Prospective association of soft drink consumption with depressive symptoms

Abstract

Objective

Consumption of soft drinks has become a serious public health issue worldwide. However, prospective evidence is limited regarding the relationship between soft drink consumption and depression, especially in Asia. Here, we investigated the prospective association of soft drink consumption with the development of depressive symptoms.

Methods

We evaluated an occupational cohort of 935 adults in Japan (2012-2016), who were free from depressive symptoms at baseline and attended a 3-year follow-up assessment. Soft drink consumption was assessed using a self-administered diet history questionnaire. Depressive symptoms were assessed using the Center for Epidemiologic Studies Depression scale. Odds ratios (OR) and 95% confidence intervals (CI) were estimated from multivariate logistic regression analysis controlling for sociodemographic, lifestyle, dietary, and occupational covariates.

Results

Over the 3-year study period, 16.9% (158 cases) of the study participants reported depressive symptoms. Higher soft drink consumption was associated with higher odds of depressive symptoms. The multivariable-adjusted OR (95% CI) was 1.91 (95% CI: 1.11-3.29; p for trend = 0.015) when comparing soft drink consumption of ≥ 4 cups/week with consumption of < 1 cup/week.

Conclusion

Our results suggest that a greater consumption of soft drinks would increase the likelihood of exhibiting depressive symptoms.

Abbreviations

BDHQ

brief self-administered diet history questionnaire

BMI

body mass index

CES-D

Center for Epidemiologic Studies Depression scale

CIs

confidence intervals

MET

metabolic equivalent

OR

odds ratios

Introduction

Depression is an important public health problem worldwide. Globally, more than 300 million people of all ages suffer from depression [1]. Depression is projected to be ranked as one of the three leading cause of disease burden in the world by the year 2030 [2]. In addition, depression imposes a substantial economic burden on society [3]. To curb the public health burden, modifiable risk factors for depression need to be identified and characterized in diverse populations. In addition to regular exercise [4], alcohol use [5], and smoking [6], dietary factors may play important roles in the development of depressive symptoms [7–11].

The excessive consumption of soft drinks has become a serious public health issue worldwide [12, 13]. A meta-analysis showed that higher consumption of sugar-sweetened beverages was positively associated with depression [14]. However, that meta-analysis included only 2 cohort studies [15, 16], both of which did not control for potentially important confounders, including dietary (e.g., coffee) and lifestyle factors (e.g., sleep duration) [15, 16]. To our knowledge, no study has prospectively investigated this association in Asian populations, which consume less soft drinks than western populations but there is a rapid increase in soft drink consumption across age groups [17, 18]. In addition, there have been no studies of an occupational population in which middle-aged adults are at high risk of developing depression, the leading cause of long-term sick leave [19]. Additional evidence from Asian workers will be crucial for characterizing the association between soft drink consumption and depression symptoms. We thus aimed to investigate the prospective association between soft drinks and the likelihood of exhibiting depressive symptoms in an occupational cohort in Japan.

Methods

Study procedure

The Furukawa Nutrition and Health Study as part of the Japan Epidemiology Collaboration on Occupational Health Study was conducted by recruiting workers from two manufacturing sites of a company in Chiba and Kanagawa Prefectures. In this current study, we evaluated participants who attended two surveys three years apart (April 2012 and 2015 in Chiba; and May 2013 and 2016 in Kanagawa) and designated the two time points at each site as study baseline and endpoint, respectively. Prior to the health check-ups, all employees (n = 2828) at these factories were invited to take part in the survey and asked to fill out survey-specific questionnaires incluing diet and health-related lifestyle in general. On the day of the health check-up, research staff checked the questionnaires for completeness, and if necessary, participants were asked to clarify answers. We obtained additional data collected during periodic health examinations, including anthropometric and biochemical measurements and information on dietary history (alcohol habits). The study protocol was approved by the Ethics Committee of the National Center for Global Health and Medicine, Japan and written informed consent was obtained from all participants.

Participants

Of the 2828 total health check-up attendees, 2162 (76%) agreed to participate in the baseline survey. Of these, 2151 participants completed baseline questionnaires, assessing diet and health-related lifestyle. We excluded 610 participants with depressive symptoms (Center for Epidemiological Studies Depression Scale [CES-D] score ≥ 16; see below). To avoid potential reverse causality that chronic illness affected lifestyle, dietary habits, and depression risk over time, we further excluded participants with histories of one or more of the following conditions at baseline (n = 58): cancer (n = 15), cardiovascular disease (n = 17), chronic hepatitis (n = 2), kidney disease including nephritis (n = 8), pancreatitis (n = 2), and mental disorders (n = 18) including depression and anxiety disorder.

Of the 1483 participants, 938 participants responded to the 3-year follow-up survey (response rate: 63%). We further excluded 3 participants with missing CES-D score data [20, 21] at the follow-up survey. In total, 935 participants (834 men and 101 women aged 21–71 years) were analysed (Figure 1).

Figure 1.

Definition of depressive symptoms

Depressive symptoms were assessed using the Japanese version [20] of the CES-D [21]. The CES-D consisted of 20 questions addressing six symptoms of depression, including depressed mood, guilt or worthlessness, helplessness or hopelessness, psychomotor retardation, loss of appetite, and sleep disturbance. Each item was scored on a scale of 0 to 3 according to the frequency of the symptom, and the scores were then summed to give the total CES-D score, ranging from 0-60. The criterion validity of the CES-D scale was well established in both Western [21] and Japanese [20] populations. Participants with a CES-D score ≥ 16 were considered to have depressive symptoms.

Assessment of beverage and food intake

Dietary intake during the month before the health check was evaluated using the Brief Self-administered Diet History Questionnaire (BDHQ), which had been previously assessed for validity [22, 23] and implemented in our study population. Soft drink consumption was assessed using one beverage item question that included cola and other sweetened juice (e.g., sports drinks) but excluded 100% vegetable and fruit juice. Levels of soft drink consumption, as well as coffee and green tea, were assessed with response options ranging from none to ≥ 4 cups/day.

According to a validation study of the BDHQ using 16-day semi-weighed dietary records as the gold standard, Spearman’s correlation coefficients for soft drink consumption was 0.39 for women and 0.49 for men [22]. Dietary intake for energy and selected nutrients, such as folate, vitamin B₆, vitamin B₁₂, magnesium, zinc, and n-3 polyunsaturated fatty acids were estimated using an ad hoc computer algorithm adopted for the BDHQ and a food composition table for Japan [24]. According to the validation study, Spearman’s correlation coefficients for energy, folate, vitamin B₆, vitamin B₁₂, magnesium, zinc, and n-3 polyunsaturated fatty acids were 0.45, 0.68, 0.61, 0.50, 0.69, 0.45, and 0.48 for women and 0.40, 0.66, 0.61, 0.60, 0.71, 0.64, and 0.53 for men, respectively [23].

Assessment of other covariates

Physical activity during work and housework or in commuting and leisure time were expressed as the sum of metabolic equivalents (METs) multiplied by the duration of time (in hours) across physical activities with different levels. Psychological stress in the work environment was assessed via the Job Content Questionnaire [25], and the job strain score was calculated according to standard procedures. Body height and weight were objectively measured to the nearest 0.1 cm and nearest 0.1 kilograms, respectively, by trained staff and used to calculate body mass index (BMI; kg/m²). Marital status, diabetes status, job grade, night and rotating shift work, overtime work, smoking, alcohol consumption, sleep duration, physical activity during work and housework or in commuting to work, and leisure-time physical activity were ascertained via questionnaires of health-related lifestyle.

Statistical analysis

We compared baseline characteristics between those who completed the follow-up and those lost during the follow-up to assess the potential bias due to selective outcome assessment. Considering the distribution of soft drink consumption in this study, we created three groups: those consuming < 1 cup/week, 1-3 cups/week and ≥ 4 cups/week of soft drinks, similar to a previously conducted Chinese study [26]. Treating this soft drink variable as an ordinal variable, participant characteristics according to soft drink consumption were evaluated with linear regression analysis for continuous variables and the Mantel-Haenszel test of trends for categorical variables. We investigated the association of soft drink consumption with the likelihood of exhibiting depressive symptoms using multiple logistic regression analysis and estimated odds ratios (ORs) and 95% confidence intervals (CIs). We constructed three models to assess the influence of adjusting for baseline characteristics of a number of covariates according to the existing literature [7–11] and our previous study [27]. First, we adjusted for age, sex, and factory location (two different sites). The second model was further adjusted for marital status (married or other), smoking (never-smoker, quitter, current smoker consuming < 20 cigarettes/day, or current smoker consuming ≥ 20 cigarettes/day), alcohol consumption (non-drinker, drinker consuming 1-3 days/month, weekly drinker consuming < 1, 1 to < 2, or ≥ 2 go/day; one go contains 23 g of ethanol), physical activity at work and housework or in commuting to work (< 3, 3 to < 7, 7 to < 20, or ≥ 20 MET-hours/day), leisure-time physical activity (not engaged, > 0 to < 3, 3 to < 10, or ≥ 10 MET-hours/week), BMI (kg/m²), sleep duration (< 6, 6 to < 7, or ≥ 7 hours/day), job grade (low, middle, or high), night or rotating shift work (yes/no), overtime work (< 10, 10 to < 30, or ≥ 30 hours/month), job strain (quartile), under treatment for or history of diabetes (yes or no), coffee consumption (< 1, 1, or ≥ 2 cups/day), green tea consumption (≤ 1 cup/week, 2 to 6 cups/week, or ≥ 1 cup/day), total energy intake (kcal/day) and nutrient intake expressed as energy density (folate, magnesium, zinc, vitamin B₆, vitamin B₁₂, and n-3 polyunsaturated fatty acids). The third model is further adjusted for CES-D score at baseline (continuous). Linear trend was assessed by assigning ordinal values (0-2) to the three categories and modelling the values as a continuous variable. For participants with missing data on smoking (missing, 0.1%), leisure-time physical activity (0.7%), overtime work (0.2%), and job strain (1.3%), we imputed the data using regression model. We first imputed the data with 20 rounds of multiple imputations and then combined the estimates from each imputed data set by using Rubin’s rule.

Additionally, we repeated the analysis by using a CES-D cut-off of ≥ 19, which may be suitable for a Japanese population [28], for sensitivity analysis. Using baseline and follow-up dietary data, we evaluated those who had consumed < 1 cup/week of soft drinks at both baseline and follow-up surveys as “maintained low consumption” in comparison to those who remained in the highest (≥ 4 cups/week) category, “maintained high consumption” (n=340). Two-sided P values < 0.05 were regarded as statistically significant. All analyses were performed using the Stata statistical software package version 14.1 (StataCorp, College Station, Texas, USA).

Results

The characteristics of study participants based on soft drink consumption categories are presented in Table 1. Approximately 20% of the participants consumed ≥ 4 cups soft drinks per week while 40% of the participants consumed < 1 cup per week. Participants with higher soft drink consumption were younger and more likely to be male, a current smoker, physically active in job or leisure-time, a night or rotating shift worker, and in a low-ranking job position; they also had higher job strain and higher CES-D scores at baseline compared with those with lower soft drink consumption, and they tended to consume a larger amount of energy and carbohydrates. Conversely, they were less likely to be married or an alcohol drinker and consumed lower amounts of protein, folate, vitamin B₆, vitamin B₁₂, magnesium, zinc, and n-3 unsaturated fatty acid compared to those with lower soft drink consumption. Compared with those who participated in the follow-up survey, those who did not participate in the follow-up were older and less likely to be a current smoker and have a lower job grade and night or rotating shift work, and consumed higher amounts of fat, protein, vitamin B₆, vitamin B₁₂, zinc, and n-3 unsaturated fatty acid, while other characteristics (e.g., soft drink consumption or depressive symptoms) were similar between them (Supplemental Table 1).

Table 1. Baseline characteristics of the study population based on categories of soft drink consumption.

	Categories of soft drink consumption			P trend **
	< 1 cup/week	1-3 cups/week	≥ 4 cups/week
No. of participants	370	371	194
Age (mean ± s.d., year)	47.3± 9.0	43.4± 9.1	39.8± 9.7	<0.001
Sex (women, %)	15.4	7.6	8.3	0.002
Site (surveyed in April 2012, %)	65.7	51.2	53.6	0.001
Marital status (married, %)	76.0	68.7	62.4	0.001
Smoking status (current, %)*	24.1	24.3	42.8	<0.001
Physical activity at work and housework or in commuting to work (≥ 20 MET-hours/day, %)	15.4	24.8	31.4	<0.001
Leisure-time physical activities (≥ 10 MET-hours/week, %)*	22.9	34.4	34.9	0.001
Body mass index (mean ± s.d., kg/m2)	22.9 ± 3.2	23.1 ± 3.0	23.2 ± 3.3	0.62
Sleep duration (< 6 hours/day, %)	34.6	35.0	40.2	0.23
Job grade (low, %)	63.2	73.3	74.2	0.002
Night or rotating shift work (yes, %)	13.2	20.8	33.0	<0.001
Overtime work (≥ 30 hours/month, %)*	22.7	26.0	23.2	0.73
Job strain (mean ± s.d.) *	0.46 ± 0.10	0.47 ± 0.11	0.49± 0.12	0.002
CES-D score (mean ± s.d.)	8.0 ± 3.7	8.6± 4.1	8.9 ± 4.2	0.03
Diabetes (yes, %)	2.4	1.6	3.6	0.54
Alcohol consumption (current †, %)	57.3	54.2	46.4	0.018
Green tea (≤ 1 cup/week, %)	25.1	27.0	30.4	0.19
Coffee (< 1 cup/day, %)	34.9	36.1	32.5	0.67
Dietary nutrient intake (mean ± s.d)
Total energy (kcal/day)	1698 ± 477	1784 ± 478	1984 ± 532	<0.001
Carbohydrates (% energy)	53.3 ±8.5	55.3 ±7.2	56.9 ± 8.4	<0.001
Fat (% energy)	24.1 ± 5.6	23.8 ± 5.1	23.5 ± 5.8	0.37
Protein (% energy)	14.1 ± 2.6	13.7 ± 2.1	13.0 ± 2.5	<0.001
Folate (μg/1000 kcal)	181 ± 65	161 ± 51	144 ± 50	<0.001
Vitamin B6 (mg/1000 kcal)	0.64 ± 0.16	0.60 ± 0.13	0.57 ± 0.14	<0.001
Vitamin B12 (μg/1000 kcal)	4.5 ± 2.2	4.3 ± 1.8	4.0 ± 2.1	0.02
Magnesium (mg/1000 kcal)	133± 28	124 ± 22	117 ± 22	<0.001
Zinc (mg/1000 kcal)	4.26 ± 0.64	4.15 ± 0.53	3.94 ± 0.65	<0.001
n-3 unsaturated fatty acid (% energy)	1.20 ± 0.34	1.15 ± 0.29	1.11 ± 0.32	0.003

Abbreviations: CES-D, Center for Epidemiologic Studies Depression Scale; MET, metabolic equivalent.

^* Participants with missing data excluded: 1 for smoking status; 7 for leisure-time physical activities; 2 for overtime work; 12 for job strain.

^**Based on the Mantel-Haenszel test for categorical variables and linear regression analysis for continuous variables.

^†Alcohol consumption at least one day per week.

At the 3-year follow-up, 158 participants (16.9%) were newly identified as having depressive symptoms. In an age-, sex-, and workplace-adjusted model, higher consumption of soft drinks was associated with a higher odds of developing depressive symptoms (Table 2). After further adjustment for a wide range of covariates including sociodemographic, lifestyle, work-related, and dietary factors, multivariable-adjusted ORs of depressive symptoms according to soft drink consumption categories of < 1 cup/week, 1-3 cups/week, and ≥ 4 cups/week were 1.00 (reference), 1.62 (1.03-2.53), and 1.91 (1.11-3.29), respectively (P for trend = 0.015). This association was somewhat attenuated after additional adjustment for baseline CES-D scores (P for trend = 0.065).

Table 2. Prospective associations of soft drink consumption with depressive symptoms.

	Categories of soft drink consumption						P trend *
	< 1 cup/week		1-3 cups/week		≥ 4 cups/week
Participants (n)	370		371		194
Cases (n)	46		68		44
Adjusted odds ratios (95% CI)
Age, sex, and site **	1.00	(Reference)	1.46	(0.96-2.21)	1.90	(1.19-3.04)	0.007
+ Marital status, lifestyle, work-related, and dietary factors	1.00	(Reference)	1.62	(1.03-2.53)	1.91	(1.11-3.29)	0.015
+ Baseline CES-D score	1.00	(Reference)	1.42	(0.89-2.27)	1.68	(0.95-2.97)	0.065

Abbreviations: CES-D, Center for Epidemiologic Studies Depression Scale; CI, confidence interval

^*Linear trends across categories of soft drinks were tested using each quartile value as an ordinal.

^**The first model adjusted for age (year, continuous), sex, and factory location (surveyed in April 2012 or May 2013).^. The second model additionally adjusted for marital status (married or other), smoking (never-smoker, quitter, current smoker consuming < 20 cigarettes/day, or current smoker consuming ≥ 20 cigarettes/day), alcohol consumption (non-drinker, drinker consuming 1-3 days/month, weekly drinker consuming < 1, 1 to < 2, or ≥ 2 go/day; one go contains 23 g of ethanol), physical activity at work and housework or in commuting to work (< 3, 3 to < 7, 7 to < 20, or ≥ 20 MET-hours/day), leisure-time physical activity (not engaged, > 0 to < 3, 3 to < 10, or ≥ 10 MET-hours/week), BMI (kg/m²), sleep duration (< 6, 6 to < 7, or ≥ 7 hours/day), job grade (low, middle, or high), night or rotating shift work (yes/no), overtime work (< 10, 10 to < 30, or ≥ 30 hours/month), job strain (quartile), under treatment for or history of diabetes (yes or no), coffee consumption (< 1, 1, or ≥ 2 cups/day), green tea consumption (≤ 1 cup/week, 2 to 6 cups/week, or ≥ 1 cup/day), total energy intake (kcal/day) and nutrient intake expressed as energy density (folate, magnesium, zinc, vitamin B₆, vitamin B₁₂, and n-3 polyunsaturated fatty acids). The third model is adjusted for CES-D score at baseline (continuous).

In a sensitivity analysis using a higher cut-off (CES-D ≥ 19) for depressive symptoms, this positive association was strengthened. The multivariable-adjusted ORs were 1.00 (reference), 1.92 (1.07-3.44), and 2.62 (1.30-5.27) (P for trend = 0.006; Supplemental Table 2). Analysing both the baseline and follow-up dietary data (n total = 340, n cases = 45), the multivariable-adjusted OR of depressive symptoms (CES-D ≥ 16) for those who maintained high consumption was 1.86 (0.60-5.77) compared with those who maintained low consumption (Supplemental Table 3). We found only a small number of participants whose soft drink consumption increased (n cases=4) or decreased (n cases=6) and therefore did not estimate ORs for the change.

Discussion

In this prospective analysis of employees in Japan, we observed that higher soft drink consumption was associated with a higher likelihood of exhibiting depressive symptoms, even after adjusting for a wide range of potential confounders. To our knowledge, this is the first prospective study to examine the association between soft drink consumption and depressive symptoms in Asia.

In our analysis, the participants who consumed ≥ 4 cups/week of soft drinks had 91% higher risk of depressive symptoms compared to those who consumed < 1 cup/week. The present findings largely agree with results of a recent meta-analysis of observational studies including cross-sectional, case-control, and cohort studies [14]. In a prospective study among older adults in the United States, soft drink consumption of ≥ 4 cans/day versus no consumption was associated with a 30% higher risk of self-reported, physician-diagnosed depression [16]. Another prospective study in Spain also reported that those who were in the highest quartile of soft drink consumption, albeit not statistically significantly, had a 12% higher risk of self-reported, physician-diagnosed depression than those in the lowest quartile [15]. While there have been no prospective studies in Asia, one cross-sectional study in China reported that adults who consumed ≥ 4 cups/week of soft drinks had a two-fold higher prevalence of depressive symptoms than those who consumed < 1 cup/week [26]. Similarly, a cross-sectional study among adolescents in China showed that students consuming soft drinks ≥ 7 times/week had significantly higher degrees of depression compared with those barely consuming soft drinks [29]. The consistent evidence from various study populations supports the hypothesis that higher consumption of soft drinks may increase the risk for depressive symptoms.

The mechanism linking soft drink consumption to depressive symptoms remains to be established, but our findings might be explained by several possible intermediate pathways. First, soft drinks may cause depressive symptoms through inflammation. In an animal experiment, rats that drank cola compared with rats that drank tap water for 6 months showed higher secretion of pro-inflammatory cytokines including interleukin-6 and tumour necrosis factor-α [30]. Higher concentrations of pro-inflammatory cytokines have also been observed in individuals with depressive disorders [31] or common mental disorders [32, 33]. Second, in another animal study, a combination of high saturated fat and refined sugar, but not high-saturated fat alone, decreased brain derived neurotrophic factor [34], which mediates synaptic plasticity and is reduced in many neurodegenerative and psychiatric diseases. Third, it has been shown in an animal experiment that long-term consumption of food with additional sugar decreased the effectiveness of somatodendritic serotonin-1A receptors. This may block feedback control over the synthesis and release of serotonin in the hypothalamus and down-regulate stress control which is critical for the development of depression [35]. Fourth, aspartame, which is a common artificial sweetener in soft drinks, may disturb amino acid metabolism, protein structure and metabolism, nucleic acid integrity, neuronal function, endocrine balance, as well as alter the brain concentrations of catecholamines causing nerves to fire excessively, which indirectly causes a very high rate of neuron depolarization [36]. Taken together, consumption of soft drinks, which contain large amounts of sugar or artificial sweeteners or both, might be involved in the pathogenesis of depression. Further clinical and observational studies are needed to establish the causal mechanism, including a dose-response relationship of the exposure to sweeteners with depressive symptoms. In Japan, soft drinks account for approximately half of all sugar consumption, such as high-fructose corn syrup or isomerized sugar, and the increased demand for sugar is due to soft drinks [13], which may lead to detrimental effects, especially on depressive symptoms.

The strengths of the present study include the prospective design of the relatively homogenous population and adjustment for a wide range of known and suspected risk factors such as work-related factors or nutrients affecting depressive symptoms. However, our study also had some limitations. First, because we did not have separate data pertaining to sugar-sweetened soft drinks versus diet soft drinks, which often use artificial sweeteners and are calorie-free, it is unclear whether these subtypes would have similar associations with depressive symptoms. In addition to the limited sample size, this uncertainty and measurement errors in beverage consumption may have contributed to the lack of precise estimates in our study. Second, 37% of the participants did not take part in the follow-up survey, which may have posed selection bias. However, most baseline characteristics including soft drink and CES-D scores did not materially differ between the participants and those did not take part in the follow-up survey. Third, dietary measurements at baseline as the primary analysis might not account for long-term habitual consumption. While a longer-term longitudinal study with repeated dietary data is of interest, this limitation does not detract from our findings: our secondary analysis of the limited subset with follow-up dietary data showed a similar strength of association compared to the primary analysis. Fourth, we used the CES-D scale to assess depressive symptoms and did not include clinically diagnosed depression; our study population may also be affected by the healthy-worker effect in particular work environments. For these reasons, our findings might not be applicable to clinical depression or broadly applied to the general population. Nonetheless, we have no reason to assume that our findings pertaining to this positive association could represent a false-positive in healthy workers at relatively low risk of developing depression.

Conclusions

In summary, this prospective study demonstrated that frequent consumption of soft drinks was associated with a higher likelihood of depressive symptoms among workers in Japan. The present findings add to evidence that higher consumption of soft drinks increases the risk of depression symptoms.