Overall, excessive weight (overweight plus obesity) was 27.7% using the WHO-2007 standard and 24.7% using the IOTF reference; whereas the prevalence of overfat by FMI using the cut-offs proposed by Alvero-Cruz et al. 
was 19.8%; and 7.7% of adolescents had abdominal obesity. Using the IOTF cut-offs, while almost all obese adolescents were overfat (95.1%) and most of them had abdominal obesity (78.7%); about 36% of overweight adolescents were misclassified on the basis of the BMI alone –being higher among girls (56.7%) than boys (17%)-. Using the WHO cut-offs, 85.2% of obese adolescents were overfat and half of them had abdominal obesity; whereas 58.2% of overweight adolescents were normal-fat –being also higher among girls (73.4%) than boys (43.9%)-. Among the normal-weight group, using both IOTF and WHO-2007 cut-offs about 1.4–2.8% of adolescents were overfat, which was higher among boys (4–6%) than girls (<1%).
The present results agree with previous studies 
which have pointed out that the IOTF reference and the WHO standard yield different results in terms of prevalence of overweight and obesity. A previous study conducted in 2004 among Canadian children and youth (n
8661, 2- to 17-year-olds) 
which compared prevalence estimates of excess weight according to three sets of BMI reference cut-offs: WHO, IOTF and the US Centers for Disease Control (CDC), found that prevalence estimate for the combined overweight/obese category was higher (35%) when based on the WHO cut-offs compared with the IOTF (26%) or CDC (28%) cut-offs. Estimates of the prevalence of obesity were similar based on WHO and CDC cut-offs (13%), but lower when based on IOTF cut-offs (8%). In the same line, a study conducted among Portuguese children and adolescents aged 10- to 18-year-olds (n
22048), the prevalence of overweight and obesity based in the IOTF cut-offs was 17.0 and 4.6% in girls, and 17.7 and 5.8% in boys, respectively; whereas WHO cut-offs resulted in overweight and obesity prevalence scores of 23.1 and 9.6% in girls, and 20.4 and 10.3% in boys, respectively 
. Monasta et al. 
have indicated that these different results are due to the different approaches used to define cut-offs and to the different criteria used to select the samples. Monasta et al. 
also suggested that at the moment, IOTF reference and cut-offs could be preferable to identify overweight and obesity both at individual and population level because they are at least based on a crude association with ill health later in life, namely the definition of overweight and obesity at 18 years. However, it is important to note that in the present study both international references also provide high different results than the FMI using the cut-offs proposed by Alvero-Cruz et al. 
. Of course BMI and FMI are different terms and although BMI performs moderately well as a proxy for these indicators, particularly at the upper end of the distribution curve 
, an important percentage of subjects classified as overweight or obese did not have really excess adiposity 
, percentage which may depend on the reference used.
The ideal definition of overweight and obesity would be based either on a close correlation with indicators of future cardiovascular and metabolic disease or on their ability to predict adverse future health outcomes 
. Therefore, despite that BMI have been suggested as a good proxy for the screening of excess body fat in adolescents when considering a whole population, as in clinical settings others criteria may be also useful in epidemiological studies in which anthropometric measurements are used to screen overweight and obesity prevalence. Moreover, a single definition useful in both epidemiological and clinical settings should be achieved because epidemiological studies should determine the magnitude of the overweight and obesity problem in a population and also to stress the need for lifestyle changes while not exaggerating risks of future obesity and cardiovascular disease.
Because it is difficult to exclude BMI from the normal-weight and obesity definition despite that provides no information regarding the composition of the weight, or its distribution; FMI and WHtR could combine BMI for a better screening and surveillance. The FMI is a useful measure to evaluate body composition parameters by effectively eliminating differences in body fat associated with height 
. Alvero-Cruz et al. 
derived cut-off points for FMI from a sample of Spanish adolescents (150 subjects, 75 boys and 75 girls) showing that the FMI had higher accuracy for overweight screening than BMI. Their results pointed out in boys that predictive positive value (meaning the diagnosis of excessively fat adolescent as overweight) were 78.1% for BMI and 89.2% for FMI; and predictive negatives value (meaning the diagnosis of lean adolescent as non-overweight) were 81.4% for BMI and 100% for FMI in them. In girls, predictive positive value for BMI were 34.8%, predictive positive value for FMI were 81.4%; and predictive negative value for BMI and FMI were 98.1% and 100%, respectively. Combining BMI and FMI, our results suggest that IOTF cut-offs have high specificity for obesity –more than WHO cut-offs-, in our adolescent population. However, there was much less evidence on the optimal definition of being normal-weight or overweight. Therefore, the present results supported that a cut-off BMI≥30 kg/m2
for age and sex 
may be a good proxy for obesity in boys and girls, and also a cut-off BMI<25 kg/m2
for age and sex for normal-weight in girls; whereas FMI may reduce misclassification among normal-weight boys and overweight adolescents. Thus, adolescents may be classified into five main groups as follows: normal-weight normal-fat, normal-weight overfat, overweight normal-fat, overweight overfat and obese.
It is also well established that central or visceral obesity is a major factor for the clustering of cardiovascular risk factors which defines the metabolic syndrome 
. The determination of adolescents with abdominal obesity in both overweight and obesity status could useful to identify adolescents who being overweight or obese have higher probability for cardiovascular risk factors. Thus, overweight overfat and obesity adolescents could be classified into type-I and type-II according to the absence or presence of abdominal obesity, respectively. Abdominal obesity should be assessed by WHtR which has been proposed because of its ability to predict cardiovascular risk factors 
and to estimate abdominal fat distribution 
, particularly in individuals who may not be classified as overweight or obese by BMI 
It is important to note that direct measurement of adiposity with sophisticated techniques is considered to be superior to indirect measures 
. However, in many circumstances it is more desirable to utilize widely available and simple techniques such as anthropometry. Therefore, it should be recommended that not only BMI but also FMI and WHtR be used whenever possible in both clinical and epidemiological settings. Both, anthropometric measurements are quick, cheap and simple which require only limited training and standardized assessment to obtain reliable data 
. Moreover, FMI and WHtR are normalized for body size making comparisons between individuals or populations, or within individuals or populations over time, to be meaningful 
The contribution of this research may lead to better methods for measuring normal-weight and overweight to support this area of public health research, at least for a more accurate classification of Spanish adolescents because further research will be needed to evaluate the FMI cut-offs proposed by Alvero-Cruz et al. 
to be generalized for international use. On the basis of our results, adolescents may be classified not only by body weight (BMI) but also adiposity (FMI) and fat distribution (WHtR). We proposed a new classification for adolescents, the Adiposity & Fat Distribution classification (AFAD-A) which classifies adolescents into the following groups: (1) normal-weight normal-fat; (2) normal-weight overfat; (3) overweight normal-fat; (4) overweight overfat (type-I and type-II, depending on the presence or absence of abdominal obesity, respectively), and (5) obesity (type-I and type-II). To facilitate the work of clinicians and epidemiologists, a questionnaire summarized as the AFAD-A classification is proposed (), despite that further research will be needed to evaluate its utility.
A proposal questionnaire to classify adolescents according to their body weight, adiposity and fat distribution: the AFAD-A questionnaire classification.
There are differences between IOTF and WHO-2007 international references and there is a misclassification when adiposity is considered. Surveillance, prevention and treatment of childhood and adolescent obesity require methods of defining obesity that are simple enough to be practical in most clinical and public health settings, but are also valid 
. However, identification of adolescents in normal-weight and overweight with excess body fat is also important not only because they have some increased risk of adiposity-related comorbid conditions 
, but also psychosocial complications derived from body fatness 
. Therefore, achieving a reliable and accurate estimation of body fatness and fat distribution is essential in both clinical and epidemiological settings. Our results support that it should be recommended that not only BMI but also FMI and WHtR be used whenever possible in both clinical and epidemiological settings.
Despite that further research will be needed to evaluate the utility of the AFAD-A questionnaire and classification, it could be the starting point towards an improvement in the traditional definition of normal-weight, overweight and obesity which may be a useful tool to surveillance adolescent’s overweight on clinical and epidemiologic settings.
Limitations of the Study
The cut-offs point considered for FMI were proposed by Alvero-Cruz et al. 
and derived from 150 subjects (75 boys and 75 girls). However, as it has been indicated by Alvero-Cruz et al. 
the average values of basic anthropometric variables (weight, height and BMI) of other 450 subjects were not significant different from those of the sample assessed, indicating that these cut-offs should be useful for overweight diagnostic in Spanish adolescents.
Certainly, it is important to note that to calculate the FMI from anthropometric measures there is an intermediate step consisting of applying an equation that allows determining the percentage of body fat, and the value depends on the applied equation that increase error misclassification. Rodríguez et al. 
compared the most commonly used equations to predict body fatness from skinfold thickness with dual-energy X-ray absorptiometry (DEXA) and found that most equations did not demonstrated good agreement compared with DEXA. However, they proposed that Slaughter et al. 
equations may be used in adolescents from both sexes to predict BF when a relative index of fatness is required in field or clinical studies. Nevertheless, the present study did not take into account pubertal development despite that chronological age may vary dramatically during this phase. Adolescents have been classified according to their pubertal stage, boys were divided into two groups: pubertal (12 to 14 y-o) and post-pubertal (15 to 17 y-o) 
Despite there is widely accepted that WHO and IOTF definitions have several limitations to define overweight and obesity and also yield different results in terms of prevalence of overweight and obesity on the same dataset, there are insufficient data to substitute these definitions for other anthropometric measurements. However, the combination of BMI and subcutaneous fat is intended to maximize specificity in identifying those adolescents who are normal-weight normal-fat or overfat, and overweight normal-fat or overfat. None the less, recommendation of AFAD-A for adolescent body composition classification must be considered as provisional because of inadequate evidence of the %BF and FMI derived from anthropometric measurements are better measurements than that from WHO or IOTF definition. Other techniques for body composition assessment such as densitometry, dual-energy X-ray absorptiometry, and magnetic resonance imaging provide more accurate information on fat and lean masses; however, they are expensive and impractical for use in routine clinics and epidemiological studies 
. Bioelectrical impedance analysis (BIA), on the other hand, is relatively cheap and easy to use 
. Direct measures should be used as a gold standard to validate indirect (anthropometric) measures of body fatness. Full studies are required before the recommendations for AFAD-A classification can be considered more than provisional.