Total Sugar-Sweetened Beverage Intake Among US Adults Was Lower When Measured Using a 1-Question Versus 4-Question Screener

Abstract

Purpose:

To compare the performance of a 1-question survey screener measuring total sugar-sweetened beverage (SSB) intake to a screener measuring SSB types separately using 4 questions.

Design:

Cross-sectional.

Setting:

Web-based 2014 SummerStyles survey.

Participants:

A total of 4167 US adults (≥18 years).

Measures:

Frequency of SSB intake measured using a 1-question screener was compared to frequency using a 4-question screener (regular soda, fruit drinks, sports/energy drinks, sweetened coffee/tea). SSB intake (number of time/day) was categorized as 0, >0 to <1, and ≥1 time/day; difference in mean intake was calculated between 4 questions versus 1.

Analysis:

Paired t tests were used, and agreement was evaluated using weighted κ and Lin’s concordance correlation coefficient (CCC).

Results:

Mean SSB intake was 1.7 (95% confidence interval [CI]: 1.65-1.79) times/day using 4 questions and 0.6 (95% CI: 0.56-0.62) times/day using 1 question (P < .001). Intake frequency based on 4 questions versus 1 was 16.0% versus 38.5% for 0 time/day, 15.6% versus 42.5% for >0 to <1 time/day, and 68.4% versus 18.9% for ≥1 time/day. There was fair agreement for the 3 SSB intake categories (κ: .27) and poor absolute agreement between the 2 continuous measures (Lin’s CCC: 0.31).

Conclusion:

Daily SSB intake was significantly lower using a 1-question screener versus a 4-question screener. Researchers should assess SSB types separately or consider that daily SSB intake is likely underestimated with 1 question.

Keywords

sugar-sweetened beverages measuring consumption screener questions food frequency questionnaire–style questions population surveillance

Purpose

Sugar-sweetened beverages (SSBs) are drinks with added caloric sweeteners such as sugars or high-fructose corn syrup. Sugar-sweetened beverages can include nondiet soft drinks/soda, fruit drinks that are not 100% fruit juice, sports drinks, energy drinks, and sweetened tea and coffee drinks.¹ Data from the National Health and Nutrition Examination Survey (NHANES) in 2014 showed that 49.3% of US adults consume 1 or more SSBs per day.² Sugar-sweetened beverages are the largest source of added sugars in the US diet,³ with adults consuming an average of 151 calories from SSBs on a given day.⁴ Frequent consumption of SSBs is an important public health concern, as daily SSB consumption has been linked to an increased risk of obesity, type 2 diabetes, hypertension, dental caries, and asthma.^5

-8

Given the adverse health consequences of daily SSB consumption, it is important to monitor SSB intake. The gold standard method of measuring usual intake of foods and beverages is to collect multiple food records or 24-hour recalls, in which respondents record, either prospectively (food record) or retrospectively (24-hour recall), all foods and beverages consumed in a 1-day period.⁹ This method, however, can be time-consuming for both the respondent and the data collector and is often prohibitively expensive for large-scale data collection.⁹ Due to costs and space limitations on surveys, it is common in SSB research and surveillance to inquire about consumption using screener questions. Screener questions are typically written in the style of food frequency questionnaires (FFQs), which ask respondents to estimate their usual intake of food or beverage items during a specified period of time.¹⁰ A previous study compared beverage intake reported as the number of times/day derived from a 7-question FFQ-style screener to beverage intake calculated as the number of servings/day derived from multiple 24-hour dietary recalls and found that screener questions are a suitable method for population-level surveillance of beverage intake, including SSBs.¹¹

Some screeners have multiple questions to inquire about different beverage types, while others use just 1 overall screener question, which asks about total consumption of all beverage types combined. However, the performance of a single screener question in assessing total consumption of SSBs is unknown. Therefore, this cross-sectional study aimed to compare the performance of a single screener question measuring total SSB intake to a screener with 4 questions, which measure the different SSB types separately, to determine whether a single question is sufficient to estimate the prevalence of daily SSB consumption among adults.

Methods

Design

Data from the summer wave of Porter Novelli’s (Washington, D.C.) 2014 Styles database were used. The Styles database is constructed from a series of web-based surveys conducted annually in the United States using GfK’s Knowledge Panel, which is established using address-based sampling methods. Respondents are included regardless of whether or not they have landline phones or Internet access, and if needed, households are provided with a laptop computer and access to the Internet. The panel is continuously replenished and maintains approximately 50 000 panelists. Styles surveys American consumers about a variety of topics, including knowledge, attitudes, and behaviors pertaining to health issues.

Sample

The summer wave of Styles is sent to a random sample of adults (≥18 years) who participated in the spring wave. The spring wave, which was completed by 6713 of the 11 018 adults who received the survey, had a response rate of 60.9%. The SummerStyles survey was sent during June to July 2014 to a random sample of 6159 adults (≥18 years) who previously completed the spring wave. The SummerStyles survey, which took approximately 36 minutes to complete, was returned by 4269 participants, for a response rate of 69.3%. The final analytic sample included 4167 respondents who had complete data for all SSB questions (n = 102, 2.4%, were excluded due to missing SSB data). The data were weighted to match the US Current Population Survey proportions for sex, age, household income, race/ethnicity, household size, education level, census region, metro status, and whether or not a respondent had Internet access prior to joining the panel. The US Centers for Disease Control and Prevention licensed the results of the 2014 SummerStyles survey from Porter Novelli, and analyses of these data were exempt from institutional review board approval because personal identifiers were not included in the data file.

Measures

SummerStyles has 2 FFQ-style screeners that ask about respondents’ consumption of SSBs during the past month. The survey has a screener with 4 individual questions, 1 for each of the beverage types: regular soda, sweetened coffee and tea, sports and energy drinks, and fruit drinks (Table 1). These questions were chosen to match the 4 questions that comprise the SSB screener in the National Health Interview Survey (NHIS), a nationally representative survey that administers a dietary module every 5 years and serves as an important source of data for surveillance of SSB intake among adults in the United States.¹² SummerStyles also contains a screener with 1 SSB question that asks about combined consumption of different beverage types over the last month (Table 1). This single question screener was placed at the end of the survey, separated by over 100 questions from the 4-question screener at the beginning of the survey, to reduce the likelihood of biasing respondents’ answers. For each of these 5 questions, participants are asked to rate the frequency of their consumption over the past month using the following response options: none, 1 to 6 times/week, 1 time/day, 2 times/day, 3 times/day, or ≥4 times/day. Response values were converted to the number of times/day that SSBs were consumed, with 1 to 6 times/week converted to 0.5 times/day (3.5 divided by 7) and ≥4 times/day converted to 4 times/day.

Table 1.

Four-Question Screener and 1-Question Screener for Assessing Sugar-Sweetened Beverage Intake: Questions, Response Options, Coding, and Intake Values—SummerStyles Survey, 2014.

Survey Questions	Response Options^a	Coding	Intake Value
Four sugar-sweetened beverage question screener
During the past month, how often did you drink regular soda or pop that contains sugar? Do not include diet soda. During the past month, how often did you drink coffee, including lattes, and tea, including bottled tea, that were sweetened with sugar or honey? Do not include drinks with things like Splenda or Equal. During the past month, how often did you drink sports and energy drinks such as Gatorade, Red Bull, and Vitamin water? During the past month, how often did you drink sweetened fruit drinks, such as Kool-Aid, cranberry, and lemonade? Include fruit drinks you made at home and added sugar to.	None 1 to 6 times/week 1 time/day 2 times/ day 3 times/day 4 or more times/day	A = 0 B = 3.5/7 C = 1 D = 2 E = 3 F = 4	0 times/day 0.5 times/day 1 time/day 2 times/day 3 times/day 4 times/day
One sugar-sweetened beverage question screener
5. During the past month, how many times did you drink sodas, fruit drinks, sports or energy drinks, and other sugar-sweetened drinks? Do not include 100% fruit juice, diet drinks, or artificially sweetened drinks.

^aResponse options shown applied to all 5 of the sugar-sweetened beverage questions in the survey.

For the 4-question screener, a composite consumption variable was then created by adding the responses to the individual SSB questions. The 4-question screener variable and the 1-question screener were categorized based on the number of times/day that SSBs were consumed. The categories were defined as 0 times/day, >0 to <1 time/day, and ≥1 times/day, to enable an assessment of daily consumption, which is consistent with other literature.^2,12,13 The frequency of SSB intake that was calculated based on the 4-question screener was then compared to the frequency calculated from the 1-question screener.

Frequencies were compared overall and by several sociodemographic factors: age (18-24 years, 25-44 years, 45-64 years, ≥65 years), sex, race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, non-Hispanic other), marital status (married/domestic partnership, not married), education (≤high school, some college, ≥bachelors), annual household income (≤$34 999, $35 000-$74 999, $75 000-$99 999, ≥$100 000), and weight status (has obesity, overweight, and underweight/normal weight, where the latter 2 categories were combined due to the low prevalence of underweight). Self-reported weight and height data were used to calculate body mass index (BMI; weight [kg]/height [m²]), and weight status was categorized as underweight/normal weight (BMI < 25 kg/m²), overweight (BMI 25 to <30 kg/m²), or has obesity (BMI ≥30 kg/m²).¹⁴

Analysis

The Rao-Scott χ² test was used to examine associations between categories of daily SSB consumption and several sociodemographic factors. The weighted κ statistic (ranging 0 to +1) was calculated, overall and within levels of the sociodemographic factors, to determine the strength of agreement between the 2 SSB screener types using the 3 categories of daily consumption. A κ of +1 indicates perfect agreement, whereas a κ of 0 indicates agreement equivalent to chance.¹⁵

The 2 screeners were also compared using a continuous variable, mean number of times/day, with comparisons made overall and by the specified sociodemographic factors. The mean difference in SSB consumption (times/day) between the 4-question screener and the 1-question screener was calculated, and paired t tests were used to determine whether there were significant differences in mean SSB intake using 4 questions versus 1 question, overall and by the sociodemographic factor levels. Independent samples t tests were used to determine whether the mean difference in SSB consumption between the 2 SSB screener types differed by levels of the sociodemographic factors. Lin’s concordance correlation coefficient (CCC) was calculated to measure the absolute agreement between daily SSB intake (as a continuous measure) using the 4-question screener and intake using the 1-question screener. Using Lin’s CCC, agreement is classified as 0.0 ≤ CCC < 0.4 (poor agreement), 0.4 ≤ CCC < 0.7 (moderate agreement), and 0.7 ≤ CCC ≤ 1.0 (good agreement).¹⁶ Analyses were conducted using Statistical Analysis Software (version 9.3, 2011; SAS Institute Inc, Cary, North Carolina) and were weighted to account for the survey design. Lin’s CCC was calculated in STATA 13.0 and was unweighted.

Results

Survey participants were 47.9% male and 52.1% female (Table 2). The majority of participants were non-Hispanic white (66.4%), married or in a domestic partnership (61.1%), and were between the ages of 25 and 44 years (34.2%) or 45 and 64 years (35.0%). Based on the 4-question screener, 16.0% of respondents consumed SSBs 0 times/day, 15.6% consumed SSBs >0 to <1 time/day, and 68.4% consumed SSBs ≥1 time/day. Based on the 1-question screener, 38.5% of respondents consumed SSBs 0 times/day, 42.5% consumed SSBs >0 to < 1 time/day, and 18.9% consumed SSBs ≥1 time/day. Using both screener types, significant differences were found in the frequency of daily SSB intake by age, sex, race/ethnicity, education, and annual household income (all P values <.05). The patterns in the distribution of daily SSB consumption categories across levels of the sociodemographic factors were similar to the overall distribution. The reporting of a lower daily SSB intake using 1 question versus 4 questions was largest among Hispanic respondents. Among Hispanics, there was a 57 percentage point difference in the prevalence of respondents who consumed SSBs ≥1 time/day using 4 questions versus 1 question, and this was the largest difference of any sociodemographic group. The overall weighted κ statistic was .27 and ranged from .18 among Hispanic respondents to .33 among respondents with obesity.

Table 2.

Prevalence of Daily Sugar-Sweetened Beverage (SSB) Consumption Categories, by SSB Screener type and Demographic Factors—SummerStyles Survey, 2014.

Characteristics	All Respondents % ± SE	Categories of Daily SSB Intake						Weighted κ^c
		4-Question Screener,^a % ± SE			1-Question Screener,^b % ± SE
		0 times/day	>0 to <1 time/day	≥1 times/day	0 times/day	>0 to <1 time/day	≥1 times/day
Total, N = 4167	–	16.0 ± 0.6	15.6 ± 0.7	68.4 ± 0.8	38.5 ± 0.9	42.5 ± 0.9	18.9 ± 0.7	.27
Age, N = 4167^d,e
18-24 years	12.4 ± 0.8	6.7 ± 1.6	16.6 ± 2.5	76.6 ± 2.8	25.3 ± 2.8	52.1 ± 3.3	22.6 ± 2.8	.19
25-44 years	34.2 ± 0.9	13.2 ± 1.2	13.4 ± 1.2	73.4 ± 1.5	31.4 ± 1.6	45.9 ± 1.7	22.7 ± 1.5	.27
45-64 years	35.0 ± 0.8	16.4 ± 1.0	16.3 ± 1.0	67.4 ± 1.2	41.2 ± 1.3	40.6 ± 1.3	18.2 ± 1.1	.26
≥65 years	18.4 ± 0.6	26.9 ± 1.6	17.7 ± 1.4	55.4 ± 1.8	55.5 ± 1.8	33.5 ± 1.8	11.1 ± 1.1	.28
Sex, N = 4167^d,e
Male	47.9 ± 0.9	14.5 ± 0.9	14.1 ± 0.9	71.4 ± 1.2	35.5 ± 1.3	43.6 ± 1.3	20.9 ± 1.1	.27
Female	52.1 ± 0.9	17.4 ± 0.9	16.9 ± 0.9	65.6 ± 1.2	41.3 ± 1.3	41.5 ± 1.3	17.1 ± 1.0	.27
Race/ethnicity, N = 4167^d,e
Non-Hispanic white	66.4 ± 0.9	18.4 ± 0.8	17.8 ± 0.8	63.8 ± 1.0	41.6 ± 1.0	41.2 ± 1.0	17.2 ± 0.8	.30
Non-Hispanic black	11.5 ± 0.6	11.1 ± 1.8	11.1 ± 1.8	77.8 ± 2.4	33.8 ± 2.7	40.6 ± 2.8	25.6 ± 2.7	.23
Hispanic	14.7 ± 0.8	8.9 ± 1.5	10.7 ± 1.7	80.3 ± 2.2	27.5 ± 2.5	49.4 ± 2.9	23.1 ± 2.5	.18
Non-Hispanic other	7.4 ± 0.6	16.5 ± 3.1	12.3 ± 2.9	71.2 ± 3.8	40.2 ± 4.2	44.0 ± 4.2	15.7 ± 2.8	.21
Marital status, N = 4167^e
Married/domestic partnership	61.1 ± 0.9	16.9 ± 0.8	15.9 ± 0.8	67.2 ± 1.0	40.1 ± 1.1	42.7 ± 1.1	17.2 ± 0.9	.27
Not married	38.9 ± 0.9	14.6 ± 1.1	15.1 ± 1.1	70.3 ± 1.4	36.0 ± 1.5	42.2 ± 1.6	21.8 ± 1.3	.27
Education, N = 4167^d,e
≤High school	41.9 ± 0.9	14.1 ± 1.0	12.5 ± 1.0	73.3 ± 1.3	36.2 ± 1.4	39.4 ± 1.5	24.3 ± 1.3	.27
Some college	28.9 ± 0.8	15.0 ± 1.1	17.6 ± 1.3	67.4 ± 1.5	37.4 ± 1.6	44.9 ± 1.7	17.7 ± 1.3	.26
≥Bachelors	29.2 ± 0.8	19.7 ± 1.2	18.0 ± 1.2	62.2 ± 1.5	42.9 ± 1.6	44.6 ± 1.6	12.5 ± 1.1	.27
Annual household income, N = 4167^d,e
≤$34 999	27.5 ± 0.8	15.0 ± 1.3	13.7 ± 1.2	71.3 ± 1.6	37.0 ± 1.7	38.1 ± 1.7	24.9 ± 1.6	.28
$35 000-$74 999	32.4 ± 0.9	14.9 ± 1.1	15.1 ± 1.1	70.0 ± 1.4	35.1 ± 1.5	42.8 ± 1.6	22.0 ± 1.4	.29
$75 000-$99 999	15.7 ± 0.7	15.6 ± 1.6	16.9 ± 1.7	67.5 ± 2.2	40.5 ± 2.3	45.1 ± 2.4	14.4 ± 1.8	.24
≥$100 000	24.4 ± 0.8	18.9 ± 1.3	17.6 ± 1.4	63.5 ± 1.8	43.5 ± 1.8	45.4 ± 1.9	11.1 ± 1.2	.26
Weight status, N = 4029^e,f
Under-/normal weight	38.3 ± 0.9	14.6 ± 1.0	16.7 ± 1.2	68.7 ± 1.5	39.1 ± 1.5	44.6 ± 1.6	16.3 ± 1.2	.24
Overweight	31.2 ± 0.9	15.1 ± 1.1	14.6 ± 1.1	70.3 ± 1.4	37.6 ± 1.6	43.9 ± 1.6	18.5 ± 1.3	.26
Obesity	30.5 ± 0.9	18.0 ± 1.2	14.9 ± 1.1	67.1 ± 1.5	36.9 ± 1.6	40.2 ± 1.6	22.9 ± 1.4	.33

Abbreviations: SE, standard error; SSB, sugar-sweetened beverage.

^aComposite consumption based on screener with 4 questions, which ask about consumption of each SSB type separately: soda, sweetened coffee/tea, sports/energy drinks, fruit drinks.

^bOne question screener which asked about the combined consumption of beverage types: soda, fruit drinks, sports/energy drinks, and other SSBs.

^cWeighted κ statistic to determine strength of agreement between the 2 SSB screeners.

^dUsing the 4-question screener, frequency of daily SSB intake differed significantly across categories of this demographic factor (Rao-Scott χ² P value <.05).

^eUsing the 1-question screener, frequency of daily SSB intake differed significantly across categories of this demographic factor (Rao-Scott χ² P value <.05).

^fWeight status was based on body mass index: <25.0 is normal or underweight, 25.0 to 29.9 is overweight, and ≥30.0 is obesity.

Using the continuous measure of SSB intake, respondents consumed SSBs an average of 1.7 times/day (95% confidence interval [CI]: 1.65-1.79) based on the 4-question screener and 0.6 times/day (95% CI: 0.56-0.62) based on the 1-question screener (Table 3). The mean difference in consumption using 4 questions versus 1 was 1.1 times/day, a difference that was statistically significant overall and within all demographic groups. The mean difference in the frequency of SSB intake was significantly higher for males (vs females), blacks and Hispanics (vs whites), and those in the 25 to 44 years or 45 to 64 years categories (vs the 65 years and older category). Also, the mean difference was significantly higher for individuals with a high school education or less (vs those with some college education or those who completed college) and among individuals in the lowest household income category (vs all higher income categories). The overall Lin’s CCC was 0.31, indicating poor absolute agreement between these 2 screeners for SSB consumption.¹⁶

Table 3.

Comparison of Mean Daily Sugar-Sweetened Beverage (SSB) Consumption (Times/Day) Between SSB Screener types, Total, and by Demographic Factors—SummerStyles Survey, 2014.^a

Characteristics	n	4-Question Screener^b Mean, 95% CI, times/day	1-Question Screener^c Mean, 95% CI, times/day	CCC^d	Difference^e,f Mean times/day
Total	4167	1.72 (1.65-1.79)	0.59 (0.56-0.62)	0.31^g	1.13^g
Age
18-24 years	258	1.80 (1.53-2.07)	0.66 (0.56-0.75)	0.30^g	1.14^g
25-44 years	1163	1.89 (1.75-2.03)	0.71 (0.64-0.77)	0.32^g	1.18^g,h
45-64 years	1839	1.74 (1.65-1.82)	0.57 (0.52-0.61)	0.30^g	1.17^g,h
≥65 years (ref)	907	1.31 (1.20-1.41)	0.37 (0.32-0.41)	0.29^g	0.94^g
Sex
Male (ref)	2032	1.85 (1.75-1.96)	0.64 (0.59-0.69)	0.31^g	1.21^g
Female	2135	1.60 (1.50-1.69)	0.54 (0.50-0.58)	0.31^g	1.05^g,h
Race/ethnicity
Non-Hispanic white (ref)	3140	1.59 (1.52-1.65)	0.55 (0.52-0.59)	0.32^g	1.03^g
Non-Hispanic black	395	2.04 (1.74-2.33)	0.71 (0.59-0.83)	0.31^g	1.33^g,h
Hispanic	403	2.13 (1.87-2.40)	0.71 (0.60-0.82)	0.25^g	1.42^g,h
Non-Hispanic other	229	1.58 (1.37-1.80)	0.48 (0.38-0.58)	0.26^g	1.10^g
Marital status
Married/domestic partnership (ref)	2705	1.67 (1.59-1.75)	0.55 (0.51-0.59)	0.30^g	1.12^g
Not married	1462	1.79 (1.67-1.92)	0.65 (0.60-0.71)	0.33^g	1.14^g
Education
≤High school (ref)	1507	2.02 (1.89-2.16)	0.71 (0.65-0.77)	0.32^g	1.31^g
Some college	1287	1.62 (1.52-1.73)	0.58 (0.52-0.63)	0.32^g	1.05^g,h
≥Bachelors	1373	1.38 (1.29-1.46)	0.43 (0.39-0.47)	0.23^g	0.94^g,h
Annual household income
≤$34 999 (ref)	1118	2.05 (1.87-2.23)	0.74 (0.66-0.82)	0.34^g	1.31^g
$35 000-$74 999	1434	1.78 (1.66-1.89)	0.64 (0.58-0.69)	0.30^g	1.14^g,h
$75 000-$99 999	617	1.48 (1.36-1.60)	0.50 (0.43-0.57)	0.26^g	0.98^g,h
≥$100 000	998	1.42 (1.32-1.52)	0.42 (0.37-0.46)	0.30^g	1.00^g,h
Weight statusⁱ
Under-/normal weight (ref)	1390	1.67 (1.55-1.79)	0.54 (0.49-0.59)	0.31^g	1.13^g
Overweight	1342	1.71 (1.60-1.82)	0.56 (0.51-0.61)	0.28^g	1.15^g
Obesity	1297	1.79 (1.66-1.93)	0.71 (0.64-0.77)	0.35^g	1.09^g

Abbreviations: CCC, concordance correlation coefficient; CI, confidence interval.

^aN = 4167.

^bComposite consumption based on a screener with 4 questions, which ask about consumption of each SSB type separately: soda, sweetened coffee/tea, sports/energy drinks, fruit drinks.

^cOne question screener which asked about the combined consumption of beverage types soda, fruit drinks, sports/energy drinks, and other SSBs.

^dLin’s CCC measuring the absolute agreement in SSB intake (times/day) between the 2 screeners.

^ePaired t tests were used to determine whether there were significant differences in mean SSB intake using the 4-question versus 1-question screener.

^fIndependent samples t tests used to determine significance of differences between demographic groups in the mean intake difference between the 4-question and 1-question screener. Reference group indicated.

^g P < .05

^hSignificant difference between groups at P <.05.

ⁱWeight status was based on body mass index: <25.0 is normal weight or underweight, 25.0 to 29.9 is overweight, and ≥30.0 is obesity.

Discussion

In the present study, SSB consumption was significantly lower when measured with a single screener question compared to a screener with 4 SSB questions; Lin’s CCC indicated poor absolute agreement between these 2 screeners in the number of times/day that SSBs were consumed. Additionally, the weighted κ statistics were within the range of only slight to fair agreement.¹⁵ The present study found that 1 SSB screener question resulted in a lower prevalence of daily SSB consumption by nearly 50 percentage points, whereas it resulted in a higher estimate for those whose daily consumption of SSBs was either none or less than daily. The average difference between these 2 screeners was more than 1 time/day that SSBs were consumed. The finding that a single screener question underestimates SSB consumption could be explained by cognitive theory, suggesting that respondents have difficulty in recalling consumption of multiple food or beverage items when asked all at once. This theory has been tested with FFQs, and the findings demonstrated that asking about several related foods in a single question produced less accurate dietary intake estimates than when those same foods were separated into multiple questions.¹⁷ Previous research comparing 19-item and 1-item fruit and vegetable screeners to multiple 24-hour recalls found that the 1-item screener underestimated intake, while the 19-item screener overestimated intake.¹⁸

The present study also found that the difference in SSB intake found using 4 questions versus 1 differed by sociodemographic groups. The mean difference between the 2 screeners in frequency of SSB intake was significantly higher for males, blacks and Hispanics, those between the ages of 25 and 44 years or 45 and 64 years, individuals with a high school education or less, and individuals in the lowest household income category. These findings suggest that when assessing disparities in total SSB intake across sociodemographic groups, using a 4-question versus 1-question screener may result in different conclusions being drawn. Previous research has found differences in the performance of FFQ questions across demographic factors like age, sex, education, and occupation.¹⁹ A limitation of FFQs is that they can require complex cognitive estimation, particularly for combined food items,⁹ and the ease of this task may vary across subgroups.

The methods used to measure and characterize daily SSB intake can influence the comparability of estimates across surveys. For example, the 4-question screener in SummerStyles produced an estimate of the prevalence of daily SSB consumption among adults (68%) that was similar to that found in the nationally representative 2010 NHIS (64%).¹² However, the SummerStyles estimate is higher than that found among adults participating in the 2011 to 2014 NHANES (49.3%), which measures SSB intake using a 24-hour dietary recall interview and characterizes daily consumption as having consumed any SSB in the previous 24 hours.² The SummerStyles estimate is also higher than that found in the 2013 Behavioral Risk Factor Surveillance System (30.1%), which is based on self-reported intake data from 23 states, measured using a 2-question screener that does not capture sweetened coffee drinks.¹³ To measure total daily SSB intake among adults, our findings suggest that using at least 4-screener questions, 1 question for each beverage type (regular soda, sweetened coffee and tea, sports and energy drinks, and fruit drinks), may be more desirable than using a single question that includes multiple types of SSBs. However, future research could validate new screener questions developed to capture additional SSB types that are introduced into the market.

The present study is subject to limitations. The SummerStyles survey is based on a sample that may not be nationally representative due to selection and nonresponse biases. The final analytic sample included responses from less than half of those who received the first survey wave in the spring. Therefore, the findings of this study might not be generalizable to the entire US adult population. However, the data were weighted using key demographic distributions in the US Current Population Survey, and the 4-question screener in SummerStyles produced an estimate of daily SSB consumption among adults that was similar to that found in the nationally representative 2010 NHIS. Another limitation is that the 2 screeners used in this study measured times/day that SSBs are consumed. Neither screener measured the ounces of SSBs consumed nor grams of sugar; therefore, the screeners cannot be used to determine the amount of added sugars contributed by the consumption of SSBs. The amount of added sugar can vary greatly based on the amount of SSB consumed and the beverage type; a 12 oz can of soda contains much more sugar than an equivalent amount of coffee with a teaspoon of sugar added. Furthermore, the present study did not have a true gold standard measurement of the amount of SSBs consumed (ie, volume) with which to assess the validity of either screener, and to our knowledge, neither screener has been validated against other methods of SSB measurement, such as 24-hour recall or food records. However, screener questions can be a suitable method for population-level surveillance of beverage intake. A previous study examined the validity of beverage screener questions in comparison to the gold standard measurement method of multiple 24-hour dietary recalls and found that the number of times/day that beverages were consumed, derived from the screener, was significantly positively correlated with the number of servings/day, derived from the 24-hour dietary recalls.¹¹ Lastly, the SummerStyles survey data are self-reported and subject to recall and social desirability bias, which may have had a differential impact on reporting for the single screener question versus the 4-question screener being compared in this study.

To our knowledge, the present study is the only one of its kind to compare the performance of a single screener question to a screener with a larger number of questions to measure adult daily SSB intake. An additional strength of the present study is that the SummerStyles survey has a large sample and allowed assessment within multiple subgroups of the population.

In this study, estimates of the prevalence of SSB intake at least once per day among adults were significantly lower with a 1-question screener compared to a screener with 4 questions, and this differed by sociodemographic groups. The impact of lower reporting for daily SSB intake using a 1-question screener could differ depending upon the aim of data collection (eg, surveillance versus studying the association between SSB intake and clinical outcomes). To improve data quality, researchers and public health professionals may need to measure each SSB type using separate questions, rather than combining all beverage types into 1 overall intake question, when conducting research or surveillance on SSB consumption. When space or time constraints limit the number of survey questions, researchers and practitioners should consider that a 1-question screener may underestimate daily SSB intake.

SO WHAT? Implications for Health Promotion Practitioners and Researchers

What is already known on this topic?

It is important to monitor sugar-sweetened beverage (SSB) intake, given the adverse health consequences of daily SSB consumption (eg, obesity and diabetes). While food frequency questionnaire–style screener questions have been found to be a suitable method for population-level surveillance of beverage intake, the performance of a single screener question to measure SSB intake is unknown.

What does this article add?

This study found poor absolute agreement between a 1-question screener measuring total SSB intake and a screener measuring SSB types separately using 4 questions. Estimates of the prevalence of SSB intake at least once per day among adults were significantly lower with a 1-question screener compared to a screener with 4 questions (18.9% vs 68.4%, respectively).

What are the implications for health promotion practice or research?

When conducting research/surveillance on SSB consumption, data quality may be improved by measuring each SSB type using separate questions, rather than combining all beverage types into 1 overall intake question. When survey space or time is limited, researchers/practitioners should consider that a 1-question screener may underestimate daily SSB intake.

Footnotes

Authors’ Note

All authors participated in the design and conceptualization of this study. EAL analyzed the data and wrote the first manuscript draft. All authors interpreted the data and reviewed and commented on subsequent drafts of the manuscript. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

US Department of Agriculture, US Department of Health and Human Services. Dietary Guidelines for Americans, 2015-2020. 8th ed. Washington, DC: US Government Printing Office; 2015.

Rosinger

Herrick

Gahche

Park

. Sugar-sweetened beverage consumption among U.S. adults, 2011-2014. NCHS Data Brief. 2017;(270):1–8

Welsh

Sharma

Grellinger

Vos

. Consumption of added sugars is decreasing in the United States. Am J Clin Nutr. 2011;94(3):726–734.

Kit

Fakhouri

Park

Nielsen

Ogden

. Trends in sugar-sweetened beverage consumption among youth and adults in the United States: 1999-2010. Am J Clin Nutr. 2013;98(1):180–188.

Dhingra

Sullivan

Jacques

. Soft drink consumption and risk of developing cardiometabolic risk factors and the metabolic syndrome in middle-aged adults in the community. Circulation. 2007;116(5):480–488.

Schulze

Manson

Ludwig

. Sugar-sweetened beverages, weight gain, and incidence of type 2 diabetes in young and middle-aged women. JAMA. 2004;292(8):927–934.

Park

Akinbami

McGuire

Blanck

. Association of sugar-sweetened beverage intake frequency and asthma among U.S. adults, 2013. Prev Med. 2016;91:58–61.

Bernabe

Vehkalahti

Sheiham

Aromaa

Suominen

. Sugar-sweetened beverages and dental caries in adults: a 4-year prospective study. J Dent. 2014;42(8):952–958.

Buzzard

. 24-Hour dietary recall and food record methods. In: Willett

, ed. Nutritional Epidemiology. 2nd ed. New York, NY: Oxford University Press; 1998.

10.

Johnson

. Dietary intake—how do we measure what people are really eating? Obes Res. 2002;10(suppl 1):63S–68S.

11.

O’Malley-Olsen

Eaton

Park

Brener

Blanck

. Comparing methods for assessing beverage intake among high school students. Am J Health Behav. 2014;38(1):114–123.

12.

Park

McGuire

Galuska

. Regional differences in sugar-sweetened beverage intake among US adults. J Acad Nutr Diet. 2015;115(12):1996–2002.

13.

Park

Town

Blanck

. Prevalence of sugar-sweetened beverage intake among adults—23 states and the District of Columbia, 2013. MMWR Morb Mortal Wkly Rep. 2016;65(7):169–174.

14.

National Heart, Lung, and Blood Institute. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults. The evidence report. NIH Publication No. 98-4083. National Institutes of Health. 1998. http://www.nhlbi.nih.gov/guidelines/obesity/ob_gdlns.pdf. Accessed October 18, 2017.

15.

Viera

Garrett

. Understanding interobserver agreement: the kappa statistic. Fam Med. 2005;37(5):360–363.

16.

Quinn

Haber

Pan

. Use of the concordance correlation coefficient when examining agreement in dyadic research. Nurs Res. 2009;58(5):368–373.

17.

Thompson

Subar

Brown

. Cognitive research enhances accuracy of food frequency questionnaire reports: results of an experimental validation study. J Am Diet Assoc. 2002;102(2):212–225.

18.

Greene

Resnicow

Thompson

. Correspondence of the NCI Fruit and Vegetable Screener to repeat 24-H recalls and serum carotenoids in behavioral intervention trials. J Nutr. 2008;138(1):200S–204S.

19.

Marks

Hughes

van der Pols

. Relative validity of food intake estimates using a food frequency questionnaire is associated with sex, age, and other personal characteristics. J Nutr. 2006;136(2):459–465.