Pairwise Agreement Among Field-Based Body-Fat Estimation Methods in Physically Active Adults: A Cross-Sectional...

Background: Field-based body-fat estimates may differ substantially across methods, but similarity in group averages does not necessarily indicate close individual-level agreement. This study characterized pairwise numerical agreement among four field-based body-fat estimation methods without designating a criterion reference. Methods: Forty physically active adults (19 women and 21 men) underwent same-session assessment using archived Carter-derived and Faulkner-derived anthropometric estimates, single-frequency bioelectrical impedance analysis (BIA; Omron HBF-514C), and multifrequency BIA (InBody 120). Agreement was evaluated using paired differences, 95% limits of agreement (LoA), bootstrap confidence intervals, and analyses of non-constant mean and variance structures. Difference-on-paired-mean patterns were additionally examined considering mathematical coupling and unequal method variances. Results: Carter-derived estimates averaged 4.16 percentage points (pp) lower than Faulkner-derived estimates, whereas Omron estimates averaged 5.34 pp higher than InBody estimates. Carter-Faulkner showed an approximately stable mean difference but increasing dispersion across the measurement range, supporting level-specific LoA. Faulkner-InBody showed a small mean difference (-1.48 pp) despite wide global LoA (-12.91 to 9.94 pp). Anthropometry-BIA comparisons showed negative difference-on-paired-mean slopes (-0.58 to -0.72); however, these patterns were algebraically consistent with mathematical coupling and unequal method variances and were not interpreted as independent evidence of adiposity-level effects. Conclusions: The methods showed systematic pairwise differences and substantial individual-level disagreement. Because no criterion method, replicate measurements, or prespecified application-specific acceptability threshold was available, the findings establish neither accuracy, superiority, nor interchangeability among the methods.