Associations of Tea Consumption With the Risk of All-Cause and Cause-Specific Mortality Among Adults With Type 2 Diabetes: A Prospective Cohort Study in China

2.1 Study Design and Participants

The data of research participants were derived from a large ongoing population-based study named “Comprehensive Research on the Prevention and Control of Diabetes” in Jiangsu Province, China. Detailed information about the study has been described previously [15, 16]. Briefly, 44 townships from Jiangsu Province (14 in Suzhou city and 30 in Huai'an city) were selected for the survey by using stratified cluster sampling method. From December 2013 to January 2014, a total of 20 053 T2DM patients who were managed by the National Basic Public Health Service were recruited. T2DM patients were defined as fasting blood glucose levels ≥ 7.0 mmol/L or a self-report T2DM history without type 1 diabetes. After the participants signed the informed consent form, the trained staff conducted face-to-face interviews using standardized questionnaires and standard body measurements. Ethical approval was obtained from the Jiangsu Provincial Center for Disease Control and Prevention (No. 2013026).

In current study, we excluded 4262 participants with cancer, coronary heart disease (CHD), or stroke at baseline to minimize confounding, avoid reverse causality, and enhance the generalizability of the results [13]. Additionally, 87 participants with missing data for tea consumption were further excluded, a total of 15 718 participants were included in the final analysis. When analyzing the various behaviors of tea consumption, the number of participants for each exposure factors was as follows: frequency of tea consumption (n = 15 718), consumption grams of tea per day (n = 15 097), consumption cups of tea per day (n = 15 557), and duration of tea consumption (n = 15 060) (Figure S1).

2.2 Assessment of Tea Consumption

Tea consumption information was assessed for all participants using a baseline questionnaire. (1) Do you have a habit of drinking tea (yes or no)? For those who answered “yes,” we further asked. (2) How many times do you drink tea in a typical week (≥ 3 times/week or 1–2 times/week)? (3) When did you start drinking tea? (4) How much tea leaves do you usually consume each time (with a picture showing the amount of tea leaves in grams)? (5) How many cups of tea do you make per day on average (each time the tea leaves are replaced, it counts as one cup of tea)? (6) Which type of tea do you consume most frequently (e.g., green tea, black tea, oolong tea, or other varieties)? Figure S2 displays images of various tea weights in both their dried and brewed states.

Based on the above question, we divided participants into three groups according to whether they drink tea and frequency of tea consumption (never, 1–2 times/week, ≥ 3 times/week), For drinkers, we further categorized participants according to the consumption grams of tea per day (1–4 g/day, > 4 g/day), consumption cups of tea per day (1 cup/day, ≥ 2 cups/day), and the duration of tea consumption (1–30 years, > 30 years), with reference to previous tea consumption studies [17, 18].

2.3 Assessment of Covariates

Covariate information for all participants was collected through baseline questionnaires, including sociodemographic characteristics (age, sex, marital status, education, annual household income), lifestyle behaviors (smoking status, drinking status, physical activity, and intakes of red meat, fresh vegetables, and fruits), and medication use history (insulin, oral antidiabetic medication). Diabetes duration was calculated by subtracting the data of first diagnosis of diabetes from the date of enrolment at the baseline assessment. Information on physical activity was self-reported by participants using the Global Physical Activity Questionnaire, which has good reliability and validity [19]. To calculate the daily level of physical activity, the metabolic equivalent tasks (METs) value for each type of activity was multiplied by the number of hours spent on each activity, and the total MET-hours for all activities were summed. Information of fruits, vegetables, and red meat consumption was obtained through food-frequency questionnaire. Height, weight and waist circumference were measured using uniform equipment. Body mass index (BMI) was calculated as weight (kg) divided by height squared (m²).

2.4 Assessment of Outcomes

Death information was obtained regularly through the local Death Certificate System. This system provides complete and reliable cause-specific death information medically certified by doctors [20]. Trained medical staff coded the underlying cause of death by applying the rules from the 10th revision of the International Classification of Diseases (ICD-10). The primary outcomes were all-cause mortality, CVD mortality (I00-I99), and cancer mortality (C00-C97). In addition, we also separately analyzed major cardiovascular mortality, including CHD (I20-I25), acute myocardial infarction (I21), stroke (I60-I61, I63, and I64), hemorrhagic stroke (I60-I61), and ischemic stroke (I63).

2.5 Statistical Analysis

Person-years were calculated from the date of enrollment to either the date of death or the end of follow-up (September 30, 2023), whichever came first. Baseline characteristics were calculated based on tea consumption, continuous variables were described by median (interquartile range) and categorical variables were described by proportions (n, %). Cox proportional hazards regression models were used to estimate the associations of different tea consumption behaviors with all-cause and cause-specific mortality risk by calculating hazard ratios (HRs) and 95% confidence intervals (95% CIs). Two multivariate adjustment models were constructed to account for potential confounding factors. Model 1 was adjusted for age (years), sex (male, female). Model 2 was further adjusted for marital status (married, unmarried), educational level (without formal education, primary and middle school, high school or above, unknown), annual household income (< 40 000 yuan, 40 000–99 999 yuan, ≥ 100 000 yuan, unknown), smoking status (never, previous, current, unknown), alcohol drinking status (never, previous, current, unknown), diabetes duration (years), oral antidiabetic medication use (no, yes), insulin use (no, yes), body mass index (kg/m²), total physical activity (MET-h/day), fruit consumption (never, 1–3 times/week, 4–6 times/week, ≥ 7 times/week, less than weekly), vegetable consumption (never, 1–3 times/week, 4–6 times/week, ≥ 7 times/week, less than weekly), red meat consumption (never, 1–3 times/week, 4–6 times/week, ≥ 7 times/week, less than weekly). Linear trends of all-cause and cause-specific mortality related to different tea consumption patterns were calculated by modeling the levels of tea consumption frequency and the amount of tea consumption years and including them as continuous variables in models.

We conducted stratified analyses according to baseline age (< 65, ≥ 65 years), sex (male, female), smoking status (never or previous, current), alcohol drinking status (never or previous, current), and diabetes duration (< 5, ≥ 5 years). The p values for interactions were estimated using a likelihood ratio test comparing. Cox proportional risk models with and without cross-product terms for stratified and exposed variables. We used multivariate cubic regression splines with 4 knots to visually explore nonlinear associations of tea consumption with all-cause and cause-specific mortality. A cutoff value was defined as the point where the curve started to level off. In the sensitivity analysis, we first compared the baseline characteristics of participants with (n = 19 966) and without (n = 15 718) major diseases, including baseline cancer, stroke, and coronary heart disease (CHD). Additionally, we excluded 348 participants who died within the first 2 years of follow-up, 193 participants who died from accidents (V01-Y89), 3479 participants with a diabetes duration of less than 1 year, 187 participants who consumed coffee, and 511 participants with baseline kidney disease. We also analyzed data from 19 966 participants, including those with baseline diagnoses of major illnesses such as cancer, coronary heart disease, and stroke. To mitigate potential misclassification of tea consumption data, we also excluded 270 participants categorized under “other” tea types.

All analyses were conducted using the R statistical package, version 4.2.1. p values for statistical tests were two-tailed and significance was defined as p < 0.05.

3.1 Baseline Characteristic

A total of 15 718 T2DM participants were included in the study with an average age of 61.98 years. Among them, 5028 participants consumed tea and 3926 participants drank tea ≥ 3 times a week. Compared to nonconsumers, T2DM participants who consumed tea were younger, more likely to be men, with a higher level of education, and drank alcohol (Table 1).

3.2 Tea Consumption and All-Cause, Cause-Specific Mortality

During a median follow-up of 9.77 years (interquartile: 9.69–9.82 years), a total of 3046 deaths were recorded, including 922 from CVD and 736 from cancer. Compared with the T2DM population who never drank tea, the fully adjusted HRs (95% CI) for all-cause mortality were 0.81 (0.73, 0.90) for those who drank tea ≥ 3 times per week, 0.77 (0.67, 0.87) for those who consumed quantity of tea > 4 g per day, 0.85 (0.74, 0.98) for those who consumed quantity of tea ≥ 2 cups per day, and 0.76 (0.68, 0.85) for those whose tea drank duration > 30 years. Similarly, we found that the association of tea consumption with CVD mortality was consistent with all-cause mortality. Among cancer mortality, we did not find a statistically significant association of tea consumption with cancer mortality (Table 2). In addition, restricted cubic spline analysis showed a nonlinear relationship of the consumption of tea grams per day with all-cause mortality (p for nonlinear = 0.004), and the exposure-response association kept declining until reaching approximately 4 g per day of tea leaves consumption and then became stabilizing (Figure 1).

3.3 Subgroup Analysis and Sensitivity Analysis

As shown in Table 3 and Table 4, tea consumption was inversely associated with the risk of stroke mortality, but not statistically significant in other subtypes of CVD. Compared with T2DM participants who never drank tea, those who drank tea ≥ 3 times per week, consumed tea > 4 g per day, consumed cups of tea 1 per day, drank duration of tea > 30 years had 28% (HR = 0.72, 95% CI: 0.53–0.97), 48% (HR = 0.52, 95% CI: 0.35–0.79), 33% (HR = 0.67, 95% CI: 0.49–0.91), and 37% (HR = 0.63, 95% CI: 0.45–0.89) lower stroke mortality risk, respectively (Table 3).

We further analyzed the relationship of tea consumption with all-cause mortality in specific population subgroups. The strength of the associations of tea consumption with the risk of stroke was largely consistent across subgroups, when classified by age, sex, smoking status, alcohol drinking status, and diabetes duration (p for interaction > 0.05) (Table S1). In the sensitivity analysis, we first compared the baseline characteristics of participants with (n = 19 966) and without (n = 15 718) major diseases, including baseline cancer, stroke, and coronary heart disease (CHD). The comparison revealed that participants with baseline major diseases had higher age, BMI, and diabetes duration, and lower physical activity levels (Table S2). Excluding participants who died within the first 2 years of follow-up (n = 348), had diabetes duration ≤ 1 year (n = 3479), died from accidents (n = 193), consumed coffee (n = 187), or had baseline kidney disease (n = 511) did not substantially affect the findings (Table S3). The analysis of data from 19 966 participants, including those with baseline diagnoses of major illnesses, showed that the association between tea consumption and the risk of all-cause mortality, CVD mortality, and cancer mortality was consistent with the primary outcomes (Table S4). Participants classified as “other” tea types were excluded, resulting in findings that remained consistent with the primary results (Table S5).