Introduction

Hypertension is highly prevalent in the adult population, and the worldwide prevalence has been estimated to increase to approximately 30% by 2025 [1]. Hypertension increases the risk of developing cardiovascular diseases and renal complications [2], and has also been identified as a leading cause of mortality and a strong contributor to the loss of disability-adjusted life-years [3]. However, hypertension can potentially be prevented by the modification of dietary and lifestyle factors [2], and a deeper understanding of the complex relations between these factors and the risk of hypertension is imperative in improving the primary prevention strategies for hypertension.

Coffee is one of the most widely consumed beverages in the world [4]. Due to its high global consumption, there is immense scientific and public health interest to understand how the intake of coffee can affect human health. Although relationship between coffee consumption and hypertension has been extensively studied in prospective studies based in Europe and the United States, results remain controversial. Some studies have observed either an inverse U-shaped association [5, 6], a positive association without clear dose–response [7] or a null association between coffee drinking and hypertension [8–10], while a meta-analysis of randomized controlled trials and cohort studies [11] has not shown any effect of coffee on risk of hypertension. Another recent study [12] showed that risk of hypertension is lower for people who consumed coffee three to four cups/day compared to those who drank less than one cup per day. Data from Asian populations, which generally have a lower coffee consumption than their Western counterparts, are scarce.

Tea, another beverage widely consumed globally [13], has been shown in epidemiologic studies to reduce the risk of cardiovascular diseases [14, 15]. The protection against cardiovascular diseases with tea drinking is thought to be largely due to the antioxidant property of tea flavonoids [14], which naturally occur in tea leaves. Both green and black teas are derived from the leaves of the same plant, Camellia sinensis, and they differ in their flavonoid content due to different manufacturing processes that affect the oxidation of tea flavonoids [13, 16]. Even though the protective association between tea consumption and cardiovascular diseases has been well documented [17–19], the effect of regular consumption of tea on risk of hypertension remains unclear [14, 20].

This study examined the association between consumption of coffee and tea and the risk of hypertension in a prospective population-based cohort study of Chinese men and women residing in Singapore. Further, we sought to understand the role of caffeine in these associations.

Methods

Study population

The design of the Singapore Chinese Health Study has been previously described [21]. Briefly, the cohort recruited 63,257 Chinese between April 1993 and December 1998. The participants were 45–74 years at the time of recruitment, and were citizens or permanent residents of Singapore residing in government housing estates (86% of Singaporeans resided in such facilities during the recruitment period). Participants were restricted to the two major dialect groups among Chinese in Singapore, the Hokkiens and Cantonese, who originated from Fujian and Guangdong provinces in China, respectively. Participants were subsequently followed-up via telephone interview during follow-up I (1999–2004) and follow-up II (2006–2010) to update information on selected lifestyle factors and medical history. A total of 52,322 participants were contacted during follow-up I and 39,528 participants were contacted during follow-up II. This study was approved by the Institutional Review Board at the National University of Singapore, and all participants gave informed consent.

Assessment of intake of coffee, tea, caffeine, and other covariates

Information such as demographics, height, weight, lifetime tobacco use, current physical activity, sleep duration, and medical history was collected using a structured questionnaire through in-person interview at recruitment. The usual diet of the participant, including alcohol, coffee and tea, was assessed using a 165-item semi-quantitative food-frequency questionnaire that was specifically developed and subsequently validated for this study population [21]. The intake frequency of a standard serving of coffee, black tea, and green tea was determined using nine predefined categories: never or hardly ever, 1–3 times a month, once a week, 2–3 times a week, 4–6 times a week, once a day, 2–3 times a day, 4–5 times a day, and 6 or more times a day. The standard serving size of one cup (237 mL) was assigned in the questionnaire. Consumption of different types of coffee, such as instant or freshly brewed, or three-in-one preparation was asked in separate questions. The amount of caffeine intake was estimated from the participant’s reported intake of coffee, tea, and other caffeinated beverages and food items. Because decaffeinated coffee or tea was rarely consumed at the time of recruitment, all coffee and tea consumed were assumed to be caffeinated in this study. Coffee and tea accounted for 84 and 12% of total dietary caffeine intake in this cohort, respectively. The remaining minor contribution (4%) came from other caffeinated food sources such as soda, cocoa drinks, and chocolate-related food items.

Assessment of hypertension

Participants were asked about history of physician-diagnosed hypertension at baseline and during the two follow-up interviews, and the age at first diagnosis was ascertained if a history of hypertension was reported. Participants were considered as prevalent cases of hypertension if they reported a history of hypertension during the baseline interview, or if their diagnosis of hypertension reported during the follow-up interviews was before study recruitment. Participants who reported a history of hypertension diagnosed any time between baseline and follow-up interviews were considered incident hypertension cases.

Statistical analysis

Our study only included the cohort participants who participated in at least one follow-up interview so that we could capture incident cases of hypertension. We further excluded participants with prevalent hypertension (n = 14,940), stroke, or coronary heart disease (n = 809) at recruitment, and the final analysis included 38,592 participants out of the original 63,257 subjects. Person years for each participant were calculated from the age at recruitment to the age at hypertension diagnosis or age at last follow-up interview, whichever occurred first. Differences in baseline characteristics between incident hypertensive cases and non-hypertensive participants were compared using Student’s t test for continuous variables and the Chi-square test for categorical variables. We used the Spearman’s partial correlation coefficient to assess correlation among the intake categories of coffee, tea, and caffeine.

We used the Cox proportional hazards models to estimate the hazard ratios (HR) and 95% confidence intervals (CI) for risk of hypertension associated with coffee, black tea, green tea, and caffeine intake. We tested proportional hazards assumptions by adding the interaction term between a function of time and the covariate of interest in the Cox model; there was no violation of Cox proportional hazards assumptions for our variables of interest. We categorized coffee intake into five categories (<weekly, weekly to <daily, one cup/day (reference), two cups/day, or ≥three cups/day), tea intake into three categories (<weekly (reference), weekly to <daily, daily), and caffeine intake into five categories (<50 (reference), 50–100, 100–200, 200–300, and ≥300 mg/day). Test for linear trend across the caffeine categories was conducted using the ordinal values of the categories as a continuous variable. We used one cup/day category and <weekly category as the reference category for coffee and tea, respectively, because these categories represented the largest proportion of subjects’ consumption and provided more stable estimates. To explore a possible U-shaped association between coffee drinking and risk of hypertension, we also assigned scores of 1, 2, 3, 4, 5 to successive categories of coffee drinking and incorporated a quadratic effect of this score in addition to a linear effect in the model in a separate analysis.

Adjustment was made for age at recruitment (years), year of recruitment (1993–1995, 1996–1998), gender, dialect group (Cantonese, Hokkien), body mass index (kg/m2), level of education (no formal education, primary school, secondary school, or higher), smoking status (never, former, current), alcohol intake (none, monthly, weekly, daily), physical activity (none, 0.5–3 h/week, ≥4 h/week), duration of sleep (≤5, 6–8, ≥9 h), and dietary intake of sodium, vegetables, fruits, and dairy products (all in quartiles) in model 1. In model 2, we additionally adjusted for simultaneous coffee, black tea, and green tea intake. For the analysis on tea, we further adjusted for caffeine intake to determine whether caffeine could explain the association between tea consumption and risk of hypertension. Since coffee was the major contributor for caffeine (84%), the coffee categories (<weekly, weekly to <daily, one cup/day, two cups/day, ≥3 cups/day) were highly correlated with caffeine intake (Spearman correlation coefficient = 0.85, p < 0.001) in this population. Hence, we did not adjust for caffeine in our coffee analysis because those with high coffee intake (≥3 cups/day) were invariably in high caffeine intake categories. We also included total caloric intake as a covariate, and all the results remained essentially the same. However, since coffee and tea do not contribute significantly to total caloric intake, we did not include this in the final models.

We also tested the non-linear association between caffeine intake and risk of hypertension by using restricted cubic splines with four knots in Cox regression analysis. We used first knot (8 mg/day) in caffeine intake as a reference to test for the linear association. In our analysis on caffeine, we included a model (Model 2) that adjusted for coffee categories to examine the effect of caffeine from other beverage and food sources.

All statistical tests were two-tailed, and values of p < 0.05 were regarded as statistically significant. The SAS statistical package version 9.3 (Statistical Analysis System Inc., Cary, NC) was used for analysis except for the restricted cubic spline plots, which were fitted in Stata version 11 (Stata Corporation, College Station, Texas, USA).

Results

Compared to participants who drank one cup of coffee per day, those who drank coffee less than weekly were less likely to be smokers, and had lower alcohol and daily sodium intake but higher dairy intake. In contrast, those who drank coffee more than two cups per day were more likely to be smokers, and had higher alcohol and daily sodium intake but lower dairy intake (Table 1). Compared to participants with less than weekly consumption of black or green tea, daily tea drinkers had a higher level of education, greater physical activity, and a larger proportion of smokers. We also observed higher intake of alcohol, physical activity, sodium, vegetables, fruits, and dairy products for those who drank black tea or green tea daily than those who drank less than weekly (Supplementary Tables 1, 2).

Table 1 Baseline characteristics of participants, n (%) for categorical variables or mean ± SD for continuous variables, according to frequency of coffee consumption
Full size table

We identified 13,658 incident cases of hypertension in our cohort over a mean follow-up of 9.5 (SD 4.1) years, and mean age at diagnosis of hypertension was 62.4 (SD 8.1) years. Compared with non-hypertensive participants, hypertensive patients were more likely to be women, never-smokers, and non-drinkers of alcohol. They were also older, less educated, had higher BMI, and more physically active (Table 2).

Table 2 Baseline characteristics of participants, n (%) for categorical variables or mean ± SD for continuous variables, between hypertensive cases and the rest of the cohort participants
Full size table

As expected, there was a high correlation between intake of coffee and caffeine; the Spearman’s partial correlation coefficient between intake categories of caffeine and coffee was 0.85 (p < 0.001). Conversely, the correlation coefficient between intake categories of caffeine and tea was lower at 0.20 (p < 0.001). The intake of coffee and tea was inversely correlated; correlation coefficient between the intake categories of coffee and tea was −0.13 (p < 0.001).

Results from multivariable analysis revealed an inverse U-shaped association between coffee intake and risk of hypertension. Compared to those drinking one cup/day, drinking coffee less frequently was associated with a reduction in risk of hypertension. Similarly, the risk was decreased in those who drank more than two cups/day (Fig. 1). In the multivariable analysis, compared to those who drank one cup/day, the HR (95% CI) of incident hypertension were 0.94 (0.88–1.00) for weekly to <daily drinkers, 0.87 (0.83–0.91) for those who drank less than weekly, 0.98 (0.94–1.02) for those who drank two cups/day, and 0.93 (0.86–1.00) for those who drank ≥3 cups per day (Table 3). The quadratic term for the coffee score in successive categories in the model that also included the linear term was statistically significant (p < 0.001). The association between coffee intake and incidence of hypertension did not vary significantly between men and women (all p for interaction >0.5).

Fig. 1
figure 1

Inverse U-shaped association between coffee intake and risk of hypertension. Model 1 was adjusted for age at recruitment (years), year of recruitment (1993–1995, 1996–1998), gender, dialect group (Cantonese, Hokkien), body mass index (kg/m2), level of education (no formal education, primary school, secondary school, or higher), smoking status (never, former, current), alcohol intake (none, monthly, weekly, daily), physical activity (none, 0.5–3 h/week, ≥4 h/week), sleep duration (≤5, 6–8, ≥9 h), and dietary intake of sodium, vegetables, fruits, and dairy products (all in quartiles). Model 2 was adjusted for Model 1 plus intake of tea (<weekly, weekly to <daily, daily)

Full size image
Table 3 HR (95% CI) for the association between coffee intake and risk of hypertension
Full size table

Participants who drank black tea daily had slightly increased risk of 9% compared to those who consumed black tea less than weekly; the multivariable HR (95%) was 1.09 (1.03–1.15). Similar results were observed for green tea; the corresponding multivariable HR (95%) was 1.06 (1.00–1.11) for green tea. However, these associations were attenuated and not statistically significant when we further adjusted for caffeine (Table 4).

Table 4 HR (95% CI) for the association between tea intake and risk of hypertension
Full size table

We observed that there was a statistically significant positive association between caffeine intake and risk of hypertension; compared with <50 mg/day caffeine intake, all the four higher categories of caffeine intake had statistically significant increase in risk of hypertension, but there was no clear dose-dependent trend among them. After adjusting for coffee, the stepwise dose–response relationship between caffeine intake and hypertension risk became more apparent; compared to the lowest intake (<50 mg/day), the multivariable HR (95% CI) for 50 to <100, 100 to <200, 200 to <300, ≥300 mg/day were 1.07 (1.00–1.15), 1.10 (1.02–1.18), 1.08 (0.98–1.19), and 1.16 (1.04–1.31) (p trend = 0.02) (Table 5). Restricted cubic spline regression with adjustment for coffee intake confirmed the linear association between caffeine intake and risk of hypertension (p for non-linearity = 0.11) (Fig. 2).

Table 5 HRs (95% CI) for the association between intake of caffeine and risk of hypertension
Full size table
Fig. 2
figure 2

Association between caffeine intake and risk of hypertension using restricted cubic spline regression. Model was adjusted for adjusted for age at recruitment (years), year of recruitment (1993–1995, 1996–1998), gender, dialect group (Cantonese, Hokkien), body mass index (kg/m2), level of education (no formal education, primary school, secondary school, or higher), smoking status (never, former, current), alcohol intake (none, monthly, weekly, daily), physical activity (none, 0.5–3 h/week, ≥4 h/week), sleep duration (≤5, 6–8, ≥9 h), dietary intake of sodium, vegetables, fruits, and dairy products (all in quartiles), and intake of coffee (<weekly, weekly to <daily, 1 cup/day, 2 cups/day, ≥3 cups/day)

Full size image

Discussion

The present study showed an inverse U-shaped relationship between coffee intake and incident hypertension. Participants who drank less than one cup of coffee weekly or more than two cups of coffee per day had a significant reduction in risk of hypertension compared to those who drank one cup per day. There was a weak positive association between daily drinking of black tea or green tea and risk of hypertension, which was attenuated and became non-significant after adjusting for caffeine. We also found that increased caffeine intake was associated with increased risk of hypertension in a dose-dependent manner.

Our results are consistent with the findings from a dose–response meta-analysis of 5 cohort studies, which also suggested an inverse U- or J-shaped association between coffee and the risk of hypertension [22]. In this meta-analysis, risk of hypertension increased with coffee consumption and peaked at three cups per day, but the risk decreased at higher levels of coffee consumption. Our findings also corresponded with individual prospective studies that found a non-linear association among cohorts of European men and women [6, 7], U.S nurses [8], and other health professionals [10]. Moderate coffee drinkers (one to three cups/day) were found to have the highest risk of hypertension compared to light drinkers who drank less than one cup per day and heavy drinkers who drank more than three cups per day [6–8, 10]. Similarly, a Dutch cohort study reported an inverse J-shaped association in which hypertension risk was reduced among those who drank coffee less than a cup per day and those who drank more than three cups per day compared to one to three cups per day [5]. Our findings extended current evidence regarding the inverse U-shaped association between coffee intake and risk of hypertension to an Asian population. It is important to note that a wide range of coffee consumption dose was observed in our participants. Compared to 37–81% of participants in previous studies [5, 6, 10], only 6% of our participants consumed three or more cups per day. In addition, 20% of our participants consumed less than one cup of coffee weekly, while 9% of them were weekly to <daily drinkers. The wide spectrum of coffee consumption dose enabled us to test the association among low coffee consumption categories, especially for those who drank less than one cup of coffee weekly.

With regards to tea consumption, observations from epidemiological studies investigating the effects of regular consumption of tea on hypertension still remain unclear. Five or more cups per day of black tea was related to a lower blood pressure in a Norwegian population [23], and habitual moderate-strength green and oolong tea consumption (120mL/day or more) for at least 1 year reduced the risk of developing hypertension in Taiwanese men and women [24]. A recent meta-analysis of ten randomized controlled trials concluded that green or black tea consumption reduced blood pressure in individuals within prehypertensive and hypertensive ranges of blood pressure [25]. Even though these associations are supported by strong evidence from experimental studies on the antioxidant and vasodilator effects of tea flavonoids, studies in human populations have yielded inconsistent results [26]. In human clinical trials where potential caffeine effect was controlled, consumption of five cups of either green or black tea per day [27] and six cups of black tea per day [28] had no effect on blood pressure. Although we found a weak association between tea intake and increased risk of hypertension, the association was attenuated after we adjusted for caffeine. Hence, this weak association among daily tea drinkers in our study may be attributed to the effect of caffeine. Additionally, in Asia, multiple cups of tea are normally consumed at one sitting from a tea pot that is routinely topped up with water using the same tea leaves. Thus, while multiple cups of tea may be consumed at one time, the actual ingested tea components could still be low due to the dilution of compounds. Hence, it can be problematic for observational studies to accurately study the effect of tea flavonoids by using the number of cups of tea as the measure of exposure.

As a common active component in coffee and tea, caffeine has been shown to raise blood pressure via mechanisms that include sympathetic over-activation, antagonism of adenosine receptors, increasing norepinephrine release via direct effects on the adrenal medulla, direct renal effects, and activation of the renin–angiotensin system [29]. Similarly, in our present study we found caffeine to be linearly associated with increased risk of hypertension after adjusting for coffee intake categories. Coffee is also a rich source of potassium, and it was estimated that five cups of coffee per day contributed to approximately 26% of daily intake of potassium [30]. Potassium has been shown to promote vascular smooth muscle relaxation and endothelium-dependent vasodilation, increase endothelial nitric oxide production, and have blood pressure-lowering effect in animals and human [31–33]. Dietary intake of chlorogenic acid, a major polyphenol in coffee, has been shown to reduce blood pressure and improve endothelial dysfunction and nitric oxide bioavailability by inhibiting reactive oxygen species production in blood vessels in spontaneously hypertensive rats [34]. In mild hypertensive patients, daily intake of chlorogenic acid has shown to have an antihypertensive effect [35, 36]. In the present study, we hypothesize that those with moderate intake of coffee (one or two cups/day) were at risk of hypertension due to the combination of high caffeine but low level of other coffee compounds at this dose of coffee intake. However, at higher doses of coffee intake, other compound in coffee, such as the aforementioned minerals and polyphenols, counteracted the effects of caffeine in lowering the risk of hypertension.

The strength of the present study includes its prospective study design in a large cohort with a long period of follow-up. This allowed us to investigate the effect of habitual coffee and tea consumption on risk of hypertension in normotensive participants over time. The availability of a wide range of lifestyle and dietary factors in this cohort, collected at baseline, also allowed us to adjust for potential confounding variables known to affect the risk of hypertension. The relatively wide spectrum of coffee consumption patterns provided a unique opportunity to investigate the association between coffee intake and risk of hypertension, especially at low intake levels.

There are several limitations in the present study. Other than those with prevalent hypertension at recruitment, our study mainly excluded those who did not participate in the follow-up interviews, and since the main reason for non-participation was mortality, those who did not participate were expectantly older and had other baseline characteristics different from those who participated. Notably, all our statistical models had therefore included these factors as covariates to remove possible confounding bias by these factors. Therefore, even though there were differences between subjects included in the study and those who were excluded due to lack of follow-up, we believe the conclusions of the present study should be valid. We also acknowledge that the coffee and tea consumption, and computation of caffeine intake at baseline were from self-reported information and subject to measurement errors. The other covariates were also self-reported and may have resulted in some misclassification and residual confounding. Furthermore, we only used the diet information at baseline and the frequency of coffee and tea consumption may change over time. However, these misclassifications are most likely to be non-differential in nature and may lead to underestimation of the risk estimates. In a follow-up interview conducted after an average of 12.7 years after the baseline interview, majority of our study cohort retained their status as daily or non-daily drinker of coffee (72.3%), black tea (85.6%), and green tea (85.2%). Although we did not have direct data on usage of supplements or energy shots that contained caffeine, in our cohort, the prevalence of weekly use of vitamins or minerals was only 6.4%. Hence, we postulate that the intake of caffeine through such products would not be high and was therefore unlikely to affect the findings on caffeine from beverage and food sources in this study. The history of physician-diagnosed hypertension was self-reported, which is subject to reporting error. However in a home visit for biospecimen collection conducted among a subgroup of 12,646 participants who have reported physician-diagnosed hypertension, 88% of them reported to take antihypertensive medications, and this indicates a reasonable validity of self-reported diagnosis. We also acknowledge that among the 24,934 non-cases, there could be some with undiagnosed hypertension, and since such a misclassification is unlikely to be related to the intake of beverages and is therefore non-differential in nature, it may likely lead to underestimation of the risk estimates.

In conclusion, our study suggests that drinking coffee less than one cup/week or more than two cups/day may lower the risk of hypertension. Caffeine may account for increased risk in daily tea drinkers and in those who drank coffee moderately (one to two cups of coffee/day). The inverse U-shaped association with coffee suggests that at higher doses, other ingredients in coffee may offset the effect of caffeine and confer benefit on the risk of hypertension. Further studies are needed to confirm our findings and to identify the bioactive compounds in coffee that may account for its antihypertensive effects at high level of consumption.