Association of Sugar-Sweetened Beverage Consumption with Prediabetes and Glucose Metabolism Markers in Hispanic/Latino Adults in the United States: Results from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL)

ABSTRACT

Background

Both the incidence of diabetes mellitus and consumption of sugar-sweetened beverages are high in the Hispanic/Latino population in the United States. The associations between consumption of sugar-sweetened beverages, artificially sweetened beverages, and 100% fruit juice with prediabetes and glucose metabolism markers in the diverse Hispanic/Latino population in the United States are unknown.

Objectives

The objective of this study was to examine the cross-sectional associations between consumption of sugar-sweetened beverages, artificially sweetened beverages, and 100% fruit juice with prediabetes and glucose metabolism markers such as fasting glucose and insulin, 2-h oral-glucose-tolerance test, HOMA-IR, HOMA index for β-cell function (HOMA-B), and glycated hemoglobin (HbA1c) among US Hispanic/Latino adults.

Methods

Using baseline data from the Hispanic Community Health Study/Study of Latinos (2008–2011), beverage consumption was ascertained using two 24-h dietary recalls and a food propensity questionnaire. Diabetes/prediabetes status was defined by self-report, antihyperglycemic medication use, and American Diabetes Association laboratory criteria. Among 9965 individuals without diabetes (5194 normoglycemia, 4771 prediabetes) aged 18–74 y, the associations of beverage consumption with prediabetes and glucose metabolism markers were analyzed using logistic and linear regressions, respectively, accounting for complex survey design.

Results

Compared with individuals who consumed <1 serving/d (<240 mL/d) of sugar-sweetened beverages, individuals who consumed >2 servings/d (>480 mL/d) had 1.3 times greater odds of having prediabetes (95% CI: 1.06, 1.61) and higher glucose metabolism markers including fasting glucose, fasting insulin, HOMA-IR, and HbA1c. Consumption of artificially sweetened beverages showed an inverse association with β-cell function (HOMA-B). Intake of 100% fruit juice was not significantly associated with prediabetes nor with glucose metabolism markers.

Conclusions

Among US Hispanic/Latino adults, higher sugar-sweetened beverage consumption was associated with increased odds of prediabetes and higher glucose metabolism markers. Public health initiatives to decrease sugar-sweetened beverage consumption could potentially reduce the burden of diabetes among Hispanics/Latinos in the United States.

Introduction

Diabetes mellitus and its complications are a major public health concern and economic burden in the United States, with a reported 30.3 million people having diabetes (1). Of particular concern is that the US Hispanic/Latino population, the nation's largest ethnic minority group, is disproportionately affected by type 2 diabetes (T2D). Data from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) indicated an overall 16.9% prevalence of diabetes, with the highest prevalence in individuals of Mexican heritage (18.3%) and of Puerto Rican and Dominican heritages (18%) among US Hispanic/Latino adults (2). The prevalence of T2D is twice as high in Hispanic/Latino adults compared with non-Hispanic whites (2).

One well-established lifestyle risk factor for T2D is the consumption of sugar-sweetened beverages, including soft drinks (soda), sweetened tea/coffee/water, and fruit concentrate drinks with added energy-containing sweeteners (3–6). As an alternative to sugar-sweetened beverages, artificially sweetened beverages are considered healthier choices for reducing sugar intake (7). Another choice is 100% fruit juice given the healthy components such as vitamins, fiber, and antioxidants, although it contains similar amounts of sugar as sugar-sweetened beverages. Unlike the consistent association between sugar-sweetened beverage consumption and diabetes or prediabetes risk (3–6), the association of artificially sweetened beverage and fruit juice consumption with risk for diabetes or prediabetes is unclear (4, 8, 9).

A meta-analysis of 17 prospective studies indicates that intake of 1 additional daily serving of sugar-sweetened beverage, artificially sweetened beverage, and fruit juice was associated with 13%, 8%, and 7% greater risk of incident diabetes, respectively (4). Individuals consuming sugar-sweetened beverages daily, compared with nonconsumers, have been shown to have a 46% greater risk of developing prediabetes, whereas artificially sweetened beverage intake was not associated with incident prediabetes risk among middle-aged white persons (10). The cross-sectional analyses in epidemiological studies also support adverse effects of sugary beverages on blood glucose metabolism biomarkers, including higher fasting insulin, glycated hemoglobin (HbA1c), and HOMA-IR (11, 12), but not for artificially sweetened beverages (11–13). Fruit juice consumption showed inconsistent associations—an inverse association with fasting glucose in the study of US white persons (11) but a positive association with HbA1c among middle-aged US white women (12).

The importance of studying the association between beverage consumption and diabetes/prediabetes among Hispanic/Latino individuals is underscored by high T2D prevalence and high sugary beverage intake in this population. The global surveys of >180 countries in 2010 and 2015 showed that many Latin American countries were ranked near the top of the list for high sugar-sweetened beverage intake (14, 15). In Mexico during 1999–2012, soft drinks, sweetened coffee/tea, and aguas frescas (made from fruit blended with sugar and water) were the top 3 beverage contributors to adult energy intake (16). Data suggest that compared with Mexicans in their country of origin, Mexican Americans appear to consume more sugar-sweetened beverages, artificially sweetened beverages, and unhealthy foods such as desserts, salty snacks, pizzas, and French fries (17). Furthermore, data show that Hispanics/Latinos consume more sugar-sweetened beverages than non-Hispanic whites or Asians in the United States (18–20). Given that the consumption of particular food choices by this segment of the population may differ from that of other US groups, more studies are needed to understand the effect of beverage consumption on diabetes/prediabetes and glucose metabolism markers in the diverse US Hispanic/Latino population.

To summarize the few available studies on Hispanics/Latinos, among women in Mexico, higher consumption of sugar-sweetened beverages has been associated with a higher incidence of diabetes (21) and with higher fasting glucose (22). Homemade 100% fruit juice intake was associated with decreased, but not significantly, fasting glucose concentrations among adults in Costa Rica (23). In the Northern Manhattan Study (NOMAS), which included 53% Hispanics/Latinos, both sugar-sweetened and artificially sweetened beverage consumption were associated with higher diabetes incidence (24). Still, dietary influences on health in the diverse US Hispanic/Latino population are not well understood, particularly in the context of the heterogeneous US Hispanic/Latino population that represents immigrants from Latin America as well as US-born individuals.

In this study, we sought to establish the associations between sugar-sweetened beverage consumption and prediabetes and glucose metabolism markers in US Hispanic/Latino adults using 2008–2011 cross-sectional data from HCHS/SOL participants who were free of diabetes. Also, we aimed to determine the associations between consumption of artificially sweetened beverages and 100% fruit juice and prediabetes and markers of glucose metabolism.

Methods

Study population and measurements

HCHS/SOL is a cohort study of Hispanic/Latino adults (n = 16,415) who were aged 18–74 y when recruited in 2008–2011 from randomly selected households in 4 US locations: Bronx, NY; Chicago, IL; Miami, FL; and San Diego, CA (25, 26). The study was approved by the institutional review boards at each location, and all study participants provided written informed consent. Data collection via interviewer-administered questionnaires included Hispanic/Latino heritage (Mexican, Cuban, Puerto Rican, Dominican, and Southern or Central American), health behaviors such as smoking and physical activity, family health history, medical history, and medication use. Clinical measurements included weight, height, and waist circumference. Blood samples (>8-h fast) were used to measure glucose, insulin, and HbA1c. Using the homeostasis model assessment method, HOMA-IR was calculated as [fasting insulin (mU/L) × fasting glucose (mg/dL)]/405, and HOMA index for β-cell function (HOMA-B) was calculated as 360 × fasting insulin (mU/L)/[fasting glucose (mg/dL) – 63] (27). Participants with a fasting plasma glucose <150 mg/dL and no previous diagnosis of diabetes completed a standard 75-g 2-h oral-glucose-tolerance test (OGTT). Two 24-h dietary recalls were obtained at baseline, and a food propensity questionnaire (FPQ) was assessed at the 1-y follow-up call. Data will be available upon review by the HCHS/SOL Steering Committee on the application for data access.

Consumption of sugar-sweetened beverages, artificially sweetened beverages, and 100% fruit juice

The first of 2 dietary recalls was administered in person at the field center at baseline visit, and the second recall was administered via telephone ∼45 d after the initial recall (20). Both recalls were administered using the Nutrition Data System for Research software (version 2011; Nutrition Coordinating Center, University of Minnesota). FPQ was administered in the following year at the annual follow-up interviews to assess intake of specific foods during the past 12 mo (available in reference 28). Data on sugar-sweetened beverages, artificially sweetened beverages, and 100% fruit juice were extracted from the items in Supplemental Table 1, and the usual beverage consumption was predicted from the 2-d recalls, incorporating FPQ as a covariate to reflect long-term patterns and episodically consumed foods, as published previously (20). The approach of incorporating FPQ to the calibration model of 24-h recalls has shown increased precision compared with use of 24-h recalls without FPQ (29). After excluding unreliable recalls according to the interviewer or extreme total energy intake (recall sex-specific <1st percentile or >99th percentile), we used MIXTRAN and INDIVINT macros version 2.1 from the National Cancer Institute (30) to predict the intake of sugar-sweetened beverages, artificially sweetened beverages, and 100% fruit juice using the 2-part model for episodically consumed foods (31). Covariates included age, sex, Hispanic/Latino heritage, field center, weekday/weekend, self-reported total intake amount (more, same, or less than usual amount), and frequencies of beverage intake from the FPQ (32). Servings for each beverage category were categorized into <1 serving/d, 1–2 servings/d, and >2 servings/d. In additional analyses, we further categorized the servings into 5 groups (<0.5, 0.5–0.99, 1–1.49, 1.5–2, and >2 servings/d) and also into the quintiles. One serving size for sugar-sweetened beverages and artificially sweetened beverages was 8 fl oz (240 mL), and 1 serving size of 100% fruit juice was 4 fl oz (120 mL).

Outcomes: diabetes status and glucose metabolism markers

Using the American Diabetes Association laboratory criteria (33), in addition to self-report of diabetes diagnosis and treatment, we defined groups with normal glucose tolerance/normoglycemia, prediabetes, and diabetes. Individuals with diabetes—defined by a fasting plasma glucose concentration ≥126 mg/dL (7.0 mmol/L), 2-h OGTT ≥200 mg/dL (11.1 mmol/L), HbA1c ≥6.5% (48 mmol/mol), or by self-report—were grouped into unrecognized/undiagnosed compared with recognized/diagnosed diabetes based on self-report of diabetes, and they were also subclassified by antihyperglycemic medication use and optimal glycemic control (HbA1c <7%). Normoglycemia was defined as a fasting plasma glucose concentration <100 mg/dL (6.1 mmol/L), 2-h OGTT <140 mg/dL (7.8 mmol/L), or HbA1c <5.7% (<39 mmol/mol) and no history of diabetes or antihyperglycemic medications. Prediabetes was defined as a fasting plasma glucose concentration of 100–125 mg/dL (5.6–7.0 mmol/L), 2-h OGTT 140–199 mg/dL (7.8–11.0 mmol/L), or HbA1c 5.7–6.4% (39–47 mmol/mol) and no history of diabetes or antihyperglycemic medications.

Additional outcomes of interest included the glucose metabolism markers: fasting plasma glucose and insulin concentrations, 2-h OGTT, HOMA-IR, HOMA-B, and HbA1c. HOMA-IR and HOMA-B were not normally distributed and were log transformed in the analysis. The variability of these markers was evaluated using repeated and duplicate measurements on a subset of HCHS/SOL participants (34). The analytical precision of the markers was acceptable (less than within-individual CV/2), but the within-subject biological variations of fasting insulin and 2-h OGTT were substantial, consistent with other studies: the within-individual CV was 25% for fasting insulin, 20.2% for 2-h OGTT, 5.2% for fasting glucose, and 2.2% for HbA1c.

Covariates

Covariates included age, sex, field center, Hispanic/Latino heritage, family history of diabetes, smoking, physical activity assessed with the Global Physical Activity Questionnaire (35), BMI (in kg/m²), waist circumference, total energy intake, and a modified Alternate Healthy Eating Index–2010 (AHEI-2010). The AHEI-2010 score is the sum of 11 individual components, with each component score ranging from 0 to 10 (vegetables, whole fruit, whole grains, sugar-sweetened beverages and fruit juice, nuts and legumes, red/processed meat, trans fat, long-chain fatty acids, PUFAs, sodium, and alcohol); however, we excluded sugar-sweetened beverages from the total AHEI-2010 score (36), ranging from 0 to 100, hence a “modified” AHEI-2010.

Analytical sample

We classified 12,345 participants by diabetes status, after excluding 1044 with prevalent cardiovascular disease, 644 with cancer, 243 with missing or unreliable 24-h diet recall, 2559 with missing FPQ, and 48 without diabetes-related measurements (lab measures and self-report). The main analysis was performed on 9965 individuals after excluding 2380 people with diabetes to examine the associations between beverage consumption and prediabetes and glucose metabolic markers. A sensitivity analysis was conducted among a larger group of 10,786 individuals for an analysis comparing participants with prediabetes plus unrecognized/undiagnosed diabetes compared with normoglycemia. A flowchart of participants is presented in Supplemental Figure 1.

Statistical analyses

Age- and sex-adjusted characteristics by beverage consumption category (<1 serving/d, 1–2 servings/d, and >2 servings/d) were computed as means and SEs (continuous variables) or percentages and SEs (categorical variables). These statistics were calculated using survey linear or logistic regressions that accounted for the stratification, clustering, and sampling weighting for sample selection and nonresponse rate in HCHS/SOL to appropriately represent the target population.

We examined the association of each type of beverage consumption with prediabetes compared with normoglycemia using survey logistic regressions adjusting for potential confounders. As a secondary outcome, a comparison was made between normoglycemic and a combined group of prediabetes plus unrecognized/undiagnosed diabetes. In addition, we examined the associations between beverage consumption and fasting glucose and insulin concentrations, 2-h OGTT, HbA1c, and log-transformed HOMA-IR and HOMA-B using a survey linear regression. Models were sequentially adjusted for 1) age, sex, field center, and Hispanic/Latino heritage; 2) family history of diabetes, smoking status, BMI, and waist circumference; 3) physical activity and modified AHEI-2010 score; and 4) total energy intake. A fifth model was adjusted for the other 2 types of beverage consumption (sugar-sweetened beverage and/or artificially sweetened beverage and/or 100% fruit juice). Tests for linear trend used median category values. Step-by-step model building is described in Supplemental Table 2. We repeated analyses according to 5 subgroups (<0.5, 0.5–0.99, 1–1.49, 1.5–2, and >2 servings/d) and quintiles of beverage consumption, presented in Supplemental Tables 3–6. Analyses accounted for sampling weights and the complex survey design using SAS version 9.4, SAS-callable SUDAAN v 11.0 (SAS Institute), and R version 3.2.4 (R Foundation for Statistical Computing).

Results

Participant characteristics

Among 9965 individuals without diabetes, 53.1% (n = 6034) were women, and the mean age was 39.0 y (SE = 0.2). Grouped by intake of <1, 1–2, and >2 servings/d of each beverage, there were 34.4% (n = 4016), 38.1% (n = 3707), and 27.5% (n = 2178) individuals by frequency of sugar-sweetened beverage consumption, respectively (Table 1). The artificially sweetened beverage consumption was grouped into 67.1% (n = 7155), 27.9% (n = 2344), and 5.0% (n = 402) individuals by servings per day (<1, 1–2, >2). The consumption of 100% fruit juice was 79.6% (n = 8131), 17.2% (n = 1559), and 3.3% (n = 263) by <1, 1–2, and >2 servings/d, respectively. Sugar-sweetened and artificially sweetened beverage intakes were modestly correlated (Pearson correlation coefficient; r = 0.42), but 100% fruit juice intake was not correlated with other beverages (r <0.05).

TABLE 1.

Age- and sex-adjusted descriptive characteristics of individuals without diabetes by beverage consumption, HCHS/SOL (2008–2011)¹

	Sugar-sweetened beverage			Artificially sweetened beverage			100% fruit juice
	<1 serving/d	1–2 servings/d	>2 servings/d	<1 serving/d	1–2 servings/d	>2 servings/d	<1 serving/d	1–2 servings/d	>2 servings/d
n (%)	4016 (34.4)	3707 (38.1)	2178 (27.5)	7155 (67.1)	2344 (27.9)	402 (5.0)	8131 (79.6)	1559 (17.2)	263 (3.3)
Sugar-sweetened beverage, servings/d2	0.66 ± 0.01	1.43 ± 0.01	2.61 ± 0.03	1.16 ± 0.02	1.95 ± 0.06	2.03 ± 0.18	1.32 ± 0.02	1.49 ± 0.05	1.39 ± 0.11
Artificially sweetened beverage, servings/d2	0.58 ± 0.01	0.74 ± 0.01	1.08 ± 0.01	0.63 ± 0.01	1.28 ± 0.01	2.35 ± 0.05	0.77 ± 0.01	0.84 ± 0.03	0.87 ± 0.04
100% fruit juice, servings/d2	0.45 ± 0.01	0.49 ± 0.01	0.52 ± 0.01	0.47 ± 0.01	0.52 ± 0.01	0.49 ± 0.02	0.43 ± 0.005	1.30 ± 0.02	2.31 ± 0.05
Age, y	46.1 ± 0.4	37.7 ± 0.3	31.9 ± 0.3	41.9 ± 0.3	32.9 ± 0.4	33.7 ± 0.9	39.8 ± 0.3	37.3 ± 0.5	36.5 ± 0.9
Women, %	75.2 ± 1.0	53.7 ± 1.2	24.3 ± 1.3	63.5 ± 0.8	32.1 ± 1.2	28.2 ± 2.9	57.3 ± 0.8	38.8 ± 1.7	26.1 ± 3.5
Hispanic/Latino background, %
Mexican	33.2 ± 2	37.6 ± 2.3	43.1 ± 2.2	33.6 ± 1.8	45.5 ± 2.1	47.9 ± 4.0	39.9 ± 1.8	30.1 ± 2.5	24.6 ± 3.9
Puerto Rican	10.2 ± 1	10.8 ± 0.8	20.5 ± 1.4	10.3 ± 0.7	18.3 ± 1.4	24.5 ± 3.0	13.3 ± 0.8	13.0 ± 1.2	11.7 ± 2.6
Dominican	13.7 ± 1.1	8.7 ± 0.9	6.6 ± 0.9	12.0 ± 0.9	6.0 ± 0.9	2.8 ± 1.1	9.0 ± 0.7	14.3 ± 1.7	6.1 ± 1.7
Cuban	26.3 ± 2	23.9 ± 2.1	12.1 ± 1.5	27.2 ± 2.1	10.0 ± 1.3	8.8 ± 2.4	21.3 ± 1.8	20.6 ± 2.3	28.2 ± 4.0
South American	6.3 ± 0.6	5.8 ± 0.6	4.6 ± 0.5	5.3 ± 0.4	6.0 ± 0.7	6.8 ± 1.3	4.6 ± 0.4	9.6 ± 1.1	9.3 ± 1.8
Central American	6.6 ± 0.7	8.4 ± 0.8	8.9 ± 1	7.5 ± 0.7	9.4 ± 1.0	5.2 ± 1.1	7.6 ± 0.6	8.3 ± 1.0	12.0 ± 2.5
More than one	3.7 ± 0.6	4.9 ± 0.6	4.2 ± 0.6	4.1 ± 0.5	4.7 ± 0.6	4.1 ± 1.3	4.1 ± 0.4	4.2 ± 0.8	8.1 ± 2.3
Center, %
Bronx, NY	29.5 ± 2.0	22.4 ± 1.5	27.6 ± 2.0	26.7 ± 1.6	25.4 ± 1.9	24.9 ± 3.1	25.5 ± 1.5	30.7 ± 2.4	20.1 ± 3.6
Chicago, IL	10.9 ± 0.9	15.5 ± 1.3	22.1 ± 1.6	12.6 ± 1.0	22.9 ± 1.7	18.6 ± 2.5	16.4 ± 1.1	13.5 ± 1.3	12.8 ± 2.6
Miami, FL	36.1 ± 2.5	35.2 ± 2.6	22.8 ± 2.3	37.9 ± 2.5	20.7 ± 2.1	17.8 ± 2.7	30.2 ± 2.2	36.3 ± 3.1	51.9 ± 4.5
San Diego, CA	23.5 ± 2.2	26.9 ± 2.5	27.5 ± 2.3	22.8 ± 1.8	31 ± 2.7	38.7 ± 4.4	27.8 ± 2.1	19.6 ± 2.4	15.2 ± 3.4
Current smoking, %	15.4 ± 1.1	19.8 ± 1.0	25.4 ± 1.5	18.6 ± 0.8	21.6 ± 1.2	24.3 ± 3.1	20.8 ± 0.7	14.7 ± 1.2	20.9 ± 3.3
Alcohol use, %
Never	20.6 ± 1.5	18.5 ± 1.1	12.9 ± 1.1	19.5 ± 1.1	13.7 ± 1.1	14.3 ± 2.9	17.2 ± 0.9	19.1 ± 1.6	20.3 ± 3.9
Former	25.3 ± 1.2	28.0 ± 1.2	31.6 ± 1.5	27.4 ± 0.9	29.6 ± 1.4	28.3 ± 3.1	28.2 ± 0.9	28.6 ± 1.5	24.1 ± 3.5
Current	54.1 ± 1.8	53.5 ± 1.5	55.5 ± 1.6	53.1 ± 1.2	56.7 ± 1.6	57.5 ± 3.5	54.6 ± 1.1	52.3 ± 1.9	55.6 ± 4.6
Total physical activity, MET-min/d	635 ± 25	621 ± 23	929 ± 23	671 ± 22	798 ± 32	750 ± 74	701 ± 18	746 ± 41	766 ± 111
BMI, kg/m2	28.6 ± 0.2	28.6 ± 0.1	29.5 ± 0.2	28.6 ± 0.1	29.2 ± 0.2	29.7 ± 0.4	29.0 ± 0.1	28.3 ± 0.2	27.9 ± 0.5
Waist circumference, cm	94.7 ± 0.4	95.6 ± 0.3	97.4 ± 0.5	95.3 ± 0.3	96.3 ± 0.4	98.3 ± 1.1	96.1 ± 0.3	94.4 ± 0.4	93.6 ± 1.2
Total energy intake, kcal/d	1869 ± 12	2025 ± 12	2163 ± 13	1998 ± 11	2039 ± 13	1996 ± 30	1992 ± 9	2055 ± 16	2182 ± 40
Alternative Healthy Eating Index without sweetened beverages and fruit juice	47.3 ± 0.3	46.4 ± 0.2	45.3 ± 0.2	46.1 ± 0.2	46.9 ± 0.2	47.3 ± 0.6	46.6 ± 0.2	45.9 ± 0.3	45.5 ± 0.5
Family history of diabetes, %	36.6 ± 1.5	36.2 ± 1.2	37.6 ± 1.6	35.1 ± 1.1	39.9 ± 1.5	40.1 ± 3.7	37.1 ± 0.9	35.4 ± 1.8	32.4 ± 3.7
Prediabetes, %	37.7 ± 1.5	40.0 ± 1.2	47.2 ± 1.6	40.1 ± 1.0	43.9 ± 1.6	41.2 ± 3.6	41.7 ± 1.0	38.9 ± 1.7	39.3 ± 4.3
Glucose metabolism
Fasting glucose, mg/dL	92.6 ± 0.2	93.4 ± 0.2	94.1 ± 0.3	93.2 ± 0.1	93.5 ± 0.2	94.1 ± 0.6	93.3 ± 0.1	93.3 ± 0.2	93.2 ± 0.6
2-h OGTT glucose, mg/dL	111.7 ± 0.9	112.8 ± 0.7	111.0 ± 0.9	112.4 ± 0.6	111.4 ± 0.9	109.7 ± 2.3	111.8 ± 0.6	112.7 ± 1.0	113.1 ± 2.6
Fasting insulin, mU/L	11.1 ± 0.3	12.2 ± 0.2	13.0 ± 0.3	12.1 ± 0.2	11.9 ± 0.3	13.0 ± 0.9	12.1 ± 0.2	12.0 ± 0.4	11.6 ± 0.8
HOMA-IR	2.58 ± 0.06	2.87 ± 0.06	3.07 ± 0.08	2.82 ± 0.05	2.78 ± 0.07	3.07 ± 0.21	2.83 ± 0.04	2.81 ± 0.08	2.74 ± 0.20
HOMA-B	140.0 ± 3.7	152.8 ± 4.5	155.2 ± 4.0	148.5 ± 2.3	149.5 ± 6.85	154.4 ± 10.7	149.5 ± 2.7	148.1 ± 4.7	139.4 ± 7.7
HbA1c, %	5.38 ± 0.01	5.40 ± 0.01	5.45 ± 0.01	5.40 ± 0.01	5.42 ± 0.01	5.41 ± 0.03	5.41 ± 0.01	5.39 ± 0.01	5.41 ± 0.03

¹Values are means ± SEs for continuous variables and percentages ± SEs for categorical variables, unless otherwise indicated, accounting for the complex study design of the HCHS/SOL. All the variables were adjusted for age and sex. The serving size, age, sex, field center, and Hispanic heritage background are unadjusted. HbA1c, glycated hemoglobin; HCHS/SOL, Hispanic Community Health Study/Study of Latinos; HOMA-B, HOMA index for β-cell function; MET, metabolic equivalent; OGTT, oral-glucose-tolerance test.

²Median servings per day ± SE. Serving sizes for sugar-sweetened beverages and artificially sweetened beverages are 8 fl oz (240 mL), and 100% fruit juice serving size is 4 fl oz (120 mL).

Individuals who consumed the most sugar-sweetened beverages tended to be younger, men, and more often of Mexican or Puerto Rican heritage; those with high sugar-sweetened beverage consumption also had a higher consumption of artificially sweetened beverages but not of 100% fruit juice. Individuals with higher sugar-sweetened beverage consumption (>2 servings/d) tended to be smokers and had higher physical activity, higher BMI, and larger waist circumference. They also had a higher energy intake and unhealthier eating index (lower modified AHEI-2010 score) (Table 1). No difference was observed related to the family history of diabetes across sugar-sweetened beverage consumption categories. Adjusted for age and sex, the high sugar-sweetened beverage group was more likely to have prediabetes and had higher fasting glucose, fasting insulin, HOMA-IR, HOMA-B, and HbA1c.

We observed similar but slightly less significant distinctions between high and low consumers of artificially sweetened beverage categories. Of note, individuals with high artificially sweetened beverage intake were more likely than others to have a family history of diabetes and to eat healthier. Individuals consuming high amounts of 100% fruit juice were younger; more likely to be men; engaged in higher self-reported physical activity; and had higher total energy intake, lower BMI, and smaller waist circumference. They were also more likely to be of Cuban or Central American heritage. After adjusting for age and sex, no associations were found between 100% fruit juice consumption and prediabetes or glucose metabolism markers, except for inverse associations with HOMA-IR and HOMA-B.

Sugar-sweetened beverage consumption and prediabetes

Among people without diabetes, after adjusting for age, sex, Hispanic/Latino heritage, and field center, consumption of >2 servings/d (>480 mL/d) of sugar-sweetened beverages was associated with higher odds of having prediabetes [compared with <1 serving/d (<240 mL/d); OR: 1.45; 95% CI: 1.19, 1.77; P-trend < 0.001; Table 2, Model 1). After additionally adjusting for obesity, family history of diabetes, and smoking, the strength of association was slightly reduced but still statistically significant (OR: 1.35; 95% CI: 1.11, 1.64; P-trend = 0.002). With further adjustment for physical activity, modified AHEI-2010, and energy intake, as well as the consumption of artificially sweetened beverages and 100% fruit juice, the association between sugar-sweetened beverages and prediabetes remained significant (OR: 1.30; 95% CI: 1.06, 1.61; P-trend = 0.011). On the other hand, no associations were observed between artificially sweetened beverage or 100% fruit juice consumption and prediabetes in fully adjusted models. Similar results were obtained after using more detailed categories for beverage consumption (<0.5, 0.5–1, 1–1.5, 1.5–2, and >2 servings/d) as well as quintiles of beverage intake (Supplemental Tables 3 and 4).

TABLE 2.

Adjusted ORs (95% CIs) for the association between sugar-sweetened beverage consumption and prediabetes compared with normoglycemia, HCHS/SOL (2008–2011)¹

	Sugar-sweetened beverage2				Artificially sweetened beverage2				100% fruit juice2
	<1 serving/d	1–2 servings/d	>2 servings/d	P for trend	<1 serving/d	1–2 servings/d	>2 servings/d	P for trend	<1 serving/d	1–2 servings/d	>2 servings/d	P for trend
Prediabetes/normoglycemia, n	2137/1879	1715/1992	886/1292		3610/3545	960/1384	168/234		3950/4181	702/857	112/151
Model 13	Ref	1.11 (0.95, 1.28)	1.45 (1.19, 1.77)	<0.001	Ref	1.15 (0.99, 1.35)	1.05 (0.78, 1.42)	0.27	Ref	0.91 (0.77, 1.07)	0.92 (0.64, 1.33)	0.31
Model 23	Ref	1.10 (0.95, 1.28)	1.35 (1.11, 1.64)	0.002	Ref	1.13 (0.96, 1.32)	0.97 (0.71, 1.34)	0.57	Ref	0.96 (0.81, 1.13)	1.01 (0.69, 1.48)	0.79
Model 33	Ref	1.10 (0.94, 1.28)	1.30 (1.06, 1.61)	0.011	Ref	1.11 (0.95, 1.30)	0.97 (0.71, 1.32)	0.67	Ref	0.95 (0.80, 1.11)	1.05 (0.71, 1.56)	0.83

¹Values are ORs (95% CI) from a survey logistic regression. AHEI-2010, Alternate Healthy Eating Index–2010; DM, diabetes mellitus; HCHS/SOL, Hispanic Community Health Study/Study of Latinos.

²Serving sizes for sugar-sweetened beverages and artificially sweetened beverages are 8 fl oz (240 mL), and 100% fruit juice serving size is 4 fl oz (120 mL).

³Model 1 is adjusted for age, sex, Hispanic/Latino heritage, and field center. Model 2 is additionally adjusted for family history of DM, smoking, BMI, and waist circumference. Model 3 is additionally adjusted for physical activity, AHEI excluding sugar and sugar-sweetened beverages, total energy intake, and the other 2 types of beverage consumption (sugar-sweetened beverage and/or artificially sweetened beverage and/or 100% fruit juice).

As a sensitivity analysis, we examined the association between beverage consumption and prediabetes plus unrecognized/undiagnosed diabetes compared with normoglycemia. The associations showed similar trends (Supplemental Table 7) for sugar-sweetened beverage consumption, and effects for artificially sweetened beverages and 100% fruit juice remained not significant.

Sugar-sweetened beverage consumption and glucose metabolism markers in individuals without diabetes

Table 3 and Supplemental Table 8 present the adjusted mean differences in glucose metabolism markers between beverage consumption categories in individuals with normoglycemia or prediabetes and also the mean concentrations of markers according to the consumption category are depicted in Figure 1. After adjusting for age, sex, Hispanic/Latino heritage, and field center (Model 1), consumption of >2 servings/d (>480 mL/d) of sugar-sweetened beverages compared with <1 serving/d (<240 mL/d) was associated with higher fasting blood glucose (β: 1.56 mg/dL; 95% CI: 0.90, 2.21 mg/dL; P-trend < 0.001), higher fasting insulin (β: 1.75 mU/L; 95% CI: 0.86, 2.64 mU/L; P-trend < 0.001), higher log(HOMA-IR) (β: 0.16; 95% CI: 0.08, 0.23; P-trend < 0.001), higher log(HOMA-B) (β: 0.09; 95% CI: 0.02, 0.16; P-trend = 0.02), and higher HbA1c (β: 0.06% or 0.7 mmol/mol; 95% CI: 0.02, 0.08% or 0.2, 0.9 mmol/mol; P-trend < 0.001), but not 2-h OGTT glucose (β: –0.73 mg/dL; 95% CI: –3.46, 2.00 mg/dL; P-trend = 0.51). These associations were similar with additional adjustment for covariates and other beverage consumption (Models 2 and 3 in Table 3) or more refined categories of sugar-sweetened beverage consumption (Supplemental Tables 5 and 6). The association between 3 categories of sugar-sweetened beverage intake and HOMA-B was not significant after additional adjustment, but using more refined intake categories (<0.5, 0.5–1, 1–1.5, 1.5–2, and >2 servings/d) showed significant associations. In contrast, artificially sweetened beverage consumption was not associated with glucose, insulin, or HbA1c but showed an inverse association with HOMA-B [mean difference in log(HOMA-B) between >2 servings/d (>480 mL/d) and <1 serving/d (<240 mL/d) is –0.06; 95% CI: –0.14, 0.02; P-trend = 0.008]. This inverse association was consistently observed with more refined categories of artificially sweetened beverage consumption [by 5 groups (<0.5, 0.5–1, 1–1.5, 1.5–2, and >2 servings/d) or by quintiles], presented in Supplemental Tables 5 and 6. Last, 100% fruit juice consumption was not associated with any glucose metabolism markers.

TABLE 3.

Adjusted difference (95% CI) in glucose metabolism markers by beverage consumption among participants without diabetes, HCHS/SOL (2008–2011)¹

	Sugar-sweetened beverage2				Artificially sweetened beverage2				100% fruit juice2
Model3	<1 serving/d	1–2 servings/d	>2 servings/d	P for trend	<1 serving/d	1–2 servings/d	>2 servings/d	P for trend	<1 serving/d	1–2 servings/d	>2 servings/d	P for trend
Fasting glucose, mg/dL
n	3995	3694	2165		7120	2332	402		8094	1554	259
Model 1	Ref	0.71 (0.17, 1.25)	1.56 (0.90, 2.21)	<0.001	Ref	0.26 (–0.28, 0.81)	0.88 (–0.30, 2.06)	0.11	Ref	0.00 (–0.54, 0.55)	–0.27 (–1.43, 0.89)	0.76
Model 2	Ref	0.71 (0.17, 1.26)	1.39 (0.74, 2.03)	<0.001	Ref	0.16 (–0.39, 0.72)	0.72 (–0.42, 1.86)	0.20	Ref	0.12 (–0.42, 0.67)	0.01 (–1.15, 1.16)	0.77
Model 3	Ref	0.70 (0.16, 1.24)	1.31 (0.63, 1.98)	<0.001	Ref	0.10 (–0.46, 0.65)	0.67 (–0.47, 1.82)	0.30	Ref	0.07 (–0.47, 0.61)	0.18 (–0.99, 1.34)	0.70
2-h OGTT glucose, mg/dL
n	3801	3565	2095		6823	2257	381		7771	1484	253
Model 1	Ref	1.04 (–1.01, 3.10)	–0.73 (–3.46, 2.00)	0.51	Ref	–1.09 (–3.24, 1.07)	–2.64 (–6.99, 1.71)	0.16	Ref	1.00 (–1.14, 3.15)	1.38 (–3.79, 6.55)	0.37
Model 2	Ref	1.18 (–0.82, 3.18)	–0.86 (–3.36, 1.65)	0.41	Ref	–1.15 (–3.18, 0.87)	–2.89 (–7.25, 1.47)	0.11	Ref	1.16 (–1.02, 3.33)	2.32 (–2.54, 7.19)	0.20
Model 3	Ref	1.58 (–0.40, 3.57)	0.73 (–1.93, 3.39)	0.66	Ref	–0.68 (–2.7, 1.35)	–2.41 (–6.76, 1.95)	0.19	Ref	1.18 (–0.96, 3.32)	3.01 (–1.92, 7.95)	0.14
Fasting insulin, mU/L
n	3990	3688	2163		7112	2328	401		8084	1552	258
Model 1	Ref	0.94 (0.38, 1.50)	1.75 (0.86, 2.64)	<0.001	Ref	–0.21 (–0.95, 0.53)	0.95 (–0.74, 2.64)	0.49	Ref	0.06 (–0.70, 0.82)	–0.65 (–2.14, 0.83)	0.61
Model 2	Ref	0.80 (0.32, 1.28)	0.97 (0.28, 1.65)	0.01	Ref	–0.50 (–1.13, 0.12)	0.21 (–1.14, 1.55)	0.68	Ref	0.49 (–0.22, 1.20)	0.14 (–0.99, 1.26)	0.25
Model 3	Ref	0.64 (0.17, 1.12)	0.81 (0.09, 1.52)	0.04	Ref	–0.50 (–1.15, 0.14)	0.29 (–1.03, 1.6)	0.69	Ref	0.45 (–0.29, 1.19)	0.32 (–0.81, 1.46)	0.21
log(HOMA-IR)
n	3990	3688	2163		7112	2328	401		8084	1552	258
Model 1	Ref	0.08 (0.03 ,0.13)	0.16 (0.08, 0.23)	<0.001	Ref	–0.04 (–0.09, 0.02)	0.03 (–0.09, 0.15)	0.89	Ref	0.00 (–0.05, 0.05)	–0.09 (–0.22, 0.05)	0.27
Model 2	Ref	0.07 (0.03, 0.11)	0.1 (0.04, 0.15)	0.002	Ref	–0.06 (–0.11, –0.02)	–0.04 (–0.13, 0.05)	0.07	Ref	0.03 (–0.01, 0.07)	–0.02 (–0.13, 0.08)	0.59
Model 3	Ref	0.06 (0.02 ,0.1)	0.09 (0.03, 0.15)	0.007	Ref	–0.06 (–0.11, –0.02)	–0.03 (–0.12, 0.05)	0.06	Ref	0.03 (–0.01, 0.07)	0.01 (–0.09, 0.11)	0.32
log(HOMA-B)
n	3988	3686	2162		7107	2328	401		8080	1551	258
Model 1	Ref	0.06 (0.01, 0.10)	0.09 (0.02, 0.16)	0.02	Ref	–0.05 (–0.10, –0.00)	–0.01 (–0.12, 0.10)	0.37	Ref	–0.01 (–0.06, 0.04)	–0.08 (–0.21, 0.05)	0.24
Model 2	Ref	0.05 (0.01, 0.08)	0.04 (–0.02, 0.10)	0.23	Ref	–0.07 (–0.11, –0.03)	–0.07 (–0.15, 0.02)	0.007	Ref	0.02 (–0.02, 0.06)	–0.02 (–0.12, 0.07)	0.81
Model 3	Ref	0.03 (–0.01, 0.07)	0.04 (–0.02, 0.10)	0.28	Ref	–0.07 (–0.11, –0.02)	–0.06 (–0.14, 0.02)	0.008	Ref	0.02 (–0.02, 0.07)	0.00 (–0.10, 0.09)	0.50
HbA1c, %
n	3991	3685	2164		7110	2329	401		8081	1555	257
Model 1	Ref	0.02 (0.00, 0.04)	0.06 (0.03, 0.09)	<0.001	Ref	0.01 (–0.01, 0.04)	–0.01 (–0.05, 0.04)	0.81	Ref	–0.01 (–0.03, 0.01)	0.01 (–0.06, 0.07)	0.56
Model 2	Ref	0.02 (–0.01, 0.04)	0.05 (0.02, 0.08)	0.001	Ref	0 (–0.02, 0.03)	–0.02 (–0.06, 0.02)	0.45	Ref	0.00 (–0.02, 0.02)	0.02 (–0.04, 0.08)	0.74
Model 3	Ref	0.01 (–0.01, 0.03)	0.03 (0.00, 0.06)	0.02	Ref	0.00 (–0.03, 0.02)	–0.02 (–0.07, 0.02)	0.28	Ref	–0.01 (–0.03, 0.01)	0.02 (–0.04, 0.08)	0.97

¹Values are mean differences (95% CIs) from a survey linear regression. AHEI-2010, Alternate Healthy Eating Index–2010; DM, diabetes mellitus; HbA1c, glycated hemoglobin; HCHS/SOL, Hispanic Community Health Study/Study of Latinos; HOMA-B, HOMA index for β-cell function; OGTT, oral-glucose-tolerance test.

²Serving sizes for sugar-sweetened beverages and artificially sweetened beverages are 8 fl oz (240 mL), and 100% fruit juice serving size is 4 fl oz (120 mL).

³Model 1 is adjusted for age, sex, Hispanic/Latino heritage, and field center. Model 2 is additionally adjusted for family history of DM, smoking, BMI, and waist circumference. Model 3 is additionally adjusted for physical activity, AHEI excluding sugar and sugar-sweetened beverages, total energy intake, and the other 2 types of beverage consumption (sugar-sweetened beverage and/or artificially sweetened beverage and/or 100% fruit juice).

FIGURE 1.

Blood markers of glucose metabolism by beverage consumption among participants without diabetes, HCHS/SOL (2008–2011). Data are the mean values of (A) fasting glucose, (B) 2-h OGTT glucose, (C) fasting insulin, (D) HOMA-IR, (E) HOMA-B, and (F) HbA1c, adjusted for age, sex, Hispanic/Latino heritage, field center, family history of diabetes, smoking, BMI, waist circumference, physical activity, modified AHEI-2010, total energy intake, and the other 2 types of beverage consumption (sugar-sweetened beverage and/or artificially sweetened beverage and/or 100% fruit juice) in survey linear regressions. Error bars indicate 95% CIs of the adjusted means. Adjusted means of HOMA-IR and HOMA-B were calculated in log-scale and then back-transformed to the original scale. P for trend was < 0.05 for the association between sugar-sweetened beverage consumption with fasting glucose, fasting insulin, log(HOMA-IR), log(HOMA-B), and HbA1c. Serving sizes for sugar-sweetened beverages and artificially sweetened beverages are 8 fl oz (240 mL), and 100% fruit juice serving size is 4 fl oz (120 mL). AHEI-2010, Alternate Healthy Eating Index–2010; HbA1c, glycated hemoglobin; HCHS/SOL, Hispanic Community Health Study/Study of Latinos; HOMA-B, HOMA index for β-cell function; OGTT, oral-glucose-tolerance test.

Discussion

Our cross-sectional study confirmed the positive association between sugar-sweetened beverage consumption and prediabetes in a diverse cohort of US Hispanic/Latino adults. High artificially sweetened beverage intake was associated with low HOMA-B but not with prediabetes or other glucose metabolism markers. Consumption of 100% fruit juice was not significantly associated with prediabetes and glucose metabolism markers. Although it has been reported that Hispanics/Latinos tend to consume high amounts of sugar-sweetened beverages, including traditional aguas frescas or fruit juices (jugos, batidos), and have higher T2D prevalence, studies examining the association of consumption of sugar-sweetened beverages in this diverse population are limited, making our study a significant contribution to the current literature.

Consistent with associations reported from other race/ethnic populations, we found significant associations of sugary beverages with prediabetes and glucose metabolism markers. In the Framingham Offspring Study of middle-aged non-Hispanic white persons, sugar-sweetened beverage consumption was associated with higher fasting insulin and HOMA-IR but not with fasting glucose (11). Following up the same cohort for a median of 14 y, 6 servings/wk of sugar-sweetened beverages was associated with a 46% higher risk of prediabetes and a greater increase of HOMA-IR compared with nonconsumers (10). In the Nurses’ Health Study of predominantly white, non-Hispanic US women, higher HbA1c and fasting insulin concentrations were associated with more frequent consumption of sugar-sweetened beverages (12). A prior study of adults in Mexico without diabetes estimated that fasting glucose was increased by 0.56 mmol/L (10 mg/dL) per additional serving of sugar-sweetened beverage (22).

In our study, the only glucose metabolism marker that was not consistently associated with sugar-sweetened beverage consumption was 2-h OGTT glucose. This may be related to high within-subject biological variation in 2-h OGTT glucose (e.g., 20.2% within-subject CV in HCHS/SOL) (34, 37). Alternatively, it is possible that people with high sugar-sweetened beverage consumption have high basal concentrations of glucose, insulin, and HbA1c, but maintain relatively normal post-load glycemic concentrations, which is something that might be investigated in future studies.

Although artificially sweetened beverages do not add calories to the diet, it has been extensively debated whether intake of artificially sweetened beverages or artificial sweeteners may be favorable or unfavorable. High artificially sweetened beverage consumption has been associated with increased stroke incidence or total and cardiovascular disease mortality in postmenopausal US women from the Women's Health Initiative and in women from the Nurses’ Health Study (38, 39). In relation to glycemic outcomes, it is difficult to draw a clear conclusion from the available study results. In our study of US Hispanic/Latino adults, artificially sweetened beverage intake was not associated with prediabetes risk or unfavorable glucose metabolism markers, although there was a trend for high consumers to have lower HOMA-B. Studies examining the association with β-cell function using HOMA-B are limited. However, 1 study in Brazilian adults using cross-sectional data, including newly diagnosed diabetes, showed that greater consumption of artificially sweetened beverages was associated with greater odds of diabetes, higher fasting glucose, and decreased β-cell function among normal-weight individuals but not among obese or overweight individuals (40). The consumption of artificially sweetened beverages was associated with increased incident diabetes in prospective epidemiological studies, including the meta-analysis of 17 prospective cohorts (4) and the NOMAS (in which 53% of participants were US Hispanic/Latino adults) (24). On the other hand, no prospective association of artificially sweetened beverage intake was found with incident prediabetes and HOMA-IR change in middle-aged US non-Hispanics in the Framingham Offspring Study (10). Similarly, no cross-sectional association was observed with prediabetes, fasting glucose, fasting insulin, or HOMA-IR among people without diabetes in the Maastricht study in Netherlands (13) or in the Framingham Offspring Study (11). In a study of US non-Hispanic white women using the cross-sectional data from the Nurses’ Health Study, artificially sweetened beverage consumption was not associated with HbA1c and fasting insulin but was associated with increased adiponectin concentrations (12). Adiponectin is the hormone involved in regulating glucose concentrations and breaking down fatty acids, playing a crucial protective role against diabetes and atherosclerosis (41). The conflicting results among studies might be explained by confounding such that the artificial beverage consumers might be more conscious of health (e.g., eat healthier). However, some data show that artificial sweeteners in the food without energy may promote hunger, and there is no long-term difference in overall energy intake despite reduced calories from artificial sweeteners (42). This may be explained by the fact that artificial sweeteners do not effectively stimulate the secretion of glucagon-like peptide-1, an incretin hormone secreted in the intestine in response to carbohydrate intake (42). Hence, additional avenues for research on artificial sweeteners may include the effects of consuming these ingredients on the gut, gut microbiota, or food preferences and the interactive effect of food consumed along with artificial sweeteners (13, 43–45).

We observed that consumption of 100% fruit juice was not associated with prediabetes or any of the glucose metabolism markers examined in this study. This lack of an association might result from the small contribution to total sugar intake by fruit juice in our study population, in which only 3.3% of participants consumed >2 servings of fruit juice in a day, whereas 27.5% of participants consumed >2 servings of sugar-sweetened beverages. Previous studies have shown inconsistent results. In a meta-analysis of prospective studies, one serving increase of fruit juice consumption was associated with a 7% increase in incident diabetes, but this association did not persist among the subset of studies that more accurately ascertained diabetes (4). Among young and middle-aged US white non-Hispanic health professional women (aged 24–44 y) followed for 8 y in the Nurses’ Health Study II (baseline in 1991), there was no prospective association between fruit juice intake and diabetes incidence (5). By contrast, among the middle-aged health professional women (aged 38–66 y) over the 18 y of follow-up in the Nurses’ Health Study (baseline in 1984 with diet questionnaire), 1 additional serving per day of fruit juice was associated with an 18% greater hazard of developing diabetes (46). Participants without diabetes consuming higher amounts of fruit juice showed increased HbA1c and decreased adiponectin based on the cross-sectional data in the Nurses’ Health Study (12). Among US non-Hispanic whites from the Framingham Offspring Study, fruit juice intake is inversely associated with fasting glucose but not fasting insulin or HOMA-IR (11). The differences in ascertainment of diabetes status, definition and measurement method of fruit juice consumption, content of fruit juice, study design, and follow-up time may account for the discordant results across studies. In particular, the amount of sugars and antioxidants such as polyphenols differs across specific fruits, and it may be challenging to isolate the effects of sugar compared with other nutrients from fruit juice, resulting in heterogeneous associations (9).

Last, in our study, participants with higher consumption of sugar-sweetened beverages (>2 servings/d) also demonstrated other unhealthy behaviors, including smoking and unhealthy or lower quality diet eating (lower modified AHEI-2010 score). This is consistent with other reports showing that unhealthy behaviors usually co-occur (47–49). The co-occurrence of unhealthy behaviors might represent a more rooted behavioral problem that might warrant interventions that go beyond culturally adapted behavioral interventions. Furthermore, clustering of unhealthy behaviors has been associated with lower socioeconomic status (47). Hispanic/Latinos in the United States are also disproportionally impacted by poverty, with 15.7% living in poverty (50). Interventions addressing clustering of unhealthy behaviors and factors that take into account income might yield a higher impact with regard to prevention of diabetes.

The findings of this study must be interpreted considering the study limitations, including cross-sectional analysis of observation study data in general and the possibility of reverse causality because some participants may have known their prediabetes status and may have made changes to their dietary habits. Furthermore, bias may be introduced by the residual confounding from the heterogeneity in health conditions, health consciousness, and genetic predisposition to diabetes or risk factors. Also, the participants’ report of beverage consumption might be subject to social desirability bias, and there may be underestimation in the dietary assessments, although 24-h recall demonstrates the least underestimation and variation compared with FPQ, food record, or food histories (51). Our study has many strengths, including a large sample with representation of the diverse Hispanic/Latino adult populations in the United States and lab measurement of glucose metabolic markers. The study used the beverage intake estimated from two 24-h recalls complemented with intake frequency in the following year from the FPQ—a method shown to improve accuracy of dietary intake (29).

In conclusion, in this diverse sample of Hispanic/Latino adults in the United States, higher consumption of sugar-sweetened beverages was associated with an increased risk of prediabetes. Given the high burden of T2D in the Hispanic/Latino population in the United States, additional studies are needed to better quantify the dose–response relation between sugar-sweetened beverage consumption and prediabetes/diabetes risk and glucose metabolism markers. Further investigation of the health effects of artificial sweeteners and/or artificially sweetened beverages in consideration of the food context (carbohydrate rich, fat rich, none, etc.) consumed together is warranted.

Notes

The Hispanic Community Health Study/Study of Latinos (HCHS/SOL) is a collaborative study supported by contracts from the National Heart, Lung, and Blood Institute (NHLBI) to the University of North Carolina (HHSN268201300001I/N01-HC-65233), University of Miami (HHSN268201300004I/N01-HC-65234), Albert Einstein College of Medicine (HHSN268201300002I/N01-HC-65235), University of Illinois at Chicago (HHSN268201300003I/N01-HC-65236 (Northwestern University), and San Diego State University (HHSN268201300005I/N01-HC-65237). The following institutes/centers/offices have contributed to the HCHS/SOL through a transfer of funds to the NHLBI: National Institute on Minority Health and Health Disparities, National Institute of Deafness and Other Communications Disorders, National Institute of Dental and Craniofacial Research, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of Neurological Disorders and Stroke, and NIH Institution–Office of Dietary Supplements.

Author disclosures: The authors report no conflicts of interest.

Supplemental Tables 1–8 and Supplemental Figure 1 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn/.

Abbreviations used: AHEI-2010, Alternate Healthy Eating Index–2010; FPQ, food propensity questionnaire; HbA1c, glycated hemoglobin; HCHS/SOL, Hispanic Community Health Study/Study of Latinos; HOMA-B, HOMA index for β-cell function; NOMAS, Northern Manhattan Study; OGTT, oral-glucose-tolerance test; T2D, type 2 diabetes.

Data Availability

Data described in the manuscript, code book, and analytic code will be made available upon request pending application and approval by HCHS/SOL in addition to a Data and Materials Distribution Agreement.