Objective: To investigate the association of adiposity measures with blood pressure (BP) in Chinese children and adolescents.
Design: A cross-sectional study.
Participants: 1330 boys and 1170 girls aged 6–18 years from a rural population-based cohort of twins studied in Anhui, China, 1998–2000.
Outcome measures: Adiposity measures included body mass index (BMI), total body fat and trunk fat assessed by dual-energy x-ray absorptiometry. BMI was divided into fat mass index (FMI) and lean mass index (LMI) in the analysis. Major outcomes included: systolic (S) and diastolic (D) BP. Both linear and logistic regressions were performed to assess gender-specific associations between various adiposity measures and BP, with adjustment for age and height. Generalised estimating equations were used to account for intra-twin pair correlations.
Results: Mean BMI and percentage body fat in children aged 6–11 years were 14.9 kg/m2 and 9.7%, respectively; corresponding measures in children aged 12–18 years were 17.8 kg/m2 and 14.2%. Adiposity measures were more strongly associated with SBP (p<0.05 in all age strata) than DBP (p<0.05 only in children aged 6–11 years). Both FMI (β = 1.26–2.37) and LMI (β = 1.00–1.71) are associated with SBP across age and gender strata after adjustment for age and height (p<0.05).
Conclusions: These results indicate that, in this relatively lean population of Chinese children and adolescents, BP, particularly SBP, is positively associated with measures of adiposity. Of all the adiposity measures, BMI is the strongest predictor of BP.
Statistics from Altmetric.com
Hypertension is a global public health problem. For example, hypertension prevalence in Chinese adults increased from 11.3% in 1991 to 27.2% in 2000.1 2 During this period, the prevalence of overweight and obesity has also increased. From 1992 to 2002, the prevalence of overweight and obesity in Chinese children aged 0–6 and 7–17 years increased 31.7% and 17.9%, respectively.3 In adults, there is a well-established positive association between body mass index (BMI) and hypertension.4 5 This association has been observed in lean Chinese adults.6 However, limited data are available on children and adolescents. It is plausible that the positive BMI–blood pressure (BP) associations seen in adults are also present in children, as BMI and BP track from childhood through adulthood.7
Although BMI correlates with direct measures of adiposity such as total body fat (BF), trunk fat (TF) and percentage body fat (%BF), it does not distinguish lean body mass (eg, muscle, tendon and bone) from fat mass. Therefore, it has been suggested that BMI be split into its two components: fat mass index (FMI) and lean mass index (LMI).8 The proportions of lean mass and fat mass dramatically change during puberty and differ between boys and girls.9 Studies are needed to examine the association between adiposity and BP using BMI and direct measures of adiposity such as dual-energy x-ray absorptiometry (DEXA) to better understand the BMI–BP relationships and underlying biological mechanisms during a critical period of development: childhood and adolescence.
This study investigated the associations between adiposity measures and BP in a population-based sample of rural Chinese twins aged 6–18 years. In China, 85% of the population resides in rural settings.10 Although there is a lower prevalence of obesity in rural areas of China, the rate of change in obesity in such areas is outpacing that in its urban counterparts.3 The prevalence of obesity in rural China will increase, as will obesity-associated morbidity. Despite the growth of obesity in Chinese youth, few contemporary studies have described BP levels in relation to adiposity measures in Chinese children,11–13 and even less information on rural children is available. In particular, we are interested in whether the associations vary by indirect versus direct measures of adiposity, and whether the observed associations vary by age and gender. Such information is useful for deriving the best predictive model for BP in children and adolescents, and for identifying individuals at high risk of developing hypertension.
Study population and ascertainment of twins
This report includes data from a population-based cohort of twin pairs enrolled in Anqing, China from 1998 to 2000. Anqing spans 80 km along the north bank of the Yangtze River and includes three urban and eight rural counties covering 15 000 km2. The annual average temperature is 15.0°C. Twins were recruited to participate in a study designed to identify environmental and genetic determinants of chronic diseases including the metabolic syndrome. Potential participants were identified through a multistage process. First, investigators from the Anhui Medical University and the Anqing Hospitals and Research Institutes held a 3-day workshop in each township to train local doctors to collect twin information and explain the purpose, scope and procedures of the study. Local doctors went back to their own villages to prepare a list of all twins (both monozygotic and dizygotic). Epidemiologists from Anhui Medical University checked all twin lists with the township/village doctors. For the larger study, twins were chosen on the basis of age (6–60 years) and only if both twins or a parent of younger subjects consented. Eligible twins were invited to a central office to complete a questionnaire interview, blood drawing, physical examination and DEXA scan. The study protocol was approved by the institutional review boards of the Children’s Memorial Hospital and Biomedical Institute, Anhui Medical University.
Of 3412 age-eligible children (6–18 years), 2500 (1330 boys and 1170 girls) with both BP and adiposity measures were included in this analysis. In this report, 912 children were excluded because of missing data for DEXA measures (n = 876), BMI measures (n = 22), BP measures (n = 1), reported history of smoking (n = 9), and drinking alcohol (n = 4). The characteristics of the excluded subjects (eg, age, gender, systolic blood pressure (SBP), diastolic blood pressure (DBP), BMI and BF measures) did not differ significantly from those of included subjects (data not shown).
BP was assessed in seated and rested subjects as previously described using standard procedures and a mercury-gravity manometer.14 15 The mean of three measurements was used in the final analysis. High SBP or DBP was defined as measurements greater than or equal to the age-specific, sex-specific and height-specific 90th centiles.16
Height and weight were measured using standard protocols, without shoes or outerwear. Height was measured to the nearest 0.1 cm on a portable stadiometer and weight to the nearest 0.1 kg. BMI was calculated as weight (kg)/height squared (m2). All values reported represent the mean of three measurements.
Body fat measurements
DEXA measures the exponential attenuation of photons emitted at two energy levels that are absorbed by various body tissues. This allows accurate measurement of fat, fat-free and bone substance17 which has been validated against other estimates.18 19 DEXA is a relatively simple technique for evaluating total and regional adiposity in children of all ages.20 21 Although it cannot distinguish between intra-abdominal and subcutaneous fat, research in children22 has demonstrated strong correlations between trunk fat mass measured with DEXA and intra-abdominal fat measured with CT or MRI. A standard whole-body DEXA scan includes total body and three regional fat measures: trunk, arms and legs. A standard software (Lunar V4.6 DPX; GE Medical Systems, Madison, WI, USA) calculation17 was used to calculate BF measured with a Lunar DPXL instrument (Lunar, Madison, WI, USA) set up in Anqing. %BF was calculated by dividing BF by body weight and multiplying this by 100. %TF was calculated by dividing TF by body weight and multiplying this by 100.
Body composition measurements
BMI was split into its two components: FMI (BF/height2) and LMI (lean mass/height2), where lean mass is calculated weight−BF, and then FMI+LMI = BMI.8
All analyses were conducted separately by gender and by two age groups: children (6–11 years) and adolescents (12–18 years) using SAS V9. All twins were treated as individuals. As measures within twin pairs tend to correlate, standard errors obtained from unmatched analysis may be artificially deflated. To account for this effect, robust estimates of the variances were calculated using the generalised estimating equation (GEE) implemented in the SAS Procedure Genmod (V6.11). Firstly, we examined the crude association between age and BP by tertile BMI, FMI and LMI using smoothing plots. Separate smoothed plots for each gender using a non-parametric SAS procedure, LOESS, were created.23 24 Next, multiple linear regression analyses were used to evaluate the associations of adiposity measures with BP, adjusted for age, age2, height, years of education, and occupation (student/others). This modelling approach is similar to previous studies on children and adolescents in which BP was evaluated by age and height.25 Lastly, we used multiple logistic regression analysis to evaluate the associations of adiposity measures with the odds of high BP using age-specific, gender-specific and height-specific SBP and DBP 90th centile values as cut-offs, with adjustment for years of education and occupation (student/others) using GEE regression. p<0.05 was used to determine significance. The analysis was repeated including only one twin per family (the first-born twin). Because results were very similar (data not shown), this report presents results using data from all of the twins.
Table 1 shows gender-specific and age group-specific characteristics, adiposity measures and BP. Girls were slightly older than boys in both age groups. In the 6–11-year-old group, boys had a slightly higher BMI, but less fat mass than girls; in the 12–18-year-old group, boys had a slightly lower BMI and considerably lower fat mass than girls. BP levels were positively associated with age in boys and girls, and levelled off around age 12. Significant gender differences were noted for SBP, beginning around 12 years of age. No gender difference was noted for DBP.
The correlation coefficients between BF and BMI were 0.56 (p<0.001) in children and 0.73 (p<0.001) in adolescents. As these variables were highly correlated, we attempted to separate the effects of BMI and BF on BP by using FMI and LMI in the graphic analysis and regression models.
Figure 1A,B shows the association between age and SBP and DBP by tertiles of BMI, FMI and LMI. On average, BP levels are higher among those with higher BMI, FMI and LMI. This association is present in both genders and across the age range except for SBP in 6–8-year-olds and DBP in 16–18-year-olds.
Table 2 shows the association between each of the adiposity measures and BP by gender and age group using multiple regression analyses. After adjustment for age, age2, height, education and occupation (in adolescents), all adiposity measures (including LMI and FMI) in children and adolescents were associated with SBP. A dose–response relationship was apparent, as the highest tertile of adiposity had the highest SBP, whereas those in the lowest tertile of adiposity often had the lowest SBP and those in the middle tertile fell in between (data not shown). Findings were not as robust for the association between adiposity measures and DBP, in which the strength of the associations was greater for the younger age groups.
A similar analysis was performed for a binary BP outcome (ie, high versus low BP), using age, gender, and height-specific SBP and DBP 90th centile values as cut-offs. As shown in table 3, BMI, FMI and LMI are significantly associated with an increased odds of high BP across age strata and gender except for the high DBP in 12–18-year-old boys.
In adults, the association between adiposity and high blood pressure is known,4 5 but data for children are limited. A study of American adolescents (11–17 years old) reported that SBP correlated significantly with BMI, the Slaughter skinfold formulas, and %BF (DEXA) in boys, but not in girls.26 Similarly, Sorof and Daniels27 found an increased prevalence of systolic hypertension as BMI increased from the 5th to the 95th centile. Findings from the present report indicate that adiposity measures are generally positively associated with BP, although the relationships were non-linear. The adiposity measures were more strongly associated with SBP than DBP, and associations with DBP were stronger in children (ages 6–11 years) than in adolescents (ages 12–18 years).
Interestingly, even among children within a normal BMI range, there is a positive relation between BMI and BP level. Our observations are consistent with the results from studies conducted in other countries. Lindsay et al28 reported that BMI, %BF and fat mass showed similar degrees of correlation with BP during childhood (5–20 years) in a sample of Pima Indians. Monyeki et al29 reported that DBP, but not SBP, was significantly and positively associated with BMI in South African children (6–13 years). In the Amsterdam Growth and Health Study,30 an increase in the subscapular/triceps skinfold ratio was significantly associated with an increase in SBP in males and females (13–27 years). Our data indicate that the risk of raised BP increased across the entire BMI range, without a clear threshold effect. This suggests that using a continuous BMI variable is an appropriate method for quantitative description of risk in children and adolescents.
Most studies have used BMI as a measure of adiposity. Limited data are available on whether BMI versus direct body fat measures confer differential associations with BP. It is the general perception that body fat is better than BMI for predicting BP. Our study shows that, in a relatively lean population of children and adolescents, both fat mass and lean mass predict BP. This is consistent with another study on body composition and BP.31 Our data showed that BMI (which includes FMI and LMI), as an indirect adiposity measure easily obtained in medical practice, is the strongest predictor of BP among the adiposity indices assessed in this study. Similar findings were obtained when BP was analysed as a binary outcome (high versus low BP, using 90th centile BP values as cut-offs). In these relatively lean Chinese children and adolescents, there appears to be no additional benefit of measuring body fat in addition to BMI in the assessment of risk of high BP. However, this may be different in the assessment of other cardiovascular risk factors, including dyslipidaemia,32 insulin resistance and glucose intolerance, which are more closely associated with central adiposity.
Interestingly, in this study, girls had lower SBP values despite having higher %BF. In contrast, there was no obvious gender difference in DBP between any age group. A similar pattern has been observed in British subjects.33 Furthermore, whereas there are significant gender differences in adiposity measures, particularly during adolescence, there is no gender difference in the association between adiposity measures and BP. The findings from previous studies on gender difference are inconsistent.26 30 34 35 He et al36 reported a significant positive relationship between BP and trunk fat only in 5–18-year-old boys who were African–American, Asian and Caucasian. These findings were consistent when two independent adiposity measures (skinfold thickness and DEXA) were used. The varied findings across studies are probably due to differences in ethnicity, physical activity, nutritional status, and environmental and demographic discrepancies.
What is already known on this topic
In adults, there is a well-established positive association between body mass index (BMI) and hypertension. This association has also been seen in lean Chinese adults.
There is significant tracking of both BMI and blood pressure from childhood to adulthood.
What this study adds
This is the first study on blood pressure in relation to adiposity among lean, rural Chinese children and adolescents, using direct body fat measures determined by dual-energy x-ray absorptiometry.
Blood pressure, particularly systolic blood pressure, is positively associated with various measures of adiposity, but most strongly with BMI.
There appears to be no additional benefit of obtaining direct measures of body fat in addition to BMI for assessing risk of high blood pressure in this relatively lean, rural Chinese population.
Also, in this cohort of Chinese children, there are few significant relationships between adiposity measurements and DBP during adolescence. Burke et al37 monitored BP over a year in adolescents and, contrary to expectations, found that high DBP was inversely associated with weight: leaner children were more likely to remain in the upper quintile of DBP than the obese ones. This suggests that characteristics other than obesity contribute to high DBP during adolescence. The reason for this finding is not clear. DBP is determined by the elasticity and resistance of arteries. Aging lowers DBP because arterial elasticity decreases. Exposure to obesity early in life may induce changes in the arteries contributing to the development of atherosclerosis in adulthood.38 A recent study suggested that increased intima media thickening occurred more in obese children39 and probably produces decreased vascular elasticity in adolescents which then tracks into adulthood.40 These findings may partially explain why adiposity measures are not related to DBP in adolescents in our study.
The present study has the following limitations. Firstly, the physical activity level of the subjects was not assessed in detail. Ribeiro et al41 reported that children 8–15 years old in the highest quartile of body fatness are likely to be physically inactive, and others have found that the risk of hypertension was significantly lower if obese/overweight children had a higher than average level of cardiorespiratory fitness.42 Subjects in our twin cohort were lean and from rural counties where physical inactivity is not common,43 which limits generalisation of our findings to urban and non-lean populations. Secondly, the study sample was drawn from another larger study of twins, and our statistical analysis did not use a twin design, but rather treated twins as individuals with adjustment for auto-correlation between the twin pairs. This may be a deficiency in the methodology to have a possible confounding factor inherent in the design of the study itself. However, we did perform an additional analysis that included one twin from each family, and the result did not differ substantially. Thirdly, although our current sample size allowed us to demonstrate an association between BP and adiposity measures, an even larger sample size would be required to detect more subtle associations.
This is a relatively lean rural Chinese population. Using the Centers for Disease Control and Prevention growth charts,44 we plotted the Chinese children’s median BMI by age and gender (data not shown). On average, the Chinese children have lower BMIs. The average difference in BMI between US and Chinese children is 1.5 kg/m2 (boys) and 1.8 kg/m2 (girls) at 6–11 years of age and 2.2 kg/m2 (boys) and 2.0 kg/m2 (girls) at 12–18 years of age. We are not certain whether the same relationships between adiposity measures and BP that we found in our study subjects will hold true in other populations, especially among overweight and obese children.
However, more research on rural populations is needed and considered with the following study strengths. China has the largest population in the world, and rural populations in China (85% of the total Chinese population)10 constitute a fast growing segment of the world’s total population. The rapid increase in obesity in China and other developing countries is accompanied by an increasing prevalence of hypertension. Identifying early precursors of hypertension and individuals at risk, particularly while young, is an essential step in developing effective primary prevention of hypertension. This large study is one of a few that have simultaneously assessed direct and indirect measures of adiposity in children and adolescents. Such information is useful for deriving the best predictive model for BP in children and adolescents and for identifying individuals at high risk of developing hypertension.
In lean Chinese children and adolescents, BP, particularly SBP, is positively associated with measures of adiposity. These associations are already apparent in children aged 6–11 years within a normal BMI range. Our data indicate that the risk of raised BP increased across a range of BMI values, without a clear threshold. This suggests that treating BMI as a continuous variable is the most appropriate method for assessing the risk of hypertension in children and adolescents. Of the assessed adiposity measures, BMI appears to be most strongly associated with BP, indicating that it is not necessary to obtain additional direct measures of body fat beyond BMI to determine the risk of high BP in this relatively lean, rural Chinese population.
We thank the study participants and their families, and the faculty and staff of Anhui Medical University. We also thank Drs Wendy Brickman, Donald Zimmerman and Katherine Kaufer Christoffel for their review and critique of the manuscript.
Funding: This study is supported in part by grant R01 HD049059, R01 HL0864619, and R01 AG032227 from the National Institute of Health and by the Food Allergy Project.
Competing interests: None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.