Body fat levels in children and adolescents: Effects on the prevalence of obesity

Summary

Background & aims

There is little information regarding the effect of different definitions of obesity on nutritional epidemiology. The aim was thus to assess: (a) the values of percentage of body fat (%BF) by gender and age; (b) the prevalence of obesity according to different %BF cut-offs; and (c) the sensitivity and specificity of BMI according to different %BF cut-offs used to define obesity.

Methods

Cross-sectional study on 2494 boys and 2519 girls aged 10–18 years from the Lisbon area. %BF was measured using a hand-held device. In a sub sample of 211 boys and 724 girls %BF was assessed using skin folds.

Results

%BF levels were higher in girls and decreased with age in both genders. Prevalence of obesity varied considerably according to the %BF cut-off used: in boys, it ranged from 4.7% (age-specific 95th percentile) to 26.5% (fixed 25% cut-off), whereas by BMI it was 5.3%. In girls, prevalence of obesity ranged from 0.4% (age-specific BMI-derived %BF values) to 25.4% (fixed 30% cut-off), whereas by BMI it was 4.7%. The specificity of BMI criteria was over 95% irrespective of the %BF cut-off used; conversely, most sensitivities were below 40%. Sensitivities over 50% were obtained for the age-specific BMI-derived %BF values in boys and the age-specific 95th %BF percentile in both genders. Using %BF derived from the skin fold measurements leads to similar results.

Conclusions

Prevalence of obesity varies considerably according to the %BF cut-off used. BMI cut-offs have a low sensitivity but a high specificity. Age- and gender-specific cut-offs for %BF should be used to define pediatric obesity.

Introduction

Child and adolescent obesity are increasing dramatically worldwide. Already in 1765, obesity was defined as excessive body fat,¹ a definition repeatedly confirmed.² Nonetheless, there is no definite agreement regarding the cut-off values of body fat to define obesity in the pediatric setting. Most publications indicate a single cut-off for the percentage of body fat to define obesity (25% in boys and 30% in girls),3, 4, 5 although other values such as the gender and age-specific 85th⁶ or 95th7, 8 percentile or estimated body fat associated with International Obesity Task Force (IOTF) cut-offs⁹ or even population-specific values10, 11 have been proposed, but it is unclear whereas those definitions actually lead to similar results regarding the prevalence of obesity in the pediatric population.

Since body fat is relatively difficult to measure in most clinical and epidemiological settings, surrogate measurements such as body mass index (BMI) have been used to define obesity, and several age- and gender-specific cut-offs have been proposed.¹² Compared to reference body fat measurements, those BMI cut-offs have shown a high specificity but a somewhat lower sensitivity in adults.3, 13. Recently, several lightweight, hand-held bioelectrical impedance devices have been proposed for the rapid estimation of body fat. Those apparatuses have been shown to be reliable by comparison to dual energy X-ray absorption (DXA).14, 15 The use of inexpensive devices to estimate body fat might also provide a better diagnostic tool for obesity in the general population than the measurement of BMI. Still, to our knowledge, no data regarding the gender and age-specific distribution of body fat measured by hand-held devices has been published.

Thus, the aims of the present study were: (a) to assess the values of body fat, as evaluated by hand-held bioelectrical impedance, in a large sample of healthy Portuguese children and adolescents; (b) to evaluate the prevalence of obesity according to different body fat cut-offs proposed in the literature; and (c) to evaluate the sensitivity and specificity of the IOTF BMI criteria according to different body fat cut-offs used to define obesity.