Carmen Piernas, Deborah F Tate, Xiaoshan Wang, and Barry M Popkin

Transcription

1 Does diet-beverage intake affect dietary consumption patterns? Results from the Choose Healthy Options Consciously Everyday (CHOICE) randomized clinical trial 1 3 Carmen Piernas, Deborah F Tate, Xiaoshan Wang, and Barry M Popkin ABSTRACT Background: Little is understood about the effect of increased consumption of low-calorie sweeteners in diet beverages on dietary patterns and energy intake. Objective: We investigated whether energy intakes and dietary patterns were different in subjects who were randomly assigned to substitute caloric beverages with either water or diet beverages (DBs). Design: Participants from the Choose Healthy Options Consciously Everyday randomized clinical trial (a 6-mo, 3-arm study) were included in the analysis [water groups: n = 106 (94% women); DB group: n = 104 (82% women)]. For energy, macronutrient, and food and beverage intakes, we investigated the main effects of time, treatment, and the treatment-by-time interaction by using mixed models. Results: Overall, the macronutrient composition changed in both groups without significant differences between groups over time. Both groups reduced absolute intakes of total daily energy, carbohydrates, fat, protein, saturated fat, total sugar, added sugar, and other carbohydrates. The DB group decreased energy from all beverages more than the water group did only at month 3 (P-group-bytime, 0.05). Although the water group had a greater reduction in grain intake at month 3 and a greater increase in fruit and vegetable intake at month 6 (P-group-by-time, 0.05), the DB group had a greater reduction in dessert intake than the water group did at month 6 (P-group-by-time, 0.05). Conclusions: Participants in both intervention groups showed positive changes in energy intakes and dietary patterns. The DB group showed decreases in most caloric beverages and specifically reduced more desserts than the water group did. Our study does not provide evidence to suggest that a short-term consumption of DBs, compared with water, increases preferences for sweet foods and beverages. This trial was registered at clinicaltrials.gov as NCT Am J Clin Nutr 2013;97: INTRODUCTION Low-calorie sweeteners (LCSs) 4 have gained attention as dietary tools that help reduce the energy contents of foods and beverages while maintaining their sweet taste (1, 2). However, little is understood about the effect of an increased consumption of LCSs on dietary patterns and energy intakes. LCSs and their benefits on energy balance and metabolic health have been questioned by many studies that yielded conflicting results. These studies were performed under laboratory conditions and used different vehicles for LCSs such as foods, beverages, or capsules and supplements (3 15). A number of longitudinal studies that did not account for the types of diets consumed by LCS consumers have linked LCSs with increased cardiometabolic risks (16 18). However, 2 recent studies suggested that dietary patterns associated with LCS consumption modify its effects on cardiometabolic risks (19, 20). None of these previous works addressed the more critical issue of whether LCS consumption enhances the consumption of sweet-tasting foods and beverages and consequently affects food intake and dietary patterns. Previous research that focused on sweet-taste familiarity and exposure reported a greater sweettaste preference in individuals who consumed sweet products (21 23). This effect was mediated by the amount of discretionary sweetener added, and it was shown to be equally influenced by both caloric sweeteners (CSs) and LCSs (23, 24). Furthermore, the consumption of sweetness coupled with or without energy needs to be considered. Although foods sweetened with LCSs provides sweetness and energy, LCS beverages usually contain little or no energy coupled when no other things are consumed at the same time. The intake of LCS beverages in the absence of energy has been hypothesized to affect appetite and energy intake by disrupting hormonal and neurobehavioral pathways that control hunger and satiety (4, 25 28). 1 From the Departments of Nutrition (CP, DFT, and BMP), Health Behavior and Health Education (DFT), and Biostatistics (XW), the Gillings School of Global Public Health (CP, DFT, XW, and BMP), and the Lineberger Comprehensive Cancer Center (DFT), The University of North Carolina at Chapel Hill, Chapel Hill, NC. 2 Supported by Nestle Waters USA (grant to BP and DFT), the Robert Wood Johnson Foundation (grant 70017), and the NIH (grants R01 HL104580, CPC 5 R24 HD050924, P30 DK56350, and P30 CA16086). Water was provided by Nestle Waters USA. However, they had no involvement in the study design, conduct, preparation, or review of this manuscript. 3 Address correspondence to BM Popkin, Carolina Population Center, University of North Carolina, 123 W Franklin Street, Chapel Hill, NC popkin@unc.edu. 4 Abbreviations used: AC, attention control; CHOICE, Choose Healthy Options Consciously Everyday; CS, caloric sweetener; DB, diet beverage; FDR, false-discovery rate; LCS, low-calorie sweetener; NDSR, Nutrition Data System for Research; PAR-Q, Physical Activity Readiness Questionnaire; RCT, randomized clinical trial. Received July 30, Accepted for publication December 7, First published online January 30, 2013; doi: /ajcn Am J Clin Nutr 2013;97: Printed in USA. Ó 2013 American Society for Nutrition

2 DIET-BEVERAGE INTAKE AND DIETARY PATTERNS 605 However, if LCSs contribute to obesity or are a consequence of it still remains under great debate. Although some authors claim an adverse effect on energy intake and body weight (29), others argue that these results could be explained by an increased consumption of LCSs in individuals who are already overweight and obese, and, thus, reflect an attempt to lose weight. To our knowledge, no research has previously addressed both the shortand long-term effect of beverages that contain LCSs (ie, diet sodas) compared with unsweetened beverages (ie, water) on dietary patterns, food selection, and sweetness consumption in free-living individuals. Participants from the Choose Healthy Options Consciously Everyday (CHOICE) study were randomly assigned to substitute caloric beverages with either water, diet beverages (DBs), or an attention control (AC), which did not receive any beverage intervention (30). The substitution of caloric beverages by lowcalorie beverages (DBs or water) resulted in average weight losses of w2 2.5%. Subjects in intervention groups, regardless of the type of beverages they consumed, were twice as likely as control subjects to achieve a 5% weight loss at 6 mo (30). In the current research, our aim is to investigate if dietary patterns of subjects differed between subjects in water and DB groups. We specifically tested the hypothesis that the short-term consumption of LCS beverages (DB group) enhances the consumption of sweet-tasting foods and beverages. SUBJECTS AND METHODS CHOICE randomized clinical trial Participants The CHOICE randomized clinical trial (RCT) includes 318 participants from the Raleigh-Durham area in North Carolina. Eligible participants were adult men and women y of age who were overweight or obese [BMI (in kg/m 2 ) range: ], consumed $280 kcal/d of caloric beverages, including sugarsweetened beverages, alcohol, sweetened flavored milk (excluding plain milk), fruit juice, fruit drinks, sweetened tea or coffee, and sports drinks, and were willing to introduce a dietary change as recommended by the study. The exclusion criteria were as follows: a recent weight loss $2.3 kg (5 lb), recent participation in another research project that involved a weightloss or physical activity program, pregnancy or lactation during the previous 6 mo or planned pregnancy in the next 6 mo, thyroid medication, diabetes treated with oral medication or insulin, cancer in the previous 5 y, history of heart disease or surgery, current psychiatric treatment or major psychiatric diagnoses or hospitalizations, alcohol dependence assessed by the Rapid Alcohol Problems Screen (Quantity-Frequency) Questionnaire (31), plans for moving, or inability to attend the monthly group meetings or carry out the study supplies at home (ie, DBs or water bottles). Because participants received information about the benefits of regular physical activity, the Physical Activity Readiness Questionnaire (PAR-Q) was administered to screen the readiness of participants to engage in exercise (32). Subjects who experienced heart problems, frequent chest pain, faintness, or dizziness (items 1 3 on the PAR- Q) were excluded from the study, whereas subjects who reported other heart problems (items 4 7 on the PAR-Q) had to obtain the consent of a physician to participate. The CHOICE RCT was approved by the Institutional Review Board at the University of North Carolina. The written informed consent of participants was obtained before enrollment and random assignment. Intervention and study design CHOICE is a 6-mo, 3-arm, single center, single-blind, weightloss RCT. Eligible participants were recruited and randomly assigned in cohorts to 2 dietary substitutions or a control. Cohort recruitment commonly occurs in large RCTs. This recruitment was also chosen to control for seasonality in beverage consumption such that cohort start dates varied throughout the year. A computer-assisted random assignment of participants into one of the 3 groups was generated after eligibility was confirmed. Details regarding the interventions and protocols of the original study are included in an article on the main outcomes of the trial (30). Briefly, treatment groups were told to substitute $2 servings/d ($200 kcal) of caloric sweetened beverages with either water (water group) or DBs (DB group) that contained LCSs. As in previous studies (33), the CHOICE RCT provided four ml (12 16 oz) of single-serving beverages per person per day plus 2 additional servings per day to account for the consumption of study beverages of family members, which was discouraged. Beverages were provided during the intervention period and were available to pick up at the monthly group meetings. Examples of beverages provided to the water group included still and nonsweetened carbonated water [eg, Perrier and Deer Park Sparkling (Nestlé Waters)]. DBs included still and carbonated LCS beverages [eg, diet versions of Coke and Sprite (The Coca-Cola Company); Pepsi, Mountain Dew, Aquafina Splash water (PepsiCo); Dr Pepper (Dr Pepper Snapple Group); Diet Lipton Tea (Unilever); Nestea (Nestlé) and low-calorie fruit drinks that contain LCSs [eg, Tropicana Lemonade (PepsiCo)]. Participants in the water and DB groups were able to order any combination of beverages from a group-specific list of available beverages on the study website. Treatment groups were masked to the true study purpose (water or DB replacement) until after the final assessment. Beverages for each intervention group were delivered to the treatment facility within 2 d of group sessions and were stored separately to mask each group to beverages that were delivered to the other groups. As part of the intervention, monthly group sessions that included a behavioral component were delivered to participants to promote adherence to beverage substitutions and avoid dietary compensation. Monthly meetings were group specific and included topics related to changing beverage patterns and beverage selection. Regarding other sources of LCSs in foods and beverages, the behavioral component of the intervention did not address intakes of other LCS foods during the study, although the water group was discouraged to use beverages that contained LCSs or flavor additives to their water that contained LCSs. Dietary intake data Dietary intake data were collected at baseline and 3 and 6 mo by trained interviewers by using 2 unannounced, telephoneadministered 24-h recalls. Dietary data included one weekday plus one weekend day within a 14-d period. The Nutrition Data System for Research (NDSR) software (version 4.03_31, 2000)

3 606 PIERNAS ET AL developed by the Nutrition Coordinating Center at the University of Minnesota, Minneapolis, MN, was used to collect and analyze dietary intake data. The NDSR software incorporates a multipass interview methodology and provides standardization and quality control during the interview process (34). Energy intake, macronutrients, and food-grouping system Total daily energy (kcal) and macronutrient intake (g) were calculated as the average of the 2 d of intake for each subject in each intervention group. Total sugars included glucose, fructose, galactose, sucrose, lactose, and maltose but excluded starch. Added sugars did not include naturally occurring sugars such as fructose in fruit but included those added in processing and preparation. Other carbohydrates (g) were calculated by subtracting total added sugars (g) from total carbohydrates (g). We also show estimated amounts of artificial sweetener consumption because the NDSR only provides estimates of the artificial sweetener content (mg) in typical products, including saccharin, aspartame, sucralose, and acesulfame K. Individual foods and beverages were grouped into 10 groups of beverages and 15 groups of foods. Beverage groups include the ones previously defined (ie, water and LCS beverages) plus CS beverages (which included sports, energy, juice drinks, and soft drinks), coffee or tea unsweetened or with LCSs, coffee or tea with sugar, milk or soy milk (including other dairy and soy-based drinks), juices (including fruit and vegetable juice), and alcoholic drinks. Food groups included dairy (including yogurt and cheese), a protein group (including meat, poultry, seafood and fish, and eggs), mixed, frozen, and fast-food meals (including fast-food sandwiches, meat- or grain-based dishes, and commercial entrees and dinners), fats and nuts, grains (including bread, cereals, salty snacks, pasta, rice, and legumes), desserts and sweeteners (including grain-based and dairy-based desserts and discretionary CSs), and fruit and vegetables. Because of the high proportion of nonconsumers of some food and beverage groups, we grouped similar items into the same beverage group (ie, coffee or tea with LCSs and unsweetened; soy milks with or without LCSs) or food group (ie, diet yogurt and regular yogurt in the dairy group; whole grains and refined grains in the grain group). Statistical analysis All analyses were performed with STATA software (version 12) (35). A 2-sided P value,0.05 was set for statistical significance. From the final sample of 315 participants, we restricted our analysis to the water and DB arms because we focused on the effect of the 2 interventions on dietary patterns over time. We use all data available for all the analyses with a final sample sizes of n = 106 in the water group and n = 104 in the DB group at baseline, n = 97 in the water group and n =93in the DB group at 3 mo, and n = 85 in the water group and n =84 in the DB group at 6 mo. Baseline demographic characteristics such as age, sex, raceethnicity (white, African American, or other), education (high school or less, some college, or college graduate plus) and BMI (weight divided by the square of height) were tested between groups by using Student s t tests for continuous variables and the chi-square test for categorical variables. The dietary intake from 2 dietary recalls was used to compute average intakes of 2 d. In cases when just 1 d of intake was provided, actual available data were used (baseline: n = 2 in the water group and n = 1 in the DB group; month 3: n = 7 in the water group and n = 10 in the DB group; month 6: n = 3 in the water group and n = 6 in the DB group). Dietary patterns were analyzed in the current work by using measures of total energy intake, macronutrient intake, and food- and beverage-group intake. Thus, we studied the following outcomes in separate statistical models: 1) energy intake (kcal), 2) macronutrient intake [kcal (%)], 3) beverage intake (ml), and 4) food intake (kcal) (Tables 1 and 2). Baseline differences in energy, macronutrient, and food and beverage intakes were tested by using Student s t test. For each outcome separately, we investigated the main effect of time and treatment-by-time interaction by using mixed-effect models at each time point with an unstructured dependence structure (36, 37). In subsequent analyses, we tested if there were differences by including the baseline energy intake as a covariate in the model by using the energy partition model, which was controlled for baseline kilocalories but with kilocalories excluded from the food group that was being modeled. We also performed additional adjustments by age and sex to the energy-adjusted mixed-effect models for the same outcomes. As a secondary analysis, we performed corrections for multiple comparisons for each set of outcomes by using the Benjamini-Hochberg method, which controlled for the false-discovery rate (FDR) (38). RESULTS A flow diagram of the eligibility, enrollment, random assignment, and follow-up of study participants is included in the main study (30). Briefly, of the 3435 potential eligible participants, 2914 subjects were ineligible for various reasons, with the most common being because their intake of sugar-sweetened beverages was,280 kcal/d (n = 1391). A total of 521 subjects were invited, and 318 subjects were eventually randomly assigned to one of the following 3 groups: water (n = 108), DB (n = 105), or AC (n = 105). Retention rates were 92% (water), 91% (DB), and 86% (AC) at 3 mo and 84% (water), 89% (DB), and 84% (AC) at 6 mo. From the original sample of 318 eligible participants who were randomly assigned, 3 subjects (one participant from the DB group and 2 participants from the water group) were excluded from the study because they were unable to receive the intervention because of either pregnancy or relocation out of state, which made the beverage delivery infeasible. We analyzed differences at baseline between completers and noncompleters at months 3 and 6. Baseline age and percentage of men were significantly different between completers and noncompleters, respectively [age: compared with y (P = 0.004); percentage of men: % compared with 7 6 4% (P = 0.05)]. Other baseline covariates such as raceethnicity, BMI, total energy intake, energy intake from beverages and foods, and total daily intakes of carbohydrates, fat, and protein were not significantly different between completers and noncompleters. Demographic characteristics of each intervention group are presented in Table 3. None of the baseline demographic variables were significantly different between groups (P. 0.05). Within each arm, subjects were predominantly of middle age,

4 DIET-BEVERAGE INTAKE AND DIETARY PATTERNS 607 TABLE 1 Changes in energy intake and macronutrient consumption between the water and DB groups 1 P value Assessment period Time Group 3 time interaction Outcome variable and groups Baseline Month 3 Month 6 Month 3 compared with baseline Month 6 compared with baseline Month 3 to baseline Month 6 to baseline Total energy intake (kcal) 2 Water 2056 (1933, 2179) (1392, 1646) 1517 (1383, 1651),0.001,0.001 DB 2283 (2160, 2407) 1753 (1624, 1882) 1601 (1466, 1735),0.001, Total beverage intake (kcal) Water 331 (297, 366) 126 (90, 162) 131 (93, 169),0.001,0.001 DB 391 (357, 426) 121 (85, 158) 135 (97, 174),0.001, Total food intake (kcal) Water 1725 (1618, 1832) 1393 (1282, 1504) 1383 (1266, 1500),0.001,0.001 DB 1892 (1784, 2000) 1632 (1519, 1744) 1464 (1347, 1582),0.001, Carbohydrates (kcal) Water 994 (930, 1058) 699 (633, 765) 703 (633, 773),0.001,0.001 DB 1098 (1033, 1162) 778 (710, 845) 716 (646, 787),0.001, Carbohydrates (%) Water 48 (47, 50) 46 (44, 47) 46 (45, 48) DB 48 (47, 50) 45 (43, 46) 44 (43, 46) Protein (kcal) 2 Water 316 (295, 336) 268 (247, 289) 257 (234, 279),0.001,0.001 DB 352 (331, 372) 314 (292, 335) 299 (277, 321) 0.003, Protein (%) Water 16 (15, 17) 18 (17, 19) 18 (17, 18), DB 16 (15, 17) 19 (18, 19) 19 (18, 20),0.001, Fat (kcal) Water 749 (695, 804) 563 (506, 619) 557 (497, 617),0.001,0.001 DB 813 (758, 868) 672 (615, 730) 587 (526, 647),0.001, Fat (%) Water 36 (35, 38) 37 (36, 38) 36 (35, 38) DB 35 (34, 37) 37 (36, 39) 37 (35, 38) Saturated fat (kcal) Water 153 (142, 165) 121 (109, 133) 122 (110, 135),0.001,0.001 DB 155 (143, 167) 135 (123, 147) 128 (115, 141) 0.004, Total sugar (kcal) 5 Water 448 (409, 488) 270 (229, 311) 282 (239, 326),0.001,0.001 DB 493 (454, 533) 285 (243, 327) 264 (221, 308),0.001, Added sugar (kcal) 6 Water 358 (322, 393) 185 (149, 222) 200 (161, 239),0.001,0.001 DB 384 (349, 419) 195 (158, 232) 176 (137, 215),0.001, Other carbohydrates (kcal) 7 Water 636 (592, 681) 514 (467, 560) 500 (452, 549),0.001,0.001 DB 714 (668, 759) 581 (534, 628) 539 (490, 588),0.001, LCSs (mg) 8 Water 522 (278, 766) 470 (222, 719) 515 (259, 771) DB 356 (110, 602) 828 (575, 1080) 939 (681, 1197),0.001,0.001,0.001, Main effects of time and the treatment-by-time interaction were tested by using mixed-effect models at each time point. Sample sizes were as follows: n = 106 in the water group and n = 104 in the DB group at baseline, n = 97 in the water group and n = 93 in the DB group at 3 mo, and n = 85 in the water group and n = 84 in the DB group at 6 mo. In subsequent analyses that used age- and sex-adjusted mixed-effect models for the same outcomes, study results and conclusions did not change. DB, diet beverage; LCSs, low-calorie sweeteners. 2 Baseline values were significantly different between groups, P, 0.05 (Student s t test). 3 Mean; 95% CI in parentheses (all such values). 4 P value was nonsignificant after correction for multiple comparisons using the method of Benjamini and Hochberg. 5 Included glucose, fructose, galactose, sucrose, lactose, and maltose but excluded starch. 6 Did not include naturally occurring sugars such as fructose in fruit but included those added in processing and preparation. 7 Other carbohydrates (kcal) were calculated by subtracting added sugars (kcal) from total carbohydrates (kcal). 8 Included saccharin, aspartame, sucralose, and acesulfame K. Values are estimates obtained from average amounts of sweeteners contained in typical food products with LCSs.

5 608 PIERNAS ET AL TABLE 2 Changes in beverage and food consumption between the water and DB groups 1 P value Assessment period Time Group 3 time interaction Outcome variable Group Baseline Month 3 Month 6 Month 3 compared with baseline Month 6 compared with baseline Month 3 to baseline Month 6 to baseline Beverage groups (ml) Water Water (546.9, 833.6) (1718.7, ) (1419.1, ),0.001,0.001 DB (438.1, 726.7) (355.3, 656.0) (347.4, 659.1) ,0.001,0.001 LCS beverages Water (8.6, 200.7) (87.4, 286.8) 79.8 (26.0, 185.6) DB (41.5, 234.7) (982.1, ) (811.1, ),0.001,0.001,0.001,0.001 CS beverages Water (352.3, 477.1) 86.2 (21.3, 151.0) 86.4 (17.3, 155.4),0.001,0.001 DB (394.0, 519.4) 87.3 (21.1, 153.4) 80.7 (11.2, 150.2),0.001, Coffee and tea unsweetened with LCSs Water (202.5, 333.2) (125.1, 259.4) (162.1, 302.1) DB (174.9, 306.5) (166.2, 302.7) (175.1, 316.0) Coffee and tea with sugar 3 Water (231.7, 320.3) 54.3 (8.3, 100.4) 57.9 (8.8, 107.1),0.001,0.001 DB (137.5, 226.5) 35.7 (211.3, 82.7) 43.9 (25.5, 93.4),0.001, Milk and soy milk Water 86.8 (57.2, 116.4) 72.1 (41.4, 102.8) 60.9 (28.4, 93.4) DB (90.2, 149.7) 74.8 (43.5, 106.1) 65.3 (32.6, 98.0) Fruit and vegetable juice Water 60.8 (39.6, 82.1) 35.1 (13.1, 57.2) 39.8 (16.3, 63.3) DB 76.7 (55.4, 98.1) 17.9 (24.7, 40.4) 33.4 (9.7, 57.0), Alcohol Water 45.6 (16.6, 74.7) 27.0 (23.1, 57.0) 27.2 (24.5, 58.9) DB (81.6, 140.0) 29.9 (20.8, 60.5) 45.8 (13.9, 77.7),0.001, Food groups (kcal) 4 Dairy Water 88.9 (69.3, 108.5) 63.9 (43.6, 84.2) 66.6 (45.0, 88.1) DB (84.6, 124.0) 64.2 (43.5, 84.9) 87.7 (66.1, 109.4) Protein group Water (325.6, 414.6) (230.2, 322.8) (199.0, 297.8) 0.003,0.001 DB (319.1, 408.5) (258.4, 352.9) (240.9, 340.2) Mixed, frozen, fast-food meals Water (386.0, 512.2) (329.2, 460.3) (290.5, 430.0) DB (446.5, 573.4) (366.1, 499.9) (356.7, 497.0) Fats and nuts Water 96.0 (72.6, 119.4) (76.0, 124.6) 79.3 (53.4, 105.1) DB (82.1, 129.1) (85.1, 134.7) 76.8 (50.8, 102.8) Grains Water (337.8, 425.2) (240.4, 331.1) (257.7, 354.1) DB (315.4, 403.3) (305.0, 397.6) (272.3, 369.3) Desserts and sweeteners 3 Water (209.1, 301.1) (127.8, 223.3) (165.8, 267.3) DB (288.1, 380.5) (201.9, 299.4) (125.9, 228.1) 0.006, Fruit and vegetables 3 Water 82.1 (63.5, 100.7) 94.9 (75.6, 114.1) (87.7, 128.6) DB (95.1, 132.5) (96.2, 135.5) 97.0 (76.5, 117.6) Main effects of time and the treatment-by-time interaction were tested by using mixed-effect models at each time point. Sample sizes were as follows: n = 106 in the water group and n = 104 in the DB group at baseline, n = 97 in the water group and n = 93 in the DB group at 3 mo, and n = 85 in the water group and n = 84 in the DB group at 6 mo. In subsequent analyses that used energy-, age-, and sexadjusted mixed-effect models for the same outcomes, study results did not change. CS, caloric sweetener; DB, diet beverage; LCS, low-calorie sweetener. 2 Mean; 95% CI in parentheses (all such values). 3 Baseline values were significantly different between groups, P, 0.05 (Student s t test). 4 Dairy included yogurt and cheese. Protein group included meat, poultry, seafood, and eggs. Mixed, frozen, and fast-food meals included fast-food sandwiches, meat- and grain-based dishes, and commercial entrees and dinners. Grains included bread, cereals, pasta, rice, salty snacks, and legumes. Desserts and sweeteners included grain- and dairy-based desserts and discretionary caloric sweeteners. 5 P value was nonsignificant after correction for multiple comparisons by using the method of Benjamini and Hochberg.

6 DIET-BEVERAGE INTAKE AND DIETARY PATTERNS 609 TABLE 3 Baseline descriptive characteristics by intervention group Diet-beverage group (n = 104) Water group (n = 106) P value 1 Age (y) Sex [n (%)] Men 22 (21) 12 (11) Women 82 (79) 94 (89) Education [n (%)] High school or less 7 (7) 8 (8) Some college 36 (35) 45 (42) College graduate plus 61 (59) 53 (50) Race-ethnicity [n (%)] White 47 (45) 36 (34) African American 52 (50) 65 (61) Other 5 (5) 5 (5) BMI (kg/m 2 ) P values relate to Student s t tests for continuous variables and chisquare tests for categorical variables (P, 0.05) 2 Mean 6 SE (all such values). 3 Race was self-reported by participants in the Choose Healthy Options Consciously Everyday study. obese (BMI $30), highly educated, and presented a larger proportion of women and similar proportions of African Americans and whites. Changes in energy and macronutrient intakes Both intervention groups significantly decreased the total daily energy intake during the study period (P, 0.001), but there were no between-group differences (Table 1). Energy from beverages decreased significantly more in the DB group than in the water group only at month 3 (2270 kcal compared with 2205 kcal, respectively; P = 0.047; nonsignificant difference after correction for the FDR), whereas the overall energy intake from food decreased in both intervention groups during the study period (P, 0.001) without between-group differences. Energy intakes (kcal) of carbohydrates, protein, fat, saturated fat, total sugar, added sugar, and other carbohydrates also decreased within each group (P, 0.001), but there were no between-group differences at month 3 or 6 (Table 1). As expected because of intervention assignments, the intake of artificial sweeteners (mg) increased significantly more in the DB group than in the water group at both months 3 and 6 (+583 compared with 27 mg, respectively, at month 6; P, 0.001), which actually decreased the consumption of artificial sweeteners. Changes in beverage and food intakes Changes in beverage consumption (ml) between water and DB groups are shown in Table 2. As expected on the basis of intervention assignments, the water group significantly increased water intake compared with that in the DB group (+1176 ml at month 3; +884 ml at month 6; interaction group and time: P-group-by-time, 0.001), whereas the DB group showed no significant changes in water intake. Compared with the water group, the DB group significantly increased the artificially sweetened beverage intake at months 3 and 6 (+946 ml at month 3; +780 ml at month 6; P-group-by-time, 0.001). Both DB and water groups reduced their intakes of sugar sweetened beverages and coffee and tea with sugar over the study period (P, 0.001), but there were no differences between both groups at any time point. Compared with the water group, the DB group showed a significant higher decrease in alcohol only at month 3 (P-group-by-time = 0.01). Changes in food consumption (kcal) between water and DB groups are shown in Table 2. Both water and DB groups reduced dairy consumption at month 3 only (P, 0.05), but there were no differences between groups. The protein-food group (meat, fish, and eggs) decreased in the water group at months 3 and 6 and in the DB group at month 6 (P, 0.05) with no differences between groups. The intake of grains significantly decreased in the water group at months 3 and 6, whereas the DB group did not decrease the intake of grains over time. Compared with the DB group, the water group significantly decreased more in grain intake at month 3 (P-group-by-time = 0.029; nonsignificant difference after correction for FDR). Desserts decreased in both water and DB groups at month 3 (P, 0.01), but the DB group decreased significantly more in desserts at month 6 (P-group-bytime, 0.01) compared with that in the water group. The intake of fruit and vegetables increased significantly more in the water group at month 6 than in the DB group (P-group-by-time, 0.05; nonsignificant difference after correction for FDR). In subsequent analyses that used energy-, age-, and sex-adjusted mixed effect models for the same outcomes, study results did not change. DISCUSSION To our knowledge, this was the first RCT to investigate changes in dietary patterns and energy and macronutrient intakes in participants who were randomly assigned to replace caloric beverages with either water or beverages that contained LCSs. The trial allowed us to examine the effect of LCS beverages on dietary intake patterns, food choices, and consumption of sweet products. In terms of energy and macronutrients, both DB and water groups reduced absolute intakes of total daily energy, carbohydrates, fat, protein, saturated fat, total sugar, added sugar, and other carbohydrates. Overall, the macronutrient composition changed in both groups without significant differences between groups over time. In both intervention groups, the percentage of carbohydrates decreased, and the percentage of protein increased, whereas the percentage of fat increased in the DB group only. A previous nonrandomized study reported a greater benefit on energy intake when sugar-sweetened beverages were replaced with water but not with beverages that contained LCSs (39). Other short-term crossover trials did not find different energy and macronutrient intakes between subjects who drank water and subject who drank DBs (6, 10, 40), despite the fact that appetite increased in subjects who consumed aspartame-sweetened beverages (4, 6). When artificial and regular sweeteners were compared, participants in a 9-wk trial reduced their intakes of daily energy and dietary sugar when they were drinking diet soda, whereas the intake of other macronutrients was not affected (9, 14). Overall, long-term trials provided evidence that supported incomplete compensation and decreased energy intake in consumers of LCSs (9, 14). Our study supported these findings; although it should be noted that participants were focused on sweetened-beverage reduction as a means for weight loss and likely had heightened awareness toward other calories in the diet.

7 610 PIERNAS ET AL We also investigated the effect of LCS beverages on participant dietary patterns. In terms of beverage consumption, the DB group decreased more energy from all beverages than the water group did at month 3. Both DB and water groups reduced their intakes of CS beverages and coffee and tea with sugar. The DB group reduced significantly more alcohol than did the water group. Overall, the DB group decreased significantly more other CS beverages than the water group did, which indicated that participants in the DB group might have a better adherence to the treatment than the water group did. In terms of food-group consumption, the water group increased in fruit and vegetables and decreased in dairy, desserts, protein group, mixed or frozen fast-food meals, and grains. The DB group decreased in dairy, protein group, and desserts. Compared with the DB group, the water group reduced significantly more carbohydrate-rich grains at month 3 and increased significantly more fruit and vegetables at month 6, although these differences were no longer significant after correction for FDR. The DB group decreased significantly more desserts than the water group at month 6. Previous works in relation to food choices and preferences reported an increased motivation to eat more foods selected on a food-preference list in individuals who consumed preloads that contained LCSs (3). A recent observational study that used a cluster analysis reported that, in DB consumers, participants in the Western pattern had significantly higher intakes (kcal) of total energy, fast-foods, fat, refined grains, meat, mixed dishes, pizza, salty snacks, and regular soda and a lower water intake than individuals who were classified in the prudent pattern. DB consumers in the Western pattern had a higher risk of some cardiometabolic outcomes (20). Another study that compared grocery-purchasing patterns of regular and diet soft drink consumers showed that diet-soda consumers made better nutrition choices, particularly regarding the energy content (41). In our study, all participants showed positive changes in dietary patterns. The water group increased the intake of fruit and vegetables and decreased intakes of protein foods, grains, and mixed, frozen, and fast-food meals; whereas the DB group showed decreases in most caloric beverage groups and specifically reduced more desserts over the study period. In the context of a weight-loss trial in which participants were highly motivated to make a dietary substitution to reduce their caloric intake, we have shown that, although the water group, compared with DB group, increased fruit and vegetables and reduced grains and other food groups, participants who drank DBs actually reduced desserts and other CS beverages at the end of the study. Previous literature has supported the notion that an increased intake of artificial sweeteners could increase sweetness preferences and intakes. The plausibility for such an effect might be explained by a repeated exposure to sweetness as a consequence of a higher consumption of DBs and, therefore, by an increased preference for sweet taste that will translate into a higher intake of sweet foods (42). Although DB participants did not decrease their intake of carbohydrate-rich grains, our study did not support the hypothesis that LCS beverages (the DB group) increased sweetness consumption because the DB group had decreased consumption of desserts at month 6, which indicated the lack of such an effect in the context of our study. In contrast, this effect could have been a consequence of the behavioral intervention that motivated all participants to eat healthier. These differential dietary patterns between DB and water groups did not result in different weight losses or energy intakes between participants in treatment groups (30). Overall, the results of the CHOICE RCT were strengthened by the randomized design, which added a stronger evidence for causality, and also, because participants were masked to the study purpose, the study included.50% racial and ethnic minorities and had strong retention rates. Also, participants did not have a caloric dietary prescription; participants only received instructions to replace 2 servings of caloric beverages with DBs or water. This procedure was important because the study is more generalizable and applicable to a free-living population. However, our results were limited by the lack of power to detect significant differences between treatment groups. Originally, this study was designed to test differences between each treatment arm and the control, which compromised the precision of our estimates in some cases. Also, the CHOICE study was designed as a weight-loss trial, and together with beverage substitutions, several guidelines to achieve and maintain a good nutritional status were given. It was expected that participants in all groups would reduce their total energy and macronutrient intake and also their intakes of caloric beverages and several other food groups. In that context, it was difficult to find meaningful differences between water and the DB groups if participants in both groups were trying to reduce their food intakes. Also, although the study collected dietary intake objectively by using two 24-h dietary recalls, this measurement tool still collected self-reported data, which might have been affected by underreporting, especially of unhealthful foods and beverages (43, 44). In relation to the food-grouping system, we showed high proportions of nonconsumers for some food groups. We aggregated foods and beverages into similar groups, although the ability to investigate changes from regular to diet versions of products (ie, from regular to diet yogurt) was lost. Finally, our estimates of LCS consumption (mg) were approximated because the NDSR does not collect exact amounts of LCSs in products but, rather, contains average estimates of LCSs in typical products (34). In conclusion, to our knowledge, the CHOICE RCT is the first trial to compare the effect of an increased consumption of beverages that contained LCSs or water on dietary patterns, food choices, and consumption of sweet products in free-living individuals. Overall, both intervention groups reduced energy and macronutrient intakes and showed positive changes in dietary patterns. The water group had an increased intake of fruit and vegetables and decreased intake of some food groups such as protein foods, grains, and mixed, frozen, or fast-food meals. Participants in the DB group showed decreases in most caloric beverage groups and specifically reduced more desserts over the study period. In general, although this study was not designed to definitively address the issue of LCS food and beverage intakes on the habituation to sweetness and desire to consume greater sweet foods, it provides preliminary evidence to counter this argument. Studies of overall diets with greater doses of LCS foods and beverages and longer consumption periods and followups are needed to address this topic more thoroughly. We thank Frances L Dancy for administrative assistance and Karen Erickson for exceptional administrative support. We also thank CHOICE study participants and the entire CHOICE study team. The authors responsibilities were as follows CP, DFT, and BMP: study concept and design, critical revision of the manuscript for important intellectual content, obtainment of funding, and study supervision; and all

8 DIET-BEVERAGE INTAKE AND DIETARY PATTERNS 611 authors: analysis and interpretation of data, drafting of the manuscript, statistical expertise, and administrative, technical, or material support. All authors had full access to all study data and took full responsibility for the integrity of data and the accuracy of the analysis. None of the authors had a conflict of interest. REFERENCES 1. Mattes RD, Popkin BM. Nonnutritive sweetener consumption in humans: effects on appetite and food intake and their putative mechanisms. Am J Clin Nutr 2009;89: Brown RJ, de Banate MA, Rother KI. Artificial sweeteners: a systematic review of metabolic effects in youth. Int J Pediatr Obes 2010;5: Rogers PJ, Blundell JE. Separating the actions of sweetness and calories: effects of saccharin and carbohydrates on hunger and food intake in human subjects. Physiol Behav 1989;45: Blundell JE, Hill AJ. Paradoxical effects of an intense sweetener (aspartame) on appetite. Lancet 1986;1: Porikos KP, Hesser MF, van Itallie TB. Caloric regulation in normalweight men maintained on a palatable diet of conventional foods. Physiol Behav 1982;29: Black RM, Leiter L, Anderson GH. Consuming aspartame with and without taste: differential effects on appetite and food intake of young adult males. Physiol Behav 1993;53: Rolls BJ, Hetherington M, Laster LJ. Comparison of the effects of aspartame and sucrose on appetite and food intake. Appetite 1988;11 (suppl 1): Mattes R. Effects of aspartame and sucrose on hunger and energy intake in humans. Physiol Behav 1990;47: Tordoff MG, Alleva AM. Effect of drinking soda sweetened with aspartame or high-fructose corn syrup on food intake and body weight. Am J Clin Nutr 1990;51: Black RM, Tanaka P, Leiter LA, Anderson GH. Soft drinks with aspartame: effect on subjective hunger, food selection, and food intake of young adult males. Physiol Behav 1991;49: Canty DJ, Chan MM. Effects of consumption of caloric vs noncaloric sweet drinks on indices of hunger and food consumption in normal adults. Am J Clin Nutr 1991;53: Drewnowski A, Massien C, Louis-Sylvestre J, Fricker J, Chapelot D, Apfelbaum M. Comparing the effects of aspartame and sucrose on motivational ratings, taste preferences, and energy intakes in humans. Am J Clin Nutr 1994;59: Blackburn GL, Kanders BS, Lavin PT, Keller SD, Whatley J. The effect of aspartame as part of a multidisciplinary weight-control program on short- and long-term control of body weight. Am J Clin Nutr 1997; 65: Raben A, Vasilaras TH, Moller AC, Astrup A. Sucrose compared with artificial sweeteners: different effects on ad libitum food intake and body weight after 10 wk of supplementation in overweight subjects. Am J Clin Nutr 2002;76: Anton SD, Martin CK, Han H, Coulon S, Cefalu WT, Geiselman P, Williamson DA. Effects of stevia, aspartame, and sucrose on food intake, satiety, and postprandial glucose and insulin levels. Appetite 2010;55: Dhingra R, Sullivan L, Jacques PF, Wang TJ, Fox CS, Meigs JB, D Agostino RB, Gaziano JM, Vasan RS. Soft drink consumption and risk of developing cardiometabolic risk factors and the metabolic syndrome in middle-aged adults in the community. Circulation 2007; 116: Lutsey P, Steffen L, Stevens J. Dietary intake and the development of the metabolic syndrome: the Atherosclerosis Risk in Communities study. Circulation 2008;17: Nettleton JA, Lutsey P, Wang Y, Lima J, Michos E, Jacobs D Jr. Diet soda intake and risk of incident metabolic syndrome and type 2 diabetes in the Multi-Ethnic Study of Atherosclerosis (MESA). Diabetes Care 2009;32: de Koning L, Malik VS, Rimm EB, Willett WC, Hu FB. Sugarsweetened and artificially sweetened beverage consumption and risk of type 2 diabetes mellitus among men. Am J Clin Nutr 2011;93: Duffey KJ, Steffen LM, Van Horn L, Jacobs DR Jr, Popkin BM. Dietary patterns matter: diet beverages and cardiometabolic risks in the longitudinal Coronary Artery Risk Development in Young Adults (CARDIA) Study. Am J Clin Nutr 2012;95: Pangborn RM, Giovanni ME. Dietary intake of sweet foods and of dairy fats and resultant gustatory responses to sugar in lemonade and to fat in milk. Appetite 1984;5: Beauchamp GK, Moran M. Dietary experience and sweet taste preference in human infants. Appetite 1982;3: Holt SHA, Cobiac L, Beaumont-Smith NE, Easton K, Best DJ. Dietary habits and the perception and liking of sweetness among Australian and Malaysian students: a cross-cultural study. Food Qual Prefer 2000;11: Available from: Mahar A, Duizer LM. The effect of frequency of consumption of artificial sweeteners on sweetness liking by women. J Food Sci 2007;72: S Davidson TL, Swithers SE. A Pavlovian approach to the problem of obesity. Int J Obes Relat Metab Disord 2004;28: Tordoff MG. How do non-nutritive sweeteners increase food intake? Appetite 1988;11(suppl 1): Swithers SE, Davidson TL. A role for sweet taste: calorie predictive relations in energy regulation by rats. Behav Neurosci 2008;122: Brown RJ, Walter M, Rother KI. Ingestion of diet soda before a glucose load augments glucagon-like peptide-1 secretion. Diabetes Care 2009; 32: Fowler SP, Williams K, Resendez RG, Hunt KJ, Hazuda HP, Stern MP. Fueling the obesity epidemic? Artificially sweetened beverage use and long-term weight gain. Obesity (Silver Spring) 2008;16: Tate DF, Turner-McGrievy G, Lyons E, Stevens J, Erickson K, Polzien K, Diamond M, Wang X, Popkin B. Replacing caloric beverages with water or diet beverages for weight loss in adults: main results of the Choose Healthy Options Consciously Everyday (CHOICE) randomized clinical trial. Am J Clin Nutr 2012;95: Cherpitel CJ. Screening for alcohol problems in the U.S. general population: comparison of the CAGE, RAPS4, and RAPS4-QF by gender, ethnicity, and service utilization. Rapid Alcohol Problems Screen. Alcohol Clin Exp Res 2002;26: ACSM. American College of Sports Medicine s guidelines for exercise testing and prescription. 6th ed. New York, NY: Lippincott, Williams & Wilkins, Ebbeling CB, Feldman HA, Osganian SK, Chomitz VR, Ellenbogen SJ, Ludwig DS. Effects of decreasing sugar-sweetened beverage consumption on body weight in adolescents: a randomized, controlled pilot study. Pediatrics 2006;117: Schakel SF. Maintaining a nutrient database in a changing marketplace: keeping pace with changing food products a research perspective. J Food Comp Anal 2001;14: Stata Corp. Stata statistical software, release 12, College Station, TX: Stata Corp LP. 36. Littell RC, Pendergast J, Natarajan R. Modelling covariance structure in the analysis of repeated measures data. Stat Med 2000;19: Peduzzi P, Henderson W, Hartigan P, Lavori P. Analysis of randomized controlled trials. Epidemiol Rev 2002;24: Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B 1995;57: Stookey JD, Constant F, Gardner CD, Popkin BM. Replacing sweetened caloric beverages with drinking water is associated with lower energy intake. Obesity (Silver Spring) 2007;15: DellaValle DM, Roe LS, Rolls BJ. Does the consumption of caloric and non-caloric beverages with a meal affect energy intake? Appetite 2005; 44: Binkley J, Golub A. Comparison of grocery purchase patterns of diet soda buyers to those of regular soda buyers. Appetite 2007;49: Sørensen LB, Møller P, Flint A, Martens M, Raben A. Effect of sensory perception of foods on appetite and food intake: a review of studies on humans. Int J Obes Relat Metab Disord 2003;27: Tooze JA, Subar AF, Thompson FE, Troiano R, Schatzkin A, Kipnis V. Psychosocial predictors of energy underreporting in a large doubly labeled water study. Am J Clin Nutr 2004;79: Heitmann BL, Lissner L, Osler M. Do we eat less fat, or just report so? Int J Obes Relat Metab Disord 2000;24: