Study population
Relevant clinical and echocardiographic data were obtained from Kangbuk Samsung Health Study (KSHS). KSHS is a cohort study to investigate the medical data of Koreans who have received medical health check-up in Kangbuk Samsung Hospital. Korea’s Industrial Safety and Health law orders that all of Korean employees should receive medical health check-up annually or biennially.
Among study participants in KSHS, we initially enrolled 136,393 subjects who had responded to semi-quantitative food frequency questionnaire (FFQ) and Center for Epidemiologic Studies Depression (CES-D) between March 2011 and December 2012. Among these subjects, we excluded 812 subjects with taking sedative or anxiolytic medications and 15,106 subjects with depressive symptom in baseline analysis. Furthermore, 15,836 subjects with missing value in covariate data (e.g. BMI, hypertension, alcohol intake) and 3236 subjects with a history of serious medical diseases (e.g. coronary heart disease, stroke, and cancer) were further excluded. Additionally, 14,288 subjects with lost to follow-up were excluded. Finally, the total number of eligible study participants was 87,115 (Fig. 1). The median period of follow-up was 5.9 years.
Ethics approvals for the study protocol and analysis of the data were obtained from the institutional review board (IRB) of Kangbuk Samsung Hospital (IRB No. KBSMC 2020-09-25). All procedures performed in studies involving human participants were in accordance with the ethical standards of the IRB of the Kangbuk Samsung Hospital and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. IRB of Kangbuk Samsung Hospital approved the exemption of informed consent for the study because we only assessed retrospective data with de-identified personal information obtained from routine health check-up.
Clinical and biochemical data collection
Study data included medical history assessed by self-administered questionnaire, anthropometric measurements and laboratory measurements. All study subjects were asked to respond to a health-related behavior questionnaire, which included the topics of alcohol consumption, smoking and exercise. The degree of physical activity was evaluated by the Korean-validated version of the International Physical Activity Questionnaire (IPAQ) short form (SF)27. High physical activity was defined on the basis of health-enhancing physically active in IPAQ as follows:1 vigorous intensity activity on three or more days per week accumulating 1500 MET-min/week,2 7 days of any combination of walking, moderate intensity, or vigorous intensity activities achieving at least 3000 MET-min/week. Subject with high education were defined as those who had a university degree or higher. Hypertension was defined as a prior diagnosis of hypertension or having a measured BP ≥ 140/90 mmHg at initial and follow up examinations. Trained nurses measured BP on sitting position by automatic device (53,000-E2, Welch Allyn, USA) three times after a 5 min rest with at least 30 s interval. Final BP levels were obtained as average of second and third BP measurements. The BMI was calculated by dividing weight (kilograms) by square of height (meters2).
Blood samples were collected after more than 12 h of fasting and were drawn from an antecubital vein. The fasting serum glucose was measured using the hexokinase method, and hemoglobin A1c (Hba1c) was measured using an immunoturbidimetric assay with a Cobra Integra 800 automatic analyzer (Roche Diagnostics, Basel, Switzerland). Serum uric acid levels were measured enzymatically using an automatic analyzer Advia 1650 Autoanalyzer, Bayer Diagnostics; Leverkusen, Germany).
Glycemic status was classified into normal glycemia, prediabetes and DM. DM was defined as one of following conditions; fasting glucose ≥ 126 mg/dL, hemoglobin A1 c (HbA1c) ≥ 6.5%, and a prior diagnosis of DM28. Fasting glucose of 100–125 mg/dl or HbA1c of 5.7–6.5% were regarded as prediabetes. Insulin resistance was evaluated by calculating homeostasis model assessment-insulin resistance (HOMA-IR) as following formula: HOMA-IR = fasting serum insulin (uU/ mL) × fasting serum glucose (mg/dl)/40529.
Assessment of FFQ data
We assessed the dietary intake of KSHS participants using the FFQ that was developed for the Korean genome epidemiologic study. The dietary data to design the FFQ were obtained from the Korea Health and Nutrition Examination Survey30,31. A detailed description of the FFQ30 and its validation in the Korean population has been described in a previous study31. The food consumption frequency was composed of nine categories (e.g., SSCB intake was categorized never or rarely, once a month, two or three times a month, once or twice a week, three or four times a week, five or six times a week, one or two times a day, three or four times a day, and more than five times a day) and three serving sizes for each food (e.g., SSCB consumption was categorized as 0.5, 1 and 2 serving. 1 serving = 200 ml). Food photographs with usual intake portions also were included to increase the understanding and study reliability in study subjects. All subjects categorized into five group according to SSCB consumption as follows: never/almost never, < 1 serving/week, 1 ≤ ~ < 3 serving/week, 3 ≤ ~ < 5 serving/week, and ≥ 5 serving/week) Total energy and nutrient intake was calculated by the Can-Pro 3.0 software developed by The Korean Nutrition Society32.
Assessment of depressive symptom
Depressive symptoms were assessed using the Korean versions of CES-D scale33. The CES-D is a self-report questionnaire designed to assess the current presence of depressive symptoms in the general population34. We used the 4-factors 20-items CES-D Scale with scores ranging from 0 to 3, with 0 indicating that the depressive symptom was experienced rarely and 3 indicating that the symptom was experienced most of the time in the past week. (e.g. “I thought my life had been a failure.” 0 = seldom (not at all or less than 1 day), 1 = sometimes (1–2 days), 2 = often (3–4 days), 3 = almost always (5–7 days)). This scale has been widely used across the world and shown the validity and reliability in the Korean general population33. Depressive symptom was defined in the total score of CES-D ≥ 16. Therefore, in baseline analysis, subjects with CES-D ≥ 16 were regarded as the presence of depressive symptom in baseline and excluded from study participants. During follow-up, the subjects who newly fulfilled CES-D ≥ 16 were determined as the cases of incident depressive symptom. Detailed descriptions of study population and data collection have been included in previous studies35.
Statistical analysis
The baseline parameters with five SSCB intake groups are presented as means ± standard deviation for continuous variables and as proportions for categorical variables. The linear regression model was used for continuous variables and Cochran-Armitage trend test was used for categorical variable to assess linear response between SSCB consumption and biochemical, health related behavior, and chronic disease.
A Cox proportional hazards model was used to calculate the age-adjusted and multivariable-adjusted hazard ratio (HR) for depressive symptom and their 95% confidence intervals (CI) in each study groups (multivariable adjusted HR [95% CI]). The models were adjusted for multiple covariates including age, BMI, sex, physical activity, alcohol intake, hypertension, smoking, marital status, total calorie intake, and HOMA-IR. The covariates of the multivariable model were selected for the presence of significant differences between groups or established risk factors for depression. The incidence cases, incidence density (incidence cases per 1000 person-years), person years of each group were calculated. Trend analysis was conducted using the median of range in each SSCB consumption category (0 in never/almost never, 0.5 in < 1 serving/week, 2 in 1 ≤ ~ < 3 serving/week, 4 in 3 ≤ ~ < 5 serving/week, and 7 in > 5 serving/week). Subgroup analyses were conducted in gender and glycemic status subgroup.
The All statistical analyses were performed using R 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria), and a value of P < 0.05 (two-sided) was considered statistically significant in all analyses.