Sources of potential energy in the US diet are varied, complex, and continually expanding owing to new developments in food science and changes in eating behavior and meal patterns across the lifespan.1,2 The report of the 2015 US Dietary Guidelines Advisory Committee (DGAC) examined the sources of energy and sugar in the US diet using data from the National Health and Nutrition Examination Survey (NHANES), 2009-2010 cycle.3 The NHANES includes an in-person 24-hour dietary recall and a second telephone 24-hour dietary recall within the subsequent 10 days plus a 30-day computer-based recall of the frequency of intake of a diversity of food and beverages.4 Nutrient values for all foods and beverages identified through the NHANES interview can be found in the Food and Nutrient Database for Dietary Studies (FNDDS) prepared by the US Department of Agriculture (USDA).5 Added sugars have been defined in the FNDDS as "sugars, syrups, or caloric sweeteners that are added to foods during food processing or food manufacturing[horizontal ellipsis]fruit juice concentrates used in foods without further dilution are also considered as added sugars.^(p65) Total sugars in the diet include those sugars that are not added, such as "sugars naturally present in dairy and fruit"^(p65) plus added sugars.6

The 2015 Dietary Guidelines for Americans (DGA) recommended that individuals consume less than 10% of calories per day from added sugars to attain a healthy eating pattern.7 In 2016, to provide more information about added sugars and total sugars for the US population, the US Food and Drug Administration released the final rule on revised Nutrition Facts label requirements that includes listing the amount in grams of the added sugars included in a product as well as changing the listing of "sugars" on the food label to "Total Sugars" for clarity.8

Based on the NHANES, the 2015 DGAC reported that major energy intake in the US diet is confined to relatively few sources, with only 3 food categories accounting for 56% of all energy consumed: mixed dishes, snacks and sweets, and beverages, excluding 100% fruit juices and milk.3 The 2015 DGAC further reviewed 32 specific subcategories of foods and described that almost half (47.1%) of the total energy consumed by the US population in 2009-2010 originated from only 7 subcategories of foods: "burgers and sandwiches; desserts and sweet snacks; sugar-sweetened beverages (SSBs); rice, pasta, and grain-based, mixed dishes; chips, crackers and savory snacks; pizza; and meat, poultry and seafood mixed dishes."3^(p74) Sugar-sweetened beverages contributed 6.5% of total energy intake, whereas, overall, beverages were found to contribute 19%.

With the initiation of the upcoming 2020-2025 DGA cycle, we describe the relative source contributions of food and various categories of beverage consumption based on the latest NHANES data (2015-2016) for calories, total sugar, and added sugar among individuals in the United States 2 years or older. We also compare our findings to values from the 2015 DGAC report, which used NHANES data from 2009-2010.3 Beverage products and portion sizes are continually changing in the marketplace and in casual and fast-food restaurants.9,10 Understanding how the US population may or may not have changed their consumption patterns since the national information used by the 2015 DGAC can potentially provide a basis for modification of national dietary intake pattern guidance by the 2020 DGAC.

Sources of energy, total, and added sugars in NHANES 2015-2016 are compared with the 2015-2020 Dietary Guidelines Advisory Committee data based on NHANES 2009-2010.

MATERIALS AND METHODS

Study Population

The NHANES is a nationally representative cross-sectional survey administered to a sample of noninstitutionalized, civilian US residents by the National Center for Health Statistics, Centers for Disease Control and Prevention, to collect dietary data using a complex, stratified, multistage probability cluster sampling design.4 The NHANES is one of the best sources of national-level diet and health information and is designed to be representative of the US population. Mobile examination units travel throughout the United States and are continuously in the field with data released in 2-year cycles. Individual study volunteers are interviewed in their own homes and also experience free physical examinations including blood samples in the mobile examination units. Each year of the 2-year cycle of data collection, the National Center for Health Statistics divides the US counties into 15 groups, and 1 county, called a Primary Sampling Unit (PSU), is selected from each group. The PSUs are selected from frames of all US counties using the most recent census data with consideration of balance of the proportions of minority populations and other demographic variables. The identification of the PSUs is the first stage in the multistage sampling procedure, with stages 2 (county segments), 3 (households), and 4 (individuals) constituting the 4-stage sampling approach. Stages 1 and 2 are proportionally selected, whereas stages 3 and 4 involve some random selections. Two-year cycles are necessary to obtain a large enough sample to be representative of the US population. The NHANES also oversamples certain subgroups in the US population to increase the "[horizontal ellipsis]reliability and precision of estimates of health status indicators for these population subgroups" (https://www.cdc.gov/nchs/tutorials/nhanes/SurveyDesign/SampleDesign/Info1.htm). Ultimately, between 9 000 and 11 000 individuals from across the country participate in each NHANES 2-year cycle. The Centers for Disease Control and Prevention makes analyzed results that they have conducted, as well as the data from the NHANES, available free of charge on their website (cf, 2015-2016: https://wwwn.cdc.gov/nchs/nhanes/ContinuousNhanes/Default.aspx?BeginYear=2015).

NHANES Dietary Interview

In the 2009-2010 and 2015-2016 NHANES dietary recall, data on food and beverage items reported consumed in the first 24 hours (between midnight and midnight, 24 hours before interview) were collected in-person by the NHANES dietary interviewer using a computer-assisted dietary interview software program (USDA's automated multiple-pass method [AMPM]). This 5-step method was developed to help respondents better remember their food and beverage intake over the previous 24 hours with all of the information being computer recorded. Specifically, the interviewer first asks for a quick list of everything the respondent can remember having eaten or drank, followed by verbal probes about possible forgotten items. Then the specific time and eating occasion for the items are recorded, followed by details of the items, amounts, and any condiments that might have been added. The fifth step is a careful final verbal probe for possible forgotten items. The interviewer uses a standard booklet containing actual-size paper models of plates and glasses, as well as rubber life-like food models, to help the interviewee accurately estimate portion sizes (https://www.ars.usda.gov/northeast-area/beltsville-md-bhnrc/beltsville-human-nut). To obtain reported consumption information for children younger than 12 years, proxy respondents were used for those 5 years or younger and proxy-assisted interviews were carried out for 6- to 11-year-olds.11 Underreporting or overreporting of portion size of food and drink items has been cited as an issue for NHANES 24-hour recall data, and especially the use of 1-day data may result in an overall underestimation or overestimation of the nutrient intake of US children and adults.12,13 However, the misreporting of food and beverage items has been reduced through use of the USDA's AMPM.14,15 In addition, in this analysis, by using data to obtain an estimate of nutrient intakes from only a single 24-hour dietary recall, results should be viewed as representing intake on a typical single day, rather than a person's usual intake over multiple days. Finally, owing to the cross-sectional design of NHANES, it is not possible to evaluate potential temporal relationships between factors.

Food and Beverage Coding Definitions and Categories

Food and beverage items consumed in the dietary interview were coded using the USDA's FNDDS, and the nutrient content was estimated.16 The NHANES dietary intake data (2015-2016) over one 24-hour period (day 1) were combined with the USDA food patterns equivalents data.17

Foods, beverages, accompaniments, and other items reported consumed were coded as follows:

* Foods included any item consumed primarily on its own and thus could represent a major contributor to the daily volume, weight, or intake but exclude accompaniments and "other items."17 Foods thus included individual food items such as asparagus as well as mixed dishes and casseroles that serve as a major component of a meal or snack.

* Beverage intake was divided into mutually exclusive liquid beverage subcategories: SSBs; 100% fruit and vegetable juices; alcoholic beverages; sweetened coffee and tea; low and no calorie sweetened (LNCS) beverages; dairy-based beverages; and the few remaining other beverages, which included value-added waters and other remaining caloric beverages that did not fall into any previously stated subcategory. Sugar-sweetened beverages included soft drinks, sport drinks, energy drinks, and fruit drinks. For the analysis, beverages that were categorized as dairy-based beverages and other beverages were combined to better align with the beverage categorizations in the 2015 DGAC report.

* Accompaniments were defined as additions to either foods or beverages that contained no substantive amounts of calories, for example, low-calorie sweetener packets, condiments labeled as no calorie, and others.

* Other Items were defined as additions to either foods or beverages that included calories, such as sugar, chocolate sauce, regular salad dressing, ketchup, mustard, and others, except when sugar was added to plain coffee or tea, in which case it was categorized as sweetened coffee and tea.

* Whether or not a food or beverage item contains added sugar is easy to code because the USDA food and beverage database states the inclusion of added sweetener or sugar within the product description.

One author (B.P.M.) completed the coding of reported foods and beverages.* We combined the categories above in the figures below so that they would be comparable to the groupings used in the DGAC report.3

Analyses

The percentage of intake, on any given day, of calories, total sugars, or added sugars for individuals 2 years or older, 2 to 19 years old, and 20 years or older using the 2015-2016 NHANES cycle was calculated. These age groups were selected to be consistent with the data reported by the 2015 DGAC. Energy, total sugar, and added sugar consumed were summed for each individual per category. Sums were then divided by overall energy, total sugar, or added sugar, respectively, to yield a percentage intake. SAS version 9.4 was used to conduct all data management and analyses (SAS Institute, Cary, North Carolina). Because we did not have access to the raw data used by the 2015 DGAC team, we could not conduct a statistical comparison between the 2009-2010 pie chart information presented in their report and the 2015-2016 data we generated for this study. We did, however, perform statistical modeling to test for differences between children (2-19 years old) and adults (>=20 years old) in the 2015-2016 NHANES survey cycle data. SAS regression for survey data was used, P values are reported, and a P value of <.05 was considered statistically significant.

Complete 24-hour dietary intake data were available for 8505 participants from the 2015-2016 NHANES cycle. To provide data that were comparable to the NHANES 2009-2010 data used by the 2015-2016 DGAC, we excluded those individuals younger 2 years (n = 583), those with incomplete data (n = 5), and women who were pregnant or lactating (n = 98). The total sample for analysis included 7819 participants 2 years or older, of which 4919 were adults 20 years or older and 2099 were children 2 to 19 years old. The Medical University of South Carolina did not require institutional review board approval as a secondary data analysis was carried out and the study was not considered human subjects research.

RESULTS

Distribution of Energy Consumption From the Total Diet Among the US Population 2 Years or Older

Figure 1A presents the distribution of energy intake by individuals 2 years or older for foods, beverages, accompaniments, and other items as reported in the NHANES 2015-2016, whereas Figure 1B presents data from the 2015 DGAC report showing the distribution from the 2009-2010 NHANES for energy intake. Figure 1B illustrates that among individuals in the US population 2 years or older in 2015-2016, foods contributed 79% of total energy intake, whereas beverages contributed 16% of total energy intake. Energy intake from reported accompaniments and other items was 0% and 5%, respectively. Foods (79%) plus accompaniments (0%) and other items (5%) comprise what the DGAC analyses term "All Foods" and represent 84% of energy intake. Beverage subcategories that contributed substantially to the total energy intake included SSBs (5%), dairy-based beverages and other beverages (4%), and alcoholic beverages (4%). One-hundred percent fruit or vegetable juices, sweetened coffee and tea, and LNCS beverages each contributed 1% or less of the total energy intake.

FIGURE 1. (A) Distribution of energy consumption from the total diet among individuals 2 years or older based on NHANES 2015-2016. Foods included any item consumed primarily on its own and thus could represent a major contributor to the daily volume, weight, or intake but exclude accompaniments and "other items."

Comparison of the NHANES 2015-2016 analyses conducted here (Figure 1A) with the NHANES 2009-2010 DGAC data (Figure 1B) indicates that reported distribution of energy intake from 2009-2010 to 2015-2016 changed slightly among the US population, with the proportion of overall energy from all foods increasing by 3% (ie, from 81% to 84%), whereas the proportion of energy from beverages decreased by a corresponding 3%. The relative distribution of beverage energy sources in the US diet remained overall relatively constant or declined slightly over this period, that is, SSBs (7% to 5%), dairy-based beverages ([almost equal to]4% at both time points), alcoholic beverages ([almost equal to]4 at both time points% to 3.7%), 100% fruit and vegetable juices (2% to 1%), and sweetened coffees and teas ([almost equal to]1% at both time points).

Percentage of total energy from all foods increased by 3%, whereas that from beverages decreased by 3%.

Distribution of Total Sugar Consumption From the Total Diet Among the US Population 2 Years or Older

Figure 2 shows the distribution of reported total sugar consumption from the US diet by individuals 2 years or older for all foods, beverages, accompaniments, and other items based on dietary reporting in the NHANES 2015-2016. Among individuals in the US population 2 years or older, in 2015-2016, foods, beverages, accompaniments, and other items contributed 49%, 43%, 1%, and 7%, respectively, of total sugar intake. Among beverages, beverage subcategories that contributed substantially to overall total sugars intake included SSBs (21%), dairy-based beverages and other beverages (9%), 100% fruit or vegetable juices (5%), and sweetened coffee and tea (4%). Alcoholic beverages and LNCS beverages each contributed 2% or less to overall total sugar intake in the US diet.

FIGURE 2. Distribution of total sugar consumption from the total diet by category among individuals 2 years or older based on NHANES 2015-2016. Foods included any item consumed primarily on its own and thus could represent a major contributor to the daily volume, weight, or intake but exclude accompaniments and "other items."

Distribution of Added Sugar Consumption From the Total Diet Among the US Population 2 Years or Older

Figure 3 illustrates the distribution of added sugar sources in the US diet based on the NHANES 2015-2016 data (Figure 3A) and the NHANES 2009-2010 data from the 2015 DGAC report (Figure 3B). In constructing Figure 3A, we removed 100% fruit and vegetable juices as well as dairy-based beverages and other beverages from the overall beverage category and included them as separate wedges in the pie chart for comparison to the 2015 DGAC graphic (Figure 3B). Based on the 2015-2016 NHANES (as defined in Figure 3A), beverages contributed 43% of added sugar to the US diet. Other items that include condiments and related food components that are added to main foods such as relishes, separate sauces, and others, and that contain calories contributed 10% of added sugar in the US diet. Accompaniments (additions to food or beverages that contain low or no calories) were a source of 1% of added sugar. Among beverages, SSBs were the largest contributor of added sugar to the US diet (32%), followed by sweetened coffees and teas (7%) and alcoholic beverages (2%) (Figure 3A). The distribution of added sugar in the US diet from the 2009-2010 NHANES, as presented in the 2015 DGAC report (Figure 3B), indicates that beverages contributed 47%. The category identified as condiments, gravies, spreads, and salad dressings in the 2009-2010 analyses contributed a lower percentage of the added sugar (2%) than what we termed "other items" in our analysis of 2015-2016 NHANES, which was more inclusive (10%). Among beverages in the 2009-2010 NHANES analyses conducted by the 2015 DGAC, SSBs were also the largest beverage source of overall added sugar (39%), followed by sweetened coffees and teas (7%) and alcoholic beverages (1%).

FIGURE 3. (A) Distribution of added sugar consumption from the total diet by category among individuals 2 years or older based on NHANES 2015-2016. Foods included any item consumed primarily on its own and thus could represent a major contributor to the daily volume, weight, or intake but exclude accompaniments and "other items."

The Appendix provides the food and beverage sources of energy, total sugars, and added sugar intake for individuals 2 years or older to represent the total population similar to the 2015 DGAC report. For comparison, data on adults alone 20 years or older and children 2 to 19 years old are included. Comparing children with adults, a higher proportion of added sugar intake in children was from food (50.9% in children compared with 40.6% in adults, P < .01), with a lower proportion of added sugar intake coming from beverages (41.8%, children; 48.0%, adults; P < .05), accompaniments (0.5%, children; 0.6%, adults; P < .05), and other items (6.8%, children; 10.9%, adults; P < .05) in children compared with adults.

DISCUSSION

As reflected in a number of recent publications,18,19 the trends in the US diet continue to change, with significant increases in energy intake from snacking among children¹⁹² and significant declines in trends of SSB consumption among both children and adults.18,20 Earlier analyses based on comparison of the 2003-2004 and 2011-2012 NHANES data found that the estimated mean daily intake of added sugars had decreased substantially among both children and adults.21 Examination of the NHANES 2013-2014 dietary data found that 42% of Americans 2 years or older met the 2015-2020 DGA recommendation of limiting added sugars intake to less than 10% of daily calories.22

We sought to compare the 2009-2010 NHANES information cited by the 2015 DGAC on contribution of dietary sources of energy, total sugar, and added sugar with the most recent information presented here based on the 2015-2016 NHANES data. The proportion of energy intake from foods, other items, and accompaniments combined seems to have increased slightly in the 6-year period, whereas the proportion of energy intake from beverages in the total diet seems to have decreased. Among the beverage subcategories, the greatest percentage decline in proportion of total energy intake has been in intake of energy from SSBs. These data on the proportional decrease in SSB-based energy intake reinforce previously reported trends in SSB intake.18

We also assessed total sugar intake in the most recent 2015-2016 NHANES data and found that whereas food and related condiments and accompaniments contribute 57% of the total sugar in the US diet, among the beverage categories, SSBs contribute 21% of total dietary sugar. There were statistically significant decreases in total sugar intake from consumption of SSBs, soft drinks, all beverages, and the total diet between 2003-2004 and 2015-2016 among the age categories: 2 years or older, 2 to 19 years old, and 20 years or older.20 The relatively low contribution of other items, which included condiments and items added to food and beverages (7%), to the total sugar intake is not surprising given that our other recent research has shown that condiments are a key source of low-calorie sweetener in the US diet.23

Evaluating the dietary sources of added sugars between the 2009-2010 and 2015-2016 NHANES among individuals 2 years or older, we found that there has been a decrease in the percentage of added sugar in the US diet from beverages-as defined by the 2015 DGAC-from 47% to 43%, with this change primarily due to a decrease in the percentage of added sugar from SSBs. Percentage of added sugar from coffees and teas has remained relatively constant at roughly 7% and the percentage of added sugar from alcoholic beverages has increased from 1% to 2%.

The percentage of added sugars from beverages, mostly from sugar-sweetened beverages, has decreased in the US diet.

Although added sugar in the US diet is often linked to concerns about obesity,24-26 research on dairy product intake among children has further indicated the complexity of the issue. For example, Nicklas et al27 have underscored the importance of breakfast as a meal for the overall nutrition of children. Nicklas et al28 further identified that children who consumed flavored milks were more likely to meet calcium recommendations even if they selected flavored milk with added sugar compared with children who did not drink milk.

Limitations of this work arise from the cross-sectional nature of NHANES and include the inability to assess potential temporal relationships between factors. In addition, the overall intake by US adults and, therefore, nutrient levels may be underestimated or overestimated owing to the use of NHANES 24-hour recall data.12 However, our study used nationally representative databases containing comprehensive dietary intake information (NHANES and FNDDS), as well as the USDA's AMPM for collection of dietary intake data in both the NHANES 2009-2010 and 2015-2016 cycles, which minimizes misreporting of food and beverage items.14,15 A specific limitation of this comparison of relative intake between the 2009-2010 data from the 2015 DGAC and the data we generated from the NHANES 2015-2016 is that we could not conduct statistical analyses because we did not have access to the dataset used by the 2015 DGAC analysis team. Future studies could expand information on beverage portion sizes across NHANES cycles to further elucidate the methods of the change in consumption. In addition, assessment of the demographics of the apparent doubling in reported proportion of added sugar from alcoholic beverage intake might benefit public health messaging.

CONCLUSION

Not unexpectedly, the overall diet has changed over the 5-year review cycle of the DGA. Sources of energy, total sugar, and added sugar have shifted slightly. The proportion of energy intake from SSBs in the total diet seems to have decreased over the period examined. Among beverages, SSBs remain the main beverage subcategory source of total and added sugar among persons 2 years or older in the United States. These data underscore the findings of the 2015 DGAC that although small changes may occur in the relative proportions of food and beverage intake among the US population, almost half of the total energy in the US diet is confined to 7 subcategories of foods and beverages. Public health messaging that targets the dietary patterns related to these subcategories, including SSBs, seems to offer potential in the continuing US fight to reduce overweight and obesity and their related life-shortening health outcomes.

REFERENCES