (0.7-6.8)0.180
      Higher7.5 (5.6-9.5)5.2 (2.9-7.5)4.1 (1.5-6.7)3.4 (1.7-5.2)3.2 (1.3-5.1)2.1 (1-3.1)2.2 (0.8-3.7)< 0.001
      Adults
      Male
      Age group (y)
      20-3929.7 (23.1-36.4)24.1 (19.4-28.7)21.1 (16.1-26.1)17 (13.3-20.6)17.6 (14.6-20.7)19.4 (15.9-22.9)14 (10.6-17.5)< 0.001
      40-5916.5 (12.3-20.7)12.7 (7.8-17.6)14.2 (9.4-19)10.5 (7.5-13.5)11.8 (8.8-14.7)11.9 (8.7-15.1)16 (11.9-20.1)Nonlinear
      Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0380).
      >601.4 (0.3-2.4)4.4 (2.7-6)2.5 (1.6-3.4)3 (1.5-4.5)4.9 (2.4-7.4)4.2 (2.4-6)4.8 (1.2-8.3)0.047
      Race/ethnicity
      Non-Hispanic White17.7 (14.4-21)13.9 (11.4-16.4)14.5 (9.8-19.3)11 (8.8-13.1)11 (9.4-12.7)11.4 (9-13.9)13.8 (11.3-16.3)Nonlinear
      Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0076).
      Non-Hispanic Black20.3 (15.3-25.2)21.1 (15.4-26.8)16.6 (12.6-20.6)13.1 (9.4-16.7)15.8 (12.5-19)14 (10.5-17.5)12.7 (6.6-18.8)0.009
      Mexican American21.7 (15.7-27.8)23.4 (16.5-30.4)13.8 (9.4-18.2)12.3 (8.2-16.4)14.7 (9-20.4)21.8 (17.2-26.4)10.1 (5.4-14.8)0.022
      Non-Mexican Hispanic11.9 (0-26.6)13.8 (5.3-22.2)17 (11.3-22.8)17.6 (14.2-21)15.1 (7.8-22.3)13.1 (6.5-19.6)14.3 (10.5-18.1)0.927
      Income
      Lower24.7 (18.8-30.7)23.4 (17.5-29.4)19.4 (14-24.8)15.8 (11.9-19.8)17.2 (13.2-21.1)23.6 (17.9-29.3)15.6 (10.9-20.4)0.089
      Higher16.3 (13.7-18.9)13.2 (10.8-15.6)12.8 (9.6-16.1)10.1 (8.3-12)11.2 (9.5-12.9)10.1 (8.1-12.1)11.8 (10.1-13.4)Nonlinear
      Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0212).
      Female
      Age group (y)
      20-3912 (9.4-14.5)9.5 (6.5-12.6)11.2 (7.5-15)11 (8.1-13.9)10.3 (7.4-13.2)9 (6.6-11.3)5.9 (4.2-7.5)Nonlinear
      Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0382).
      40-597.8 (3.9-11.7)6.5 (5-8)6.6 (3.8-9.4)5.6 (3.5-7.6)7.1 (4.1-10.2)7.1 (3.6-10.5)7.4 (5.2-9.7)0.900
      >602.1 (1.3-2.9)1.6 (0.3-2.8)1.8 (1-2.5)2.7 (1.4-4)0.7 (0-1.4)2.3 (0.7-3.8)3.4 (1.3-5.4)0.260
      Race/ethnicity
      Non-Hispanic White7 (5.1-8.8)5.8 (4.3-7.3)7.1 (4.1-10)6.2 (4.8-7.6)6.3 (4.5-8.1)6.1 (3.8-8.4)6.2 (3.9-8.6)0.668
      Non-Hispanic Black15.5 (11.1-19.8)10.4 (7.4-13.4)9.4 (6.5-12.4)9.9 (7.4-12.4)10.3 (6.8-13.8)10.4 (7-13.8)6.2 (3.7-8.7)0.005
      Mexican American9.6 (3.6-15.5)5.2 (1.6-8.9)5.6 (3.1-8.2)10.6 (7-14.2)7.3 (2.9-11.6)4.6 (1.9-7.2)6.5 (5.1-7.8)0.343
      Non-Mexican Hispanic3.8 (-1.2-8.8)6.2 (0.6-11.8)6.6 (2.7-10.6)7 (3.1-10.9)5 (2-7.9)7.8 (2.5-13.2)4.4 (2.6-6.3)0.953
      Income
      Lower10.6 (7.9-13.3)10.1 (5.7-14.5)11.3 (6.9-15.7)12.2 (9-15.4)11.1 (8.9-13.3)11.2 (8.4-13.9)8.6 (4.4-12.8)0.672
      Higher6.9 (4.8-8.9)5.1 (4-6.3)5.6 (3.8-7.4)5.1 (3.9-6.3)5.1 (3.4-6.9)5 (3-6.9)5 (3.7-6.3)0.171
      a To obtain trend estimates, separate models were fitted among each subgroup, adjusting for all other covariates (eg, model was fit among non-Hispanic White girls, adjusting for survey year, age category, and income). Results are not reported for children and adults of other race/ethnicity due to small sample sizes. Negative predicted values were truncated at 0. NA indicates non-estimable due to small sample sizes.
      b Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0357).
      c Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0206).
      d Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0380).
      e Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0076).
      f Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0212).
      g Evidence of a nonlinear trend in heavy SSB intake over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0382).
      Table 5Per capita intake (kilocalories) by sugar-sweetened beverage (SSB) subtype among heavy SSB drinkers (≥500 kcal/day) between 2003-2004 and 2015-2016 for children and adults participating in the National Health and Nutrition Examination Survey. This table provides point estimates and 95% confidence intervals for the results presented in Figure 3
      Variable
      Negative predicted values were truncated at 0.
      SodaFruit drinksEnergy/sport drinksLow calorie SSBOther SSB
      point estimate (95% CI)
      Children
      2003-2004444 (408-479)202 (173-232)26 (7-44)0 (0-1)64 (38-90)
      2005-2006389 (293-485)192 (150-235)77 (56-97)1 (0-3)109 (57-162)
      2007-2008387 (341-433)131 (91-171)72 (32-113)6 (0-15)95 (57-133)
      2009-2010429 (328-530)166 (111-222)48 (18-78)16 (4-29)161 (66-256)
      2011-2012232 (141-323)204 (150-257)55 (28-83)6 (0-11)226 (17-435)
      2013-2014352 (237-467)78 (48-108)96 (25-167)9 (1-17)249 (122-376)
      2015-2016303 (233-373)101 (42-160)40 (15-64)3 (0-7)254 (150-358)
      P value for linear trend0.001< 0.0010.199Nonlinear
      Evidence of a nonlinear trend over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0162).
      < 0.001
      Adults
      2003-2004483 (430-537)119 (83-154)13 (1-24)11 (0-22)146 (118-173)
      2005-2006429 (383-475)113 (91-135)30 (18-42)2 (0-3)187 (130-244)
      2007-2008422 (348-496)97 (78-116)53 (32-74)5 (2-8)178 (135-220)
      2009-2010395 (359-432)125 (104-145)30 (15-44)7 (4-9)229 (186-272)
      2011-2012312 (264-360)132 (98-165)60 (29-91)12 (3-22)268 (224-312)
      2013-2014386 (338-434)53 (35-71)55 (35-75)15 (3-28)294 (250-337)
      2015-2016364 (301-426)35 (24-47)67 (42-93)3 (0-8)327 (271-382)
      P value for linear trend< 0.001Nonlinear
      Evidence of a nonlinear trend over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0001).
      < 0.001Nonlinear
      Evidence of a nonlinear trend over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P =0.0153).
      < 0.001
      a Negative predicted values were truncated at 0.
      b Evidence of a nonlinear trend over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0162).
      c Evidence of a nonlinear trend over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P = 0.0001).
      d Evidence of a nonlinear trend over time, as indicated by statistically significant joint Wald test of the quadratic and cubic terms for survey year (P =0.0153).

      References

        • Mesirow M.S.
        • Welsh J.A.
        Changing beverage consumption patterns have resulted in fewer liquid calories in the diets of US children: National Health and Nutrition Examination Survey 2001-2010.
        J Acad Nutr Diet. 2015; 115: 559-566
        • Welsh J.A.
        • Sharma A.J.
        • Grellinger L.
        • Vos M.B.
        Consumption of added sugars is decreasing in the United States.
        Am J Clin Nutr. 2011; 94: 726-734
        • Bleich S.N.
        • Vercammen K.A.
        • Koma J.W.
        • Li Z.
        Trends in beverage consumption among children and adults, 2003-2014.
        Obesity. 2018; 26: 432-441
        • Malik V.S.
        • Li Y.
        • Pan A.
        • et al.
        Long-term consumption of sugar-sweetened and artificially sweetened beverages and risk of mortality in US adults.
        Circulation. 2019; 139: 2113-2125
        • Malik V.S.
        • Popkin B.M.
        • Bray G.A.
        • Després J.P.
        • Willett W.C.
        • Hu F.B.
        Sugar-sweetened beverages and risk of metabolic syndrome and type 2 diabetes: A meta-analysis.
        Diabetes Care. 2010; 33: 2477-2483
        • Malik V.S.
        • Schulze M.B.
        • Hu F.B.
        Intake of sugar-sweetened beverages and weight gain: A systematic review.
        Am J Clin Nutr. 2006; 84: 274-288
        • Han E.
        • Powell L.M.
        Consumption patterns of sugar-sweetened beverages in the United States.
        J Acad Nutr Diet. 2013; 113: 43-53
        • Mendez M.A.
        • Miles D.R.
        • Poti J.M.
        • Sotres-Alvarez D.
        • Popkin B.M.
        Persistent disparities over time in the distribution of sugar-sweetened beverage intake among children in the United States.
        Am J Clin Nutr. 2018; 109: 79-89
        • Healthy Food America
        Sugary beverages: Map and chart the movement.
        http://www.healthyfoodamerica.org/map
        Date accessed: September 22, 2019
      1. California Legislative Information. Senate Bill No. 1192: Children’s meals.
        • Taylor K.
        People are drinking less Pepsi and Coke than ever—and it reveals the power of the ‘biggest marketing trick of the century.’.
        • Ogden C.L.
        • Kit B.K.
        • Carroll M.D.
        • Park S.
        Consumption of sugar drinks in the United States, 2005-2008.
        NCHS Data Brief. 2011; 71: 1-8
        • National Center for Health Statistics
        NHANES survey methods and analytic guidelines.
        • Wang Y.C.
        • Bleich S.N.
        • Gortmaker S.L.
        Increasing caloric contribution from sugar-sweetened beverages and 100% fruit juices among US children and adolescents, 1988-2004.
        Pediatrics. 2008; 121: e1604-e1614
        • Park S.
        • Blanck H.M.
        • Sherry B.
        • Brener N.
        • O'Toole T.
        Factors associated with sugar-sweetened beverage intake among United States high school students.
        J Nutr. 2012; 142: 306-312
        • White A.H.
        • James S.A.
        • Paulson S.W.
        • Beebe L.A.
        Sugar sweetened beverage consumption among adults with children in the home.
        Front Nutr. 2018; 5: 1-8
        • Willett W.
        Nutritional Epidemiology.
        Oxford University Press, New York, NY2012
        • Tooze J.A.
        • Midthune D.
        • Dodd K.W.
        • et al.
        A new statistical method for estimating the usual intake of episodically consumed foods with application to their distribution.
        J Acad Nutr Diet. 2006; 106: 1575-1587
        • Barrera C.M.
        • Moore L.V.
        • Perrine C.G.
        • Hamner H.C.
        Number of eating occasions and source of foods and drinks among young children in the United States: NHANES, 2009–2014.
        Nutrients. 2019; 11: 897
      2. Stata [computer program]. Version 16.0. College Station, TX: StataCorp; 2019.

        • Ingram D.D.
        • Malec D.J.
        • Makuc D.M.
        • et al.
        National Center for Health Statistics guidelines for analysis of trends: Data evaluation and methods research.
        https://stacks.cdc.gov/view/cdc/53672
        Date accessed: August 7, 2020
        • Falbe J.
        • Thompson H.R.
        • Becker C.M.
        • Rojas N.
        • McCulloch C.E.
        • Madsen K.A.
        Impact of the Berkeley excise tax on sugar-sweetened beverage consumption.
        Am J Public Health. 2016; 106: 1865-1871
        • Colchero M.A.
        • Salgado J.C.
        • Unar-Munguía M.
        • Molina M.
        • Ng S.
        • Rivera-Dommarco J.A.
        Changes in prices after an excise tax to sweetened sugar beverages was implemented in Mexico: Evidence from urban areas.
        PloS ONE. 2015; 10e0144408
        • Colchero M.
        • Guerrero-López C.M.
        • Molina M.
        • Rivera J.A.
        Beverages sales in Mexico before and after implementation of a sugar sweetened beverage tax.
        PloS ONE. 2016; 11e0163463
        • Colchero M.A.
        • Molina M.
        • Guerrero-López C.M.
        After Mexico implemented a tax, purchases of sugar-sweetened beverages decreased and water increased: Difference by place of residence, household composition, and income level.
        J Nutr. 2017; 147: 1552-1557
        • Lee M.M.
        • Falbe J.
        • Schillinger D.
        • Basu S.
        • McCulloch C.E.
        • Madsen K.A.
        Sugar-sweetened beverage consumption 3 years after the Berkeley, California, sugar-sweetened beverage tax.
        Am J Public Health. 2019; 109: 637-639
        • Roberto C.A.
        • Lawman H.G.
        • LeVasseur M.T.
        • et al.
        Association of a beverage tax on sugar-sweetened and artificially sweetened beverages with changes in beverage prices and sales at chain retailers in a large urban setting.
        JAMA. 2019; 321: 1799-1810
        • Healthy Food America
        Taxing sugary drinks.
        • Koplan J.P.
        • Liverman C.T.
        • Kraak V.I.
        Preventing childhood obesity: health in the balance: Executive summary.
        J Acad Nutr Diet. 2005; 105: 131-138
        • Frelier J.M.
        • Moran A.J.
        • Vercammen K.A.
        • Jarlenski M.P.
        • Bleich S.N.
        Trends in calories and nutrients of beverages in US chain restaurants, 2012-2017.
        Am J Prev Med. 2019; 57: 231-240
        • Harris J.
        • Schwartz M.B.
        • LoDolce M.
        • et al.
        Sugary Drink FACTS 2014: Some Progress but Much Room for Improvement in Marketing to Youth.
        Rudd Center for Food Policy and Obesity, New Haven, CT2014
        • Cuffey J.
        • Beatty T.K.
        • Harnack L.
        The potential impact of Supplemental Nutrition Assistance Program (SNAP) restrictions on expenditures: A systematic review.
        Public Health Nutr. 2016; 19: 3216-3231
        • Basu S.
        • Seligman H.K.
        • Gardner C.
        • Bhattacharya J.
        Ending SNAP subsidies for sugar-sweetened beverages could reduce obesity and type 2 diabetes.
        Health Aff. 2014; 33: 1032-1039
        • US Department of Agriculture, Food and Nutrition Service
        Characteristics of Supplemental Nutrition Assistance Program households: Fiscal year 2017.
        • Franckle R.L.
        • Moran A.
        • Hou T.
        • et al.
        Transactions at a Northeastern supermarket chain: Differences by Supplemental Nutrition Assistance Program use.
        Am J Prev Med. 2017; 53: e131-e138
        • Barnhill A.
        Impact and ethics of excluding sweetened beverages from the SNAP program.
        Am J Public Health. 2011; 101: 2037-2043
        • Kurian A.K.
        • Cardarelli K.M.
        Racial and ethnic differences in cardiovascular disease risk factors: A systematic review.
        Ethn Dis. 2007; 17: 143-152
        • Ogden C.L.
        • Carroll M.D.
        • Fryar C.D.
        • Flegal K.M.
        Prevalence of obesity among adults and youth: United States, 2011–2014.
        NCHS Data Brief. 2015; 219: 1-8

      Biography

      K. A. Vercammen is a doctoral degree candidate, Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA.
      L. Kennedy-Shaffer is an assistant professor, Department of Mathematics and Statistics, Vassar College, Poughkeepsie, NY; at the time of the study, he was a doctoral degree candidate, Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA.
      M. J. Soto is a programmer, Department of Health Policy and Management, Harvard T.H. Chan School of Public Health, Boston, MA.
      A. Moran is an assistant professor, Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD.
      S. N. Bleich is a professor, Department of Health Policy and Management, Harvard T. H. Chan School of Public Health, Boston, MA.