A Population Specific Regression Model for Estimating Visceral Fat from BMI in Bengali Older Women

Original Research Article

Abstract

The older adult population of India is experiencing a rising prevalence of metabolic complications, with increased body fat, particularly visceral fat being a primary contributor. Early screening and detection of elevated visceral fat levels can facilitate timely lifestyle interventions and reduce the risk of metabolic disorders. The study aimed to develop and internally validate a population-specific linear regression model to estimate bioelectrical impedance-derived visceral fat from body mass index (BMI) among Bengali older women. This cross-sectional study included 217 Bengali older women aged 60-85 years, selected through simple random sampling from southern West Bengal. Standard anthropometric measurements were recorded, and visceral fat was estimated using bioelectrical impedance analysis (BIA). A simple linear regression model was developed using BMI as the predictor variable, with the sample split into training (70%) and testing (30%) subsets for internal validation. Bootstrapping with 1,000 resamples was applied to improve the robustness of parameter estimates. The regression model was statistically significant (F = 2067.43, p < 0.001) and explained 93% of the variance in BIA-derived visceral fat (R² = 0.93, adjusted R2 = 0.93). The derived regression equation was: VF = −16.889 + 1.065 × BMI. The internally validated model provides a cost-effective approach for estimating visceral fat at the population level among Bengali older women and may aid preliminary metabolic risk screening in resource-limited settings. Keywords: Bengali older women, metabolic risk screening, visceral fat, linear regression model, ageing, biological anthropology