Accuracy and Precision of Opportunistic Measures of Body Composition from the Tanita DC-430U

Abstract

Background:

It is essential to quantify the accuracy and precision of bioelectrical impedance (BIA)-estimated percent body fat (%BF) to better interpret community-based research findings that utilize opportunistic measures.

Methods:

Study 1 measured the accuracy of a new dual-frequency foot-to-foot BIA device (Tanita DC-430U) compared with dual-energy X-ray absorptiometry (DXA) among healthy elementary school-aged children (N = 50). Study 2 examined the precision of BIA %BF estimates within and between days among children and adults (N = 38).

Results:

Regarding accuracy, Tanita DC-430U underestimated %BF by 8.0 percentage points compared with DXA (20.6% vs. 28.5%), but correctly ranked children in terms of %BF. Differences in %BF between BIA and DXA were driven by lower BIA-estimated fat mass (7.8 kg vs. 9.9 kg, p < 0.05) and higher BIA-estimated fat-free mass (25.3 kg vs. 24.1 kg, p < 0.05). The absolute agreement between BIA and DXA for estimated %BF was moderate (concordance correlation coefficients = 0.53). Regarding precision, measures taken at the same time, but on different days (root mean square standard deviation [RMSD] = 0.42–0.74) were more precise than the measures taken at different times within a single day (RMSD = 1.04–1.10).

Conclusion:

The Tanita DC-430U substantially underestimated %BF compared with DXA, highlighting the need to assess accuracy of new BIA devices when they are introduced to the market. Opportunistic measures of %BF estimates were most precise when taken at consistent times and in the morning, but may be utilized throughout the day with an understanding of within- and between-day variability.

Introduction

BMI is widely used to approximate adiposity in surveillance studies and clinical trials because it is noninvasive, fast to collect, and can be readily deployed in field-based research settings (e.g., schools).^1–3 While BMI and BMI z-score are strongly to moderately correlated (r = 0.69–0.90) with children's body composition at a population level,^4,5 they provide no direct information on children's fat distribution, muscle mass, or lean body mass. Supplementing BMI and BMI z-score with other noninvasive measures of body composition could produce greater certainty about adiposity in research studies, informing efforts to assess the prevalence of obesity and help target intervention efforts.

BIA is a portable noninvasive technique that can predict body composition, including percent body fat (%BF), which is based on conduction of an electrical current in the human body. BIA-estimated %BF is impacted by technique, total body water, food and liquid intake, type of exercise modality and timing, clothing, and hormonal fluctuations.^6–8 Nonetheless, commercially available, multifrequency, foot-to-foot BIA devices have demonstrated good agreement with reference measures of body composition such as air displacement plethysmography, dual-energy X-ray absorptiometry (DXA), and the gold standard 4-compartment body composition model.^9,10 Thus, these devices have the potential to complement and extend traditional BMI measures by producing indirect estimates of fat-free mass (FFM), fat mass (FM), and %BF.

Because BIA-estimated %BF measurements may be influenced by numerous factors and because BIA has inherent error in estimating %BF, it is essential to examine the accuracy of the estimates from these tools. Furthermore, as technology advances and new devices become available for research, studies exploring the accuracy of these devices are needed to ensure that estimates of %BF from new devices are valid and produce comparable estimates with previous BIA devices. Tanita recently released the DC-430U mobile BIA device, which to date has not been independently validated.

Similar to standard weight, BIA-estimated %BF fluctuates within and between days.^11,12 These fluctuations can be due to actual fluctuations in %BF but are likely mostly due to error (i.e., of technical error, user error, and/or prediction error). To minimize the error, BIA protocols recommend measurement first thing in the morning following a 4–10-hour fast and after voiding the bladder and bowels.¹³ However, conducting measures first thing in the morning after a fast and bladder/bowel void may be untenable for community-based studies that rely upon opportunistic measurement occasions, such as measures occurring during physical education, and studies with children. This may explain why despite a large body of literature validating BIA devices, BIA has not been widely adopted in field-based public health research. For instance, a recent systematic review of BIA protocols for the estimation of %BF in children (0–17 years old) only identified 71 studies between 1988 and 2016.⁸

Studies that did use BIA for the estimation of %BF in children were also limited in their reporting of BIA measurement protocols. For example, 79% of studies did not report protocols related to time of day. Furthermore, 44%, 72%, and 75% of studies did not report protocols related to fasting, voiding, and exercise, respectively. Thus, it is critically important to explore the upper bounds of within- and between-day fluctuations in %BF estimates to appropriately interpret findings from the majority of studies utilizing BIA-estimated %BF as a measure.

In the case of BIA, precision refers to the variability of different BIA estimates taken from the same individual over a period of time. There is a large body of evidence related to the precision of BIA estimates of %BF.^14,15 However, this literature has largely not focused on the fluctuation of BIA-estimated %BF in opportunistic field measures within and between days. Rather, studies have typically focused on the precision of BIA estimates that occur under ideal conditions and in rapid succession. While this approach does provide evidence of device precision (all but ensuring that any variation in %BF is due to error and not actual %BF fluctuation), it does little to inform the precision of opportunistic measures (i.e., variability that can be expected due to error) that are collected at different times of day when the participant is not fasted and has not voided the bowels or bladder.

We are aware of only two studies that have explored the precision of BIA over the course of a single day and these studies focused on the electrical current aspects of impedance.^16,17 These studies stopped short of exploring how the variability in impedance ultimately impacted BIA-estimated %BF, which limits the interpretability of these studies to applied public health research. Therefore, establishing the magnitude of the fluctuation in BIA-estimated %BF estimates within a day and between days is essential for interpreting studies that use opportunistic measures of BIA.¹⁸

Therefore, the purpose of this project was to examine the accuracy and precision of BIA-estimated %BF using the Tanita DC-430U. To accomplish this aim, two separate studies were conducted. The purpose of study 1 was to examine the accuracy of BIA-estimated %BF using a new dual-frequency foot-to-foot BIA device (Tanita DC-430U) compared with DXA in a sample of healthy children (5–12 years). The purpose of study 2 was to examine the precision of the same BIA device (Tanita DC-430U) among children and adults by assessing the within- and between-day variability of %BF estimates collected at home to simulate opportunistic measures that are commonly used in applied public health settings (i.e., in schools, community, workplace).

Materials and Methods

Participants and Sample Size Considerations

Study 1 (accuracy)

Participants were eligible if they were elementary school aged and had no medical conditions that would prevent them from completing a DXA scan. Participants were recruited from June to August 2019 in the greater metropolitan area of southeastern United States surrounding the institution in which the study was conducted. With a sample of 50 participants, calculations completed using G*Power (v. 3.1.7; Los Angeles, CA, USA) indicated that the study had power of 0.8, with an α = 0.05 and correlation between measures of 0.85, to detect a 2% difference in body composition constructs. This was determined to be adequate power based on past childhood obesity interventions that measured %BF¹⁹ and a recent systematic review of changes in %BF as a result of intervention.²⁰

Study 2 (precision)

Participants were eligible if they were elementary school aged (5–12 years) or adults (>18 years). Participants were recruited from May to December 2020 in the same southeastern area as study 1. Power calculations were conducted as described above using G*Power (v. 3.1.7), which indicated the study had a power of 0.9, assuming an α = 0.05 and correlation between measures of 0.8, to detect a 2% difference in measures within and between days. Similar to study 1, this was determined to be adequate power based on past studies.^19,20

Protocol

Before the first participant's enrollment, procedures for study 1 and study 2 were approved by the lead author's institutional review board (IRB).

Study 1 (accuracy)

Appointment times were set between the hours of 6–9 am with the parents to complete the study at the lead author's university. Consistent with the manufacturer's guidelines, parents/guardians were asked to arrive at the institution with their child on the morning of their appointment following an overnight fast (8–10 hours). Parents/guardians were also asked to have their child refrain from any moderate-to-vigorous physical activity before completion of study measures. Participants were asked to wear light, “gym-like clothing” containing no metal zippers to the appointment. Upon arrival, participants were asked to void their bladder and bowels. Participants were then asked to remove shoes and jewelry. All measurements were taken in a single 30-minute testing window. First, research assistants collected a single measurement of anthropometrics and BIA-estimated %BF, FM, and FFM (∼20 minutes). Second, a trained radiation technologist completed the DXA measurement (∼10 minutes).

Study 2 (precision)

Due to the COVID-19 pandemic school closures and social distancing requirements, and to simulate real-world opportunistic measures, all measures were collected by the participant or the participant's parent/guardian at the participant's home over a 4-day period (see Table 1). On day 1, a trained research staff delivered the Tanita BIA system to the participant's house and provided training on how to use the equipment along with written instructions on how to operate the device. The trained research assistant then collected a single height, weight, and BIA measurement for the participant and/or parent/guardian to observe, outdoors while following social distancing protocols. The participant then collected a single BIA measurement before bed on day 1. On days 2 and 3, fasted measurements were conducted each morning as BIA protocols recommend.¹³

Table 1.

Data Collection Schedule of Bioelectrical Impedance Measures for Study 2

	Wake	9:00 am	11:00 am	1:00 pm	5:00 pm	Bed
Day 1						✓
Day 2	✓	✓	✓	✓	✓	✓
Day 3	✓	✓	✓	✓	✓	✓
Day 4	✓

Fasted measurements were collected by the participants (i.e., for adults) or their parent/guardian (i.e., for children) when the participants woke in the morning after voiding their bladder and bowels. On days 2 and 3, additional measurements were collected throughout the day at 9 am, 11 am, 1 pm, 5 pm, and immediately before bed. On day 4, participants completed a single fasted measure after waking in the morning and voiding their bowels and bladder. Participants received a reminder text to complete their BIA measurements at all measurement time points.

Measurements

Demographics

For both studies, age (i.e., years), sex (i.e., male or female), race (i.e., Black, White, other) were obtained by participant (i.e., for adults) or parent/guardian (i.e., for children) report.

BMI

For both studies, participants' heights were obtained by trained research assistants using a portable stadiometer (Model S100; Ayrton Corp., Prior Lake, MN, USA.). All heights were estimated to the nearest 0.1 cm, with participants wearing light clothing and no footwear.^21,22 Weights were collected to the nearest 0.1 kg using the BIA device.

Dual-energy X-ray absorptiometry

For study 1, a whole-body DXA scan (GE Healthcare Lunar Prodigy, Madison, WI, USA) was completed to determine a reference measure of %BF.¹⁶ Total body scans were conducted by trained radiation technologists following the manufacturer's specified instructions. Software provided by the manufacturer calculated whole-body %BF using adapted formulas for the age of the participants. The radiation technologist instructed participants to lie still on the DXA scanner for the duration of the scan, ∼10 minutes.

BIA-estimated %BF

For both studies, BIA was measured with the novel Tanita DC-430U portable dual-frequency body composition analyzer. Participants' height, age, and sex were entered into the Tanita Health Ware^® software by a research assistant. Once the information for each participant was entered into the software, participants were instructed to stand on the device as still as possible for ∼10 seconds as the full-impedance analysis was completed. Whole-body %BF was calculated using the Tanita age-specific propriety algorithms.

Statistical Analyses

All statistical analyses were performed using Stata software (Version 16; StataCorp, College Station, TX, USA). Descriptive statistics were calculated for participants before running study-specific statistical analyses.

Study 1 (accuracy)

Spearman rank-order and Lin's concordance correlation coefficients (CCC) examined the relationship between BIA and DXA.^17,18 Separate Bland–Altman plots for girls and boys were constructed to assess mean bias and limits of agreement for BIA body composition estimates when compared with DXA.²³ Regression analysis with bias (i.e., difference between BIA and DXA body composition estimate) as the dependent variable and the mean of DXA- and BIA-estimated body composition as the independent variable examined trends in agreement between DXA- and BIA-estimated body composition. All regression models included age and weight as covariates.

Study 2 (precision)

Root mean square standard deviations (RMSSD) were calculated to examine precision.^24,25 RMSSD were calculated within days (i.e., all time points within a day) and between days (i.e., same time point across days). RMSSD were calculated separately for children and adults. Least significant change (LSC) was then calculated by multiplying the RMSSD by 2.77, consistent with the International Society for Clinical Densitometry.²⁵ To explore if precision varied between time points within and between days, a secondary analysis was conducted. Multilevel mixed-effect regressions (observations nested within participants) estimated within-day (i.e., measures on the same day compared with fasted) and between-day (i.e., measures at the same time on different days) variations in BIA-estimated %BF.

The dependent variable was the difference between fasted and the current measure (i.e., 9 am, 11 am, 1 pm, 5 pm, bed) for the within-day models and the difference between the measure on the first and second day in the between-day model. The independent variable was the time of measure (9 am, 11 am, 1 pm, 5 pm, bed for the within-day model and fasted, 9 am, 11 am, 1 pm, 5 pm, bed for between-day model). Regressions were run separately for children and adults.

Results

Descriptive Data

Descriptive data including means and standard deviations (±1) of the sample are shown in Table 2.

Table 2.

Demographics

	Mean (SD)	Median (IQR)	Mean (SD)	Median (IQR)	Mean (SD)	Median (IQR)
Study 1	Total (N = 50)		Boys (N = 22)		Girls (N = 28)
Age (years)	9.1 (2.2)	9.4 (7.3, 11.3)	8.8 (2.0)	8.9 (6.7, 10.3)	9.3 (2.3)	9.6 (7.4, 11.6)
Height (cm)	134.9 (14.0)	133.0 (124.0, 147.3)	134.1 (13.1)	131.8 (124.0, 146.9)	135.5 (14.8)	134.1 (125.5, 148.4)
Weight (kg)	33.1 (14.2)	29.4 (24.1, 39.9)	32.1 (15.1)	28.9 (22.7, 35.8)	34.0 (13.7)	29.5 (24.6, 43.5)
% White	74.0	—	81.8	—	67.9	—
% Black	16.0	—	9.1	—	21.4	—
BMI	17.7 (5.6)	16.3 (14.5, 18.7)	17.2 (4.6)	16.0 (14.2, 18.5)	18.1 (6.3)	16.6 (14.7, 19.9)
BMI percentile (%)	48.4 (31.7)	49.3 (13.7, 75.3)	47.0 (32.6)	56.5 (11.4, 67.4)	49.5 (31.2)	0.45 (0.29, 0.76)
BMI z-score	−0.01 (1.20)	−0.02 (−1.09, 0.68)	−0.12 (1.16)	0.16 (−1.20, 0.45)	−0.03 (1.23)	−0.12 (−0.55, 0.71)
% Body fat
BIA	20.6 (9.0)	19.6 (15.9, 24.1)	18.7 (8.3)	17.1 (14.5, 20.1)	22.1 (9.4)	21.4 (17.1, 25.9)
DXA	28.5 (6.5)	27.4 (24.2, 31.9)	26.6 (6.7)	25.1 (21.6, 29.4)	30.0 (6.0)	29.6 (25.5, 32.7)
Mean difference	7.9 (4.9)	8.3 (5.8, 10.5)	7.8 (3.9)	8.0 (5.1, 10.2)	7.9 (5.6)	8.4 (6.0, 11.0)
Absolute mean difference	8.5 (3.5)	8.3 (6.2, 10.9)	7.9 (3.7)	8.0 (5.1, 10.2)	9.0 (3.3)	9.0 (6.3, 11.1)
FM (kg)
BIA	7.8 (8.2)	5.7 (3.3, 9.0)	7.0 (8.7)	4.4 (3.3, 6.5)	8.5 (7.9)	6.7 (3.5, 10.6)
DXA	9.9 (6.8)	7.7 (5.8, 12.2)	9.2 (8.1)	7.0 (6.2, 8.4)	10.5 (5.8)	10.1 (5.6, 13.9)
Mean difference	2.1 (2.4)	2.4 (1.5, 3.2)	2.2 (1.3)	2.5 (1.3, 3.1)	2.0 (3.0)	2.2 (1.7, 3.4)
Absolute mean difference	2.7 (1.8)	2.5 (1.5, 3.2)	2.3 (1.1)	2.5 (1.3, 3.1)	2.9 (2.1)	2.5 (1.8, 3.6)
FFM (kg)
BIA	25.3 (7.4)	23.4 (19.5, 31.7)	25.1 (7.5)	23.7 (19.0, 29.8)	25.5 (7.4)	23.2 (20.3, 33.0)
DXA	24.1 (8.0)	21.5 (18.8, 29.1)	23.9 (7.7)	22.7 (17.6, 28.7)	24.3 (8.3)	20.5 (19.0,31.1)
Mean difference	−1.2 (2.4)	−1.5 (−2.2, −0.6)	−1.2 (1.3)	−1.5 (−2.0, −2.5)	−1.2 (3.0)	−1.4 (−2.5, −0.8)
Absolute mean difference	2.7 (1.8)	1.6 (0.7, 2.3)	1.5 (1.0)	1.6 (0.6, 2.2)	2.1 (2.4)	1.6 (0.9, 2.7)

Study 2	Total (N = 38)		Adults (N = 17)		Children (N = 21)
Age (years)	20.6 (14.9)	18 (7.5,31.0)	35.3 (10.0)	32.0 (28.0,37.0)	8.6 (2.4)	8.0 (7.0,11.0)
% White	84.6	—	87.4	—	82.2	—
% Black	9.8	—	0.0	—	17.8	—
Height (cm)	150.6 (21.3)	155.0 (132.2, 164.1)	167.6 (10.7)	164.1 (161.0, 175.0)	135.5 (16.4)	132.3 (121.7, 149.6)
Weight (kg)	49.1 (20.3)	51.5 (30.6, 64.3)	66.4 (13.1)	65.9 (55.3, 73.3)	35.2 (13.1)	31.7 (24.0, 43.2)
BMI	19.9 (5.0)	19.3 (15.5, 23.2)	23.3 (3.2)	22.4 (21.3, 24.5)	17.1 (4.5)	15.7 (14.7, 16.9)
BIA %BF (%)	20.1 (7.4)	17.9 (15.1, 21.1)	21.2 (5.7)	19.3 (15.3, 22.5)	19.2 (8.4)	16.2 (15.0, 20)

Mean differences were calculated by taking the difference of the criterion DXA from the BIA measure for each body composition construct.

%BF, percent body fat; BIA, bioelectrical impedance analysis; DXA, dual-energy X-ray absorptiometry; FFM, fat-free mass; FM, fat mass; IQR, interquartile range; SD, standard deviation.

Study 1 (accuracy)

Fifty children (9.1 ± 2.2 years) participated in the study. The sample was predominately female and non-Hispanic White. Children in the sample were generally of healthy weight.

Study 2 (precision)

Twenty-one children (8.6 ± 2.4 years) participated in the study. The sample was predominately female and non-Hispanic White. Children and adults in the sample were generally of healthy weight.

Study 1: Accuracy of BIA-estimated %BF

BIA underestimated %BF by 7.9 percentage points when compared with DXA (28.5% ± 6.5% vs. 20.6% ± 9.0%), representing a relative difference between DXA and BIA of 29.8%, with similar differences for boys and girls (i.e., 7.9 and 9.0 percentage point differences, respectively). Spearman rank-order correlations (see Table 3) between DXA- and BIA-estimated %BF were strong for %BF (r = 0.83). However, %BF CCCs (see Table 3) were moderate (CCC = 0.53). The Bland–Altman plots comparing DXA and BIA estimated %BF (see Fig. 1) and indicated a bias for girls of 7.9% with limits of agreement ranging from −3.0% to 18.8%, and a bias for boys of 7.8% with limits of agreement ranging from 0.1% to 15.5%.

Figure 1.

Study 1: Bland–Altman for difference in agreement, % body fat of BIA analysis vs. DXA for (a) girls and (b) boys. BIA, bioelectrical impedance; DXA, dual-energy X-ray absorptiometry.

Table 3.

Study 1: Correlations for Bioelectrical Impedance and Dual-Energy X-Ray Absorptiometry

Spearman rank-order correlations	Total (N = 50)			Boys (N = 22)			Girls (N = 28)
% Body fat	0.83			0.55			0.88
FM	0.96			0.89			0.97
FFM	0.99			0.98			0.98
Lin's CCC
	Total (N = 50)			Boys (N = 22)			Girls (N = 28)
	Coef	95% CI		Coef	95% CI		Coef	95% CI
% Body fat	0.53	0.40	0.65	0.55	0.34	0.71	0.50	0.30	0.64
FM	0.91	0.86	0.94	0.95	0.91	0.98	0.86	0.77	0.92
FFM	0.94	0.90	0.97	0.97	0.94	0.99	0.92	0.83	0.96

CCC, concordance correlation coefficients; CI, confidence interval.

The slope of the regression line examining the difference between the two measures against the mean was 0.15 for girls and 0.08 for boys and did not reach statistical significance for either sex (girls 95% confidence interval [CI] = −0.22 to 0.53; boys 95% CI = −0.19 to 0.34). Other analyses stratified by sex were similar to the overall sample with the exception of Spearman rank-order correlations (see Table 3), which were different for boys (0.55) and girls (0.88).

Study 2: Precision of estimated %BF from opportunistic BIA measures

The within- and between-day RMSSD for children and adults are presented in Table 4.

Table 4.

Study 2: Precision in Bioelectrical Impedance-Estimated Percent Body Fat

		Root mean squared standard deviation	95% CI	LSC
Kids within-day variability		1.04	0.74 to 1.34	2.88
Kids between-day variability	Fasted	0.57	0.31 to 0.84	1.58
	9 am	0.42	0.24 to 0.60	1.16
	11 am	0.65	0.37 to 0.94	1.80
	1 pm	0.95	−1.58 to 1.36	2.63
	5 pm	0.74	0.42 to 1.06	2.05
	Bed	0.72	0.41 to 1.03	1.99
Adults within-day variability		1.10	0.74 to 1.46	3.05
Adults between-day variability	Fasted	0.61	0.32 to 0.90	1.69
	9 am	0.69	0.33 to 1.05	1.91
	11 am	0.50	0.24 to 0.76	1.39
	1 pm	0.71	0.34 to 1.08	1.97
	5 pm	0.58	0.29 to 0.88	1.61
	Bed	0.78	0.41 to 1.14	2.16

Within-day referent: fasted; between-day referent: same time across days.

LSC, least significant change.

The within-day RMSSD of BIA-estimated %BF suggested that the Tanita DC-430U is more precise in children 1.04 (95% CI = 0.74–1.34) compared with adults 1.10 (95% CI = 0.74–1.46). The between-day RMSSD of BIA-estimated %BF indicated that the Tanita DC-430U is more precise in the morning compared with the afternoon (see Table 4). LSC of BIA-estimated %BF suggested that change in %BF would have to be greater than 1.16%–2.63% to be a change beyond measurement error in children between days. For adults, between-day LSC of BIA-estimated %BF suggested that change in %BF would have to be greater than 1.61%–2.16% to be a change beyond measurement error. Within-day LSC of BIA-estimated %BF suggested that change in %BF would have to be greater than 2.88% and 3.05% to be considered a change beyond measurement error, in children and adults, respectively.

For children, regression analysis indicated a positive trend at 9 am and a negative trend in bias as %BF increased throughout the day, within day (see Supplementary Table S1). For adults, regression analysis indicated a positive and statistically significant trend throughout the day, within day (see Supplementary Table S1). Between-day linear regression of the bias indicated a positive trend at 11 am, but a negative trend in bias at all other time points as %BF increased for children and adults (see Supplementary Table S1).

Discussion

The overall aim of this project was to assess the accuracy and precision of the Tanita DC-430U BIA-estimated %BF. The aim of study 1 was to assess the accuracy of the new Tanita DC-430U BIA device against DXA. Findings indicated that the Tanita DC-430U underestimates %BF compared with DXA by 8.0 percentage points in elementary school-aged children. Differences in %BF between BIA and DXA were driven by lower BIA-estimated FM (7.8 kg vs. 9.9 kg, p < 0.05) and higher BIA-estimated FFM (25.3 kg vs. 24.1 kg, p < 0.05). Absolute agreement for estimated %BF was moderate (CCC 0.53) between DXA and BIA. Bland–Altman analyses also showed large biases and limits of agreement. These differences are substantially larger than previous studies comparing different Tanita BIA models with reference measures.^10,26,27

These findings highlight the need for the assessment of the accuracy of new BIA devices when they are introduced to the market. Based on the findings of this study, the DC-430U is less comparable with DXA than previous Tanita devices. Thus, it may be problematic to compare findings using this device with previous Tanita devices. However, Spearman rank-order correlations were strong (r = 0.83), which provides evidence that the Tanita DC-430U can accurately rank participants in terms of relative %BF. The utility of ranking participants given the large limits of agreement may be limited, however, because of the large differences between BIA and DXA and the wide limits of agreement.

Study 1 does have limitations that should be considered when interpreting the results. This was a relatively small sample of healthy, predominately non-Hispanic White children with little variability in body mass, which may not be representative of all children. Ultimately, future studies might use larger more heterogeneous samples to examine the concordance of BIA-estimated %BF with multiple reference measures of body composition. Although participants in study 1 were fasted at measurement, future research may need to also account for additional factors such as hormonal fluctuations or duration of fasting that might differentially impact the agreement between BIA-estimated %BF and DXA. Estimates of %BF also need to be considered in light of inherent technical measurement errors from using the Tanita DC-430U.

The aim of study 2 was to examine the precision of the Tanita DC-430U among children and adults by assessing the within- and between-day variability %BF estimates collected at home to simulate opportunistic measures that are commonly used in applied public health settings. The findings from the current study suggest that opportunistic measures of BIA-estimated %BF using the DC-430U are more precise within a day in children compared with adults (RMSSD of 1.04 vs. 1.10) and between days in the morning rather than the afternoon. The differences in precision between children and adults may be due to a larger overall mass in adults.²⁸ Other potential sources of differences in precision between children and adults include developmental stage, fitness levels, and dietary intake. These findings suggest that studies using opportunistic measures of BIA-estimated %BF should standardize the time of day that measures are taken to minimize measurement error.

Understanding the precision, by examining the within-day variability, of BIA-estimated %BF for studies that use opportunistic measures using the Tanita DC-430U is also critical because it can help to establish the upper bound of variation in %BF that may be expected due to error. Ideally the LSC would be calculated for each measurement device; practically this is rarely done in applied research. Thus, this study can help to inform interpretation of previous and future studies that use opportunistic measures of BIA using the DC-430U to predict %BF. For example, in this study, the maximum LSC between fasted and a following measure during the same day was 2.63 and 2.16, for children and adults, respectively.

This finding suggests that studies using opportunistic measures of BIA-estimated %BF using the Tanita DC-430U, and that observe changes of less than 2.63 and 2.16 for children and adults, respectively, are likely due to measurement error and not true change. To place this finding in context, a recent systematic review and meta-analysis of childhood obesity interventions identified 29 effects from 24 studies that measured %BF, with nearly all of those studies using opportunistic measures of BIA.²⁹ The mean difference of change in %BF between intervention and control was greater than 2.63 in only three of the identified effects. This suggests that most of the differences in change did not fall outside of what would be expected due to measurement error, and thus, may not be due to true change.

Study 2 was the first study to examine the real-world performance of opportunistic measures of %BF in children and in adults using the Tanita DC-430U. The current study should be interpreted in the context of its limitations, including the inherent technical, user, and prediction error from using opportunistic measures of the Tanita DC-430U in the real world.

However, this study is one of the first studies to quantify and report the upper bounds of that error. The sample size was also relatively small, healthy in terms of BMI and %BF, and predominately non-Hispanic White individuals, and so, findings may not generalize to all children or adults. Furthermore, this study used BIA measurements using the Tanita DC-430U, a foot-to-foot multifrequency device, and findings may not generalize to other BIA techniques (i.e., single-frequency, multifrequency, or BIA spectroscopy) or devices. In addition, protocols took place at the participant's home due to the COVID-19 pandemic, and so, procedures could not be overseen by research staff to ensure measurement protocol fidelity.

Future research might include a more heterogenous sample to increase generalizability and uncover any differences by sex or ethnicity. Participants in study 2 were instructed to be fasted and voided for their first measurement; however, future studies may be conducted in a more controlled environment to account more accurately for fasting duration, food and liquid intake, exercise, and clothing worn.

Conclusions

Ideally, body composition measurements would be taken with the participant in a fasted state; however, the Tanita DC-430U may provide clinically useful information to supplement BMI and BMI z-score even when opportunistic measures are used. Opportunistic measures may be best suited for children and in the morning but may be taken throughout the day with the understanding of the variability within and between days. To benefit population health and more widely utilize BIA as a means to supplement BMI and BMI z-score, it is essential to understand the variability in BIA-estimated %BF if studies use opportunistic measures.

Impact Statement

Current protocols for bioelectrical impedance (BIA) require a 4–10-hour fast and voiding of bladder and bowels. However, community-based research studies often do not follow these protocols measuring BIA opportunistically. Understanding the accuracy and precision of opportunistic measures of BIA-estimated percent body fat allows for better interpretation of results from community-based research.

Funding Information

No funding was received.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

References

Kuczmarski

. 2000 CDC Growth Charts for the United States: Methods and Development. Department of Health and Human Services, Centers for Disease Control and Prevention; 2002.

Zemel

, Riley

, Stallings

. Evaluation of methodology for nutritional assessment in children: Anthropometry, body composition, and energy expenditure. Annu Rev Nutr, 1997; 17:211–235.

World Health Organization. BMI-for-age (5–19years) 2021. https://www.who.int/tools/growth-reference-data-for-5to19-years/indicators/bmi-for-age (Last accessed January 17, 2022).

Freedman

, Horlick

, Berenson

. A comparison of the Slaughter skinfold-thickness equations and BMI in predicting body fatness and cardiovascular disease risk factor levels in children. Am J Clin Nutr, 2013; 98:1417–1424.

Wohlfahrt-Veje

, Tinggaard

, Winther

, et al. Body fat throughout childhood in 2647 healthy Danish children: Agreement of BMI, waist circumference, skinfolds with dual X-ray absorptiometry. Eur J Clin Nutr, 2014; 68:664.

Tyo

, Nicks

. Effect of moderate intensity exercise on body composition measurements. Meas Phys Educ Exerc Sci, 2020; 24:172–180.

Brantlov

, Ward

, Jødal

, et al. Critical factors and their impact on bioelectrical impedance analysis in children: A review. J Med Eng Technol, 2017; 41:22–35.

Brantlov

, Jødal

, Lange

, et al. Standardisation of bioelectrical impedance analysis for the estimation of body composition in healthy paediatric populations: A systematic review. J Med Eng Technol, 2017; 41:460–479.

Lim

, Hwang

, Lee

, et al. Cross-calibration of multi-frequency bioelectrical impedance analysis with eight-point tactile electrodes and dual-energy X-ray absorptiometry for assessment of body composition in healthy children aged 6–18 years. Pediatr Int, 2009; 51:263–268.

10.

Lee

L-W

, Liao

Y-S

, Lu

H-K

, et al. Validation of two portable bioelectrical impedance analyses for the assessment of body composition in school age children. PLoS One, 2017; 12:e0171568.

11.

Kyle

, Bosaeus

, De Lorenzo

, et al. Bioelectrical impedance analysis—Part II: Utilization in clinical practice. Clin Nutr, 2004; 23:1430–1453.

12.

Böhm

, Heitmann

. The use of bioelectrical impedance analysis for body composition in epidemiological studies. Eur J Clin Nutr, 2013; 67(S1):S79.

13.

Yanovski

, Hubbard

, Heymsfield

, et al. Bioelectrical impedance analysis in body composition measurement: National Institutes of Health Technology Assessment Conference Statement. Am J Clin Nutr, 1996; 64:524S–32S.

14.

Hamilton-James

, Collet

T-H

, Pichard

, et al. Precision and accuracy of bioelectrical impedance analysis devices in supine versus standing position with or without retractable handle in Caucasian subjects. Clin Nutr ESPEN, 2021; 45:267–274.

15.

Ward

. Bioelectrical impedance analysis for body composition assessment: Reflections on accuracy, clinical utility, and standardisation. Eur J Clin Nutr, 2019; 73:194–199.

16.

Goran

, Driscoll

, Johnson

, et al. Cross-calibration of body-composition techniques against dual-energy X-ray absorptiometry in young children. Am J Clin Nutr, 1996; 63:299–305.

17.

Puth

M-T

, Neuhäuser

, Ruxton

. Effective use of Pearson's product–moment correlation coefficient. Anim Behav, 2014; 93:183–189.

18.

Jinyuan

, Wan

, Guanqin

, et al. Correlation and agreement: Overview and clarification of competing concepts and measures. Shanghai Arch Psychiatry, 2016; 28:115.

19.

Gutin

, Yin

, Johnson

, et al. Preliminary findings of the effect of a 3-year after-school physical activity intervention on fitness and body fat: The Medical College of Georgia Fitkid Project. Int J Pediatr Obes, 2008; 3(sup1):3–9.

20.

Kelley

, Kelley

, Pate

. Are there inter-individual differences in fat mass and percent body fat as a result of aerobic exercise training in overweight and obese children and adolescents? A meta-analytic perspective. Child Obes, 2020; 16:301–306.

21.

Nihiser

, Lee

, Wechsler

, et al. BMI measurement in schools. Pediatrics, 2009; 124(Supplement 1):S89–S97.

22.

Ogden

, Carroll

, Kit

, et al. Prevalence of obesity and trends in body mass index among US children and adolescents, 1999-2010. JAMA, 2012; 307:483–490.

23.

Bland

, Altman

. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet, 1986; 327:307–310.

24.

Hudda

, Owen

, Rudnicka

, et al. Quantifying childhood fat mass: Comparison of a novel height-and-weight-based prediction approach with DXA and bioelectrical impedance. Int J Obes, 2021; 45:99–103.

25.

Densitometry ISfC. Precision Assessment & Calculator FAQs 2020. https://iscd.org/knowledge-base/precision-assessment-calculator-faqs/ (Last accessed January 5, 2022).

26.

Wang

, Hui

SS-C

. Validity of four commercial bioelectrical impedance scales in measuring body fat among Chinese children and adolescents. BioMed Res Int, 2015; 1–8.

27.

Kabiri

, Hernandez

, Mitchell

. Reliability, validity, and diagnostic value of a pediatric bioelectrical impedance analysis scale. Childhood Obes, 2015; 11:650–655.

28.

Jensky-Squires

, Dieli-Conwright

, Rossuello

, et al. Validity and reliability of body composition analysers in children and adults. Br J Nutr, 2008; 100:859–865.

29.

Salam

, Padhani

, Das

, et al. Effects of lifestyle modification interventions to prevent and manage child and adolescent obesity: A systematic review and meta-analysis. Nutrients, 2020; 12:2208.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB