Assessing the dietary sources of caffeine among US children and adults has implications for public health policy. Caffeine, a mild stimulant [1
], is a frequently consumed dietary constituent, which is an added ingredient or is naturally present in varying amounts in both beverages and solid foods [3
]. Caffeine is recognized by the US Food and Drug Administration as safe for adults in amounts < 400 mg/day [8
]. The 2015 Dietary Guidelines Advisory Committee (DGAC) report concluded that there was only limited evidence to link high-dose caffeine intakes with health risks for children and adults [9
]. However, whereas the high dose for adults was set at 400 mg/day, the dose for children was not determined [9
]. Some concerns remain about the consumption of energy drinks by children and teenagers, particularly due to their high caffeine content, but also due to their energy content [10
Most dietary caffeine is provided by beverages [3
], with foods such as chocolate providing relatively small amounts. Coffee contains more caffeine than do most beverages and the amounts consumed depend on age [3
]. Adults drink coffee, but young children generally do not [3
]. In contrast, caffeinated energy drinks are most often consumed by teenagers and young adults [2
]. The consumption of carbonated sugar sweetened beverages (SSBs) begins in childhood, peaks in young adult life, and declines thereafter [14
]. As a result, the chief sources of dietary caffeine show sharp differences by age [3
]. To complicate matters, age-driven patterns of beverage consumption are subject to secular trends [3
]. Whereas the consumption of energy drinks has increased in recent years, the consumption of caffeinated sugar-sweetened beverages has, to the contrary, declined [6
Based on recent analyses of 24-h recalls from the nationally representative NHANES 1999–2010 database, 70% of children and adolescents (aged < 22 years) consumed some caffeine on any given day [11
]. However, no net increases in caffeine intakes were observed during that time [11
]. Analyses of seven-day food diaries from the Kantar Worldpanel [3
] reported that 85% of the US population aged ≥ two years consumed at least one caffeinated beverage per day. Mean caffeine intake across all age groups was estimated at 165 mg/day, largely driven by coffee consumption.
The present analyses were based on seven consecutive NHANES cycles (1999–2012) with a total of 56,207 respondents aged ≥ four years. First, detailed analyses of caffeine sources by beverage type and age group were based on the most recent 2011–2012 NHANES cycle (n = 7456). First, the contribution of different beverages and foods to caffeine consumption for each age group was assessed both in terms of absolute amounts (mg/day) and in terms of percent contribution to total intakes. Second, we examined fourteen-year trends in total caffeine consumption, overall and by beverage type. Lastly, analyses of time trends by purchase location for children and adults were based on NHANES 2003–2012. By including both children and adults, the present analyses complement and extend earlier reports to identify the major sources of caffeine by beverage type and purchase location.
2.1. Dietary Intake Data
The current study was based on 7 cycles of the nationally representative National Health and Nutrition Examination Survey (NHANES) corresponding to years 1999–2012 [16
]. Data from the most recent cycle (2011–2012) for 2655 children (4–19 years) and 4801 adults (≥20 years) were used to characterize caffeine consumption overall and beverage type, using the most current nationally representative dietary data. For analysis of trends from 1999–2000 through 2011–2012, data for 22,276 children and 33,931 adults was used.
Caffeine consumption data were based on the first 24-h dietary recall, conducted in person by trained interviewers [20
]. The interview probed for all foods and beverages consumed from midnight-to-midnight in the preceding 24 h. A second-day telephone recall was available; however, consistent with other studies [11
], this study used the first recall only so that the data were comparable across all survey cycles.
For the NHANES dietary recall, recalls for children aged 4–5 years were completed by a parent/guardian. For children aged 6–11 years, the child was the primary respondent, but the parent was present and able to assist. Children aged 12–19 years were the primary respondents but could be assisted by an adult. The necessary ethics approval for NHANES had been obtained by the National Center for Health Statistics [19
] and all data used was publicly available.
Dietary recalls were transformed into energy and nutrient intakes using the US Department of Agriculture Food and Nutrient Database for Dietary Studies (FNDDS) which contains nutrient composition data, including caffeine, for all foods and beverages consumed by participants [21
]. FNDDS includes more than 7000 individual foods and beverages, including approximately 50 coffee/coffee drinks, 30 types of tea, and caffeinated and decaffeinated versions of soda. The database is updated every two-years to account for changes in the food supply and reformulation of foods. The depth of FNDDS permits the in-depth disaggregation of foods/beverages.
2.2. Classification of Caffeine Sources
For analyses of caffeine consumption by beverage type we adapted the What We Eat in America Food Categories, developed by USDA [22
]. Beverages were first classified into 15 broad categories and those that included caffeine-containing beverages were flagged. The revised categories included coffee, tea, soda (i.e.
, carbonated SSBs or low-calorie beverages), energy drinks, milk and other beverages. The other beverages included alcohol and meal replacement beverages, which contributed small amounts of caffeine, and fruit juices, fruit drinks, sports drinks, vegetable juice, water and flavored waters, which contained no caffeine. Caffeine in milk was from modest amounts of caffeine in chocolate milk. As some caffeine also comes from solid foods, namely products containing cocoa, we created a foods category consisting of all non-beverage items. As the present study focused on dietary sources of caffeine, data from dietary supplements were not included.
2.3. Purchase Locations of Origin
The NHANES database includes information on the locations of origin (e.g., where the food was purchased or otherwise obtained). Overall, the primary locations were supermarkets or other food stores, fast food restaurants (FFR, which includes pizza takeout/delivery and takeaway coffee stores), full service restaurants (FSRs), and schools [23
]. The other locations of origin included someone else (i.e.
, gifts), common pot/common source (e.g., shared coffee pot or fruit basket at work, herein referred to as “common pot”), vending machines, other types of cafeterias (e.g., workplace), tavern/bar, or sporting/cultural/entertainment events [20
]. The store category did not separate grocery stores, supermarkets, convenience stores, pharmacies and specialty food stores. For this study, the primary purchase locations were limited to grocery stores, FFRs, FSRs, common coffee pot, and a combined “other” category. This information was first collected in the 2003–2004 NHANES cycle, and has been collected in every subsequent cycle, allowing for the evaluation of trends in the origin of caffeine from 2003 to 2012.
2.4. Statistical Analysis
Descriptive analyses examined caffeine consumption using NHANES 2011–2012 data, overall, and by age group (4–8 years, 9–13 years, 14–19 years, 20–34 years, 35–49 years, 50–70 years, ≥71 years), sex, race/ethnicity (non-Hispanic white, non-Hispanic black, non-Hispanic Asian, Mexican-American, and other Hispanic), family income-to-poverty ratio (<1.3 (generally eligible for Supplemental Nutrition Assistance Program (SNAP) and Women Infant’s and Children (WIC) program), 1.3–1.849 (eligible for WIC, not SNAP), 1.85–2.99 and ≥3), educational attainment (<high school, high school graduate, some college and college graduate or higher), and employment status (employed vs. not employed). Family income-to-poverty ratio represents the ratio of family income to the federal poverty guidelines. For example, in 2012 for a family of four, the poverty threshold was $23,050. Pairwise comparisons using a survey-weighted t-test were made comparing caffeine consumption at each variable level to a reference group. Analyses by education and employment status were not conducted for children/adolescents. Using the 2011–2012 data we also examined the absolute and proportional amount of caffeine from seven sources (i.e., coffee, tea, soda, energy drinks, milk, other beverages and foods), overall and by age groups.
Overall trends in caffeine consumption, and by type of beverage, were evaluated for children and adults separately. Additional analyses evaluated heterogeneity in trends by age group, which was evaluated by fitting a model with an interaction term for age group and survey cycle. In the evaluation of overall trends in caffeine consumption, limited sensitivity analyses were conducted by removing the few respondents who consumed >1000 mg/day of caffeine (>10.5 cups of coffee) to determine if extreme values could explain any observed trends. We also evaluated whether changing demographics in the US from 1999 to 2012 could explain caffeine consumption trends, by including age and race/ethnicity as covariates in the linear regression model.
From 2003 to 2004 onward, trends in caffeine consumption by location of origin were evaluated for children and adults, separately. Locations of origin evaluated included stores, fast food restaurants/coffee shops, full-service restaurants, common coffee pot, and other sources. Trends in each source of caffeine were evaluated using a survey-weighted linear regression with survey cycle/year as a continuous variable. Data from five cycles including location of origin data (e.g., 2003–2004 through 2011–2012) were pooled to provide a summary estimate of sources of caffeine over the study period. Additional analyses evaluated trends in caffeine consumption stratified by location of origin (e.g., stores or fast food/coffee restaurant) and type of beverage (e.g., coffee, tea, or soda), with trends evaluated in a similar manner.
All analyses accounted for the complex survey-design of NHANES data and were conducted using Stata 13.1 (College Station, TX, USA, 2013).
The present analyses of current caffeine consumption patterns and time trends between 1999 and 2012 add to the growing literature on the dietary sources of caffeine in the US [3
]. The present analyses, based on the most recent 2011–2012 NHANES, confirm the downward trend in caffeine intakes from caffeine containing beverages and foods. Whereas one recent study based on NHANES 2001–2010 [24
], estimated caffeine consumption at 211 mg/day for men and 161 mg/day for women, the present estimates based on the most recent NHANES 2011–2012 data were 196 mg/day for men and 151 mg/day for women. For children and adolescents (ages 4–19 years), caffeine intakes were estimated at 35 mg/day. The bulk of caffeine was supplied by store-bought beverages, particularly coffee, tea and soda. The observed decline in caffeine intakes was driven largely by a decline in soda consumption, particularly by teenagers and young adults.
Previous consumption estimates, based on 37,603 persons aged ≥ two years in the Kantar Worldpanel 2010–2011 [3
] were 164 mg/day caffeine, with 64% coming from coffee and 17% each from tea and sodas. Energy drinks accounted for 2.0% of total caffeine. The present estimates for persons > four years in 2011–2012 were 135 mg/day caffeine, with 66% coming from coffee, 18.5% from tea, 15.6% from soda, and 2.0% from energy drinks. The Kantar Worldpanel was based on seven day diaries for 37,602 consumers of caffeine (out of 42,851) collected from October 2010 through September 2011. There was good agreement between the present data and the Kantar Worldpanel, even with substantial differences in the methodologies. Our US-based estimates are also consistent with reports from elsewhere. For example, in Germany coffee and tea were observed to be the primary sources of caffeine among adults, while tea and cola were the predominant source of caffeine among those 10–19 years [25
]. However, given the comparatively high consumption of soda, including both sugar-sweetened and low-calorie sweetened varieties, soda contributed greater amounts of caffeine in the US.
Analyses of NHANES consumption trends by age showed that total caffeine consumption increased several-fold across the lifespan, consistent with other reports [3
]. Whereas soda was the principal source of caffeine for children and adolescents, coffee was the major source among those aged > 20 years. The absolute amounts of caffeine need to be considered: 10 mg of caffeine from soda provided 38% of a total intake of 26 mg in the 9–13 years age group—but 20 mg of caffeine from soda provided only 9% of total intake of 213 mg in the 51–70 years age group.
Analyses of consumption patterns confirmed that the proportion of beverage consumers varied by both age and beverage type. Whereas the percentages of coffee and tea drinkers increased with age, soda consumption peaked in young adult life and then declined. In previous studies, a large proportion of children < 19 consumed some caffeine on a given day (71%–73%) [10
]. Peak frequency of energy drink consumption (10.3% in the Kantar data) was observed among teenagers [3
]. The percentage of respondents consuming soda ranged from 40% to 50% depending on age. Between 10% and 20% consumed tea, and less than 10% consumed coffee or energy drinks.
The present analyses of time trends across multiple NHANES cycles were consistent with past reports. First, coffee and tea now represent a greater proportion of caffeine intakes as soda consumption has declined across all age groups. Similarly, Branum [11
] showed that mean caffeine intakes have been stable but declined among young children. As the contribution of soda to caffeine consumption declines, the relative contribution of coffee and to a lesser extent energy drinks has increased. Because both low-calorie and sugar-sweetened sodas may contain caffeine, trends in both sugar-sweetened beverages and low-calorie sweetened beverages may influence intake. The decline in sugar-sweetened soda consumption in the US is well-documented, and there is some evidence of increases in consumption of low-calorie sweetened beverages [26
Other studies have come to similar conclusions. Recent data on caffeine intakes and caffeinated food and beverages intakes were presented in a FDA report with data from a consumer panel database and other surveys [7
]. An earlier study by Frary [27
], based on the Continuing Survey of Food Intakes by Individuals (CSFII), also estimated caffeine consumption from soda, tea and coffee, leading to similar conclusions. Kit et al.
] reported that the consumption of soda significantly declined, but the consumption of energy drinks has increased. These studies add to data collected in the late 1980s and early 1990s [1
The 2015 Dietary Guidelines for Americans Advisory Committee concluded that among no age group did caffeine intakes exceed safe levels [8
]. However, they also noted the lack of consensus regarding safe levels among children and adolescents [8
]. Therefore, ongoing and context-specific surveillance of caffeine intakes among children/adolescents should be a priority area for dietary surveillance. Understanding the beverage/food sources of caffeine is essential, but understanding the source of caffeine, as done here, is also critical. Identifying trends by food source (e.g., store, fast food/coffee shop) may have implications for the implementation of interventions should caffeine intakes become a concern.
A number of limitations should be noted. First, like any dietary study, some foods/beverages may be under-reported or portion sizes estimated with error. The under-reporting of some beverages, namely sugar-sweetened beverages, might result in under-estimates of caffeine consumption. Second, for each single food/beverage item reported only a single estimate of caffeine content was used. The dietary database used here contains more than 7000 foods/beverages with numerous options that differentiate decaffeinated and caffeinated beverages. For example, the database includes more than 50 varieties of coffee/coffee drinks, 30 types of tea and numerous caffeinated and decaffeinated versions of soda. Variability in caffeine content may lead to error; this is particularly a concern for brewed coffee and hot tea [25
]. Lastly, this analysis focused on dietary sources of caffeine, therefore caffeine from supplements were not included. Despite these limitations, the current data are the best suited to examine US trends in caffeine consumption in a large and nationally representative sample.