Anthropometric Estimations of Percent Body Fat in NCAA Division I Female Athletes: A 4-Compartment Model Validation : The Journal of Strength & Conditioning Research

Secondary Logo

Journal Logo

Original Research

Anthropometric Estimations of Percent Body Fat in NCAA Division I Female Athletes: A 4-Compartment Model Validation

Moon, Jordan R1; Tobkin, Sarah E1; Smith, Abbie E1; Lockwood, Chris M1; Walter, Ashley A2; Cramer, Joel T2; Beck, Travis W2; Stout, Jeffrey R1

Author Information
Journal of Strength and Conditioning Research 23(4):p 1068-1076, July 2009. | DOI: 10.1519/JSC.0b013e3181aa1cd0
  • Free

Abstract

Introduction

Accurate assessments of body composition in college athletes can be used to identify general health and athletic performance. For instance, increasing body fat has been shown to hinder athletic performance (34). Accurate body composition measurements can be used to evaluate the effectiveness of training and nutrition regimens. Frequently, field methods, such as skinfold measurements, are used to estimate body composition and percent fat (%fat) in athletes, due to the low cost of skinfold calipers and their ease of use. However, many of the suggested skinfold equations used to estimate %fat have been derived from hydrostatic weighing (HW) (12,13). Specifically, generalized equations of Jackson et al. (13) have been used in the past with the sum of 4 skinfold equation, producing consistently valid results in female athletes compared with HW and dual x-ray absorptiometry (DXA) (10,11,25,29). However, 2-compartment models, such as HW, have inherent error because of assumptions of fat-free mass (FFM) components, such as hydration and bone mineral content (16,26,33). Specifically, Lohman (16) calculated the estimated error of a 2-compartment model, based on the work by Siri (27), to be 3.9%fat in the general population. Moon et al. (20) in agreement with this determined that individual errors from HW were ±3.8%fat in college-aged nonathletic females compared with a 3-compartment model. Such errors in 2-compartment models could potentially be seen in anthropometric equations derived from these criterion methods. Therefore, it is suggested that a criterion method consist of more than 2 compartments (16,20,33).

Multiple-compartment (multi-compartment) models (more than 2 compartments) have become the favored criterion methods due to the added measurements of total body water (TBW) and bone mineral content (BMC). Wang et al. (33) reported that, when compared with the 6-compartment model, the most accurate methods were multi-compartment models that included the estimate of TBW, such as a 4-compartment model. Recently, Evans et al. (4) generated skinfold equations from males and females and black and white athletes using a 4-compartment model for use in NCAA Division I athletes. However, the final equations were derived from the entire sample of athletes, which included football, basketball, volleyball, gymnastics, swimming, and track athletes, and have yet to be validated. Regression analysis in the Evans et al. (4) investigation indicated that males and females and blacks and whites produced significantly different slopes for %fat. Therefore, the final Evans et al. (4) equations used sex and race coefficients and included only 44 white female NCAA Division I athletes. Additionally, Fornetti et al. (5) derived an equation from DXA using height and weight to estimate FFM and, subsequently, %fat. Although DXA has been categorized with an error similar to that of 2-compartment models because both models assume a constant FFM hydration (0.73%) (34), the validity of an equation containing height and weight deserves attention due to its potential ease of use and lesser degree of required training compared with skinfold measurements.

Although developing anthropometric equations from multi-compartment models could potentially reduce errors in estimating %fat, error still exists in any criterion method. In 2002, Wang et al. (31) developed a 4C model that accounts for both bone mineral and soft tissue mineral components in an effort to reduce the small error associated with the 4C model. However, to date, the validity of anthropometric equations to predict percent fat in female white NCAA Division I athletes compared with the updated Wang et al. (32) 4C model is unknown. Additionally, the most resent skinfold equations of Evans et al. (4) have not been validated in any population compared with any criterion model.

Many coaches and trainers are familiar with the Jackson et al. (13) equations and may be using one of the Jackson et al. (13) equations in everyday body composition assessments of athletes and nonathletes. However, if the updated Evans et al. (4) equations were to produce more accurate results than the Jackson et al. (13) equations, the Evans et al. (4) equations would become the suggested equations for estimating %fat. In addition, the Fornetti et al. (5) equation requires the estimation based on height and weight only, thus reducing the technical skill and errors associated with skinfold measurements. Furthermore, if the Fornetti et al. (5) equation produced similar %fat estimations compared with skinfold-based equations, coaches and trainers would need to measure height and weight only for an accurate estimation of %fat.

Methods

Experimental Approach to the Problem

The purpose of the current investigation was 2-fold: (a) to validate the newly developed Evans et al. (4) skinfold equations, along with the height and weight-based equation of Fornetti et al. (5), to the most recent multi-compartment model (4C); and (b) to compare the generalized equations of Jackson et al. (13) with the newly developed equations and with the 4C model.

The Jackson et al. (13), Evans et al. (4), and Fornetti et al. (5) equations were selected as the dependent variables, which included the generalized sum of 3, 4, and 7 (JPW3, JPW4, and JPW7) equations of Jackson et al. (13); the newly developed sum of 3 and 7 equations (E3 and E7) of Evans et al. (4), and the newly developed height and weight-based equation of Fornetti et al. (5) (FHW). Skinfold thickness measurements along with height and weight were selected as the dependent variables because of common usage in a collegiate setting due to ease of use and low cost. Additionally, the criterion model should be the most advanced as well as accurate; therefore, the Wang et al. (31) 4C model was used as the criterion method in this investigation.

It was hypothesized that (a) compared with the 4C model, the Evans et al. (4) prediction equation would result in more accurate estimations of %fat, (b) the FHW equation would produce the largest error because of the inclusion of only height and weight in the equation, and (c) all 3 Jackson et al. (13) equations and the FHW equation would produce errors no greater than those produced by HW alone (3.9%fat) (16).

Five skinfold-based and 1 height and weight-based equations were validated against the criterion %fat from the 4-compartment model (4C) (31). The Jackson et al. (13) (JPW3, JPW4, and JPW7) equation values were converted from body density to %fat using the revised formula of Brozek et al. (2). The Evans et al. (4) (E3 and E7) equations produced %fat values and the Fornetti et al. (5) (FHW) equation values were converted to %fat from FFM. Before HW, skinfold measurements were taken along with TBW and BMC in no particular order. All body composition assessments were performed on the same day after a 12-hour fast (ad libitum water intake was allowed). The subjects were also instructed to avoid exercise for at least 12 hours before testing. Height (HT) was measured to the nearest 0.5 cm using a calibrated stadiometer. Body weight (BW) was measured using a calibrated clinical scale to nearest 0.01 kg with subjects wearing bathing suits.

Subjects

Twenty-nine white female NCAA Division I collegiate athletes (mean ± SD = 20 ± 1 years) volunteered to participate in the study. Descriptive characteristics of the subjects can be found in Table 1. All 29 subjects were currently participating in one of 3 collegiate level sports: volleyball (n = 7), softball (n = 16), or track and field (n = 6). Next to professional athletes, NCAA Division I collegiate athletes represent the highest caliber. Additionally, all athletes in the current investigation were varsity level and currently enrolled in an off-season strength and conditioning program and had been training and participating in said sport for at least 1 year before study participation. The purpose of the study and a description of the testing protocol were explained to each subject. Subjects were informed of the experimental risks and signed an informed consent document before the investigation. The investigation was approved by an institutional review board for use of human subjects.

T1-3
Table 1:
Descriptive characteristics of subjects (n = 29).

Procedures

Four-Compartment Model Calculations

Criterion percent body fat (%fat) was estimated using the 4C model described by Wang et al. (32). The equation includes measurements of body volume (BV), total body water (TBW), total body bone mineral (Mo), and BW. The equations for fat mass (FM) and %fat are as follows:

Propagation of Error

Although multi-compartment models are recommended over 2C models for assessing and tracking changes in body composition, the potential propagation of errors because of the inherent measurement error of each device used to assess each variable may offset the improved accuracy of 4C model estimates of body composition (32). Wang et al. (32) suggested calculating the propagated error, sometimes referred to as the total error of measurement (TEM) (20), to account for the accuracy of the 4C equation. The standard errors of measurement (SEM) from the reliability data for the measurement of BV, TBW, and Mo were used to calculate propagated errors for %fat (32). In the current study, TEM was 0.59%fat, which is similar (less than 1%fat) to values reported for the 4C and 5C models in other laboratories (0.70-0.89%fat) (24,36). The TEM for the 4C model was calculated from the following equations (32):

Additionally, the test-retest measurements of 10 men and women measured 24-48 hours apart for the 4C equation resulted in an intraclass correlation coefficient (ICC) of 0.99 with a standard error of measurement (SEM) of 0.67%fat, which confirms the TEM results, indicating an error of less than 0.68%fat. The following paragraphs provide a description of the procedures used to estimate each variable included in the 4C equation and the estimation of %fat from the anthropometric equations.

Dual-Energy X-Ray Absorptiometry

Dual x-ray absorptiometry (software version 10.50.086; Lunar Prodigy Advance, Madison, WI) was used to estimate total body bone mineral content. Bone mineral content (BMC) was converted to total body bone mineral (Mo) using the following equation: Mo = total body BMC × 1.0436 (7). Each day before testing, a quality assurance phantom was performed and passed. Before each test, the subjects' height, weight, sex, and race were entered into the computer program. The subjects were positioned supine on the DXA table with hands pronated and flat on the table. Total body mode was selected for each scan, and scanning thickness was determined by the DXA software. All DXA scans were performed by a certified enCORE software operator. Previous test-retest scans of 11 men and women measured 24-48 hours apart for Mo produced an SEM of 0.05 kg with ICCs greater than 0.99.

Total Body Water

Bioimpedance spectroscopy (BIS) was used to estimate TBW following the procedures recommended by the manufacturer (Imp SFB7; ImpediMed Limited, Queensland, Australia). This technique, explained elsewhere (17), uses a range of frequencies, encompassing both low and high ranges that allow electrical current to pass around and through each cell, and has produced valid estimates of TBW when compared with a criterion method, such as deuterium oxide (17,21,30). In addition, Moon et al. (21) determined that the BIS device used in the current investigation is a valid, nonbiased measurement tool for determining TBW in college-aged females when compared with a criterion method to estimate TBW (deuterium oxide). Furthermore, BIS has been used to assess TBW for multi-compartment equations in previous validation studies (19,20,23). Measurement of TBW was taken while the subject was lying in a supine position on a table with arms ≥30° away from the torso and legs separated. Electrodes were placed at the distal ends of each subject's right hand and foot following the manufacturer guidelines. Before electrode placement, excess body hair was removed, and the skin was cleaned with alcohol at each site. The average of 2 trials within ±0.05 L was used as the representative TBW. Before analysis, each subject's HT, BW, age, and sex were entered into the BIS device. Previous test-retest measurements of 11 men and women measured 24-48 hours apart for TBW using the Imp SFB7 BIS produced an SEM of 0.48 L and an ICC greater than 0.99.

Hydrostatic Weighing

Body volume was assessed from HW as previously described by our laboratory and others (3,9,20,22,28,33). Residual volume was determined with the subject in a seated position using the oxygen dilution method of Wilmore (35) via a metabolic cart with residual volume software (True One 2400; Parvo-Medics, Inc., Provo, UT.). Subjects completed a minimum of 2 trials, and the average of the closest 2 trials within 5% was used to represent residual volume.

Underwater weight was measured to the nearest 0.025 kg in a submersion tank in which a seat was suspended from a calibrated Chatillon 15-kg scale (Model # 1315DD-H, Largo, FL). The average of the 3 highest values (6-10 trials) was used as the representative underwater weight. Previous test-retest reliability data for HW from our laboratory indicated that for 11 young adults (24 ± 2 years) measured on separate days, the ICC was greater than 0.99 with an SEM of 0.8%fat and 0.34 L for BV. These values are comparable to those reported by other laboratories (10,20,28).

Anthropometric Measurements

Skinfold thickness measurements were taken on the right side of the body with a calibrated Lange caliper by an investigator who had previously demonstrated a test-retest reliability of an ICC >0.95 and an SEM < 0.52%fat. Measurements were taken according to the recommendations of Jackson and Pollock (12) at the sites for chest, axilla, triceps, abdomen, suprailium, subscapular, and thigh. Body density values were calculated using the generalized skinfold equation of Jackson et al. (13) (Table 2). Percent body fat for the Jackson et al. (13) equations (JPW3, JPW4, JPW7) was calculated from BD using the revised formula of Brozek et al. (2). For this investigation, the Evans et al. (4) equations (E3 and E7) used the same skinfold sites as the Jackson et al. (13) equations to estimate %fat (Table 2); however, the Evans et al. (4) equations solve for %fat and did not require a conversion equation. The height and weight equation of Fornetti et al. (5) was used to estimate FFM using the height and weight measurements previously described. Fat-free mass was converted to %fat using the formula located in Table 2.

T2-3
Table 2:
Anthropometric equations.

Statistical Analyses

The validity of the %fat estimates (JPW3, JPW4, JPW7, E3, E7, and FHW) was based on the evaluation of predicted values versus the criterion or actual value from the 4C model by calculating the constant error (CE = actual [4C] − predicted %fat), Pearson product-moment correlation coefficient (r), standard error of estimate

and total error

(8). The mean difference (CE) between the predicted and actual (4C) %fat values was analyzed using dependent t-tests with the Bonferroni alpha adjustment (p ≤ 0.008) (14). Additionally, the method of Bland and Altman (1) was used to identify the 95% limits of agreement between the criterion and predicted %fat values.

Results

Compared with the 4C model, all anthropometric-based equations estimated %fat with TE values of less than 3.39%fat (Table 3). The height and weight-based equation of Fornetti et al. (5) produced the lowest r value (0.78) and largest SEE (2.97%fat) and TE (3.38%fat) values compared with the 4C model. The skinfold-based anthropometric equations produced similar r, SEE, and TE values (Table 3) compared with the 4C model. The E7 equation (mean = 26.56%fat) produced the only significantly different CE value compared with the 4C model (mean = 24.98%fat). The individual estimates for %fat derived from the anthropometric-based equations were analyzed using Bland and Altman plots, and the results are shown in Figures 1-3. The 95% limits of agreement (CE ± 1.96 SD of residual scores [predicted − actual]) were largest for the FHW equation (1.55 ± 6.00%fat) compared with 4C model, whereas the skinfold-based equations (less than ± 5.35%fat) produced tighter limits of agreement. Additionally, the height and weight-based equation of Fornetti et al. (5) produced the only significant trend (r = 0.59) compared with the 4C model, indicating an underestimation of %fat as total body fat increased (Figure 3). All the skinfold-based anthropometric equations produced similar individual estimation errors indicated by the 95% limits of agreement (less than ± 5.35%fat), with the E7 equation producing the tightest limits (±4.39%, −5.97 to 2.81%fat) (Table 3).

T3-3
Table 3:
Validation of methods for predicting % body fat compared with the Wang 4C model(n = 29).*
F1-3
Figure 1:
Bland and Altman plots comparing %fat estimations by the E3 and E7 equations with the 4C model (n = 29). The solid lines represent the upper and lower limits of agreement (±1.96 SD). The dotted/dashed line represents the constant error or mean bias. The dashed regression line represents the trend between the differences of methods and the mean of both methods.
F2-3
Figure 2:
Bland and Altman plots comparing %fat estimations by the JPW3, JPW4, and JPW7 equations with the 4C model (n = 29). The solid lines represent the upper and lower limits of agreement (±1.96 SD). The dotted/dashed line represents the constant error or mean bias. The dashed regression line represents the trend between the differences of methods and the mean of both methods.
F3-3
Figure 3:
Bland and Altman plots comparing %fat estimations by the FHW equation with the 4C model (n = 29). The solid lines represent the upper and lower limits of agreement (±1.96 SD). The dotted/dashed line represents the constant error or mean bias. The dashed regression line represents the trend between the differences of methods and the mean of both methods.

Discussion

In agreement with our hypothesis, the primary findings of the current study indicated that all skinfold-based anthropometric equations provided accurate estimations of %fat compared with the 4C model of Wang et al. (31) in white female NCAA Division I athletes. The height and weight-based anthropometric equation of Fornetti et al. (5) produced acceptable estimates of %fat in this population but were less accurate (TE = 3.38%fat) than the estimates of %fat from the skinfold-based anthropometric equations (TE < 2.72%fat). The Evans et al. (4) (E3 and E7) skinfold equations (less than ±4.87%fat) produced more accurate individual estimation of %fat compared with the Jackson et al. (13) equations (greater than ±4.90%fat). However, the Evans et al. (4) equations produced larger CE values (E3 = −1.20%fat, E7 = −1.58%fat) compared with the Jackson et al. (13) equations (CE < 0.93%fat), suggesting a systematic overestimation of more than 1% for both the E3 and E7 equations. However, it should be noted that CEs indicate systematic errors in equations or methods and that these errors can potentially be reduced by adjusting for these CE values.

The possible causes of the greater than 1.0%fat CE values for the Evans et al. (4) equations compared with the 4C model could be related to several factors. Specifically, the Evans et al. (4) equations were designed using Harpenden calipers, whereas the current investigation utilized Lange calipers. Gruber et al. (6) determined that Harpenden calipers produce significantly smaller values than Lange calipers, resulting in an underestimation of ≈1.5%fat. Based on these findings, if the current investigation were to use Harpenden calipers rather than Lange calipers, the CE values for the Evans et al. (4) equations may not have produced an overestimation of %fat. Specifically, if the CE values were corrected for the 1.5%fat overestimation due to the use of Lange calipers, the resulting CE values for the Evans et al. (4) equations would have been −0.08%fat for the E7 equation and 0.30%fat for the E3 equation, compared with the current CE values of −1.58%fat for the E7 equation and −1.20%fat for the E3 equation. Therefore, when using Lange calipers in the Evans et al. (4) equations, greater accuracy may be accomplished by correcting for the −1.58%fat or −1.20%fat CE values associated with the E7 equation and the E3 equation, respectively. However, the type of caliper used may not be the only discriminating factor in the CE values of greater than 1.0%fat.

The Evans et al. (4) equations measured skinfold thicknesses at slightly different locations than those reported by Jackson and Pollock (12), which were used in the current investigation. Specifically, the abdominal and axilla skinfold measurements in the current investigation were measured using vertical skinfolds, whereas the Evans et al. (4) investigation used horizontal skinfold measurements for these sites. However, the results of vertical skinfolds compared with horizontal skinfolds at the axilla and abdominal sites and their effects on %fat when using skinfold equations in white female NCAA Division I athletes is not known. Nonetheless, the current results suggest that the brand of skinfold caliper may be more important than the orientation of the abdominal and axilla skinfold measurements; however, this hypothesis warrants further investigation.

Another contributing factor for the CE values greater than 1.0%fat could be related to the multiple-compartment model used. Specifically, Evans et al. (4) used an older 4C model (15) compared with the most recent 4C model of Wang et al. (31), which accounts for variations in soft tissue minerals. However, Wang et al. (31) suggests that the updated 4C model is similar to the Lohman-Selinger model (15), which was used in the investigation by Evans et al. (4). Therefore, the impact of the 4C model used is not considered to be the cause of the CE >1.0%fat. Additionally, our 4C model included BIS rather than deuterium oxide for the TBW measurements and a Lunar Prodigy Advance DXA device compared with a Hologic QDR-1000W DXA, which were employed in the Evans et al. (4) investigation. However, due to the accurate and valid estimations of %fat using the current, most recent 4C model compared with the Evans et al. (4) equations and the known deviations associated with caliper brands (Harpenden vs. Lange), the CE values >1.0% are most likely due to the brand of caliper used in the current investigation. Thus, it is hypothesized that the caliper brand may be the largest contributing factor; however, more research is required to prove or disprove this hypothesis.

Although the CE values for the Evans et al. (4) equations may have been greater than 1.0%fat, the SEE and TE values are considered “very good” to “ideal” (8), and to the best of our knowledge, this is the first investigation to validate either of the Evans et al. (4) skinfold-based equations in any population. Furthermore, the equations produced by the Evans et al. (4) investigation incorporated males, females, black, and white athletes with 44 of the athletes being white females, which were included in the final %fat estimation equations. Nonetheless, the Evans et al. (4) equations, which were based not only on 44 white female athletes accounting for both sex and race, produced accurate and valid estimations of %fat in the current investigation in an independent sample of 29 white female Division I athletes, suggesting that both the E3 and E7 skinfold equations are valid for estimating %fat in this population.

Additionally, this is the first investigation to compare the Jackson et al. (13) and Fornetti et al. (5) anthropometric equations with the most recent 4C model of Wang et al. (31). However, the 3- and 4-site Jackson et al. (13) skinfold equations have been compared with other criterion methods with varying results. Similar to the current investigation, Houtkooper et al. (11) compared the SPW3 and SPW4 equations with DXA in elite American heptathletes (females, mean 19.66 years) and found that both equations produced similar SEE (JPW3 = 2.4%fat, JPW4 = 2.4%fat) and TE (JPW3 = 2.8%fat, JPW4 = 2.8%fat) values compared with the current investigation SEE (JPW3 = 2.56, JPW4 = 2.56) and TE (JPW3 = 2.65, JPW4 = 2.68). However, several other studies, including the original equation development study of Jackson et al. (13), reported higher SEE (greater than 2.99%fat) and TE (greater than 3.22%fat) values compared with HW in college-aged and high school-aged female athletes (10,18,26,30,38). This is in contrast to the current investigation (SEE <2.57%fat, TE <2.69%fat) compared with the Wang et al. (31) 4C equation in white female Division I athletes. These discrepancies can be attributed to several factors, including the brand of skinfold caliper used, age of athletes, type of athletes (sports played), level of athlete (high school to elite), and equation used to convert HW body density values to %fat values. All these variables could have attributed to the lack of agreement between past investigations and the current study.

Specifically, the 4C model used in the current investigation does not depend on FFM hydration status, bone mineral content, or soft tissue mineral deviations, which are assumed in HW models such as the Brozek et al. (2) and Siri (26) models. However, the results of this study suggest that there is no difference between the “reference body” of Brozek et al. (2), which was used in the equation development investigation of Jackson et al. (13) and the white female Division I athletes included in this study. This hypothesis was tested by comparing the FFM density, water, mineral, and protein content of the “reference body” of Brozek et al. (2) with the female athletes' values in the current investigation using 1-sample t-tests. Results indicated that compared with the reference body cadavers of Brozek et al. (2), the white female athletes used in the current study demonstrated nonsignificant (p > 0.05) values for FFM density, water, mineral, and protein content. These findings suggest that both the generalized Jackson et al. (13) and population-specific Evans et al. (4) equations accurately predict %fat values in white female Division I athletes and include similar values for FFM density, water, mineral, and protein content. Furthermore, the individual errors associated with the Evans et al. (4) equations, indicated by the 95% limits of agreement, were tighter than the equations of Jackson et al. (13) when comparing the same number of skinfold sites (E3 vs. JPW3 and E7 vs. JPW7) (Table 3). Specifically, the E3 limits of agreement were −6.06%fat to 3.66%fat (± 4.86%fat), the JPW3 limits were −5.21%fat to 5.35%fat (± 5.28), the E7 limits of agreement were −5.97%fat to 2.81%fat (±4.39%fat), and the JPW7 limits were -3.99%fat to 5.83%fat (± 4.91%fat). These findings suggest that the population-specific equations of Evans et al. (4) may reduce the errors associated with the generalized equations of Jackson et al. (13). Nonetheless, both equations appear to be valid in this population, but because the Evans et al. (4) equations solve for %fat and are designed for this population specifically, they are preferred over the generalized equations of Jackson et al. (13). However, all skinfold estimation equations resulted in individual errors of more than ± 4.38%fat, resulting in an accurate estimation range of greater than 8.78%fat. Therefore, the skinfold equations used in the current investigation should be used with caution when estimating %fat in individuals or small groups or teams.

The height and weight-based equation of Fornetti et al. (5) resulted in an individual accurate estimation range of 12%fat (± 6.00%fat) compared with the 4C model and also resulted in a significant trend to underestimate %fat as %fat increased (Figure 3). Therefore, because of these findings and lack of validation studies using the height and weight-based equation of Fornetti et al. (5), the FHW equation is not the recommended method to estimate %fat in this population, and because height and weight alone do not take into consideration %fat as a part of the body's composition, this method may not accurately estimate %fat in individuals, small groups, or teams.

All skinfold estimation equations (E3, E7, JPW3, JPW4, and JPW7) produced accurate group and individual %fat values compared with the 4C model and were more accurate than the height and weight-based equation (FHW) of Fornetti et al. (5). The newly developed, population-specific equations of Evans et al. (4) resulted in tighter limits of agreement compared with the 4C model but produced CE values greater than 1.0%, suggesting that correcting for the CE values in the Evans et al. (4) equations when using Lange calipers may improve the accuracy of the equations. However, although all skinfold-based equations were valid group %fat estimators, the limits of agreement suggest the individual errors can be greater than ±4.39%fat. Therefore, multi-compartment models are suggested for use when assessing individual estimates of %fat in body composition in this population.

If group or team %fat estimations are desired, the skinfold equations used in the current investigation may provide accurate estimations; however, the ability of these equations to track changes in body composition warrants further investigation. The use of the Evans et al. (4) equations may provide an easier and potentially more accurate method for estimating %fat because the equations solve directly for %fat and have the least individual error compared with the 4C model. Nonetheless, the Jackson et al. (13) equations produced valid percent fat estimates in this population and produced the smallest constant error values compared with the 4C model. Furthermore, using 7 skinfold sites did not improve the group %fat estimations compared with the 3-site equations for the generalized Jackson et al. (13) or the population-specific Evans et al. (4) equations. Therefore, the Jackson et al. (13) 3-site equation (JP3) and the Evans et al. (4) 3-site equation (E3) are recommended for use in white female Division I athletes and the E3 equation may produce more accurate results when using Lange calipers if the following CE-corrected equation is used: %fat = 7.787 + 0.24658 × [triceps + abdomen + thigh (skinfold thickness in mm)].

Practical Applications

Trainers and coaches currently using the Jackson et al. (13) equations used in the current investigation should continue to use these equations in white female Division I athletes but need to understand that individual errors may be as large as ± 5.34%fat. Three skinfold measurement sites are all that are needed to produce accurate %fat estimations; thus, the JPW3 (Table 2) equation is recommended in this population. However, if Lange calipers are used and a simplified %fat equation is preferred over the more complex body density JPW3 equation with the Brozek et al. (2) %fat equation, the CE-corrected Evans et al. (4) equation above is suggested. Finally, the use of the height and weight-based equations of Fornetti et al. (FHW) is only recommended for large groups of athletes because of the large individual errors and significant trend to underestimate %fat as %fat increases.

Acknowledgments

We thank all the athletes who participated in this investigation and Vincent J. Dalbo for his efforts during data collection. Additionally, we thank Sarah Cahill for her dedication and support as a strength coach. Results of the present study do not constitute endorsement of the products by the authors or NSCA.

References

1. Bland, JM and Altman, DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307-310, 1986.
2. Brozek, J, Grande, F, Anderson, JT, and Keys, A. Densitometric analysis of body composition: Revision of some quantitative assumptions. Ann NY Acad Sci 110: 113-140, 1963.
3. Clasey, JL, Kanaley, JA, Wideman, L, Heymsfield, SB, Teates, CD, Gutgesell, ME, Thorner, MO, Hartman, ML, and Weltman, A. Validity of methods of body composition assessment in young and older men and women. J Appl Physiol 86: 1728-1738, 1999.
4. Evans, EM, Rowe, DA, Misic, MM, Prior, BM, and Arngrimsson, SA. Skinfold prediction equation for athletes developed using a four-component model. Med Sci Sports Exerc 37: 2006-2011, 2005.
5. Fornetti, WC, Pivarnik, JM, Foley, JM, and Fiechtner, JJ. Reliability and validity of body composition measures in female athletes. J Appl Physiol 87: 1114-1122, 1999.
6. Gruber, JJ, Pollock, ML, Graves, JE, Colvin, AB, and Braith, RW. Comparison of Harpenden and Lange calipers in predicting body composition. Res Q Exerc Sport 61: 184-190, 1990.
7. Heymsfield, SB, Wang, J, Heshka, S, Kehayias, JJ, and Pierson, RN. Dual-photon absorptiometry: comparison of bone mineral and soft tissue mass measurements in vivo with established methods. Am J Clin Nutr 49: 1283-1289, 1989.
8. Heyward, VH and Wagner, DR. Applied Body Composition Assessments. Champaign, IL: Human Kinetics, 2004. pp. 17-20.
9. Housh, TJ, Johnson, GO, Housh, DJ, Cramer, JT, Eckerson, JM, Stout, JR, Bull, AJ, and Rana, SR. Accuracy of near-infrared interactance instruments and population-specific equations for estimating body composition in young wrestlers. J Strength Cond Res 18: 556-560, 2004.
10. Housh, TJ, Johnson, GO, Housh, DJ, Eckerson, JM, and Stout, JR. Validity of skinfold estimates of percent fat in high school female gymnasts. Med Sci Sports Exerc 28: 1331-1335, 1996.
11. Houtkooper, L, Mullins, VA, Going, SB, Brown, CH, and Lohman, TG. Body composition profiles of elite American heptathletes. Int J Sport Nutr Exerc Metab 11: 162-173, 2001.
12. Jackson, AS, and Pollock, ML. Practical assessment of body composition. Phys Sport Med 13: 76-90, 1985.
13. Jackson, AS, Pollock, ML, and Ward, A. Generalized equations for predicting body density of women. Med Sci Sports Exerc 12: 175-181, 1980.
14. Keppel, G, and Wickens, TD. Design and Analysis: A Researchers Handbook (4th ed.). Upper Saddle River, NJ: Prentice Hall, 2004. pp. 117-118.
15. Lohman, TG. Applicability of body composition techniques and constants for children and youths. Exerc Sport Sci Rev 14: 325-357, 1986.
16. Lohman, TG. Advances in Body Composition Assessment. Champaign, IL: Human Kinetics, 1992. pp. 3-22.
17. Matthie, J, Zarowitz, B, De Lorenzo, A, Andreoli, A, Katzarski, K, Pan, G, and Withers, P. Analytic assessment of the various bioimpedance methods used to estimate body water. J Appl Physiol 84: 1801-1816,1998.
18. Mayhew, JL, Clark, BA, McKeown, BC, and Montaldi, DH. Accuracy of anthropometric equations for estimating body composition in female athletes. J Sports Med Phys Fitness. 25: 120-126, 1985.
19. Minderico, CS, Silva, AM, Keller, K, Branco, TL, Martins, SS, Palmeira, AL, Barata, JT, Carnero, EA, Rocha, PM, Teixeira, PJ, and Sardinha, LB. Usefulness of different techniques for measuring body composition changes during weight loss in overweight and obese women. Br J Nutr 99: 432-441, 2008.
20. Moon, JR, Hull, HR, Tobkin, SE, Teramoto, M, Karabulut, M, Roberts, MD, Ryan, ED, Kim, SJ, Dalbo, VJ, Walter, AA, Smith, AE, Cramer, JT, and Stout, JR. Percent body fat estimations in college women using field and laboratory methods: A three-compartment model approach. J Int Soc Sports Nutr 4: 16, 2007.
21. Moon, JR, Tobkin, SE, Roberts, MD, Dalbo, VJ, Kerksick, CM, Bemben, MG, Cramer, JT, and Stout, JR. Total body water estimations in healthy men and women using bioimpedance spectroscopy: A deuterium oxide comparison. Nutr Metab (Lond) 5: 7, 2008.
22. Prior, BM, Modlesky, CM, Evans, EM, Sloniger, MA, Saunders, MJ, Lewis, RD, and Cureton, KJ. Muscularity and the density of the fat-free mass in athletes. J Appl Physiol 90: 1523-1531, 2001.
23. Sardinha, LB, Silva, AM, and Teixeira, PJ. Usefulness of age-adjusted equations to estimate body fat with air displacement plethysmography in male adolescent athletes. Acta Diabetol 40: 51-55, 2003.
24. Silva, AM, Minderico, CS, Teixeira, PJ, Pietrobelli, A, and Sardinha, LB. Body fat measurement in adolescent athletes: multicompartment molecular model comparison. Eur J Clin Nutr 60: 955-964, 2006.
25. Sinning, WE and Wilson, JR. Validation of “generalized” equations for body composition analysis in women athletes. Res Q Exerc Sport 55: 153-160, 1984.
26. Siri, WE. The gross composition of the body. Adv Biol Med Phys 4: 239-280, 1956.
27. Siri, WE. Body composition from fluid spaces and density. Analysis of methods. In: Techniques for Measuring Body Composition. J. Brozek and A. Henschel, eds. Washington, DC: National Academy of Sciences, 1961. pp. 223-244.
28. Stout, JR, Housh, TJ, Eckerson, JM, Johnson, GO, and Betts, NM. Validity of methods for estimating percent body fat in young women. J Strength Cond Res 10: 25-29, 1996.
29. Thorland, WG, Johnson, GO, Tharp, GD, Fagot, TG, and Hammer, RW. Validity of anthropometric equations for the estimation of body density in adolescent athletes. Med Sci Sports Exerc 16: 77-81, 1984.
30. Van Loan, MD, Withers, P, Matthie, J, and Mayclin, PL. Use of bioimpedance spectroscopy to determine extracellular fluid, intracellular fluid, total body water, and fat-free mass. Basic Life Sci 60: 67-70, 1993.
31. Wang, Z, Pi-Sunyer, FX, Kotler, DP, Wielopolski, L, Withers, RT, Pierson, RN Jr, and Heymsfield, SB. Multicomponent methods: Evaluation of new and traditional soft tissue mineral models by in vivo neutron activation analysis. Am J Clin Nutr 76: 968-974, 2002.
32. Wang, Z, Shen, W, Withers, RT, and Heymsfield, SB. Multicomponent molecular-level models of body composition analysis. In: Human Body Composition. Heymsfield, SB, Lohman, TG, Wang, Z, and Going, SB, eds. Champaign, IL: Human Kinetics, 2005. pp. 163-176.
33. Wang, ZM, Deurenberg, P, Guo, SS, Pietrobelli, A, Wang, J, Pierson, RN Jr, and Heymsfield, SB. Six-compartment body composition model: Inter-method comparisons of total body fat measurement. Int J Obes Relat Metab Disord 22: 329-337, 1998.
34. Welch, BE, Riendeau, RP, Crisp, CE, and Isenstein, RS. Relationship of maximal oxygen consumption to various components of body composition. J Appl Physiol 12: 395-398, 1958.
35. Wilmore, JH, Vodak, PA, Parr, RB, Girandola, RN, and Billing, JE. Further simplification of a method for determination of residual lung volume. Med Sci Sports Exerc 12: 216-218, 1980.
36. Withers, RT, LaForgia, J, Pillans, RK, Shipp, NJ, Chatterton, BE, Schultz, CG, and Leaney, F. Comparisons of two-, three-, and four-compartment models of body composition analysis in men and women. J Appl Physiol 85: 238-245, 1998.
37. Withers, RT, Whittingham, NO, Norton, KI, La Forgia, J, Ellis, MW, and Crockett, A. Relative body fat and anthropometric prediction of body density of female athletes. Eur J Appl Physiol Occup 56: 169-180, 1987.
    Keywords:

    multi-compartment model; skinfold; field method; body composition; sports

    © 2009 National Strength and Conditioning Association