Study population
Between 2004 and 2008, 512,891 participants aged 30–79 years were recruited from five urban and five rural regions across China to the CKB cohort and surveyed for baseline information. Briefly, study design of the CKB and characteristics of the study participants have been previously described [12, 13]. During 2013–2014, a re-survey involving ~ 5% of randomly chosen surviving participants was conducted, with the administrative unit of the rural village or urban residential committee as the primary sampling unit. Apart from a few new questions and physical measurements, the data collection and survey procedures were much the same as in the baseline survey. Signed informed consent was obtained from all participants at both surveys.
The present study was performed in 25,045 participants who finished the re-survey during 2013–2014, in which calcaneal QUS measurements were first added to the survey. We excluded participants who had missing data on any measures of BMD (n = 360) and waist or hip circumferences (n = 9) at re-survey; those who reported having a history of kidney disease (n = 306), osteoporosis (n = 544), fracture (n = 2290), or rheumatic arthritis (n = 1165) at re-survey; and those who developed multiple myeloma and malignant plasma cell neoplasms (10th revision of the International Statistical Classification of Diseases, ICD-10: C90; n = 4) since enrollment. The reason to exclude participants with kidney or bone diseases is that their pathology of bone might have differed from the general population. We also excluded participants who reported to be a new tea consumer since baseline, yet had a starting age of tea consumption earlier than their baseline age (n = 258). The final analyses included 20,643 participants.
Assessment of tea consumption
Tea consumption was assessed by an interviewer-administered questionnaire at both baseline and re-survey. Asking the same questions at an average interval of 8.0 years allowed us to know potential changes in tea consumption habit over a long period of time. All CKB participants reported their frequency of tea consumption during the past 12 months (never, only occasionally, only at certain seasons, monthly but less than weekly, and at least once a week). Participants who consumed tea weekly were further asked about days consuming in a typical week (1 to 2 days, 3 to 5 days, or almost every day), cups (in 300 mL size) of tea consumed and amount (in gram) of tea leaves added in one consuming day, type of tea consumed most commonly (green tea, oolong tea, black tea, or others), and age when starting their tea consumption habit. To help with gauging the amount of tea leaves, a pictorial guide was provided to each participant. Years of tea consumption was calculated as the difference between age at re-survey and age of starting tea consumption.
In the present analyses, all participants were grouped into five categories according to their reported tea consumption at baseline and re-survey: never consumers, former consumers, recent new consumers, prolonged consumers, and prolonged weekly consumers (Table 1). Prolonged weekly tea consumers were further categorized, according to the amount of tea leaves added per consuming day reported at re-survey, as < 3.0, 3.0–5.9, and ≥ 6.0 g/day
Table 1 Classification of participants according to frequencies of tea consumption Assessment of covariates
Covariate information collected from the re-survey was used in the present study. Questionnaire information included sociodemographic characteristics (age, sex, region, education, occupation, household income, and marital status), lifestyle behaviors (alcohol consumption, tobacco smoking, physical activity, and intakes of red meat, fresh fruits, vegetables, milk, yoghurt, other dairy foods, fish oil or cod liver oil, vitamins, calcium, iron, or zinc supplements, and coffee), personal medical history (diabetes, chronic hepatitis or cirrhosis, peptic ulcer, and gallstones or cholecystitis, etc.), and women’s menopausal status. Daily level of physical activity was calculated by multiplying the metabolic equivalent tasks (METs) value for a particular type of physical activity by hours spent on that activity per day and then summing the MET-hours for all activities. Habitual dietary intake in the past year was assessed by a qualitative food frequency questionnaire. Trained staff undertook measurements of body weight, height, waist, and hip circumferences. Body mass index (BMI) was defined as the body weight divided by the square of the height, and waist-hip ratio as the ratio of waist circumference to hip circumference.
Assessment of BMD
At re-survey, participants had two feet measured for BMD in a seated position based on a calcaneus QUS measurement, using an Achilles EXPII Operator (GE Medical Systems Lunar, USA). Trained staff implemented daily quality control and cleaning procedures following manufacturer’s instruction. Parameters directly measured from QUS included broadband ultrasound attenuation (BUA, in dB/MHz) and speed of sound (SOS, in m/s). The parameter of stiffness index (SI) was calculated from BUA and SOS in the Achilles system by the following equation: SI = (0.67 × BUA + 0.28 × SOS) − 420. For the present analyses, the QUS measurements were averaged between both calcanei for each participant.
Statistical analysis
General linear regression was used to estimate the regression coefficients (β) and 95% confidence intervals (CIs) for the associations between tea consumption and BMD measures. Multivariable models were adjusted for established and potential confounders: age, sex, study regions, marital status, occupation, household income, alcohol consumption, smoking status, physical activity, intake frequencies of red meat, fresh fruits, vegetables, milk, yoghurt, and other dairy foods, menopausal status (for women only), BMI, and waist-hip ratio. Tests for linear trend were only conducted in prolonged weekly consumers by assigning the median amount of tea leaves added to each category and then modeling tea consumption as a continuous variable in a separate model. Analyses were performed in all participants and separately in men and women.
Further, we examined the associations between tea consumption and BMD measures according to three types of tea (green tea, oolong tea, or black tea) and years of tea consumption (≤ 20, or > 20), all as compared with those never consumers who did not consume tea during the past 12 months at both baseline and re-survey. We also examined whether the associations of prolonged weekly tea consumption with BMD measures differed according to several characteristics at re-survey: age, study regions, alcohol consumption, smoking status, the level of physical activity, dairy food consumption, BMI, and menopausal status (in women). In the subgroup analyses, we included only never and prolonged weekly consumers and tested for interaction using likelihood ratio test, which involved comparing models with and without interaction term between the stratifying variable and tea consumption as an ordinal variable.
The statistical analyses were performed with Stata (version 14.0, Stata). All P values were two-sided, and statistical significance was defined as P < 0.05. For testing of three outcomes, a Bonferroni correction was applied to the significance level that divided 0.05 by 3.
Data availability
Study protocol: Cohort description and questionnaires are available at www.p3gobservatory.org/questionnaire/list.htm. Statistical code: Available from Dr. Lv (e-mail, lvjun@bjmu.edu.cn). Data set: See study Web Site (www.ckbiobank.org) for data access policy and procedures.