Reliability of body composition assessment using A-mode ultrasound in a heterogeneous sample

Accurate and reliable assessment of body composition is crucial in various fields, including sports medicine and clinical disciplines focused on weight management. While methods like DXA, MRI, UWW, and ADP are considered accurate, they are expensive and require specialized equipment. Field methods such as anthropometry, BIA, and US offer cost-effective alternatives, but their validity and reliability need thorough evaluation across diverse populations. Ultrasound (US) shows promise as a body composition assessment tool, with advancements in hardware and software. Existing studies on US validity have yielded mixed results, highlighting the need for further investigation. While several studies have addressed the validity of US for body composition assessment, fewer have evaluated its reliability, especially using a heterogeneous sample and considering the impact of different prediction formulas. This study aimed to address this gap by assessing the reliability of A-mode US for %BF estimation using four different formulas in a large, heterogeneous sample, and to determine if reliability differed between genders.

Several studies have explored the validity of ultrasound for body composition assessment, with varying results. Some studies showed good agreement between US measurements and DXA or other gold-standard methods, particularly when combined with anthropometry or in specific athletic populations. However, other studies revealed significant differences between US and DXA or other techniques, suggesting inconsistencies possibly related to the device used or the specific prediction formulas employed. The accuracy of US in measuring subcutaneous adipose tissue layer thickness has been validated in studies using excised tissues and cadavers. While the validity of US is being established, its reliability, especially for tracking treatment progress, is equally vital. Previous reliability studies on US for %BF were often limited by small, homogeneous samples and focused only on a few prediction formulas. The present study aimed to overcome these limitations.

This study recruited 144 healthy adults (81 men, 63 women) aged 18-70 years via social networks and flyers. Participants provided written informed consent, and the study was approved by the institutional Committee of Research Ethics. Body mass (BM) and height were measured using calibrated equipment. Subcutaneous fat layer thickness was measured at eight anatomical sites using a BodyMetrix™ BX2000 A-mode ultrasound instrument at 2.5 MHz. Two testers, each with about one year of experience, performed triplicate measurements on each subject. Percent body fat (%BF) was calculated using four formulas from BodyView™ software: 7-sites Jackson and Pollock (JP7), 3-sites Jackson and Pollock (JP3), 3-sites Pollock (P3), and 1-point biceps (BIC). Data analysis used MATLAB, including Bland-Altman analysis for assessing agreement between measurements, and calculation of intraclass correlation coefficient (ICC), technical error of measurement (TEM), standard error of measurement (SEM), and minimal detectable change (MDC) to quantify reliability. The impact of gender on reliability was also analyzed.

The 7-sites Jackson and Pollock (JP7) formula yielded the highest reliability for %BF estimates, with intraclass correlation coefficients (ICCs) of 0.979 for Tester 1 and 0.985 for Tester 2. Technical error of measurement (TEM) values were 1.07% BF for Tester 1 and 0.89% BF for Tester 2. Minimal detectable change (MDC) values were 2.95% BF for Tester 1 and 2.47% BF for Tester 2. The intertester bias was -0.5% BF, with an ICC of 0.972 and an MDC of 3.43% BF for the entire sample. Reliability was generally higher for men than women. Analysis of consecutive pairs of trials showed no clear trend suggesting learning effects. The number of sites used in the prediction formula influenced reliability, with more sites leading to higher reliability. A paired t-test revealed a significant underestimation of subcutaneous adipose tissue layer thickness by Tester 1 compared to Tester 2, particularly in women, contributing to the intertester bias.

This study demonstrates the high reliability of A-mode ultrasound for %BF assessment, particularly when using the 7-sites Jackson and Pollock formula. The findings support the use of this technique for monitoring changes in body composition, especially in men. However, the lower reliability in women and the observed intertester variability highlight the importance of standardized training and procedures to minimize examiner-related errors. The influence of the number of measurement sites underscores the importance of selecting appropriate formulas for accurate and reliable %BF estimates. Future research could focus on improving the training protocols for examiners to reduce inter-tester variability and improve precision, particularly in women. Further investigation of the factors contributing to the observed gender differences in reliability could provide valuable insights into optimizing US-based body composition assessment.

A-mode ultrasound is a highly reliable method for assessing percent body fat, especially when using formulas incorporating multiple anatomical sites. While precise for men, further improvements are needed to enhance reliability in women. Minimizing examiner-related variability through standardized training protocols is crucial for improving the technique's precision. This technique is suitable for tracking moderate changes in body composition.

The study's sample, while heterogeneous, may not be fully representative of all populations. The use of only two testers limits the generalizability of inter-tester reliability findings. The study focused solely on reliability and did not assess the validity of the method against a gold-standard technique.