This study used data from National Health and Nutrition Examination Surveys (NHANES) between the years 2013 and 2014 for subjects 20 to 80 years of age. The NHANES is an on-going cross-sectional survey conducted by the National Center for Health Statistics (NCHS) in 2-year cycles, which designed to evaluate the health and nutritional status of adults and children in the United States. The NHANES employs a stratified, multistage probability sampling design and includes demographic, socioeconomic, dietary, health-related information, and examination and laboratory tests administered by highly trained medical personnel.
There were 5,476 participants aged 20 yeare and older from the years 2013-2014. Of these, we excluded those with missing information on body-mass index (BMI, calculated as weight [kg] divided by height [m] squared), race/ethnicity, educational level, smoking status, blood pressure, high density lipoprotein, glycohemoglobin, physical activity, family history of cardiovascular disease, family history of diabetes or sitting time. Meanwhile, we removed people who had daily energy intake < 500 kcal or > 4500 kcal and pregnant and/or lactating females. Finally, we used 4,162 samples to do the following analysis.
MetS was defined according to the updated International Diabetes Federation (IDF) criteria  as follows: central obesity (waist circumference ≥102 cm for males or ≥88 cm for females) plus at least two of the following: 1) serum triglyceride level of >150 mg/dL,or current drug treatment for elevated TG; 2) HDL cholesterol level , <40 mg/dL for males and , <50 mg/dL for females, or current drug treatment for reduced HDL cholesterol; 3) fasting glucose level >100 mg/dL or use of antidiabetic medications (insulin or oral agents) or diatebes; or 4) systolic blood pressure >90 mmHg and/or diastolic blood pressure >140 mmHg, or use of antihypertensive medications.
Healthy Beverage Index (HBI)
HBI was used to assess the quality of beverage intake based on the Beverage Guidance System and recommalesdations from the Dietary Guidelines for Americans, which consisted of 10 components including water, SSBs, total beverage energy, unsweetened coffee and tea, and alcohol. A full description of the individual components can be found in Additional file 1: Tabel S1. The HBI score ranges from 0 to 100, with a higher score indicating better adherence to beverage guidelines and healthier beverage intake pattern .
Assessment of beverage intake, food intake and nutrients
Participants' food intake for two non-consecutive days through 24-hour dietary recall interviews were collected in the NHANES survey. The first dietary recall was conducted in-person and the second dietary recall was conducted 3-10 days later via telephone. Beverage intake and energy intake was estimated using the USDA’s Food and Nutrient Database for Dietary Studies (FNDDS). The mean values of nutrient intakes for day one and day two of the 24 h dietary recall were used in analyses.
Potential covariables were consisted of age (years), gender (male/female), race/ethnicity (Mexican American/Other Hispanic/Non-Hispanic White/Non-Hispanic Black/Other race), socioeconomic status (high/medium/low), the family history (no/either/both), BMI (under 18.5 kg/m2/18.5-25 kg/m2/25-30 kg/m2/over 30 kg/m2), physical activity (under 600 MET-mins/wks/600-3999 MET-mins/wks/4000-7999 MET-mins/wks/over 8000 MET-mins/wks), smoking status (current smokers/former smokers/never smokers), sitting time (under 4 h/day/4-6 h/day/6-8 h/day/over8 h/day), sleep duration (under 5 h/day/5-7 h/day/ 7-9 h/day/over 9 h/day), abdominal obesity (yes/no), DASH scores and non-HDL cholesterol (under 100 mg/dL/100-145 mg/dL/145-185 mg/dL/185-220 mg/dL/over 220 mg/dL).
We used mean ± SE or proportions to describe the characteristics of the study subjects by quartiles of HBI. We used binary logistic regression to estimate the associations of MetS and stroke prevalence with HBI scores, respectively. We used nested models to adjust for confounding that included: model 1, no adjustment; model 2, age, gender, race/ethnicity, socioeconomic status; model 3, age, gender, race/ethnicity, socioeconomic status, BMI, the family history of CVD or Diabetes, abdominal obesity; model 4, model 3 with the addition of smoking status, Non-HDL cholesterol, physical activity, sleep duration, sitting time and DASH scores. To access effect modification, we constructed stratified models to access modification by gender.
In order to assessed the sensitivity of our results, we performed several sensitivity analyses. We tested for effect modification by age, smoking status, physical activity and adequate intake (AI) using subgroup analysis. Meanwhile, sensitivity analyses were conducted among participants who did not drink tap water and who did not have diabetes mellitus (DM). And we did the same sensitivity analyses among males and females, respectively. All sensitivity analyses were conducted using the final adjustment model (model 4).
All statistical analyses were performed using SPSS 21.0 (SPSS Inc., Chicago, IL, USA). Graphic production was carried out by adopting R version 4.0.5 (The R Foundation for Statistical Computing, Vienna, Austria). Two-sided P <0.05 was considered statistically significant.