The association of wrist circumference with hypertension in northeastern Chinese residents in comparison with other anthropometric obesity indices

Study design and participants

This is a cross-sectional study based on urban community-dwelling residents in Shenhe district, Shenyang City, Liaoning Province, China. Details of the methods of enrollment have been reported previously (Fan et al., 2017; Yang et al., 2017). In brief, in the first stage, all 14 blocks of Shenhe district were selected and eight residential communities were randomly selected from each of the blocks. In the second stage, systematic sampling was conducted to select 25 households in each residential community. In the third stage, within each household, residents older than 18 years old who lived in the district for at least 2 years were randomly selected without replacement. Finally, 2,761 local residents were recruited to finish the questionnaires and accepted the relevant physical examinations and biochemical tests. Among 2,761 participants, those who did not meet the inclusion criteria in this study were excluded as following: 74 participants missing either information of weight, height, WC, hip circumference, NC or WrC; 2 participants younger than 20 years old; and 354 participants who used to be diagnosed as either cancer or CVD (Fig. 1). As a result, 2,331 participants (761 male and 1,570 female participants) were remained for final analyses. Since all participants were randomly selected from communities without the consideration of gender, the total numbers of males and females were different.

To confirm the sample size involved in final analysis is enough or not, we calculate the minimum sample size required according to the following formula: $n=Z_{1-\alpha ∕2}^2\times p\left(1-p\right)∕d^2$ (Charan & Biswas, 2013). In this formula, Z_1−α∕2 is standard normal variate (at 5% type I error (P-value <0.05), Z_1−α∕2 is 1.96), p is the expected proportion in population based on previous studies, and d is the absolute error or precision (we decided d as 0.05 in our study). As a previous survey from northern Chinese population showed, the prevalence of hypertension was 48.9% and 30.8% in male and female participants, respectively (Sun et al., 2017). Therefore, the calculated minimum sample size for the current study were as following: n (male) = 1.96² × 0.489 (1–0.489)/0.05² = 384; n (female) = 1.96² × 0.308 (1–0.308)/0.05² = 328. The final sample size we selected for analysis was large enough compared to the number we calculated.

This study was in accordance with the World Medical Association Declaration of Helsinki-Ethical Principles for Medical Research Involving Human Subjects and was approved by the Ethics Committee of China Medical University (No: CMU62073024; July 15th, 2008). A written informed consent form was obtained from each participant after they were informed of the objectives and benefits of this study.

Obesity indices

In this study, body weight, height and other circumferences were all measured twice using a vertical weight scale and a metric scale with standardized protocols, and to the nearest 0.1 kilogram (kg) and 0.1 centimeter (cm), respectively. All participants were asked to accept measuring of body weight and height wearing light clothes without shoes. BMI was then calculated as weight (kg) divided by the square of the height (m). WC was measured at the level of the midpoint between the lower rib margin and the iliac crest, and hip circumference was measured at the level of maximal protrusion of the gluteal muscles. WHR was calculated as WC (cm) divided by the hip circumference (cm). WHtR was calculated as WC (cm) divided by height (cm). NC was measured at the level of the laryngeal prominence while participants were asked to be in the standing position with the head held erect and eyes facing forward. WrC was measured when participants were asked to hold their arm on a flat surface, and the superior border of the tape measure was placed just distal to the prominences of radial and ulnar bones.

Measuring of BP and the definition of hypertension

BP was measured using a standard mercury-column sphygmomanometer after 15 min of rest in the sitting position based on the standardized procedural guideline (Kirkendall et al., 1981). Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were determined by the first and the fifth Korotkoff sounds. We measured BP on both right and left arm firstly, and the arm with higher blood pressure would be selected to obtain three consecutive measurements with a time interval of at least two minutes (Pickering et al., 2005). Then, the average on the selected arm would be recorded as the final BP. Before measuring, participants were also need to be asked for not doing following behaviors, including drinking alcohol, tea or coffee, smoking, or taking any exercise for at least 30 min before measuring BP. In this study, hypertension was defined as systolic blood pressure (SBP)/diastolic blood pressure (DBP) ≥140/90 mmHg or anti-hypertensive medication use (Pickering et al., 2005).

Other risk factor measurements

All participants were asked for permission to collect a blood sample after an overnight fast of longer than 8 h. The concentrations of total cholesterol (TC), triglycerides (TG), high density lipid cholesterol (HDL-C), and fasting glucose were examined using Mindray Autoanalyzer (BS 380 type; Mindray Ltd.; Shenzhen, China) in local community health service centers. All assays were performed according to the manufacturer’s instructions. In addition, self-administered questionnaires were also used to obtain information on demographic characteristics, medical history, medication use, smoking habits, and other pertinent factors. Trained staff members confirmed the reported information with each participant. In this study, smoking status was categorized as current smokers (smoking during the last one year or quit smoking for less than six months), past smokers (quit smoking for more than six months), and never. Drinking status were categorized as current drinkers (at least twice per week for male and once per week for female), past drinkers (quit drinking for more than six months), and never.

Statistical analysis

We have tested distributions of all continuous variables using Shapiro–Wilk test, and found that all these variables followed normal distributions (p-value >0.05). Therefore, we showed values of all continuous variables in mean and standard deviation (SD). Categorical variables were presented as percentages. Student t test and chi-square test were respectively used to estimate the difference of continuous and categorical variables between male and female. Six obesity indices were estimated in this study as following: BMI, WC, WHR, WHtR, NC and WrC. All six anthropometric obesity indices were standardized by being divided by their SD, respectively, so that crude and adjusted slopes per SD increase of each obesity index could be compared with each other. Multivariable linear regression models were used to estimate the crude and adjusted associations of BP (SBP and DBP) with per SD increase of each obesity index among participants without any anti-hypertensive medication use. Odds ratio (OR) and 95% confidence interval (95%CI) were used to estimate the risk of hypertension per SD increase of each obesity index using multivariable logistic regression models. In this study, we used three models in above-mentioned linear and logistic regressions: Model 1 (unadjusted), Model 2 (adjusted for age and gender), Model 3 (adjusted for smoking, alcohol drinking, fasting glucose, TG, TC and HDL-C in addition to Model 2). The ability of predicting hypertension using each obesity index together with other risk factors (confounders in Model 3) was examined by receiver operative characteristic (ROC) curves. We thereafter compared areas under ROC curves (AUC) of other 5 obesity indices to that of the index whose OR was largest in multivariable logistic regression model. The Bonferroni adjustment for multiple comparisons was used to adjust p-values. The subgroup analysis for gender was also performed, and p-values of interactions by gender on associations of obesity indices with BP and hypertension were estimated. All statistical analyses were performed by SAS 9.4 software (SAS Institute, Inc., Cary, North Carolina), and p-value <0.05 was considered as statistical significance.

Baseline characteristics

Baseline characteristics of 2,331 participants are showed in the Table 1. As it is shown, male participants were generally older than female participants. Except WHtR, the other five obesity indices (i.e., BMI, WC, WHR, NC, WrC) in male subjects were larger than that in female subjects (P-values were 0.293 for WHtR, and <0.001 for other five indices). Also, male participants showed higher levels of SBP and DBP than female participants (p-values were <0.001). The prevalence of hypertension among male participants (42.4%) was also found higher than that in females (35.5%) (P-value was 0.001). Additionally, male participants showed dramatically higher rates of current smokers (24.3%) and drinkers (51.5%) (p-values were <0.001).

The associations of WrC and the other obesity indices with BP

Table 2 shows the crude (Model 1) and adjusted (Model 2, 3) associations of SBP and DBP with per SD increase of each obesity indices. In the Model 1, WrC and other 5 obesity indices were all found positively associated with both SBP and DBP (p-values were <0.001). These positive associations remained present in the Model 2, which further adjusted for age and gender (p-values of associations of SBP with obesity indices were 0.043 for WrC, and <0.001 for other five indices; p-values of associations of DBP with obesity indices were 0.011 for WrC, 0.031 for WHR, and <0.001 for other four indices). However, in the Model 3, the association between WrC and SBP was not statistically significant (P-value was 0.062). In contrast, the other five obesity indices remained the positive associations with both SBP and DBP (p-values of association of SBP with five obesity indices but WrC were <0.001; p-values of associations of DBP with obesity indices were 0.031 for WHR, and <0.001 for other four indices but NC, whose P-value was 0.001). Furthermore, comparison of the increases of BP per SD of each obesity indices showed that BMI was the largest, with 2.52 mmHg for SBP and 1.59 mmHg for DBP in Model 2 and 2.45 mmHg for SBP and 1.53 mmHg for DBP in Model 3. In gender stratified analyses, no associations of WrC with SBP and DBP were found in either male or female participants after adjustment for all risk factors (Fig. 2) (p-values for SBP were 0.068 and 0.336 in male and female; p-values for DBP were 0.117 and 0.090 in male and female). On the contrary, the positive associations of BMI, WC and WHtR with SBP and DBP were found, respectively in both male and female participants even after consideration of all risk factors (Fig. 2) (p-values of association of SBP with BMI were <0.001 in both male and female; p-values of association of SBP with WC were 0.006 and <0.001 in male and female; p-values of association of SBP with WHtR were 0.012 and <0.001 in male and female; p-values of association of DBP with BMI were <0.001 in both male and female; p-values of association of DBP with WC were 0.005 and <0.001 in male and female; p-values of association of DBP with WHtR were 0.010 and <0.001 in male and female). Besides, we also found stronger associations of both WC and WHtR with SBP in female than that in male (p-values for interactions by gender were 0.014 and 0.031 for WC and WHtR, respectively) (Fig. 2A).

The associations of WrC and the other obesity indices with hypertension

The associations of the prevalence of hypertension and obesity indices are shown in the Table 3. The findings suggested positive associations of five obesity indices but WHR with hypertension even after adjustments for all risk factors in Model 3 (P-values were 0.024 for WrC, 0.064 for WHR and <0.001 for other four indices). However, OR values of WrC kept the smallest (OR was 1.16, 95 CI% [1.07–1.27] in Model 1; OR was 1.11, 95 CI% [1.02–1.21] in Model 2; and OR was 1.11, 95 CI% [1.01–1.21] in Model 3, respectively) compared with the other five indices. The association of BMI with hypertension was strongest after adjustments for risk factors (OR was 1.59, 95 CI% [1.44–1.75] in Model 2; and OR was 1.57, 95 CI% [1.42–1.73] in Model 3). Gender stratified analyses showed that WrC was only associated with hypertension in male after adjustment for all risk factors (P-value was 0.035), while no interaction by gender was found (Fig. 3) (P-value for interaction was 0.252). In contrast, BMI, WC and WHtR were found positively associated with hypertension in both male and female participants after adjusting for all risk factors (Fig. 3) (P-values for BMI were <0.001 in both male and female; p-values for WC were 0.019 and <0.001 in male and female; p-values for WHtR were 0.019 and <0.001 in male and female). Additionally, the associations of WC and WHtR with hypertension in female were found strongly higher than that in male (p-values for interaction by gender were 0.004 and 0.019 for WC and WHtR, respectively) (Fig. 3).

Furthermore, we compared the AUC of each obesity indices in predicting hypertension under the control of risk factors. As Table 4 shows, the AUC of BMI was the largest (0.709) among the 6 obesity indices. Compared to BMI, the AUC of WHR (AUC = 0.674), NC (AUC = 0.681) and WrC (AUC = 0.674) were found significantly lower (adjusted p-values were <0.001). Similar results were also suggested in female participants (Table 4).