Journal of Human Hypertension

https://doi.org/10.1038/s41371-018-0033-6

ARTICLE

Comparison of body mass index, waist circumference, conicity

index, and waist-to-height ratio for predicting incidence of
hypertension: the rural Chinese cohort study
Xu Chen1 Yu Liu2 Xizhuo Sun2 Zhaoxia Yin2 Honghui Li2 Kunpeng Deng3 Cheng Cheng1 Leilei Liu1
● ● ● ● ● ● ● ●

Xinping Luo4 Ruiyuan Zhang4 Feiyan Liu4 Qionggui Zhou4 Chongjian Wang1 Linlin Li1 Lu Zhang1
● ● ● ● ● ● ●

Bingyuan Wang1 Yang Zhao1 Junmei Zhou4 Chengyi Han1 Hongyan Zhang1 Xiangyu Yang1 Chao Pang5
● ● ● ● ● ● ●

Lei Yin5 Tianping Feng5 Jingzhi Zhao5 Ming Zhang4 Dongsheng Hu 1

● ● ● ●

Received: 11 September 2017 / Revised: 20 November 2017 / Accepted: 8 January 2018

© Macmillan Publishers Limited, part of Springer Nature 2018

Abstract
This study compared the ability of body mass index (BMI), waist circumference (WC), conicity index, and waist-to-height
1234567890();,:

ratio (WHtR) to predict incident hypertension and to identify the cutoffs of obesity indices for predicting hypertension in
rural Chinese adults. This prospective cohort study recruited 9905 participants aged 18–70 years during a median follow-up
of 6 years in rural China. Logistic regression and receiver operating characteristic (ROC) curve analyses were used to assess
the association, predictive ability, and optimal cutoffs (in terms of hypertension risk factors) of the four obesity indices: BMI,
WC, conicity index, and WHtR. The 6-year cumulative incidence of hypertension was 19.89% for men and 18.68% for
women, with a significant upward trend of increased incident hypertension with increasing BMI, WC, conicity index, and
WHtR (P for trend < 0.001) for both men and women. BMI and WHtR had the largest area under the ROC curve for
identifying hypertension for both genders. The optimal cutoff values for BMI, WC, conicity index, and WHtR for predicting
hypertension were 22.65 kg/m2, 82.70 cm, 1.20, and 0.49, respectively, for men, and 23.80 kg/m2, 82.17 cm, 1.20, and 0.52,
respectively, for women. BMI, WC, conicity index, and WHtR cutoffs may offer a simple and effective way to screen
hypertension in rural Chinese adults. BMI and WHtR were superior to WC and conicity index for predicting incident
hypertension for both genders.

Introduction
Electronic supplementary material The online version of this article
(https://doi.org/10.1038/s41371-018-0033-6) contains supplementary Hypertension has accounted for a grave and increasing
material, which is available to authorized users. burden of the global health, with an increasing prevalence
* Dongsheng Hu
and concomitant risks of cardiovascular and kidney disease
hud@szu.edu.cn [1]. China has the largest absolute disease burden of
hypertension in the world, with an estimated prevalence of
1
Department of Epidemiology and Health Statistics, College of 265 million people (aged 20–79 years) in 2010 [2].
Public Health, Zhengzhou University, Zhengzhou, Henan,
Epidemiological studies have shown that overweight and
People’s Republic of China
2
obesity are predominant underlying risk factors for hyper-
The Affiliated Luohu Hospital of Shenzhen University Health
tension [3], and the prevalence of overweight and obesity
Science Center, Shenzhen, Guangdong, People’s Republic of
China has increased tremendously in rural China with the rapid
3 socioeconomic developments and lifestyle changes [4, 5].
Yantian Entry-exit Inspection and Quarantine Bureau,
Shenzhen, Guangdong, People’s Republic of China Many obesity indices have been proposed, including body
4 mass index (BMI), waist circumference (WC), conicity
Department of Preventive Medicine, Shenzhen University Health
Science Center, Shenzhen, Guangdong, People’s Republic of index, and waist-to-height ratio (WHtR) because of their
China simplicity of measurement and cost-effectiveness. In gen-
5
Department of Prevention and Health Care, Military Hospital of eral, BMI and WC are the most widely used measures in
Henan Province, Zhengzhou, Henan, People’s Republic of China epidemiologic studies to define general and central obesity,

respectively [6]. Both indices have been found associated
with hypertension risk in Chinese adults [7, 8]. However,
BMI is limited by its inability to differentiate between fat
mass and lean body mass [9], and WC cannot account for
the effect of height on disease risk. Thus other obesity
indices are needed. WHtR was calculated to meet this need
and has been found better provide predictive validity of
hypertension than BMI and WC [10]. Another anthropo-
metric index of obesity is conicity index, introduced in 1991
by Valdez [11], that evaluates WC in terms of height and
weight. In a cross-sectional study performed in Bahia,
Brazil, conicity index was the best anthropometric index to
discriminate high coronary risk among people of African
descent with a high prevalence of hypertension [12].
Nevertheless, which obesity index (BMI, WC, conicity
index, and WHtR) indicates the highest risk for hyperten-
sion is controversial and which index has the best dis-
criminatory power for distinguishing people with
hypertension is unknown. Most studies investigated the
relationship between obesity indices and hypertension using
a cross-sectional design, and only a few investigated people
living in rural areas [13, 14].
To compare the ability of different obesity indices (BMI,
WC, conicity index, and WHtR) to predict incident hyper-
tension, this study was based on data from a rural Chinese
prospective cohort study. In addition, we identified the
optimal cutoffs for each index in this rural Chinese popu-
lation. To our knowledge, this is the first prospective study
to use receiver operating characteristic (ROC) curve ana-
lysis to compare the predictive ability of obesity indices for
hypertension and to identify optimal BMI, WC, conicity
index, and WHtR cutoffs for incident hypertension in rural
Chinese adults.
Methods
Study population
In total, 20,194 participants were selected by using cluster
random sampling method from a rural district in Henan
Province of China. Participants aged >18 years were
investigated by questionnaire interview, anthropometric
measurements, and laboratory measurements from July to
August 2007 and from July to August 2008 at baseline. A
total of 17,265 participants (response rate 85.5%) were re-
investigated and underwent a follow-up examination from
July to August 2013 and from July to October 2014. We
excluded people aged >70 years (n = 1441), with diagnosis
of hypertension (n = 4745), and with missing height and
weight and WC values (n = 5) at baseline as well as people
who died before the follow-up examination (n = 345) or
had unknown hypertension at follow-up (n = 824). Finally,
X. Chen et al.
9905 eligible participants (6039 women) were included in
the present study. The study was conducted according to the
Declaration of Helsinki and was approved by the Zhengz-
hou University Ethics Committee. All participants gave
their informed consent before any collection of samples or
data.
Data collection
Baseline examination, performed from July to August of
2007, included a standard questionnaire to gather data on
sociodemographic characteristics (gender, age, and educa-
tion level), behavioral measures (smoking and drinking),
history of antihypertensive drugs, family history, and pre-
sent disease history by face-to-face interviews. Smoking
was defined as having smoked 100 or more cigarettes
during the lifetime. Drinking was defined as drinking any
alcohol >12 times during the last year. Physical activity was
classified as low, moderate, or high according to the Inter-
national Physical Activity Questionnaire [15].
Before the day of investigation, participants were asked
to fast overnight. Height, WC, and weight were measured
twice according to a standard protocol [16], and the average
was used. Height and WC were measured to the nearest 0.1
cm by using a metric scale; weight was measured to the
nearest 0.5 kg by using a vertical weight scale. The intra-
class correlation coefficient (ICC) was used to assess
reliability and was 1 for height, weight, and WC. BMI,
conicity index, and WHtR were calculated as follows:
weight ðkgÞ
BMI ¼ ;
height2 ðmÞ
WC ðmÞ
Conicity index ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi;
0:109 weight ðkgÞ=height ðmÞ
WC ðcmÞ
WHtR ¼ :
height ðcmÞ
During one visit, after participants rested for 5 min in the
seated posture, blood pressure (BP) was assessed three
times in the right arm by using an electronic sphygmoman-
ometer (Omron, HEM-770AFuzzy, Kyoto, Japan), with 30
s between each measurement, according to a standard
protocol [17]. The average of the three closest readings was
used as the BP value [the ICC was 0.936 for systolic blood
pressure (SBP) and 0.925 for diastolic blood pressure
(DBP)]. Smoking, drinking a stimulant (tea, coffee, alcohol
etc.), and excessive exercise were prohibited 30 min before
measuring BP. Hypertension was defined as SBP ≥ 140 mm
Hg or DBP ≥ 90 mm Hg or use of blood pressure-lowering
drugs [18].
Blood samples were collected in the morning after an
overnight fast of at least 8 h and were used for measuring
levels of glucose (GLU), total cholesterol (TC), triglycer-
ides (TG), and high-density lipoprotein cholesterol (HDL-
Obesity indexes and hypertension

Table 1 Baseline characteristics of rural Chinese adults by blood pressure status at follow-up overall and by normotension and hypertension
Variables Total (n = 9905) Normotension (n = Hypertension (n = P-value
8008) 1897)

Men 3866 (39.0) 3097 (38.7) 769 (40.5) 0.135

Age (years) 47.00 (39.00–56.00) 46.00 (38.00–55.00) 52.00 (43.00–59.00) <0.001
Education level
High school or 1120 (11.3) 940 (11.7) 180 (9.5) 0.005
above
Smoking 2794 (28.2) 2260 (28.2) 534 (28.1) 0.950
Drinking 1264 (12.8) 1039 (13.0) 225 (11.9) 0.191
Physical activity 0.241
Low 5243 (52.9) 4212 (52.6) 1031 (54.3) 0.169
Moderate 2197 (22.2) 1802 (22.5) 395 (20.8) 0.113
High 2465 (24.9) 1994 (24.9) 471 (24.8) 0.948
Family history of 3007 (34.5) 2345 (29.3) 662 (34.9) <0.001
hypertension
BMI (kg/m2) 23.58 (21.40–25.94) 23.33 (21.22–25.67) 24.62 (22.43–26.84) <0.001
WC (cm) 80.25 (73.75–82.35) 79.55 (73.05–86.50) 83.75 (76.78–90.46) <0.001
CI 1.20 (1.15–1.26) 1.20 (1.15–1.25) 1.23 (1.17–1.28) <0.001
WHtR 0.51 (0.46–0.55) 0.50 (0.46–0.55) 0.53 (0.49–0.57) <0.001
SBP (mm Hg) 115.67 113.67 125.33 <0.001
(107.67–124.00) (106.33–121.33) (118.33–131.67)
DBP (mm Hg) 74.00 (68,67–79.33) 72.67 (67.67–78.00) 79.33 (74.33–84.00) <0.001
GLU (mmol/L) 5.29 (4.95–5.69) 5.27 (4.93–5.66) 5.41 (5.05–5.85) <0.001
TC (mmol/L) 4.27 (3.74–4.91) 4.24(3.71–4.87) 4.43 (3.88–5.06) <0.001
TG (mmol/L) 1.29 (0.92–1.85) 1.26 (0.90–1.81) 1.43 (1.01–2.07) <0.001
HDL-C (mmol/L) 1.14 (0.99–1.32) 1.15 (1.00–1.33) 1.12 (0.97–1.29) <0.001
LDL-C (mmol/L) 2.50 (2.00–2.90) 2.40 (2.00–2.90) 2.60 (2.10–3.00) <0.001
Data are presented as number (%) or median (range)
P-values: normotension compared with hypertension
BMI body mass index, WC waist circumference, CI conicity index, WHtR waist-to-height ratio, SBP systolic blood pressure, DBP diastolic blood
pressure, GLU glucose, TC total cholesterol, TG triglycerides, HDL-C high-density lipoprotein cholesterol, LDL-C low-density lipoprotein
cholesterol

C) by using a Hitachi 7060 automatic biochemical analyzer
(Tokyo). Low-density lipoprotein cholesterol (LDL-C) level
was quantified by the Friedewald formula [19].
Follow-up examination
Data collection at follow-up was the same as baseline
during July to August 2013 and from July to October 2014.
The incidence of hypertension was ascertained by BP value
and use of antihypertensive drugs at follow-up [18].
Statistical analysis
Continuous data (not normally distributed) were presented
as median (range) and were analyzed by Mann–Whitney U-
test. Categorical variables were presented as percentages
and were analyzed by chi-square test. Multiple logistic
regression models were used to calculate odds ratios (OR)
and 95% confidence intervals (CI) for the relationship
between incident hypertension and quartiles of BMI, WC,
conicity index, and WHtR, with adjustment for age, gender,
education level, smoking, drinking, physical activity, family
history of hypertension, and levels of GLU, TG, TC, HDL-
C and LDL-C in three models.
To compare data overall and for men and women, ROC
curve analysis was used to compare the ability of the obe-
sity indices to predict incident hypertension and to calculate
optimal cutoff values by using the maximum Youden index
(sensitivity+specificity−1). Furthermore, we calculated the
area under the ROC curve (AUC) and 95% CI for the
obesity indices for predicting incident hypertension and
used the method of DeLong et al. (1988) to test the statis-
tical significance of differences among the indices [20].
ROC analysis involved use of MedCalc v10.1.6.0 (MedCalc
Software, Ostend, Belgium) and other analyses involved
 X. Chen et al.

Fig. 1 Association of obesity indices and risk of hypertension. BMI history of hypertension, and levels of glucose, triglyceride, total cho-
body mass index, WC waist circumference, WHtR waist-to-height lesterol, high-density lipoprotein cholesterol, and low-density lipo-
ratio, CI confidence Interval, OR odds ratios. a Adjusted for age, protein cholesterol at baseline
gender, education level, smoking, drinking, physical activity, family

Table 2 Optimal obesity index cutoffs for predicting incidence of hypertension overall and by gender
Cutoff Sensitivity (%) Specificity (%) Youden index

Total BMI (kg/m2) 23.80 60.36 55.46 0.16

WC (cm) 82.70 54.35 62.30 0.17
CI 1.20 62.57 53.95 0.17
WHtR 0.51 60.62 57.65 0.18
Men BMI (kg/m2) 22.65 69.18 48.11 0.17
WC (cm) 82.70 56.44 58.93 0.15
CI 1.20 62.94 51.60 0.15
WHtR 0.49 63.98 54.25 0.18
Women BMI (kg/m2) 23.80 64.98 51.27 0.16
WC (cm) 82.17 55.05 62.90 0.18
CI 1.20 62.32 55.43 0.18
WHtR 0.52 66.93 52.15 0.19
BMI body mass index, WC waist circumference, CI conicity index, WHtR waist-to-height ratio

SPSS 21.0 (Chicago, IL, USA). All statistical tests were
two-sided, with significance level of 0.05.
Results
Table 1 shows the demographic characteristics of 9905
eligible participants aged 18–70 years at baseline stratified
by BP status at follow-up. Participants with hypertension
were older; had a positive family history of hypertension;
and had higher levels of GLU, TC, TG, and LDL-C and
lower level of HDL-C as compared with normal participants
(P<0.001). The median values for BMI, WC, conicity
index, and WHtR were higher with hypertension than nor-
motension (24.62 kg/m2, 83.75 cm, 1.23 and 0.53 vs 23.33
kg/m2, 79.55 cm, 1.20 and 0.50, respectively; P<0.001).
Alcohol drinking and smoking status as well as education
and physical activity level did not differ between groups (all
P > 0.05). Similar results are presented in supplementary
material 1, showing the main characteristics of the study
participants at follow-up.
During a median follow-up of 6 years, the cumulative
incidence of hypertension stratified by the 3 groups for
overall, men, and women were 19.15% (1897 cases),
19.89% (769 cases), and 18.68% (1128 cases), respectively.
Figure 1 presents a positive association among BMI, WC,
conicity index, WHtR, and hypertension with adjustment
for potential confounding factors. The highest risk of
hypertension was with WHtR ≥ 0.55 overall (adjusted OR
[aOR]: 2.52; 95% CI: 2.11, 2.99), BMI ≥ 25.31 kg/m2 for
men (aOR: 2.79; 95% CI: 2.14, 3.63), and WC ≥ 86.55 cm
for women (aOR: 2.56; 95% CI: 1.85, 3.54). BMI, WC,
conicity index, and WHtR were significantly associated
with hypertension in all subgroups, except conicity index
Obesity indexes and hypertension

Table 3 Area under the receiver operating characteristic (ROC) curve (AUC) values for obesity indices in relation to hypertension overall and by
gender
Total Men Women P-value
AUC 95% CI AUC 95% CI AUC 95% CI

Model 1 BMI 0.600d 0.590, 0.610 0.609bc 0.593, 0.624 0.599bd 0.586, 0.611 0.169
d ad
WC 0.608 0.599, 0.618 0.595 0.579, 0.610 0.616ad 0.604, 0.629 0.639
CI 0.604d 0.594, 0.613 0.585ad 0.569, 0.601 0.615d 0.603, 0.627 0.502
abc bc abc
WHtR 0.619 0.609, 0.629 0.616 0.600, 0.631 0.629 0.617, 0.642 0.771
Model 2 BMI 0.660c 0.650, 0.669 0.651bc 0.636, 0.666 0.669c 0.657, 0.681 0.742
WC 0.656c 0.647, 0.666 0.640acd 0.625, 0.656 0.668c 0.656, 0.680 0.609
abd abd
CI 0.644 0.634, 0.653 0.626 0.610, 0.641 0.656abd 0.644, 0.668 0.371
c bc c
WHtR 0.660 0.651, 0.669 0.649 0.634, 0.664 0.670 0.658, 0.682 0.701
Model 3 BMI 0.670c 0.660, 0.679 0.657 0.641, 0.672 0.687c 0.675, 0.699 0.675
WC 0.666c 0.657, 0.676 0.648d 0.632, 0.663 0.684c 0.672, 0.696 0.571
CI 0.659abd 0.650, 0.669 0.645d 0.630, 0.661 0.675abd 0.663, 0.687 0.547
WHtR 0.669c 0.659, 0.678 0.658bc 0.643, 0.673 0.683c 0.671, 0.695 0.635
Model 1, unadjusted
Model 2, adjusted for age
Model 3, adjusted for age, gender, education level, smoking, drinking, physical activity, family history of hypertension, and levels of glucose,
triglyceride, total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol at baseline
P-value, comparing men and women with the method of DeLong et al. (1988)
BMI body mass index, WC waist circumference, CI conicity index, WHtR waist-to-height ratio, CI confidence interval
a
Significant differences with BMI (P < 0.05)
b
Significant differences with WC (P < 0.05)
c
Significant differences with CI (P < 0.05)
d
Significant differences with WHtR (P < 0.05)

1.15–1.20 overall, WC 74.50–81.10 cm, conicity index
1.16–1.21, and conicity index 1.21–1.26 for men and
conicity index 1.15–1.20 and BMI 21.67–23.95 kg/m2 for
women. We found a significant upward trend of increased
incident hypertension with increasing BMI, WC, conicity
index, and WHtR (P for trend < 0.001) overall and for men
and women (data not shown).
Table 2 shows the optimal cutoffs and corresponding
sensitivity, specificity, and Youden index for each obesity
index for predicting incident hypertension. The Youden
index for WHtR was highest overall (0.18) and for men and
women (0.18 and 0.19).
To identify which obesity index was the best predictor of
incident hypertension, we calculated AUC values for the
indices in terms of incident hypertension and P-values for
pairwise comparisons (Table 3). ROC curve analysis of the
association between the four indices and hypertension
overall and for men and women are shown in supplemen-
tary material 2.
ROC curve analysis showed that among the 4 obesity
indices, WHtR consistently had the largest AUC value
(0.619) for incident hypertension in the crude model and the
differences were all statistically significant (P<0.05). After
adjusting for age (Model 2), BMI and WHtR had the largest
AUC values (both 0.660). However, pairwise comparison
of AUC values indicated significant differences only
between BMI and conicity index and between WHtR and
conicity index. WHtR had the largest AUC value (0.669)
after adjusting for potential confounding factors (Model 3).
For men, among the four obesity indices, WHtR had the
largest AUC value after adjustment for potential con-
founding variables. Differences were statistically significant
only between WHtR and conicity index and between WHtR
and WC in Model 3. For women, in the crude model, WHtR
had the largest AUC value (0.616), whereas in Models 2
and 3, the AUC value was slightly larger for BMI than for
the other obesity indices. Models 2 and 3 demonstrated
nonsignificant statistically differences between BMI and
WHtR (P = 0.762 and 0.140, respectively). AUC values for
each obesity index were consistently similar between men
and women (all P > 0.05).
Discussion
In this longitudinal population-based study, BMI, WC,
conicity index, and WHtR were independently associated
with risk of incident hypertension. AUC values were larger

for BMI and WHtR than other obesity indices after
adjusting for age and other covariates and could be better
tools for predicting hypertension in both genders. The
optimal cutoffs for BMI and WHtR were 23.80 kg/m2 and
0.51 overall, 22.65 kg/m2 and 0.49 for men, and 23.80 kg/
m2 and 0.52 for women.
Population-based studies have addressed the value of the
obesity indices in predicting and preventing hypertension at
the population level [13]. However, which obesity index is
most closely associated with BP values and conveys the
highest risk for hypertension is still controversial [21]. A
cross-sectional study of 3006 Chinese adults aged 20–74
years (1522 men) found BMI associated stronger with
hypertension than with WC and WHtR [22]. A longitudinal
study indicated that, among 1441 men and 1812 women
(aged 18–65 years), BMI may be a better indicator of
hypertension than other obesity indices at 5-year follow-up
in China [23]. The ATTICA epidemiologic study suggested
that, for 3042 non-hypertensive Greek adults, WC was the
most significant predictor for incident hypertension at 5-
year follow-up [24]. Ethnicity may explain the differences
in these results: different ethnic groups usually have dif-
ferent socioeconomic status and lifestyle-relevant risk-factor
exposure, and ethnicities may differ in combinations of
genes associated with hypertension as well as
gene–environment interactions, which lead to changes in
BP [25]. Another explanation may be that some of these
studies used a cross-sectional design, and some results were
not from large-scale studies.
We examined a positive and significant association
between obesity indices and incident hypertension in this
rural Chinese adult population. Remarkably, we observed a
significantly strong association of obesity with hyperten-
sion. With increasing BMI, WC, conicity index, or WHtR,
the risk of hypertension increased substantially for both
genders (all P for trend < 0.001). The possible mechanisms
may be that BMI is related to an increase in body fluid
volume, peripheral resistance, and cardiac output [23]; WC,
conicity index, or WHtR are linked with an increase in
visceral fat, which leads to increased leptin and insulin
resistance and worsened lipid profiles [26]. We found
WHtR the best predictor of hypertension overall and BMI
and WC the best predictors for men and women after
adjusting for age and other covariates. Moreover, these
results agreed with ROC curve analyses, except for women.
The current study showed that all obesity indices can be
used to predict the incidence of hypertension because all
AUCs were >0.5. Hence, anthropometric indicators of
overall obesity (BMI) and abdominal obesity (WC, conicity
index, and WHtR) can be used for screening hypertension
for rural Chinese adults. By comparison, WHtR performed
best in identifying hypertension in the unadjusted model;
similar results were reported in several previous studies [10,
X. Chen et al.
27]. Nevertheless, the AUC for WHtR was similar to values
for BMI and WC overall and for women after adjustment
for potential confounding variables. According to the AUC
values, we found no significant differences between BMI
and WHtR in predicting hypertension for men. Silva et al.
found that, as compared with BMI, WC, and WHtR, coni-
city index was the best indicator to identify hypertension in
black women [28]. In our exploratory analysis, for both men
and women, conicity index was not better than BMI, WC,
and WHtR in predicting incident hypertension (P > 0.05).
As compared with Western populations, our Asian popu-
lation tends to have a higher total body fat and visceral fat at
a lower BMI and WC [29]. A study involving dual energy
X-ray absorptiometry revealed greater percentage of body
fat for Asians than for African-Americans and whites with
similar BMI [30]. Moreover, our study was not similar to
findings from a previous study [13] finding no significant
difference in AUC values among men and women for each
obesity index (P > 0.05). Some of the previous studies were
cross-sectional, which implies less power for a causal
inference on the association of AUC values and hyperten-
sion between men and women. In addition, some did not
adjust for age and other covariates or the sample sizes were
small. Therefore, further prospective studies are needed to
clarify the association.
We propose a series of optimal cutoffs for these obesity
indices, which could provide suggestions for similar studies
and populations. Overall, BMI and WHtR had similar
optimal cutoff values with a previous study in China (BMI:
23.24 kg/m2 for men and 23.64 kg/m2 for women; WHtR:
0.48 for men and 0.51 for women) [31]. However, the
optimal WC cutoff values to predict the risk of hypertension
in our study were 82.17 cm for men and 82.70 cm for
women, which differs from values in the literature. The
criteria for Chinese adults suggested by the International
Diabetes Federation uses WC ≥ 90 cm (men) and ≥ 80 cm
(women) as the best cutoff values to identify obesity and
health risks [32]. Moreover, in a study conducted in Chinese
urban areas, the most suitable cutoffs for conicity index to
detect metabolic syndrome risk were 1.21 for men and 1.23
for women [33]. These cutoff values are a little higher than
in our study (1.20 for both genders). The differences in
these values could be attributed to differences in lifestyle
and economic factors between urban and rural areas. For
instance, rural residents usually pay less attention to their
health status and have less information on hypertension than
do urban residents, which might be due to lower education
level or affordability as well as lack of adequate basic
healthcare as compared with the care available to urban
dwellers, thereby increasing the risks of hypertension [34].
Furthermore, we found BMI and WHtR better than other
obesity indices in identifying hypertension for both men and
women. In addition, a recent systematic review and meta-
Obesity indexes and hypertension
analysis suggested a global cutoff for WHtR of 0.5 for both
genders [10]; in our study, the optimal WHtR cutoffs were
0.49 for men and 0.52 for women. Moreover, in 2002, the
Working Group on Obesity in China developed a cutoff
value for overweight by using BMI (24.0 kg/m2) for the
general Chinese population [35], which was similar to our
study.
Strengths and limitations
The strengths of the study include its prospective design,
large community-based sample, and extensive follow-up. In
addition, the standardized high-quality anthropometric
measurements and laboratory data reduced the potential
biases and measurement errors. However, the participants
were from a relatively restricted rural area in a central
Chinese province, which may not be a representative sam-
ple of the whole rural area in China. Second, the AUC
values for the four obesity indices ranging from 0.585 to
0.682 were not high enough for clinical use. To obtain
larger AUCs, further studies need to combine the obesity
indices and consider a synthesis score to evaluate body
shape (“chilli”, “pear”, “apple”, or “pear-apple”) and simul-
taneously identify the specific obesity categories defined by
overall and central obesity measurements.
Conclusion
We found BMI, WC, conicity index, and WHtR positively
associated with follow-up incident hypertension among a
sample of rural Chinese adults. BMI and WHtR may be
better predictors of incident hypertension than other indices
for both genders after controlling for age and/or other
covariates. These results have important public health
implications for rural areas in China. Obesity indices offer
the prospect of a simple and effective approach for first-
level screening for hypertension.
Summary Table
What is known about the topic?
● Epidemiological studies based on cross-sectional sur-
veys have addressed the value of obesity indices in
predicting and preventing hypertension at the population
level.
● However, the obesity index most closely associated with
blood pressure values and incident hypertension is still
controversial.
● Little information is available on the association of
obesity indices and hypertension in rural China.
What this study adds?
● This study focused on rural adults, and its results have
important public health implications for developing
countries.
● To our knowledge, this is the first prospective study to
use receiver operating characteristic (ROC) curve
analysis to compare the predictive ability of obesity
indices for hypertension and to identify optimal cutoff
values for body mass index, waist circumference,
conicity index, and waist-to-height ratio for predicting
incident hypertension in rural Chinese adults.
Acknowledgements We thank the dedicated participants and all
research staff of the study.
Funding This study was supported by the National Natural Science
Foundation of China (grant numbers 81373074, 81402752, and
81673260); the Natural Science Foundation of Guangdong Province
(grant number 2017A03013452); the Medical Research Foundation of
Guangdong Province (grant number A2017181); and the Science and
Technology Development Foundation of Shenzhen (grant numbers
JCYJ20140418091413562, JCYJ 2016030715570, JCYJ
20170302143855721, and JCYJ20170412110537191).
Compliance with Ethical Standards
Conflict of interest The authors declare that they have no conflict of
interest.
References
1. Kearney PK, Whelton M, Reynolds K, Muntner P, Whelton PK,
He J. Global burden of hypertension--analysis of worldwide data.
Lancet. 2005;365:217–23.
2. Li D, Lv J, Liu F, Liu P, Yang X, Feng Y, et al. Hypertension
burden and control in mainland China: analysis of nationwide data
2003-2012. Int J Cardiol. 2015;184:637–44.
3. Kotsis V, Nilsson P, Grassi G, Mancia G, Redon J, Luft F, et al.
New developments in the pathogenesis of obesity-induced
hypertension. J Hypertens. 2015;33:1499–508.
4. Reynolds K, Gu D, Whelton PK, Wu X, Duan X, Mo J, et al.
Prevalence and risk factors of overweight and obesity in China.
Obes (Silver Spring). 2007;15:10–8.
5. Guo X, Li Z, Guo L, Zheng L, Yu S, Yang H, et al. An update on
overweight and obesity in rural Northeast China: from lifestyle
risk factors to cardiometabolic comorbidities. BMC Public Health.
2014;14:1046.
6. Choo V. WHO reassesses appropriate body-mass index for Asian
populations. Lancet. 2002;360:235.
7. Luo W, Chen F, Guo Z, Wu M, Hao C, Zhou Z, et al. Comparison
of the suitability of 2 years change in waist circum- ference and
body mass index in predicting hypertension risk: a prospective
study in Chinese-Han. Iran J Public Health. 2014;43:1212–20.
8. Shuger SL, Sui X, Church TS, Meriwether RA, Blair SN. Body
mass index as a predictor of hypertension incidence among
initially healthy normotensive women. Am J Hypertens.
2008;21:613–9.
9. Weber DR, Leonard MB, Shults J, Zemel BS. A comparison of fat
and lean body mass index to BMI for the identification of

metabolic syndrome in children and adolescents. J Clin Endocri-
nol Metab. 2014;99:3208–16.
10. Ashwell M, Gunn P, Gibson S. Waist-to-height ratio is a better
screening tool than waist circumference and BMI for adult car-
diometabolic risk factors: systematic review and meta-analysis.
Obes Rev. 2012;13:275–86.
11. Valdez R. A simple model-based index of abdominal adiposity. J
Clin Epidemiol. 1991;44:955–6.
12. Pitanga FJG, Lessa I. [Anthropometric indexes of obesity as an
instrument of screening for high coronary risk in adults in the city
of Salvador--Bahia]. Arq Bras Cardiol. 2005;85:26–31.
13. Jiang J, Deng S, Chen Y, Liang S, Ma N, Xu Y, et al. Comparison
of visceral and body fat indices and anthropometric measures in
relation to untreated hypertension by age and gender among
Chinese. Int J Cardiol. 2016;219:204–11.
14. Ko K-P, Oh D-K, Min H, Kim C-S, Park J-K, Kim Y, et al.
Prospective study of optimal obesity index cutoffs for predicting
development of multiple metabolic risk factors: the Korean gen-
ome and epidemiology study. J Epidemiol. 2012;22:433–9.
15. Craig CL, Marshall AL, Sjöström M, Bauman AE, Booth ML,
Ainsworth BE, et al. International physical activity questionnaire:
12-Country reliability and validity. Med Sci Sports Exerc.
2003;35:1381–95.
16. Geographical variation in the major risk factors of coronary heart
disease in men and women aged 35-64 years. The WHO MON-
ICA Project. World Health Stat Q. 1988;41:115–40.
17. Perloff D, Grim C, Flack J, Frohlich ED, Hill M, McDonald M.
et al. Human blood-pressure determination by sphygmomano-
metry. Circulation. 1993;88:2460–70.
18. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA,
Izzo JL, et al. Seventh report of the Joint National Committee on
Prevention, Detection, Evaluation, and Treatment of High Blood
Pressure. Hypertension. 2003;42:1206–52.
19. Bairaktari E, Hatzidimou K, Tzallas C, Vini M, Katsaraki A,
Tselepis A, et al. Estimation of LDL cholesterol based on the
Friedewald formula and on apo B levels. Clin Biochem.
2000;33:549–55.
20. Delong ER, Carolina N. Comparing the ch. Biometrics.
1988;44:837–45.
21. Nyamdorj R, Qiao Q, Lam TH, Tuomilehto J, Ho SY, Pitkäniemi
J, et al. BMI compared with central obesity indicators in relation
to diabetes and hypertension in Asians. Obesity.
2008;16:1622–35.
22. Dong X, Liu Y, Yang J, Sun Y, Chen L. Efficiency of anthro-
pometric indicators of obesity for identifying cardiovascular risk
factors in a Chinese population. Postgrad Med J. 2011;87:251–6.
23. Ren Q, Su C, Wang H, Wang Z, Du W, Zhang B. Prospective
study of optimal obesity index cut-off values for predicting
X. Chen et al.
incidence of hypertension in 18-65-year-old Chinese adults. PLoS
ONE. 2016;11:1–14.
24. Panagiotakos DB, Chrysohoou C, Pitsavos C, Skoumas J, Lentzas
Y, Katinioti A, et al. Hierarchical analysis of anthropometric
indices in the prediction of 5-year incidence of hypertension in
apparently healthy adults: the ATTICA study. Atherosclerosis.
2009;206:314–20.
25. Nguyen TT, Adair LS, He K, Popkin BM. Optimal cutoff values
for overweight: using body mass index to predict incidence of
hypertension in 18- to 65-year-old Chinese adults. J Nutr.
2008;138:1377–82.
26. Norman MKaplan. Primary hypertension: pathogenesis. In:
Kaplan NM, editor. Kaplan’s Clinical Hypertension. Philadelphia,
PA: Lippincott Williams & Wilkins; 2006. p. 50–121.
27. Browning LM, Hsieh SD, Ashwell M. A systematic review of
waist-to-height ratio as a screening tool for the prediction of
cardiovascular disease and diabetes: 0.5 could be a suitable global
boundary value. Nutr Res Rev. 2010;23:247–69.
28. Silva DAS, Petroski EL, Peres MA. Accuracy and measures of
association of anthropometric indexes of obesity to identify the
presence of hypertension in adults: a population-based study in
Southern Brazil. Eur J Nutr. 2013;52:237–46.
29. Tuan NT, Adair LS, Stevens J, Popkin BM. Prediction of
hypertension by different anthropometric indices in adults: the
change in estimate approach. Public Health Nutr. 2010;13:639.
30. Gallagher D, Heymsfield SB, Heo M, Jebb SA, Murgatroyd PR,
Sakamoto Y. Healthy percentage body fat ranges: an approach for
developing guidelines based on body mass index. Am J Clin Nutr.
2000;72:694–701.
31. Yu J, Tao Y, Tao Y, Yang S, Yu Y, Li B, et al. Optimal cut-off of
obesity indices to predict cardiovascular disease risk factors and
metabolic syndrome among adults in Northeast China. BMC
Public Health. 2016;16:1079.
32. Alberti KGMM, Zimmet P, Shaw J. The metabolic syndrome - a
new worldwide definition. Lancet. 2005;366:1059–62.
33. Wang H, Liu A, Zhao T, Gong X, Pang T, Zhou Y, et al. Com-
parison of anthropometric indices for predicting the risk of
metabolic syndrome and its components in Chinese adults: a
prospective, longitudinal study. BMJ Open. 2017;7:e016062.
34. Liu X, Gu W, Li Z, Lei H, Li G, Huang W. Hypertension pre-
valence, awareness, treatment, control, and associated factors in
Southwest China. J Hypertens. 2017;35:637–44.
35. Bei-Fan Z. Predictive values of body mass index and waist cir-
cumference for risk factors of certain related diseases in Chinese
adults: study on optimal cut-off points of body mass index and
waist circumference in Chinese adults. Asia Pac J Clin Nutr.
2002;11(Suppl 8):S685–S693.