WAIST CIRCUMFERENCE CUT-OFF POINTS OR BODY MASS INDEX FOR PREDICTING FAT MASS IN COLLEGE STUDENTS

§ ABSTRACT: Obesity affects millions of people worldwide, becoming a public health problem with serious implications for human life quality. This is why low-cost prevention and diagnostic strategies that are accessible to the entire population are needed. Low-cost and easy-to-use strategies include the measurement of Body Mass Index (BMI) and measurement of waist circumference (WC). The objective was to determine whether the percentage of body fat in Colombian college students can be predicted through BMI or WC, establishing WC cutoff points. This was a descriptive cross-sectional study with a simple random sample. The study included 687 students aged 18-35 years old from both sexes. Anthropometric measures of weight, length, waist circumference at navel level and skin folds (bicipital, tricipital, subscapular and suprailiac) were obtained. The sensitivity and specificity of BMI and the waist circumference at navel level (WCN) were determined to predict the estimated percentage of body fat by measuring skin folds. The cutting points of the waist circumference for men and women were determined by drawing up a Receiver Operating Characteristic Curve (ROC). Greater sensitivity and specificity of WCN were found in comparison with BMI, to predict the percentage of body fat.


Introduction
Obesity is defined as the excess of body fat (KAVAK et al., 2014). The interest in its study and physiopathology has increased over the last few years, given the physical and psychosocial implications and its impact on people's health. According to data from the latest National Survey of Nutritional Situation in Colombia (ENSIN, 2010), one out of two adults is overweight and one out of six is obese (FONSECA-CENTENO, 2011). Obesity is linked to cardiovascular complications and the development of diabetes Mellitus type 2, among other diseases (ROMAIN et al., 2012). Furthermore, the abdominal fat gain is associated with increased risk in the mentioned circumstances. This increase can be determined by measuring the WC (WALL-MEDRANO et al., 2016).
The BMI and the WC measurement are the two most commonly used anthropometric tools for determining overweight and obesity in adults, due to their simplicity in implementation and low cost, favoring their use in daily clinical practice (RAMIREZ-VELEZ et al., 2017). While BMI is used to assess thinness and obesity, its usefulness is limited since it does not determine the fat mass and its distribution in the body (NUTTALL, 2015). On the other hand, an increase in WC reflects an increase in the amount of intra-abdominal fat including visceral adipose tissue. A correlation has been reported between the increase in visceral fat tissue and the risk of cardiovascular disease (GONZÁLEZ-MUNIESA et al., 2017) and the development of certain cancers (OTTAIANO et al., 2018). Waist circumference values have been established by the International Diabetes Federation as a diagnostic criterion for metabolic complications (HU et al., 2016;ALBERTI et al., 2005).
This study aims to determine whether the percentage of body fat in Colombian college students can be predicted through BMI or WC.

Subjects
This is a cross-sectional descriptive study, carried out in the city of Manizales (Colombia) from 1995 to 2013, in a population of male and female Colombian college students from the Universidad de Caldas, aged between 18 and 35 years. The population was recruited through announcements, published in leaflets and on the University's website. The study was carried out following the guidelines of the Helsinki Declaration. The project was approved by the Ethics Committee of the Health Faculty. Participants, after signing informed consent, were surveyed to obtain data of socio-demographic importance: age, gender, curriculum, identification document and city of origin. All participants were provided with comprehensive verbal explanations about the nature, objectives and experimental risks of the study. Pregnant women, not university students, people with walking impairments, a clinical or medical diagnosis of diseases such as hypothyroidism, hyperthyroidism, type 1 and 2 diabetes, malnutrition, drug or alcohol abuse history, chronic inflammatory conditions such as systemic lupus erythematosus, rheumatoid arthritis, and people who were taking medications that affected body composition, such as multivitamins or steroids were excluded. The information was obtained by self-report and the medical history of the participants.

Methods
Measurements were made at the University's nutritional assessment laboratory. The participants removed their shoes and wore only their underwear and a hospital gown. The waist circumference was measured at the navel level (WCN) using a measuring tape, with accuracy of ± 1 mm and without skin compression. The height measurement was performed with the subject standing, using a Heightronic 235 digital stadiometer from Seca Corp., with an accuracy of ±0.5 cm. The weight was established using the Icon-Detecto PP 2000 digital scale, with an accuracy of ± 0.1 kg. The BMI was calculated as the weight (kg) divided by height, in square meters (m2). Skin folds were measured at the bicipital, tricipital, subscapular and suprailiac levels with a Skyndex II ® digital plicometer, according to the method described by Lohman et al., (1992). The fat percentage was estimated according to the formula of Durnin and Womersley, which is included in the plicometer software (DURNIN and WOMERSLEY, 1974). Obesity was defined from a percentage of body fat >30% in women and >25% in men (YANG et al., 2006). For the analysis of the results, SPSS version 15, a license on behalf of the Universidad de Caldas, was used. The sensitivity and specificity of BMI and WCN were calculated for predicting body fat percentage using the ROC curves. These curves evaluate the accuracy of a diagnostic test and its potential to establish the presence or absence of a condition that interferes with the health of individuals (KAVAK et al., 2014).

Results
The characteristics of the subjects are presented in Table 1. The values of the area under the ROC curve for the cut-off points of sensitivity and specificity are presented in Table 2. Both BMI and WCN showed predictive ability to identify individuals with increased adiposity in both genders, compared to the percentage of body fat. The areas below the curve of both indicators for men and women are shown in Figures 1 and 2, respectively. Confidence intervals are present in tables 2 and 3. The area under the BMI curve was smaller than WCN in both sexes (p <0.05). Furthermore, the data obtained from the ROC curves allowed determining cut-off points of WC for this population (Table 4).

Discussion
The high global prevalence of obesity and the associated metabolic syndrome has led the International Diabetes Federation (IDF) to develop guidelines for its diagnosis and treatment. Among the recommendations, the measurement of waist circumference has been proposed as a simple, low-cost parameter for estimating abdominal obesity.
Visceral fat surrounds the abdominal organs that have endocrine activity. These organs release fatty acids into the bloodstream. Those acids promote the production of very-low-density lipoproteins (VLDL) and low-density cholesterol (LDL-c). These cholesterol levels promote the formation of atherosclerotic plaques, increasing cardiovascular risk. Fatty tissue also promotes the formation of pro-inflammatory cytokines 1 and 6, as well as the release of tumour necrosis factor Alpha. Visceral fat also stimulates the formation of reactive C protein and is an activating factor of plasminogen type 1 (PAI 1) in the liver, contributing to the generation of inflammatory and prothrombotic processes (STANLEY and GRINSPOON, 2015). Given these phenomena, measurement of the waist circumference can determine the risk of cardiovascular disease and metabolic syndrome .
The literature reports different anatomical points for the measurement of waist circumference. The most referenced sites in the literature are the navel (28%) and the midpoint between the last rib and the iliac crest (29%). The remaining percentage is divided among the other anatomical sites (KLEIN et al., 2007). In a systematic review of 120 studies, eight protocols were reported for the measurement of waist circumference (ROSS et al., 2008). Studies comparing different protocols have reported that absolute waist circumference measurement values vary, depending on the anatomical site being evaluated (SHARIATPANAHI and VAHDAT, 2016). These discrepancies cause the cutoff points obtained to change between protocols. Thus, it is necessary to call for a global consensus to establish an anatomical point and a defined protocol. In this way, the real differences that exist by gender and between different populations can be demonstrated (RAMIREZ-VELEZ et al., 2016).
In Latin America, there are few studies on this subject and there are no defined cut-off points as a reference for the measurement of the waist circumference. Despite this, the IDF recommends using Asian population cuts: 90 cm for men and 80 cm for women (ALBERTI et al., 2005;HERNÁNDEZ et al., 2013). Anthropometric similarities seem to exist between the Asian population and the Latin American community (BUENDÍA et al., 2016). However, South America has particular characteristics because its population is the product of mixing between Europeans, Amerindians, and Africans. Besides, the cut-off points of both BMI and WC are influenced by gender and age (RAO et al., 2015). These factors create the need to establish our cut-off points for a population rich in ethnic and cultural diversity (RIVERA et al., 2014). Buendia et al. (2016) carried out a study with 501 Colombian subjects, taking as anatomical points the one used for the Asian population. The values found by them (91 cm in men and 89 cm in women) are close to those reported for Asian men but not for women, with which there would be a difference of 9 cm.
Another study in Colombia, with the participation of 83.220 subjects, used the same anatomical site of Buendia. Cutting points of 96.6 cm were found for men and 91.0 cm for women (RAMIREZ-VELEZ et al., 2016). The difference could be explained by the diverse sample sizes and differences within the same Colombian population, where there is ethnic diversity by region. A third study, with a Colombian population of 800 subjects, used the HOMA (Homeostasis Model Assessment of Insulin Resistance) as the reference method. The anatomical site used was the narrowest area between the rib cage border and the iliac crest. The authors reported different values: 92 cm for men and 84 cm for women. In this case, the difference in the points could be partly explained by the reference method used and the characteristics of the sample (ALEMAN-MATEO et al., 2014).
In the present study, the cut-off points found (both for men and women) are lower than the previous ones. Discrepancies with the studies done in Colombia may be due to differences in the anatomical site and the reference method used (BUENDÍA et al., 2016). It can also be explained by the fact that the sample was obtained between 1995 and 2013, a period during which epidemiological changes related to overweight and obesity have occurred (RAMIREZ-VELEZ et al., 2017).
It is estimated that over the last few decades there has been an increase in the average BMI worldwide, of 0.63 kg/m2 for men and 0.59 kg/m2 for women. However, it should be stressed that over time they may weigh an additional 1.5 kg (NCD-RISC, 2016). This same transition in the prevalence of obesity has occurred in both developed and developing countries (GARCIA, 2017). A similar phenomenon has occurred in Colombia. The 2010 National Health and Nutrition Survey (ENSIN) showed that the Colombian population had increased its levels of overweight and obesity by 5.6% about the 2005 survey (FONSECA-CENTENO, 2010). Therefore, it is likely that the people evaluated in 1995 were thinner than those of the three Colombian studies already mentioned, carried out in 2013, 2015 and 2016 respectively.
The results of this study are similar to those of others that show a positive correlation between BMI and WC in both sexes (SMITH and HASLAM, 2007); although they differ from another study showing positive correlation for women but not for men (FREEDMAN and FORD, 2015). BMI and WC have good sensitivity and specificity to predict body fat percentage. However, in our study, the highest level of prediction was obtained with WCN. This result is consistent with studies showing that WC is a major predictor for indirectly determining intra-abdominal fat (CABRAL et al., 2015). These data are in line with those of other studies showing limitations of BMI, especially as BMI does not distinguish between lean body mass and fat body mass (NUTTALL, 2015;SILVA et al., 2012). Thus, it is necessary to add the estimation of total fat percentage and body distribution using WC measurement to obtain a more accurate evaluation of the body composition of each subject.
Our study has some limitations. The fact that the sample is for convenience prevents extrapolating the results to the rest of the population. It should also be noted that other variables that could influence body composition, such as levels of physical activity, ethnicity or socioeconomic status, were not included. These factors modulate growth and levels of adiposity (19).
The present study contributes to a better knowledge of the diverse Colombian population, allowing us to understand the differences between the cut-off points of other regions of the country and even other nations. In Colombia, obesity can reach prevalence levels of 62% (RUIZ et al., 2012). It is, therefore, necessary to define new cut-off points for the waist circumference due to the epidemiological changes that occur in the populations. This should be done not only because of its effectiveness and low cost but also because it can contribute to a better approach to this public health problem by allowing to carry out early screening using affordable anthropometric instruments. Finally, suitable cut-off points are required for the Colombian and Latin American population, since there are still divergences in different studies (LIMPAWATTANA et al., 2014).

Conclusions
WCN measurement has greater sensitivity and specificity than BMI to predict body fat percentage. Cutting points for the WCN were found to be 70 cm for men and 64 cm for women. These points can be compared with other studies in Colombia and serve as a reference for college students with similar characteristics. The practical value of our study at present is to reinforce the idea that both researchers, as well as, clinicians should measure and report not only BMI but also the waist circumference and this should preferably be measured at the same parameterized level. Concerning future research, it ought to be worth working with samples that represent wider populations either at regional or even at the national level.