Discussion

This study evaluated the relationship between the CI and BIA-derived VF% among adolescents with overweight and obesity and compared its performance with conventional anthropometric indices. The principal finding was that although the CI was statistically correlated with VF%, its correlation and predictive contribution were substantially weaker than those of WC, AVI, and BMI.

WC and AVI showed the strongest correlations with VF%, followed closely by BMI. This finding supports the established role of waist-based measures as practical markers of central adiposity [12]. WC directly reflects abdominal size and is therefore more closely related to visceral fat accumulation than composite indices that mathematically adjust for weight and height. Although BMI does not distinguish between fat and lean mass [10], it remained a strong predictor of VF% in this cohort of adolescents with overweight and obesity; one plausible, though not directly tested, explanation is that overall adiposity was uniformly high across this restricted sample.

Contrary to our a priori hypothesis, the CI did not outperform conventional indices; instead, it showed the weakest association with VF% among the indices examined. This weak correlation should not be interpreted as evidence that the CI lacks any clinical value; several, likely overlapping, explanations are plausible, including the indirect, algorithm-based nature of BIA-derived VF% relative to imaging methods, rapid and heterogeneous changes in body composition during adolescent growth and puberty, possible ethnic-specific patterns of fat distribution, and the restriction of this cohort to adolescents already classified as overweight or obese, which narrows the range of adiposity and may attenuate observed correlations. As a post hoc hypothesis rather than a directly tested finding, we suggest that the CI may capture body shape or abdominal geometry rather than true adiposity burden. Mechanistically, the CI’s comparatively weak performance may partly reflect its dependence on total body weight in its denominator: during puberty, fat and lean mass tend to increase together, which may dilute the specificity of a weight-normalized waist index for visceral fat, unlike WC, which reflects abdominal girth directly. The finding is also important because the CI was originally developed using an adult geometric model in which central fat accumulation changes the body shape from cylindrical to biconical [14]. Adolescents, however, undergo rapid and heterogeneous changes in height, weight, lean mass, and fat distribution, particularly during puberty [16]. These developmental changes may reduce the validity of adult-derived anthropometric assumptions in pediatric populations. Prior evaluations of WC, WHR, and the CI as screening tools for trunk fat mass in children and adolescents aged three to 19 years, using dual-energy X-ray absorptiometry as the reference method, similarly found that WC substantially outperformed the CI [22]. Likewise, a study of 314 children found that WHtR was a good predictor of excess body fat measured by BIA analysis, whereas the CI demonstrated poor predictive performance (area under the curve <0.70) [23]. Collectively, these studies indicate that the CI has generally been less accurate than WC or WHtR for identifying adiposity in children, although differences in age groups, adiposity measures, and study populations may contribute to variability in findings; future studies could usefully compare the CI directly against newer composite anthropometric indices in adolescent cohorts. Because this cohort was drawn exclusively from a single tertiary-care center and restricted to adolescents already classified as overweight or obese, these findings should not be generalized to unselected, community-based adolescent populations, in whom the correlation structure between the CI and visceral adiposity may differ.

The regression analysis further supports this interpretation. The base model containing age, sex, and BMI explained more than half of the variance in VF%. The addition of WC substantially improved model performance, whereas the addition of the CI provided a smaller improvement. Importantly, when WC was already included, the incremental gain from adding the CI was minimal. This indicates that the CI does not offer substantial additional clinical information beyond BMI and WC.

AVI performed similarly to WC, but this is expected because it is mathematically derived largely from WC [18]. Therefore, despite comparable statistical performance, AVI may not provide a practical advantage over the simpler measurement of WC. Similarly, WHtR, WHR, and BAI did not significantly improve prediction when BMI was already included in the regression model.

The findings have practical clinical implications. In resource-limited or routine clinical settings, BMI and WC remain simple, inexpensive, and reproducible tools for identifying adolescents with increased visceral adiposity [10,11]. WC, in particular, is easy to measure and directly reflects central adiposity [12]. The CI may be used as an adjunctive research measure, but the present findings do not support replacing conventional anthropometric measures with the CI in adolescent obesity assessment [22,23].

This study has several strengths. It focused specifically on adolescents who were overweight and obese, a clinically important group at increased risk for future cardiometabolic disease. It compared multiple anthropometric indices within the same cohort and used standardized measurement procedures.

However, several limitations should be acknowledged. First, the cross-sectional design precludes causal inference. Second, BIA is an indirect method and is less precise than reference imaging techniques such as computed tomography or magnetic resonance imaging [20]; body composition estimates were not validated against an imaging reference in this cohort, and the device’s prediction equations have been validated predominantly in adult populations. Third, the hospital-based sampling strategy may limit generalisability to community-based adolescent populations and may introduce selection bias; detailed participant-flow data (numbers screened, excluded, and refusal rate) were not systematically logged during recruitment and are not reported. Fourth, biochemical cardiometabolic markers such as fasting glucose, insulin resistance indices, and lipid profile were not included, limiting assessment of metabolic risk beyond anthropometry and body composition. Fifth, although sexual maturity rating was recorded, it was not entered as a covariate in the regression models, and pubertal stage may confound the relationship between anthropometric indices and visceral fat. Sixth, dietary intake, physical activity, and socioeconomic status-all of which influence body composition-were not assessed and could not be adjusted for. Seventh, formal multicollinearity diagnostics (e.g., variance inflation factors) and residual diagnostics (e.g., Cook’s distance, residual plots) were not performed for the regression models; given the correlation observed between WC, AVI, and BMI (Figure 1), this is a relevant caveat, and the model was not internally validated by bootstrapping or cross-validation, nor externally validated in an independent cohort, so its predictive performance should be interpreted as exploratory rather than confirmed. Eighth, formal intra-/inter-observer reliability statistics for repeated anthropometric measurements were not calculated. Finally, although the sample size was adequately powered for the primary correlation analysis, it was not calculated to support the secondary, sex-stratified comparisons reported in Table 1, which should therefore be interpreted with caution. Future research should include longitudinal, community-based cohorts with biochemical markers, pubertal-stage adjustment, and reference imaging methods to validate anthropometric indices across different stages of adolescent development. Establishing age-, sex-, and ethnicity-specific cut-offs for composite indices such as the CI may also help clarify whether they have clinical utility in pediatric populations.