No use for waist-for-height ratio in addition to body mass index to identify children with elevated blood pressure

Abstract

Background. In children, waist-for-height ratio (WHtR) has been proposed to identify subjects at higher risk of cardiovascular diseases. The utility of WHtR to identify children with elevated blood pressure (BP) is unclear. Design. Cross-sectional population-based study of schoolchildren. Methods. Weight, height, waist circumference and BP were measured in all sixth-grade schoolchildren of the canton de Vaud (Switzerland) in 2005/06. WHtR was computed as waist [cm]/height [cm]. Elevated BP was defined according to sex-, age- and height-specific US reference data. The area under the receiver operating characteristic curve (AUC) statistic was computed to compare the ability of body mass index (BMI) z-score and WHtR, alone or in combination, to identify children with elevated BP. Results. 5207 children participated (76% response) [2621 boys, 2586 girls; mean (± SD) age, 12.3 ± 0.5 years; range: 10.1–14.9]. The prevalence of elevated BP was 11%. Mean WHtR was 0.44 ± 0.05 (range: 0.29– 0.77) and 11% had high WHtR (> 0.5). BMI z-score and WHtR were strongly correlated (Spearman correlation coefficient r = 0.76). Both indices were positively associated with elevated BP. AUCs for elevated BP was relatively low for BMI z-score (0.62) or for WHtR (0.62), and was not substantially improved when both indices were considered together (0.63). Conclusions. The ability of BMI z-score or WHtR to identify children aged 10–14 with elevated BP was weak. Adding WHtR did not confer additional discriminative power to BMI alone. These findings do not support the measurement of WHtR in addition to BMI to identify children with elevated BP.

Key Words::

• Body mass index
• children
• obesity
• waist

Introduction

Obesity is associated with hypertension and with an increased risk of cardiovascular diseases (CVD) in adults (1) and in children (2,3). While body mass index (BMI) is a common metric for identifying individuals with obesity, and because abdominal adiposity may be more closely related to CVD risk than peripheral adiposity, waist circumference has been proposed as a better indicator of health risk than BMI, at least in adults (4).

In children, a high waist circumference is associated with CVD risk factors (5). To account for differences in height, waist-for-height ratio (WHtR) has been proposed as a simple metric to identify children at increased cardio-metabolic risk, with WHtR above 0.5 being considered as high for children irrespective of sex and age (6–8). However, the evidence on the utility of WHtR to identify children with elevated blood pressure (BP) is inconclusive. In a systematic review including 10 studies, the ability of BMI and of waist circumference and derived metrics to identify children with CVD risk factors was comparable (5). One of these studies reviewed suggested inferiority of WHtR vs BMI to identify children aged 5–17 years with elevated systolic BP (9). On the other hand, in a cohort of 630 children aged 8–10 years, we found that WHtR was no better than BMI in identifying children with elevated systolic BP or with elevated diastolic BP (10).

Using data from a large representative cross- sectional population-based study in Switzerland (2), we assessed whether WHtR had a better ability than BMI to identify children aged 10–14 with elevated BP.

Methods

Study population and measurement methods

Weight, height, waist circumference and BP were measured in all sixth-grade schoolchildren of the canton de Vaud (Switzerland) in 2005/06 (2,11). Children were measured without shoes and in light garments in a quiet and tempered room by trained clinical officers. Weight and height were measured with precision electronic scales (at 0.1 kg) and fixed stadiometers (at 0.1 cm). Waist circumference was measured at mid-distance between the last floating rib and the iliac crest at the end of normal expiration with a standard tape measure (at 0.1 cm). BP was measured on the right arm. The mid-arm circumference was measured and the cuff width adapted accordingly. Three measurements of BP were taken at 1-min intervals after a rest of at least 3 min, in a seated position, using a clinically validated oscillometric device (Omron M6; Omron Healthcare Europe BV, Hoofddorp, The Netherlands) (12). The study was approved by the ethics research committee of the Faculty of Biology and Medicine, University of Lausanne. Consent was sought from directors of all schools. Signed consent of one of the parents and of the child were obtained.

Statistical analysis

BMI z-score was derived from the sex- and age- specific BMI criteria of the CDC and overweight and obesity were defined for a BMI ≥ 85th and ≥ 95th corresponding percentile, respectively (13). WHtR was considered high if above 0.5 (6,7). The average of the last two BP readings was used. Elevated BP was defined as systolic BP and/or diastolic BP equal to or above the US reference sex-, age- and height-specific 95th percentile (14).

The Spearman correlation was assessed between BMI z-score and WHtR. A logistic regression model was used to assess the association of high BMI z-score and high WHtR with elevated BP, after adjustment for sex and age. The receiver operating characteristic (ROC) curve analysis is a method for evaluating the performance of different measures to discriminate subjects with and without an outcome of interest (in our case, elevated BP) (10,15). The area under the curve (AUC) statistic represents the probability of correctly classifying children with and without elevated BP, and ranges from 0.5 (no discrimination; no better than chance) to 1.0 (perfect discrimination) (10,16). AUC were computed for BMI, BMI z-score, waist circumference, WHtR and BMI z-score together with WHtR and compared using the ROCCOMP command in Stata version 11.

Results

A total of 5207 children participated (76% response) [2621 boys, 2586 girls; mean (± SD) age, 12.3 ±  0.5 years; range: 10.1–14.9]. Some 75% of the participants had at least one Swiss national parent and 25% had no Swiss national parent. The prevalence of elevated BP was 11%, of overweight (including obesity) 14.3%, and of obesity 3.6%. Mean WHtR was 0.44 ± 0.05 (range: 0.29–0.77] and 11% had a high WHtR (> 0.5).

BMI z-score and WHtR were strongly associated, with a correlation coefficient of 0.76. Both indices were positively associated with BP. In a multivariate logistic model with adjustment for age and sex, the odds ratio of having elevated BP was 2.6 (95% CI 2.0–3.5; p < 0.001) for high BMI (overweight or obesity) and 1.5 (95% CI 1.1–2.0, p = 0.008) for high WHtR.

The sensitivity, specificity and positive predictive value of overweight for elevated BP were 31%, 88% and 26%, respectively. For high WHtR, these values were 24%, 91% and 26%, respectively. AUCs for elevated systolic or diastolic BP were relatively low for BMI z-score, WHtR and BMI z-score together with WHtR (Table I and Figure 1). These AUCs were not significantly different in pair wise comparisons. Similar AUCs were found for BMI alone (0.63; 95% CI 0.60–0.66) or waist circumference alone (0.61; 95% CI 0.58–0.64) to identify elevated BP.

Table I. Area under the receiver operating characteristic (AUC).

	Elevated BP		Elevated systolic BP		Elevated diastolic BP
	AUC	95% CI	AUC	95% CI	AUC	95% CI
BMI z-score	0.62	0.60–0.65	0.62	0.60–0.65	0.61	0.54–0.67
Waist-for-height	0.62	0.59–0.64	0.62	0.59–0.64	0.61	0.54–0.67
BMI z-score and waist- for-height	0.62	0.59–0.65	0.62	0.59–0.64	0.61	0.55–0.67

BP, blood pressure; CI, confidence interval.

Figure 1. Receiver operator curves (ROC) for body mass index (BMI) z-score alone or together with waist-to-height ratio (WHtR). BP, blood pressure; AUC, area under the curve.

Discussion

The ability of BMI z-score and WHtR to identify children aged 12 with elevated BP was weak. Adding WHtR did not confer additional discriminative power to BMI alone. These findings do not support the measurement of waist-for-height in addition to BMI to discriminate children with and without elevated BP.

A simple anthropometric measure that help identify individuals at increased risk of CVD factors, including hypertension, would potentially improve screening and early preventive interventions (17). WHtR has been advocated as a better screening tool for CVD risk factors, including hypertension, than BMI (6). The simplicity of a unique cut-off of 0.5 in adults and in children of all ages is also a strong argument for using WHtR (18). However, even in adults, a systematic review indicated that the gain in discriminative power was modest with no real practical clinical significance: for hypertension, in men, the AUC was 0.69 for WHtR and 0.65 for BMI; in women, the AUC was 0.73 and 0.69, respectively (6). Furthermore, in age- and sex-specific strata of US children aged 2–18 years, considerable residual correlation was found between height and WHtR (correlation between −0.29 and + 0.36), which means that dividing waist circumference by height might not be sufficient to have an anthropometric index of central adiposity independent of height (17). To render an index of central adiposity independent of height is not necessary if the goal of the index is to convey risk adequately in children (19). Still, our study indicates that WHtR was no better than BMI to convey the risk of elevated BP in children.

The high correlation between WHtR and BMI z-score explains that both measurements have a similar discriminative power (9) and that this power is not improved when using both indices together. In Canadian children aged 8–10, we have also shown that waist circumference, WHtR and BMI were highly correlated (r ≥ 0.93) and also had a relatively weak ability to identify children with elevated BP (10). In the Bogalusa study, in 2498 children aged 5–17 years old, the discriminative power for elevated BP was similar for BMI z-score and WHtR, and both indices were highly correlated (r = 0.88) (9). More broadly, the fact that a characteristic – in our case BMI or WHtR – has to be very strongly associated with a condition – elevated BP – to be a worthwhile screening test is not well recognized (20).

One limitation of our study is that we analyzed BP readings obtained at only one visit. Children with elevated BP were identified but we have previously shown that most of them did not have sustained elevated BP at following visits, i.e. did not have hypertension (2).

In conclusion, while BMI and WHtR were relatively strongly associated with elevated BP in children, our study and others (5,9,10) indicate that the association with BP may not be strong enough for both indices to be used as efficient tools to identify elevated BP in children (20). Our findings do not support the use of WHtR in addition to BMI to help identify children with elevated BP.

Disclosure

The authors declared no conflict of interest.