Abstract
Objectives. To evaluate 2-year changes in soda consumption, weight, and waist circumference.
Methods. We followed 11 218 women from the Mexican Teachers’ Cohort from 2006 to 2008. Dietary data were collected using a semiquantitative food frequency questionnaire. Weight was self-reported, and waist circumference was self-measured. We used linear regression to evaluate changes in sugar-sweetened and sugar-free soda consumption in relation to changes in weight and waist circumference, adjusting for lifestyle and other dietary factors.
Results. Compared with no change, a decrease in sugar-sweetened soda consumption by more than 1 serving per week was associated with less weight gain (−0.4 kg; 95% confidence interval [CI] = −0.6, −0.2). Conversely, relative to no change, an increase in sugar-sweetened soda by more than 1 serving per week was associated with a 0.3-kilogram (95% CI = 0.2, 0.5) increase in weight. An increase of 1 serving per day of sugar-sweetened soda was associated with a 1.0 kg (95% CI = 0.7, 1.2; P < .001) increase in weight. The results for waist circumference were similar.
Conclusions. Moderate changes in consumption of sugar-sweetened soda over a 2-year period were associated with corresponding changes in weight and waist circumference among Mexican women.
In the era of the global obesity epidemic, identifying effective policies aimed at preventing weight gain has become a priority. Growing evidence has shown that consumption of sugar-sweetened beverages is associated with weight gain and increased risk of obesity.1 Reducing consumption of these beverages in the population is currently regarded as a key component of any comprehensive obesity prevention strategy.2,3 Because new evidence from Mexico4 and the United States5 has shown that excise taxes on sugar-sweetened beverages affect sales and lower consumption of these beverages shortly after their introduction, it is important to understand the medium-term impact of changes in sugar-sweetened beverage consumption on weight.
Previous experimental and observational studies on sugar-sweetened beverage intake and weight may not have directly addressed the effects of changes in beverage intake when considering population-level policies. Two recent randomized trials on sugar-sweetened beverages relied on highly selected populations of children and obese and overweight adults.6,7 Moreover, most prospective observational studies have evaluated baseline sugar-sweetened beverage consumption and its association with weight gain instead of changes in sugar-sweetened beverage consumption.8 Evaluating changes in sugar-sweetened beverage intake on changes in adiposity in a population-based study may more appropriately address the temporal relation between limiting sugar-sweetened beverage consumption and adiposity in the general population. We used data from an ongoing large prospective study in Mexico that uniquely assessed diet on 2 occasions in a 2-year period to investigate the relation between changes in sugar-sweetened and sugar-free soda consumption and changes in body weight and waist circumference in women.
METHODS
The Mexican Teachers’ Cohort is a prospective study of 115 315 female teachers aged 25 years or older, initiated in 2006 among women from 2 states and expanded in 2008 to include women from 10 additional states across Mexico.9 At baseline, participants responded to questionnaires on sociodemographic characteristics, reproductive history, diet, lifestyle, and medical conditions. The average enrollment rate was 64%, and the median age at enrollment was 44 years. In its initial stage in 2006, 27 992 female teachers from Veracruz and Jalisco were enrolled, and 19 130 completed a follow-up questionnaire in 2008.
We excluded at baseline women with diabetes, cancer, or heart disease (n = 1481) and those aged 65 years or older (n = 80) because of difficulties in interpreting body weight in older individuals. We also excluded women with inadequate dietary information in either 2006 or 2008 (energy intake < 500 or > 3500 kcal/day, response to ≤ 70 items on the dietary questionnaire, or missing cereal section; n = 4394) and women with missing information on soda consumption in either 2006 or 2008 (n = 136). Women for whom body mass index (BMI; weight in kilograms divided by the square of height in meters) could not be calculated because of missing height or weight were also excluded (n = 1821). The final study population consisted of 11 218 women.
Assessment of Soda Consumption
Diet was measured using a 139-item semiquantitative food-frequency questionnaire (FFQ) derived from a previously validated 116-item FFQ in Mexico City. Correlation coefficients for total energy, carbohydrate, protein, and total fat intakes between the FFQ and four 4-day 24-hour recalls were 0.52, 0.57, 0.32, and 0.63, respectively.10 We added 23 food items to our questionnaire to capture regional and secular differences in food consumption.
For each food item, women were asked to specify how often, on average, they had consumed a specified commonly used unit or portion size of the food or beverage over the previous year. Ten multiple-choice frequencies of consumption were possible: 6 or more per day, 4 to 5 per day, 2 to 3 per day, 1 per day, 5 to 6 per week, 2 to 4 per week, 1 per week, 2 to 3 per month, 1 or less per month, and never. Intake of energy was calculated by multiplying the content of specified portion sizes by the frequency of consumption using a food composition table derived from a database developed by the National Institute of Nutrition and Medical Sciences in Mexico11 and the US Department of Agriculture nutrient database.12
The 2006 and 2008 FFQs included 2 items on consumption of sugar-sweetened soda and 1 item on sugar-free soda. We converted frequency responses of sugar-sweetened and sugar-free soda consumption to servings per day. We imputed consumption to zero when women reported in the same FFQ consumption of sugar-sweetened soda but had missing values for consumption of sugar-free soda and vice versa (2006: sugar-sweetened soda, n = 56; diet soda n = 1070; 2008: sugar-sweetened soda, n = 34; sugar-free soda, n = 786). Changes in beverage consumption were calculated by subtracting 2008 consumption from 2006 consumption. Participants with baseline or changes in sugar-sweetened and sugar-free soda consumption lower than the 5th percentile or higher than the 95th percentile of the distribution were assigned this value to minimize the influence of extreme values.
Assessment of Weight and Waist Circumference
On the 2006 and 2008 questionnaires, participants self-reported height (in cm) and weight (in kg) and were provided a plastic measuring tape and instructions to assess their waist circumference (in cm). We previously evaluated the reproducibility and validity of self-reported anthropometry in a subset of 3413 participants. Standardized technician measurements were well correlated with self-reported weight (r = 0.92), height (r = 0.86), and waist circumference (r = 0.78).9 We calculated changes in weight and waist circumference by subtracting self-reported measures in 2008 from those in 2006.
Assessment of Covariates
The 2006 and 2008 questionnaires asked participants to report age; lifestyle habits, such as smoking status, alcohol use, physical activity, and oral contraceptive use; postmenopausal status; hormone therapy use; and any recent physician-diagnosed disease. We assessed physical activity in 2006 by asking women to report the amount of time spent walking and doing recreational physical activity in the previous year. In 2008, the physical activity questionnaire was expanded to include the amount of time spent in either work or recreation, including walking, household cleaning, and sports. We calculated total metabolic equivalent hours per week in 2006 and 200813 on the basis of the amount and intensity of reported physical activity and categorized them into tertiles (low, medium, and high). We were unable to calculate changes in recreational physical activity because of differences in the 2006 assessment compared with the 2008 assessment. Alcohol consumption7 was calculated as the sum of the amount of alcohol in any wine, beer, and liquor consumed and converted to servings per day. On the basis of previous reports on dietary associations with weight change,14,15 we created 11 food groups in 2006 and 2008 and calculated change in consumption between 2006 and 2008. We included the following food groups: red meat (processed and unprocessed), total dairy, fruits, vegetables, nuts, yogurt, white bread, tortillas, orange or grapefruit juice, and homemade sweetened beverages (fruit-, hibiscus-, seed- or rice-based beverages known as aguas frescas). BMI was also calculated.
Statistical Analysis
We categorized 2-year changes in sugar-sweetened soda consumption on the basis of the distribution of reported consumption as decrease (< −1 servings/week), no change (−1 to +1 servings/week), and increase (> 1 servings/week). Only 20% of the women reported sugar-free soda consumption; therefore, we used serving per month instead of servings per day and categorized sugar-free soda consumption as a decrease, no change, and an increase (< −1, −1 to +1, and > 1 servings/month, respectively). To minimize the influence of extreme values of change in weight and waist circumference, we assigned the values of the 5th and 95th percentiles of the distribution to participants with values beyond this range. We used linear regression models to examine 2-year change in weight and waist circumference, comparing women who increased or decreased soda consumption with women who did not change their soda consumption over the same 2-year time interval. We modeled sugar-sweetened soda and sugar-free soda separately. However, the sugar-sweetened soda model was adjusted for baseline (continuous) and changes in sugar-free soda consumption (continuous) and vice versa. We also explored the linear relation between change in consumption of sugar-sweetened and sugar-free soda (servings/day) and 2-year changes in weight and waist circumference using a multivariable linear model.
As a result of missing waist circumference values in either the 2006 or the 2008 questionnaire (n = 1924), the waist circumference analyses included 9294 women from the analytical sample. We included the following baseline covariates in all models: age (continuous), state (Jalisco, Veracruz), smoking status (current, past, nonsmokers), alcohol consumption (consumer, nonconsumer), physical activity (low, medium, high), oral contraceptive use (ever, never), postmenopausal status (yes, no, unknown), and hormone therapy use (ever, never). Final models also included 2-year changes in smoking status (nonsmokers, smokers, quitters, starters); alcohol consumption (increased, decreased, no change); and consumption of red meat, dairy, fruits, vegetables, nuts, yogurt, white bread, tortillas, orange or grapefruit juice, and homemade sweetened beverages (continuous). These variables were chosen on the basis of expert knowledge of risk factors for weight change that could be related to changes in soda consumption. Total energy was not included in the model because we considered it to be a mediator of the relation between soda consumption and changes in weight and waist circumference.
We conducted stratified analyses to evaluate whether the observed relation between sugar-sweetened soda consumption and change in weight and waist circumference differed by baseline age (median baseline age, < 43 vs ≥ 43 years) or BMI (< 25, 25–29.9, ≥ 30). We tested heterogeneity by including cross-product terms of sugar-sweetened soda or sugar-free soda as continuous variables with categories of age and BMI. A Wald test with P < .05 was considered statistically significant. Individuals with chronic conditions may have changes in body weight. We conducted a sensitivity analysis in which women with incident diabetes, cardiovascular disease, and cancer (n = 475) were excluded from the analysis.
We imputed to the median for the following continuous variables with less than 5% missing values: changes in consumption of red meat, dairy, fruits, vegetables, nuts, yogurt, white bread, tortillas, and homemade sweetened beverages. We created a missing-value category for continuous variables with more than 5% missing values or for categorical variables, including changes in juice consumption, smoking, alcohol consumption, physical activity, oral contraceptive use, postmenopausal status, and hormone therapy use. None of these variables had more than 10% missing values. All analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC).
RESULTS
At baseline, average sugar-sweetened soda consumption was less than half a serving per day (0.4 servings/day; SD = 0.5), and only 20% reported sugar-free soda consumption (mean = 0.1 servings/week; SD = 0.1). The average age of participants at baseline was 43.3 (SD = 5.2) years. Mean baseline BMI was 27.2 (SD = 4.4), and 65% of women were overweight or obese (Table 1). Women who decreased and increased consumption of sugar-sweetened soda were more likely to be obese than women who did not change consumption. Women whose sugar-sweetened soda consumption increased were more likely to consume alcohol at baseline, yet more likely to reduce alcohol consumption over the 2-year period. Participants whose sugar-sweetened soda intake increased were more often smokers and more likely to increase dairy and red meat consumption than women who did not increase consumption (Table A, available as a supplement to the online version of this article at http://www.ajph.org).
TABLE 1—
Baseline Characteristics and 2-Year Average Change in Women: Mexican Teachers’ Cohort, 2006–2008
| Characteristic | 2006, % or Mean (SD) | 2-Year Change, % or Mean (IQR) |
| Age, y | 43.3 (5.2) | |
| State of residency | ||
| Jalisco | 22.5 | |
| Veracruz | 77.5 | |
| Menopausal status | ||
| Premenopausal | 77.3 | |
| Postmenopausal | 14.1 | |
| Unknown | 8.6 | |
| Postmenopausal hormone therapy usea | ||
| Ever | 37.0 | |
| Never | 59.1 | |
| Missing | 3.9 | |
| Oral contraceptive use | ||
| Ever | 50.6 | |
| Never | 48.7 | |
| Missing | 0.8 | |
| BMI | 27.2 (4.4) | |
| BMI category (kg/m2) | ||
| Normal (< 25) | 35.0 | |
| Overweight (25–29.9) | 42.0 | |
| Obese (≥ 30) | 23.0 | |
| Physical activityb | ||
| Low | 28.7 | 33.8 |
| Medium | 37.3 | 32.4 |
| High | 33.7 | 32.8 |
| Missing | 0.3 | 1.1 |
| Alcohol intake | ||
| Nondrinkers | 34.3 | |
| Drinkers | 62.1 | |
| Missing | 3.6 | |
| Change in alcohol intake | ||
| No change | 41.4 | |
| Increase | 26.5 | |
| Decrease | 26.3 | |
| Missing | 5.8 | |
| Smoking | ||
| Current | 8.1 | |
| Past | 9.1 | |
| Never | 81.3 | |
| Missing | 1.5 | |
| Change in smoking | ||
| Starters | 1.7 | |
| Quitters | 2.0 | |
| Smokers | 5.6 | |
| Nonsmokers | 86.4 | |
| Missing | 4.3 | |
| Energy, kcal/d | 1756 (614) | |
| Changes in food and beverage intake, servings/wk | ||
| Dairy foods | 14.9 (10.1) | −0.6 (−18.8–16.4) |
| Yogurt | 2.1 (2.4) | −0.4 (−5.2–3.0) |
| Fruit | 19.3 (15.0) | 3.2 (−22.1–29.4) |
| Vegetable | 24.6 (17.2) | −1.0 (−29.0–25.9) |
| Red meat | 6.4 (3.9) | −0.1 (−6.4–6.1) |
| Nuts | 1.0 (1.4) | 0.2 (−0.25–0.43) |
| White bread | 2.8 (2.7) | −0.2 (−4.9–5.0) |
| Corn tortillas | 10.7 (8.9) | 0.9 (−14.5–14.5) |
| Flour tortillas | 0.8 (1.4) | 0.0 (−2.4–2.5) |
| Homemade sweetened beverages | 10.6 (11.0) | −4.9 (−40.0–14.6) |
| Orange/grapefruit juice | 1.7 (2.0) | 0.3 (−4.0–4.9) |
Note. IQR = interquartile range. n = 11 218.
Among postmenopausal women only.
Values in the 2-year change column represent 2008 physical activity levels.
Between 2006 and 2008, study participants on average slightly decreased consumption of both sugar-sweetened soda (mean change of −0.6 servings/week; SD = 2.4) and sugar-free soda (mean change of −0.5 servings/month; SD = 2.2). Over the same period, women increased in weight on average by 1.1 kilograms (SD = 3.6). After adjusting for baseline and changes in dietary and lifestyle factors, women who decreased their consumption of sugar-sweetened sodas (< −1 serving/week) gained less weight (−0.4 kg; 95% confidence interval [CI] = −0.6, −0.2) relative to women who did not change intake. Conversely, women who increased their consumption of sugar-sweetened soda (> +1 servings/week) gained an average of 0.3 kilograms more weight (95% CI = 0.2, 0.5) than did women who did not change intake. In a linear model, an increase in sugar-sweetened soda of 1 serving per day was associated with a significant weight gain of 1.0 kilogram (95% CI = 0.7, 1.2). Changes in consumption of sugar-free soda were not associated with weight change (Table 2).
TABLE 2—
Weight Change According to Changes in Sugar-Sweetened and Sugar-Free Soda Consumption Between 2006 and 2008 in Women: Mexican Teachers’ Cohort
| Mean 2-Year Change (SD) |
|||||
| Changes in No. of Servings | No. | Intake, Servings | Weight, kg | Adjusted Weight Change,a kg (95% CI) | P |
| Sugar-sweetened soda, changes in servings/wk | |||||
| Decreased, < −1 | 3075 | −3.7 (2.0) | 0.8 (3.6) | −0.4 (−0.6, −0.2) | |
| No change, −1 to +1 | 6409 | −0.1 (0.4) | 1.1 (3.5) | 1.0 (Ref) | |
| Increased, > 1 | 1734 | 2.7 (1.1) | 1.5 (3.6) | 0.3 (0.2, 0.5) | |
| Increase in 1 serving/d | 1.0 (0.7, 1.2) | < .001 | |||
| Sugar-free soda, changes in servings/mo | |||||
| Decreased, < −1 | 2270 | −3.5 (3.1) | 1.2 (3.8) | −0.1 (−0.3, 0.1) | |
| No change, −1 to +1 | 7437 | 0.0 (0.0) | 1.0 (3.4) | 1.0 (Ref) | |
| Increased, > 1 | 1511 | 1.8 (0.7) | 0.9 (3.7) | −0.2 (−0.4, 0.0) | |
| Increase in 1 serving/d | 0.0 (−1.3, 1.4) | .98 | |||
Note. CI=confidence interval. The sample size was n = 11 218.
Adjusted for baseline sugar-sweetened soda consumption (continuous); baseline sugar-free soda consumption (continuous); age (continuous); state (dichotomous); 2006 and 2008 physical activity (low, medium, high); baseline smoking status (current, past, nonsmokers); alcohol consumption (consumer, nonconsumer); oral contraceptive use (ever, never); menopausal status (premenopausal, postmenopausal); postmenopausal hormone therapy use (ever, never); and changes in smoking status (nonsmokers, smokers, quitters, starters), alcohol consumption (increased, decreased, no change), and consumption of red meat, dairy, yogurt, fruit, vegetables, nuts, white bread, flour tortillas, corn tortillas, orange or grapefruit juice, and homemade sweetened beverages (continuous). Models were mutually adjusted for sugar-sweetened and sugar-free soda.
For waist circumference, we observed an overall mean 2-year increase of 1.1 centimeters (SD = 6.0). Decreasing and increasing sugar-sweetened soda consumption were also associated with significant changes in waist circumference (Table 3). Increasing sugar-sweetened soda consumption by 1 serving per day was associated with a nearly 1-centimeter increase in waist circumference (0.9 cm; 95% CI = 0.5, 1.4). Sugar-free soda consumption was inversely associated with waist circumference (−2.7 cm change per 1 serving/day increase; 95% CI = −5.2, −0.1), independent of sugar-sweetened soda consumption (Table 3).
TABLE 3—
Waist Circumference Change According to Changes in Sugar-Sweetened and Sugar-Free Soda Consumption Between 2006 and 2008 in Women: Mexican Teachers’ Cohort
| Mean 2-Year Change (SD) |
|||||
| Changes in No. of Servings | No. | Intake, Servings | Waist Circumference, cm | Adjusted Waist Circumference Change,a cm (95% CI) | P |
| Sugar-sweetened soda, changes in servings/wk | < .001 | ||||
| Decreased, < −1 | 2538 | −3.7 (2.0) | 0.7 (6.0) | −0.5 (−0.9, −0.1) | |
| No change, −1 to +1 | 5350 | −0.1 (0.4) | 1.1 (5.8) | 1.0 (Ref) | |
| Increased, > 1 | 1406 | 2.8 (1.1) | 1.5 (6.3) | 0.3 (0.1, 0.6) | |
| Increase in 1 serving/d | 0.9 (0.5, 1.4) | ||||
| Sugar-free soda, changes in servings/mo | .04 | ||||
| Decreased, < −1 | 1856 | −3.5 (3.1) | 1.2 (6.1) | −0.1 (−0.5, 0.3) | |
| No change, −1 to +1 | 6220 | 0.0 (0.0) | 1.2 (5.9) | 1.0 (Ref) | |
| Increased, > 1 | 1218 | 1.8 (0.7) | 0.4 (6.2) | −0.7 (−1.1, −0.3) | |
| Increase in 1 serving/d | −2.7 (-5.2, −0.1) | ||||
Note. CI=confidence interval. The sample size was n = 9294.
Adjusted for baseline sugar-sweetened soda consumption (continuous); baseline sugar-free soda consumption (continuous); age (continuous); state (dichotomous); 2006 and 2008 physical activity (low, medium, high); baseline smoking status (current, past, nonsmokers); alcohol consumption (consumer, nonconsumer); oral contraceptive use (ever, never); menopausal status (premenopausal, postmenopausal); postmenopausal hormone therapy use (ever, never); and changes in smoking status (nonsmokers, smokers, quitters, starters), alcohol consumption (increased, decreased, no change) and consumption of red meat, dairy, yogurt, fruit, vegetables, nuts, white bread, flour tortillas, corn tortillas, orange or grapefruit juice, and homemade sweetened beverages (continuous). Models were mutually adjusted for sugar-sweetened and sugar-free soda.
When we stratified analyses by age, we found no evidence that the relation between sugar-sweetened soda and change in weight and waist circumference differed by age (Table 4). However, we observed evidence for a stronger association between sugar-sweetened soda consumption and changes in weight with increasing BMI (P = .003). Women who were obese (BMI ≥ 30) at baseline gained 1.5 kilograms (95% CI = 1.0, 2.0) for each serving per day increase in sugar-sweetened soda consumption. For overweight women (BMI ≥ 25–29.9), the increase in weight was 1.0 kilogram (95% CI = 0.6, 1.3). Normal-weight women (BMI < 25) gained significantly less weight (0.6 kg; 95% CI = 0.3, 1.1) for each serving per day increase in sugar-sweetened soda consumption. This relationship was not observed for waist circumference. Exclusion of women who developed chronic conditions in the 2-year period did not result in relevant changes in our estimates.
TABLE 4—
Weight and Waist Circumference Change With a 1-Serving Increase in Sugar-Sweetened Soda Consumption Between 2006 and 2008 According to Baseline Age and BMI in Women: Mexican Teachers’ Cohort
| Weight Changea |
Waist Circumference Changea |
|||||
| Baseline | No. | kg (95% CI) | P for Heterogeneity | No. | cm (95% CI) | P for Heterogeneity |
| Baseline age | .05 | .67 | ||||
| ≥ 43 y | 5691 | 1.2 (0.8, 1.5) | 4699 | 1.0 (0.4, 1.7) | ||
| < 43 y | 5527 | 0.8 (0.4, 1.1) | 4595 | 0.9 (0.2, 1.5) | ||
| Baseline BMI category | .003 | .54 | ||||
| Obese (≥ 30 kg/m2) | 2580 | 1.5 (1.0, 2.0) | 2063 | 1.2 (0.3, 2.2) | ||
| Overweight (25–29.9 kg/m2) | 4715 | 1.0 (0.6, 1.3) | 3911 | 1.1 (0.4, 1.8) | ||
| Normal (< 25 kg/m2) | 3923 | 0.6 (0.3, 1.0) | 3320 | 0.6 (−0.2, 1.4) | ||
Note. BMI = body mass index; CI = confidence interval. The sample size was n = 11 218.
Adjusted for baseline sugar-sweetened soda consumption (continuous); baseline sugar-free soda consumption (continuous); age (continuous); state (dichotomous); 2006 and 2008 physical activity (low, medium, high); baseline smoking status (current, past, nonsmokers); alcohol consumption (consumer, nonconsumer); oral contraceptive use (ever, never); menopausal status (premenopausal, postmenopausal); postmenopausal hormone therapy use (ever, never); and changes in smoking status (nonsmokers, smokers, quitters, starters), alcohol consumption (increased, decreased, no change), and consumption of red meat, dairy, yogurt, fruit, vegetables, nuts, white bread, flour tortillas, corn tortillas, orange or grapefruit juice, and homemade sweetened beverages (continuous). Models were mutually adjusted for sugar-sweetened and sugar-free soda.
DISCUSSION
We found that moderate changes in consumption of sugar-sweetened soda over a 2-year period were associated with changes in weight. Decreasing consumption of sugar-sweetened soda was associated with less weight gain, and increasing consumption of sugar-sweetened soda had an opposite association. These results were similar when waist circumference was used as a measure of adiposity. The impact of changes in sugar-sweetened soda intake on weight appeared to be stronger among women who were overweight or obese at baseline relative to women who were of normal weight. Changes in sugar-free soda consumption were not associated with weight change. However, we observed an inverse association for changes in sugar-free soda intake and changes in central adiposity, although this association had a wide confidence interval.
Animal studies, randomized trials in humans, and epidemiological evidence have indicated that consumption of sugar-sweetened beverages is associated with weight gain and obesity.1 Liquid calories may suppress appetite and energy intake less than do calories consumed as solid foods.16–18 In animal models, intake of liquid calories is not fully compensated by reductions in intake of solid foods, resulting in a positive caloric balance and obesity. In short-term human feeding trials, consumption of liquid calories appeared to affect subsequent food and energy intake only minimally, in contrast to consumption of solid foods.19–21 In a blinded randomized trial in children that compared sugar-sweetened beverages with noncaloric beverages, body weight increased less in children randomized to the sugar-free beverage.7
Evidence from prospective observational studies has consistently shown that sugar-sweetened beverage consumption promotes weight gain in children and adults.8 However, most studies compare individuals’ usual sugar-sweetened beverage intake at baseline only. We uniquely determined dietary intake on 2 occasions in a 2-year period, thus directly assessing the impact of medium-term changes in consumption of sugar-sweetened soda on weight change. Our results are consistent with those of another study that followed this analytic approach. A large-scale cohort study of US women and men with a 20-year follow-up evaluated sugar-sweetened beverage consumption every 4 years.22 However, our study conducted with a Hispanic population revealed a stronger association than what was observed in that study. We observed that a 1-serving-per-day increase in consumption of sugar-sweetened soda was associated with a 1-kilogram weight gain over a 2-year period. By contrast, the study conducted in a mostly White US population observed a 0.4-kilogram increase over a 4-year period for the same increase in sugar-sweetened beverages.
It is important to note that our study only looked at sugar-sweetened soda and not sugar-sweetened beverages as a whole. Differences between studies may indicate differences in the distribution of genetic or environmental factors across populations that could influence the effect of sugar-sweetened beverages and weight gain. One such factor could be obesity. We observed that the association of sugar-sweetened soda and weight gain was stronger among women who were overweight or obese at baseline, compared with those of normal weight. Recent evidence has suggested that waist circumference may not be a better indicator of cardiometabolic risk in Hispanics relative to overall adiposity, as previously thought.23,24 Regardless, the consistency of the relationship between changes in sugar-sweetened soda intake and weight and waist circumference adds robustness to our results. We observed an inverse association for sugar-free soda consumption and waist circumference. The impact of noncaloric sweeteners on weight and waist circumference is unclear, controversial, and difficult to evaluate because increasing sugar-free soda intake may be linked to a multitude of behaviors aimed at weight loss. In addition, although there is evidence to support that sugar-free soda consumption might result in weight loss by limiting energy intake,25,26 some studies have suggested that intake of noncaloric sweetener could result in metabolic abnormalities and weight gain.27,28
Mexico and several cities in the United States recently introduced excise taxes on sugar-sweetened beverages. The tax that was introduced in Mexico appears to have reduced average sugar-sweetened beverage consumption by 12 milliliters per capita per day, or 0.03 servings per day.4 Our data show that a reduction in sugar-sweetened sodas of at least 0.14 servings per day is associated with a 2-year 0.4-kilogram (or almost 1-lb) less weight gain. Even though modest reductions in weight accumulated over time have an impact on noncommunicable disease risk,29–32 Mexico’s current 10% excise soda tax is expected to have a modest impact on the population’s metabolic health.33 Thus, there is a need to further reduce sugar-sweetened beverage consumption to reduce the incidence of type 2 diabetes and cardiovascular disease. In the face of a dramatic increase in the global burden of diabetes and other noncommunicable diseases, understanding the relation of health behaviors and obesity within a given population is key to support potential interventions for that specific population.
We uniquely evaluated the impact of 2-year changes in soda intake on anthropometry in a large population of adult women in a middle-income country using repeated dietary assessments. Our study used a previously validated dietary questionnaire for Mexico10 in a population of women for whom self-reported anthropometry has been shown to be highly reliable and valid.9 We used an analytic strategy evaluating changes in soda intake in relation to changes in anthropometric measures that minimizes the impact of measurement error, confounding, and reverse causation,34 and we were able to adjust for changes in several lifestyle factors that could affect weight gain.
Limitations
A number of limitations need to be considered. Unlike most disease outcomes that may be unknown to individuals, weight is observed and often monitored daily. Slight changes in perceived weight could cause individuals to modify their dietary habits or lifestyle (e.g., women who notice weight increase would decrease sugar-sweetened soda intake, increase sugar-free soda intake, or both), resulting in reverse causation. However, this change in beverage choice would likely attenuate the association we observed for sugar-sweetened soda. Health-seeking behavior is difficult to capture and consider in analyses of observational studies. Our finding that sugar-free soda consumption was inversely associated with central adiposity should be taken with care, because increasing sugar-free soda consumption may be strongly correlated with other behaviors that result in weight loss. Our dietary assessment cannot capture the quantity and type of sweeteners used in homemade drinks and did not assess consumption of other alternatives to soda, such as commercial fruit drinks and sports and energy drinks, which may be substitutes for soda consumption.
Last, misclassification of beverage consumption was unavoidable. Both random variation in beverage consumption and underreporting of food consumption among heavier women would also result in an underestimation of the true effect.35 We excluded a substantial number of women (about 35%) because of inadequate or incomplete data on diet or anthropometry. We compared women who remained in the study and those who were excluded and found no differences in major characteristics (Table B, available as a supplement to the online version of this article at http://www.ajph.org). We observed that the association could be stronger among women who were overweight or obese. Thus, the strength of the association may only be generalizable to populations with a relatively similar prevalence of overweight and obesity. Although our study only evaluated 2-year changes, it is likely that changes in weight and waist circumference would be larger if the changes in sugar-sweetened soda consumption are sustained over time. Although we adjusted for several lifestyle and dietary behaviors, unmeasured or residual confounding cannot be excluded. Finally, the study population consisted of Hispanic women living in Mexico, which may limit the generalizability to other populations.
Public Health Implications
Moderate changes in sugar-sweetened soda intake over a 2-year period were associated with differential weight gain among Mexican women. Results of this study provide support for existing and proposed policies to reduce sugar-sweetened soda consumption to lower the impact of obesity in the Mexican population. There is an urgent need to understand effective behavioral strategies to encourage lower consumption of sugar-sweetened sodas and to evaluate the impact of an increasing number of nonsoda sugar-sweetened beverages on health.
ACKNOWLEDGMENTS
R. López-Ridaura and M. Lajous have a nonrestricted investigator-initiated grant from AstraZeneca. D. Stern, R. López-Ridaura, and M. Lajous received limited salary support from Bloomberg Philanthropies through an institutional grant to the National Institute of Public Health in Mexico. M. Lajous received support from the Bernard Lown Scholars Program in Cardiovascular Health. This work is also supported by the American Institute for Cancer Research (05B047), Consejo Nacional de Ciencia y Tecnología (S0008-2009-1:000000000115312), and National Institutes of Health (T32 CA09001 and T32 ES07069).
We thank Antonio García-Anaya and Adriana Monge for exceptional assistance with the data management. We are extremely grateful to the Mexican Teachers’ Cohort participants. Without their participation, this study would not have been possible.
HUMAN PARTICIPANT PROTECTION
The study was approved by the institutional review board at the National Institute of Public Health in Mexico.
Footnotes
See also Popkin, p. 1702.
REFERENCES
- 1.Hu FB. Resolved: there is sufficient scientific evidence that decreasing sugar-sweetened beverage consumption will reduce the prevalence of obesity and obesity-related diseases. Obes Rev. 2013;14(8):606–619. doi: 10.1111/obr.12040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Barquera S, Campos I, Rivera JA. Mexico attempts to tackle obesity: the process, results, push backs and future challenges. Obes Rev. 2013;14(suppl 2):69–78. doi: 10.1111/obr.12096. [DOI] [PubMed] [Google Scholar]
- 3.Holub CK, Elder JP, Arredondo EM et al. Obesity control in Latin American and US Latinos: a systematic review. Am J Prev Med. 2013;44(5):529–537. doi: 10.1016/j.amepre.2013.01.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Colchero MA, Popkin BM, Rivera JA et al. Beverage purchases from stores in Mexico under the excise tax on sugar sweetened beverages: observational study. BMJ. 2016;352:h6704. doi: 10.1136/bmj.h6704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Falbe J, Thompson HR, Becker CM et al. Impact of the Berkeley excise tax on sugar-sweetened beverage consumption. Am J Public Health. 2016;106(10):1865–1871. doi: 10.2105/AJPH.2016.303362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Ebbeling CB, Feldman HA, Chomitz VR et al. A randomized trial of sugar-sweetened beverages and adolescent body weight. N Engl J Med. 2012;367(15):1407–1416. doi: 10.1056/NEJMoa1203388. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.de Ruyter JC, Olthof MR, Seidell JC et al. A trial of sugar-free or sugar-sweetened beverages and body weight in children. N Engl J Med. 2012;367(15):1397–1406. doi: 10.1056/NEJMoa1203034. [DOI] [PubMed] [Google Scholar]
- 8.Malik VS, Pan A, Willett WC et al. Sugar-sweetened beverages and weight gain in children and adults: a systematic review and meta-analysis. Am J Clin Nutr. 2013;98(4):1084–1102. doi: 10.3945/ajcn.113.058362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Ortiz-Panozo E, Yunes-Díaz E, Lajous M, Romieu I, Monge A, López-Ridaura R. Validity of self-reported anthropometry in adult Mexican women. Salud Publica Mex. 2017;59(3):266–275. doi: 10.21149/7860. [DOI] [PubMed] [Google Scholar]
- 10.Hernández-Avila M, Romieu I, Parra S et al. Validity and reproducibility of a food frequency questionnaire to assess dietary intake of women living in Mexico City. Salud Publica Mex. 1998;40(2):133–140. doi: 10.1590/s0036-36341998000200005. [DOI] [PubMed] [Google Scholar]
- 11.National Institute of Nutrition Salvador Zubirán. Composición de alimentos Mexicanos, composición química aminoácidos cálculo de aportes dietarios. Mexico City, Mexico: National Institute of Nutrition Salvador Zubirán; 1999. [Google Scholar]
- 12.US Department of Agriculture. USDA food composition databases. Available at: https://ndb.nal.usda.gov/ndb/search/list. Accessed November 5, 2015.
- 13.Ainsworth BE, Haskell WL, Whitt MC et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc. 2000;32(9, suppl):S498–S516. doi: 10.1097/00005768-200009001-00009. [DOI] [PubMed] [Google Scholar]
- 14.Mozaffarian D, Hao T, Rimm EB et al. Changes in diet and lifestyle and long-term weight gain in women and men. N Engl J Med. 2011;364(25):2392–2404. doi: 10.1056/NEJMoa1014296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Bertoia ML, Mukamal KJ, Cahill LE et al. Changes in intake of fruits and vegetables and weight change in United States men and women followed for up to 24 years: analysis from three prospective cohort studies[published correction appears in 2016;13(1):e1001956] PLoS Med. 2015;12(9):e1001878. doi: 10.1371/journal.pmed.1001878. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cassady BA, Considine RV, Mattes RD. Beverage consumption, appetite, and energy intake: what did you expect? Am J Clin Nutr. 2012;95(3):587–593. doi: 10.3945/ajcn.111.025437. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Reid M, Hammersley R, Duffy M et al. Effects on obese women of the sugar sucrose added to the diet over 28 d: a quasi-randomised, single-blind, controlled trial. Br J Nutr. 2014;111(3):563–570. doi: 10.1017/S0007114513002687. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Shearrer GE, O’Reilly GA, Belcher BR et al. The impact of sugar sweetened beverage intake on hunger and satiety in minority adolescents [corrigendum appears in Appetite 2016;100:272] Appetite. 2016;97:43–48. doi: 10.1016/j.appet.2015.11.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Beridot-Therond ME, Arts I, Fantino M et al. Short-term effects of the flavour of drinks on ingestive behaviours in man. Appetite. 1998;31(1):67–81. doi: 10.1006/appe.1997.0153. [DOI] [PubMed] [Google Scholar]
- 20.De Castro JM. The effects of the spontaneous ingestion of particular foods or beverages on the meal pattern and overall nutrient intake of humans. Physiol Behav. 1993;53(6):1133–1144. doi: 10.1016/0031-9384(93)90370-u. [DOI] [PubMed] [Google Scholar]
- 21.Rolls BJ, Kim S, Fedoroff IC. Effects of drinks sweetened with sucrose or aspartame on hunger, thirst and food intake in men. Physiol Behav. 1990;48(1):19–26. doi: 10.1016/0031-9384(90)90254-2. [DOI] [PubMed] [Google Scholar]
- 22.Pan A, Malik VS, Hao T et al. Changes in water and beverage intake and long-term weight changes: results from three prospective cohort studies. Int J Obes (Lond) 2013;37(10):1378–1385. doi: 10.1038/ijo.2012.225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ghandehari H, Le V, Kamal-Bahl S et al. Abdominal obesity and the spectrum of global cardiometabolic risks in US adults. Int J Obes (Lond) 2009;33(2):239–248. doi: 10.1038/ijo.2008.252. [DOI] [PubMed] [Google Scholar]
- 24.Qi Q, Strizich G, Hanna DB et al. Comparing measures of overall and central obesity in relation to cardiometabolic risk factors among US Hispanic/Latino adults. Obesity (Silver Spring) 2015;23(9):1920–1928. doi: 10.1002/oby.21176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Miller PE, Perez V. Low-calorie sweeteners and body weight and composition: a meta-analysis of randomized controlled trials and prospective cohort studies. Am J Clin Nutr. 2014;100(3):765–777. doi: 10.3945/ajcn.113.082826. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rogers PJ, Hogenkamp PS, de Graaf C et al. Does low-energy sweetener consumption affect energy intake and body weight? A systematic review, including meta-analyses, of the evidence from human and animal studies. Int J Obes (Lond) 2016;40(3):381–394. doi: 10.1038/ijo.2015.177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Fowler SP, Williams K, Hazuda HP. Diet soda intake is associated with long-term increases in waist circumference in a biethnic cohort of older adults: the San Antonio Longitudinal Study of Aging. J Am Geriatr Soc. 2015;63(4):708–715. doi: 10.1111/jgs.13376. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Suez J, Korem T, Zeevi D et al. Artificial sweeteners induce glucose intolerance by altering the gut microbiota. Nature. 2014;514(7521):181–186. doi: 10.1038/nature13793. [DOI] [PubMed] [Google Scholar]
- 29.Willett WC, Manson JE, Stampfer MJ et al. Weight, weight change, and coronary heart disease in women. Risk within the “normal” weight range. JAMA. 1995;273(6):461–465. doi: 10.1001/jama.1995.03520300035033. [DOI] [PubMed] [Google Scholar]
- 30.Colditz GA, Willett WC, Rotnitzky A et al. Weight gain as a risk factor for clinical diabetes mellitus in women. Ann Intern Med. 1995;122(7):481–486. doi: 10.7326/0003-4819-122-7-199504010-00001. [DOI] [PubMed] [Google Scholar]
- 31.Rexrode KM, Hennekens CH, Willett WC et al. A prospective study of body mass index, weight change, and risk of stroke in women. JAMA. 1997;277(19):1539–1545. doi: 10.1001/jama.1997.03540430051032. [DOI] [PubMed] [Google Scholar]
- 32.Eliassen AH, Colditz GA, Rosner B et al. Adult weight change and risk of postmenopausal breast cancer. JAMA. 2006;296(2):193–201. doi: 10.1001/jama.296.2.193. [DOI] [PubMed] [Google Scholar]
- 33.Sánchez-Romero LM, Penko J, Coxson PG et al. Projected impact of Mexico’s sugar-sweetened beverage tax policy on diabetes and cardiovascular disease: a modeling study. PLoS Med. 2016;13(11):e1002158. doi: 10.1371/journal.pmed.1002158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Smith JD, Hou T, Hu FB et al. A comparison of different methods for evaluating diet, physical activity, and long-term weight gain in 3 prospective cohort studies. J Nutr. 2015;145(11):2527–2534. doi: 10.3945/jn.115.214171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Willett WC. The search for truth must go beyond statistics. Epidemiology. 2008;19(5):655–656. doi: 10.1097/EDE.0b013e318181b877. discussion 657–658. [DOI] [PubMed] [Google Scholar]