texto en 
Inglés (pdf)
Articulo en XML
Referencias del artículo
Traducción automática
Permalink
INTRODUCTION
Obesity is a chronic condition characterized by the abnormal or excessive accumulation of body fat, responsible for compromising the health of individuals1-3. Bariatric surgery (BS) is considered an important strategy for the treatment of severe refractory obesity4. However, during the first six months after BS, the period of the rapid weight loss phase, individuals undergoing BS may present significant changes in body composition5,6. In this context, from a clinical and public health perspective, the use of simple, low-cost methods are of great relevance for the assessment of the percentage of body fat (%BF) of these individuals7.
The bioelectrical impedance (BIA) is a method that estimates the %BF in a non-invasive, safe and fast way, which consists in the application of a low frequency electric current by the body of the evaluated, and is based on the fact that the body tissues offer different oppositions (impedance) to the passage of electric current8-10. Studies demonstrate the accuracy of BIA as a method for determining %BF when compared to methods considered gold standard, such as hydrostatic weighing and dual energy x-ray emission densitometry (DXA)11,12. However, BIA has the disadvantage of depending on control factors such as diet, hydration level, menstrual cycle, ambient temperature, among others, which allow measurement errors13. In addition, the cost per assessment is relatively higher than those based on anthropometric parameters, which would make its use in larger groups or even in clinical practice unfeasible.
As an alternative tool, %BF predictive equations based on anthropometry are easy to use and low cost. Among these, the equations developed by Deurenberg et al.14, Lean et al.15 and Gómez-Ambrosi et al.16 use the variables height, weight, BMI, age and sex, simple parameters to be evaluated. The equation developed by Woolcott & Bergman17 uses only the variables gender, waist circumference (WC) and height.
The equations proposed by Woolcott & Bergman17, Gómez-Ambrosi et al.16and Deurenberg et al.14 were developed from a significant sample of individuals (12,581, 6,123 and 1,229 participants, respectively). Although most of the evaluated participants were eutrophic, the samples were also composed of overweight individuals, who showed a wide variation in %BF14-17. In addition, all formulas were compared with methods considered reference in the analysis of body composition, such as hydrostatic weighing15,16, DXA17 and plethysmography16. However, studies evaluating the relationship between the %BF estimated by these formulas and BIA during BS-induced body fat loss are less known.
Based on this information, the present study aimed to evaluate the relationship between predictive equations based on anthropometric parameters and BIA to estimate body fat in individuals undergoing BS.
MATERIAL AND METHODS
Sample and study design
A longitudinal and prospective study, carried out with adult individuals of both sexes, enrolled in the Bariatric and Metabolic Surgery Program of the Cassiano Antônio Moraes University Hospital (HUCAM), Espírito Santo, Brazil. The sample was selected for convenience, according to the criteria for performing Roux-en-Y Gastric Bypass: age between 18-60 years, BMI > 40kg/m or >35Kg/m with associated comorbidities. Pregnant women, individuals with pacemakers and those with metallic structures and/or silicone prostheses were excluded18. The study was approved by the Research Ethics Committee of the Hospital (CAAE nº 59075722.7.0000.5071), and the research participants consented to participate by signing the Free and Informed Consent Form. All procedures were performed in accordance with the World Medical Association’s code of ethics (Declaration of Helsinki).
Participants were evaluated at three times: approximately 1 month before (T0), 2nd (T1) and 6th months (T2) after surgery. Anthropometric assessment and BIA were performed during the program’s clinical follow-up consultations.
Anthropometric parameters
Body weight was measured on an anthropometric scale, with a capacity of 300 kg and precision of 0.05 kg; height was measured with the aid of a wall stadiometer, with a graduation of 0.1 cm. BMI was calculated using the formula: body weight (kg)/height2 (m). Waist circumference (WC) was measured in centimeters, over the umbilical scar.
Electrical bioimpedance
The assessment of body fat by BIA was performed in tetrapolar equipment, brand Biodynamics®, model 450, following the recommendations of the European Society of Clinical Nutrition and Metabolism18. Fat-free mass was calculated using the formula for people with obesity, proposed by Segal et al.9, and fat mass was calculated as the difference in total body weight, and expressed as a percentage.
Predictive equations
The following predictive equations were used: Woolcott & Bergman17, Deurenberg et al.14, Lean et al.15 and Gómez – Ambrosi et al.16. The equations for calculating the %BF are shown in table 1.
Table 1 : Predictive equations of body fat percentage used in the study
| Reference | Sex | Equation |
|---|---|---|
| Woolcott & Bergman (2018) | M | %BF = 64-(20 x height/WC) |
| F | %BF= 76-(20 x height /WC) | |
| Deurenberg et al. (1991) | M | %BF = 1.2 x BMI + 0.23 x (age) – (10.8 x 1) -5.4 |
| F | %BF = 1.2 x BMI + 0.23 (age) - (10,8 x 0) - 5.4 | |
| Lean et al. (1996) | M | %BF = (1.33 x BMI) + (0.236 x age) -20.2 |
| F | %BF = (1.21 x BMI) + (0.262 x age) - 6.7 | |
| Goméz-Ambrosi et al. (2012) | M | %BF = -44.988 + (0.503 x age) + (10.689 x 0) + (3.172 x BMI) - (0.026 x BMI2) + (0.181 x BMI.0) - (0.02 x BMI x age) - (0.005 x BMI2 x 0) + (0.00021 x BMI2 x age) |
| F | %BF = -44.988 + (0.503 x age) + (10.689 x 1) + (3.172 x BMI) - (0.026 x BMI2) + (0.181 x BMI x 1) - (0.02 x BMI x age) - (0.005 x BMI2 x 1) + (0.00021 x BMI2 x age) |
F: female; M: male; WC: waist circumference; BMI: body mass index; BF: body fat.
Statistical analysis
Data were analyzed by the Shapiro Wilk normality test, and later by the one-way ANOVA for repeated measures with Holm-Sidak post hoc or Friedman test with Tukey post hoc, according to the data distribution. The correlations between the data obtained by the BIA and the equations at each moment were evaluated using the Pearson or Spearman correlation, according to the data distribution, and classified as: weak (0.30 to 0.50); moderate (0.50 to 0.70); strong (0.70 to 0.90); very strong (>0.90)19. The Statistical Package for the Social Sciences - SPSS, version 22.0 software was used. The significance level adopted was 5% (p < 0.05).
RESULTS
Twenty-five individuals who attended the three scheduled assessments participated in the study. A predominance of females (75%) was observed. Participants were, on average, 41.2 ± 7.8 years old and 162.0 ± 8.7 cm tall. The evaluations took place at approximately 24.0 ± 20.5 days before (T0), 72.0 ± 19.5 (T1) and 189.0 ± 12.2 (T2) days after the BS. The values of body weight, BMI, WC, and % BF estimated by BIA and by the equations of Woolcott & Bergman17 and Gómez-Ambrosi et al.16showed normal distribution, while the values of % BF estimated by the equation of Deurenberg et al.14 and Lean et al.15 showed non-normal distribution.
Table 2 shows the changes in anthropometric parameters over time. Weight, BMI and WC significantly reduced at all times (p<0.05). Table 3 shows the % BC estimated by the BIA and the predictive equations used. The % BF estimated by BIA and by the equation by Goméz-Ambrosi et al.16 significantly reduced at T1 and remained at T2 (p<0.05). The % BF estimated by the Woolcott & Bergman17 equation was significantly reduced only in T2 (p<0.05), while the %BF estimated by the equations of Deurenberg et al.14and Lean et al.15 differed significantly at all times (p< 0.05).
Table 2 : Anthropometric parameters evaluated at different times
| Parameter | T0 | T1 | T2 | p value |
|---|---|---|---|---|
| Weight (kg) | 110,2 | 93,6 | 80,7 | <0,001 |
| (98,5 - 126,7)a | (83,0 - 108,1)b | (71,9 - 92,5)c | ||
| BMI (kg/m2) | 43,3 | 35,2 | 31,6 | <0,001 |
| (39,2 - 48,0)a | (32,5 - 41,2)b | (28,5 - 34,9)c | ||
| WC (cm) | 119 | 106,5 | 96,9 | <0,001 |
| (111,0 - 128,1)a | (103,7 - 120,3)b | (86,7 - 106,4)c |
BMI: body mass index; WC: waist circumference. N = 25. T0: 24.0 ± 20.5 days before surgery; T1: 72.0 ± 19.5 days after surgery; T2: 189.0 ± 12.2 days after surgery. Values expressed as median (interquartile range) and analyzed using Friedman’s test and Tukey’s post hoc test.
Table 3 : Percentage of body fat estimated by BIA and predictive equations at different times
| Method | T0 | T1 | T2 | P value |
|---|---|---|---|---|
| BIA (%)* | 47,9 | 43,9 | 39,6 | <0,001 |
| (44,6 - 49,5)a | (39,8 - 46,0)b | (35,9 - 43,3)b | ||
| Woolcott & Bergman (2018) (%)* | 47,4 | 44,8 | 39,5 | <0,001 |
| (44,0 - 50,4)a | (40,6 - 48,1)a | (36,2 - 44,7)b | ||
| Deurenberg et al. (1991) (%)** | 54,32 ± 7,35a | 45,97 ± 7,24b | 40,4 ± 6,94c | <0,001 |
| Lean et al. (1996) (%)** | 55,11 ± 7,48a | 46,51 ± 7,39b | 40,8 ± 7,1c | <0,001 |
| Gómez-Ambrosi et al. (2012) (%)* | 52,5 | 46,3 | 42,2 | <0,001 |
| (49,7 - 55,0)a | (43,6 - 50,6)b | (37,9 - 46,2)b |
BIA: bioelectrical impedance. N=25. T0: 24.0 ± 20.5 days before surgery; T1: 72.0 ± 19.5 days after surgery; T2: 189.0 ± 12.2 days after surgery. *Values expressed as median (interquartile range) and analyzed using Friedman’s test and Tukey’s post hoc test. **Values expressed as mean ± Standard deviation and analyzed by the ANOVA test for repeated measures and Holm-Sidak post hoc. Different letters on the same line represent significantly different values between moments (p<0.05).
Correlations between BIA estimated %BF values and predictive equations are shown in Figures 1–3. Before surgery (figure 1), the %BF estimated by the equation of Gómez–Ambrosi et al.16 showed the strongest correlation with BIA (r=0.9198; p<0.001), followed by the equations of Woolcott & Bergman17(r=0.8215; p<0.001), Deurenberg et al.14 (r=0.7792; p <0.001) and Lean et al.15 (r=0.6949; p<0.001), a result that remained for approximately two months after BS (figure 2). The equation by Gómez–Ambrosi et al.16 continued to show the strongest correlation with BIA values about six months after BS (r = 0.9294; p<0.001) (figure 3), followed by Deurenberg’s equations et al.14, Lean et al.15 and Woolcott & Bergman17. At all times, the Gómez–Ambrosi et al.16 equation showed a very strong correlation with BIA.
Figure 1 : Correlation between body fat percentage (%BF) estimated by bioelectrical impedance (BIA) and predictive formulas (A) Gómez-Ambrosi et al., (B) Woolcott & Bergman, (C) Deurenberg et al. and (D) Lean et al. at T0, 24.0 ± 20.5 days before surgery
Figure 2 : Correlation between body fat percentage (%BF) estimated by bioelectrical impedance (BIA) and predictive formulas (A) Gómez-Ambrosi et al., (B) Woolcott & Bergman, (C) Deurenberg et al. and (D) Lean et al. at T1, 72.0 ± 19.5 after surgery
DISCUSSION
In the present study, with the exception of the equation by Lean et al.15 before BS, all formulas showed a strong correlation with BIA, and the values obtained by the equation by Gómez-Ambrosi et al.16showed the highest correlation strength in the three moments evaluated.
The predictive equations used in the present study use simple and easy-to-collect anthropometric data, being useful for use in clinical practice, without the need for sophisticated equipment, and in cases in which skinfold measurements are more difficult, such as in individuals with severe obesity20. These equations were developed and validated through samples with a large number of individuals with a wide age range, and compared with reference methods such as plethysmography16, hydrostatic weighing14,15and DXA17.
The high correlation values observed between the equation by Gómez-Ambrosi et al.16 and BIA, before and after BS, corroborate other studies carried out using different reference methods, such as DXA and plethysmography12,13. The equation by Gómez-Ambrosi et al.16, also described as CUN-BAE (Clinic Universidad de Navarra-Body Adiposity Estimator), was developed with the aim of increasing accuracy in estimating body fat. The formula was derived from data from 6,123 participants with a mean BMI of 31.6 kg/m2, a broad age range (18-80 years) and adiposity (2.1 – 69.6%). Furthermore, Gómez-Ambrosi et al.16 included individuals with a high BMI (maximum of 72.8 kg/m2), while Deurenberg et al.14 and Lean et al.15 used individuals with a maximum BMI of 40.9 kg/m2 and 41.2kg/m2, respectively. In addition, the sample used by Gómez-Ambrosi et al.16 was composed mostly of women (68%), similarly to the present study. These factors may, in part, explain the greater strengths of correlation found between the formula by Gómez-Ambrosi et al.16 and the BIA.
The equation proposed by Woolcott & Bergman17 was developed from the evaluation of 6,320 men and 6,261 women, with an estimated average percentage of total body fat of 28 and 40%, respectively. In the present study, despite the %BF estimated by this formula having presented greater numerical similarity with the %BF estimated by the BIA, the strength of the correlation of this formula with the BIA values was greater after the BS, moments in which the participants presented %BF smaller and more similar to the profile of the 12,581 individuals evaluated by the aforementioned authors, suggesting that, as the %BF decreases, there is a tendency towards an increase in the precision of the formula, also known as Relative Fat Mas. Guzmán-León et al.21 also found a strong correlation between the %BF estimated by the Relative Fat Mass and the DXA in 61 eutrophic young Mexicans. However, excess abdominal fat can make it difficult to measure WC in morbidly obese individuals, which could interfere with the estimation of body composition.
Similarly, there was an increase in the strength of the correlation between the Deurenberg et al.14 equation and BIA after BS, especially after six months, when the participants had lower BMI values. In a previous study, Martins et al.13 observed a strong correlation between the % BF values estimated by the Deurenberg et al.14 equation with data obtained by DXA in individuals with grade I obesity. The lowest correlation values presented by the Deurenberg et al.14 with BIA at times when individuals had a higher %BF may also be partly related to the wide variation in the age range of the population in which this equation was validated, which was between 7 and 83 years of age14. However, the results of the present study corroborate the results of the study by Lopes et al.11, who evaluated % BF by BIA in 27 young people with obesity, and observed a discrepancy in the values estimated by the equation by Deurenberg et al.14.
Regarding the equation by Lean et al.15, greater differences were observed in relation to BIA to estimate the %BF in individuals with a high degree of adiposity (before BS), suggesting that the equation of Lean et al.15 can be more indicated for eutrophic or overweight individuals, as observed by Silveira et al.22 who compared different formulas with DXA, and observed that the best agreement in overweight men was with the equation proposed by Lean et al.15.
The reduction in anthropometric measurements over time reinforces the role of BS in the treatment of morbid obesity, as observed in other studies23-25. As a consequence of changes in body composition, the reduction in adiposity promoted by BS results in significant changes in the inflammatory state associated with obesity, with an improvement in the general health status of individuals26,27.
The use of equations based on simple, easy-to-measure anthropometric parameters can be advantageous in the assessment of larger population groups, or even in follow-up consultations for bariatric patients, since they do not require sophisticated equipment and do not depend on skinfold measurements, which may be impractical in morbidly obese individuals. However, some limitations should be mentioned: the evaluations were carried out in the follow-up consultations, which made it difficult for them to occur at the planned times. In addition, participants’ dropouts throughout the study contributed, in part, to the small sample size.
CONCLUSION
In view of the results obtained, it is concluded that all the equations used showed a strong correlation with the BIA, with the exception of Lean et al.15 before the BS. The equation by Gómez-Ambrosi et al.16 presented the highest correlation strengths with BIA at the three evaluated moments, being, in the present study, the best alternative to the use of BIA to assess changes in %BF during the follow-up of patients undergoing BS for the treatment of severe obesity.
