Excess body fat or adiposity is an important cardio-metabolic risk factor than excess body weight per se. Since BMI does not differentiate between fat and lean, other surrogate measures of body fat distribution such as WC, triceps skin fold thickness (TSFT), waist to height ratio (WHtR) are presently being evaluated for their association with metabolic risk [7-10]. While WC is a crude measure of intra abdominal fat, TSFT is predictive of body fat and metabolic risk in children and adolescents [11-13]. Measurement of Wrist circumference (WrC) is also an easy-to-detect clinical marker to identify at risk children [14]. Although there are ethnic-specific definitions for general and central obesity [15,16], few studies have compared ability of various adiposity indices and provided age-gender specific cut offs for screening children and adolescents for the risk of hypertension.

Therefore, the objectives of the present study were (i) to investigate relationships of BMI, WC, WHtR, TSFT and Wrist circumference with BP; and (ii) to provide optimal cut off points for these indices for detecting hypertension.

The study was performed in schools from five major Indian cities (Chennai, Delhi, Kolkata, Pune and Raipur), between July 2011 and January 2012). Schools were randomly selected from the list of private schools catering to well-off families in each city. The mean yearly fees of the schools were INR 26,000 (15,000 to 54000). All apparently healthy children from 6 to 18 years of age from the selected school were included after informed written consent from parents and assent from children. The exclusion criterion was children with pre-existing serious illnesses. The study was approved by the ethics committee of the Hirabai Cowasji Jehangir Medical Research Institute.

A single team led the data collection at each site and equipments were calibrated daily. Mean inter-observer and intra-observer coefficients of variation were both < 0.01(1%) for height (Ht), Weight (Wt), Waist Circumference (WC), Tricep Skin fold Thickness (TSFT), Wrist circumference and Blood Pressure measurements, and differences between observers were not statistically significant.

Children with persistently high BP underwent a further diagnostic workup for causes of hypertension including familial, renal, obesity, etc.; children with pre hypertension were managed on diet and lifestyle changes, while those with blood pressure above the 95th percentile were either advised the same and followed up or were initiated on therapy with anti-hypertensive treatment [5].

Statistical methods: SPSS version 20.0 (Chicago, 2011) was used for analysis. All results are expressed as mean (SD). Correlations were estimated (unadjusted and after age adjustment) separately for both genders to examine association of anthropometric measurements with BP. Level of significance was set at P<0.05. Two separate multiple logistic regression models adjusted for age and gender were used to examine relationship of hypertension with BMI categories, TSFT classes and Wrist with WC categories in the first model, and with WHtR categories in the second model to avoid multi-collinearity. Receiver-operating characteristic (ROC) curve analysis was performed to evaluate the accuracy of each index (i.e., BMI, WC, WHtR, TSFT, WrC) to discriminate between presence or absence of hypertension. The area under each ROC curve (AUC) and 95% confidence intervals (CI) were estimated to compare the relative ability of various anthropometric indices to identify risk of high BP. Optimal cut-off points for each anthropometric indicator were determined [26,27]. The differences between area under ROC curves for BMI, WC, WHtR, TSFT and WrC to determine the best predictor for hypertension was tested [28].

Age-and gender-wise anthropometric characteristics of children aged 6-18 years (total 6380, 3501 boys) are described in Table I. In each age group, boys had higher mean values of Ht, Wt, BMI, WC, WrC and BP than girls (P<0.05) except for TSFT.

TABLE I Characteristic of Study Population According to Age Groups

A significant positive association (r=0.85) was observed between BMI and WC in boys and girls within each age group (P<0.001). Correlations of BMI with TSFT and WHtR were 0.78 to 0.80, respectively across age groups in both genders (P<0.001). BMI showed higher correlation with Wrist (r= 0.70) in younger age group which reduced to 0.6 and then to 0.45 in older groups (P<0.001). WC was positively correlated with TSFT (r = 0.67 - 0.74, P<0.001) and Wrist (r= 0.41 - 0.78, P<0.001) in both genders across all groups (Data not shown).

Overall prevalence of hypertension was higher in boys (6.7%) than girls (4.2%) (P<0.05) with higher percentage in the younger age groups than older age group. Prevalence was also higher in overweight/obese than normal girls (P<0.05) which decreased with increasing age. In boys, prevalence of hypertension amongst overweight/obese was higher from 13-15yrs (Fig. 1 and 2).

Fig. 1 Prevalence of hypertension across BMI category in boys and girls.

Fig. 2 Prevalence of hypertension across the waist centile categories in boys and girls.

Gender wise correlations between SBP, DBP and anthropometric indices are presented in Table II. All correlations were positive and significant (P<0.001). In both genders, correlation coefficients for BMI and WC with SBP were around 0.5 (P<0.001). WHtR showed a low correlation with SBP and DBP in both genders.

TABLE II Correlation Between Systolic BP (Diastolic BP) and Anthropometric Indices in Boys and Girls

Multiple logistic regression model 1 (adjustment: age, gender) indicated that odds ratios (OR) for BMI categories, waist centile classes, TSFT centile classes, and Wrist against high BP were statistically significant (P<0.001) (Web Table I). Overweight children showed double risk of hypertension and obese children 7 times higher risk than normal weight children. TSFT >95th centile showed almost 3 times risk and between 85th-95th double risk than normal weight children. Higher WC (>90th centile) also exhibited 1.5 times risk and larger Wrist 1.26 times higher risk of hypertension. The model 2 (WHtR in place of WC) showed that estimated risk of hypertension was similar as 1.5 times in children with high WHtR (>0.44) than normal WHtR (<0.44) children (P<0.0001). Thus all these indices showed ability to predict risk of hypertension.

Table III describes the age- and gender-specific optimal cut-off values and for each of the five anthropometric indices in detecting the risk of high BP. With growing age, all the body measurements showed increasing trend which is reflected in higher cut offs for BMI, WC, WHtR, TSFT and WrC in older age groups in both genders. Sensitivity and specificity of all the indices were similar ranging from 60 to 90%. The Area under curve was also significantly high, different from 0.5 for BMI, WC, WHtR, TSFT and Wrist for both genders (Web Table II) indicating the ability of these anthropometric indices for detecting the risk of high BP. Overall comparison of the five indices in different age-gender groups suggests that BMI, WC and TSFT are better indicators of risk of hypertension.

Table III Cut-off Values of Anthropometric Indices for Prediction of Hypertension in Children

All five indices showed significant positive association with BP and indicated that obese children were seven times at risk of hypertension than normal-weight children. We have also proposed age-gender specific cut-off values for these five anthropometric indices. For boys, BMI, WC and TSFT showed similar predictive power while in girls all five indices performed equally well.

In the present study, we have provided cut-off values for various anthropometric indices i.e. BMI, waist circumference, WHtR, TSFT and for the first time for Wrist circumferences for detecting risk of hypertension in Indian children. In a study on Brazilian adolescents, cut-offs for BMI were in line with our cut-offs for girls (14-18 yrs), though they were lower for boys [31]. For Iranian adolescents, cut-offs for boys and girls were lower than those reported in our study [32].

A limitation of the study is that children were classified in age groups on consideration of conventional pubertal development years. It was not possible to assess Tanner staging for each child in the present study due to logistic reasons. Second limitation of the study was non-availability of biochemical measurements, therefore examining utility of these indices for screening other cardio-metabolic risk factors was not possible. Thirdly we did not record family history of hypertension, which may have some effect on the BP status of children. Although even after adjusting for parental influence the risk in children would be significant [33]. However, for community surveys, screening for hypertension with non-invasive body measurements would go a long way in prevention of risk of cardiovascular disease in adulthood.

To conclude, all five measures of adiposity were significantly associated with risk of hypertension in a multi-centric sample of Indian children and adolescents. Age-gender specific optimal cutoffs for BMI, TSFT, WC, wrist circumference and WHtR measurements presented in the study may be useful in screening for risk of hypertension.

Acknowledgment: We thank Dr Archana Arya, Dr Sanwar Agarwal, Dr Anjan Bhattacharya, Ms Veena Ekbote, Ms Lavnya Parthasarathy and Mrs Surbhi Shah Kulkarni for their help throughout the study. We acknowledge the help from the school authorities, and the parents and students.

Contributors: NAK: designed the data collection instruments and coordinated and supervised data collection at the sites, carried out the analysis, drafted the initial manuscript; AVK: conceptualized and designed the study, coordinated and supervised data collection, carried out the initial analyses, reviewed and revised the manuscript. She will act as guarantor of the study; SAC: conceptualized and designed the study, carried out the analyses, reviewed and revised the manuscript; VVK: conceptualized and designed the study, critically reviewed the manuscript. The final manuscript was approved by all authors.

Funding: Novo Nordisk India Pvt. Ltd.

Competing interests: None stated.