Total Sugar-Sweetened Beverage Intake Among US Adults Underreported When Using One Question Instead of Four Questions to Assess Intake

Abstract

Although surveys commonly use one question to assess total sugar-sweetened beverage (SSB) intake for brevity, the performance of a single question is unknown. We compared the performance of one question measuring total SSB intake to four questions measuring SSB types separately. We used the 2014 SummerStyles online survey with 4,167 U.S. adults (≥18 years). We compared frequency of SSB intake using one question measuring total SSB consumption to frequency using a composite of four questions (regular soda, fruit drinks, sports/energy drinks, sweetened coffee/tea). We calculated mean difference between the two and used multiple linear regression to examine demographic predictors of differences. SSB intake categories were: 0, >0–<1, and ≥1 time/day. We evaluated agreement using Kappa, ranging from 0 (no agreement) to +1 (perfect agreement), and sensitivity of 1 question to identify SSB consumption ≥1 time/day compared to 4 questions. Mean (median) SSB intake was 0.6 (0.1) times/day using 1 question and 1.7 (1.0) times/day using 4 questions. Intake frequency based on 1 vs. 4 questions, respectively, was 38.5% vs. 16.0% for no intake, 42.5% vs. 15.6% for >0–<1 time/day, and 18.9% vs. 68.4% for ≥1 time/day (p<0.0001). sensitivity of 1 question was 26.3%. Kappa was 0.20 (p<0.0001). mean difference in SSB consumption using 1 vs. 4 questions was -1.1 times/day. Sex, age, education, and race/ethnicity were significant predictors of this difference. Estimates of daily SSB intake were lower when one question was used compared to four. Researchers should consider assessing SSB types separately when measuring total SSB intake.