Visceral adipose tissue and incident cardiovascular disease among individuals without diabetes: an analysis of 15 prospective cohorts from 7 countries in Latin America

Figures

• 

  Fig. 1. Flowchart of the included sample from the CC-LAC cohorts considering the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Abbreviations: eVAT, Estimated visceral adipose tissue; BMI, Body mass index; CC-LAC, Cohorts Consortium of Latin America and the Caribbean; METS-VF, Metabolic Score for Visceral Fat.

• 

  Fig. 2. Incidence of fatal and non-fatal CVD across eVAT quartiles. Panel a: Unadjusted cumulative incidence Kaplan–Meier curves of fatal and non-fatal CVD events stratified by eVAT quartile at follow-up. Panel b: Sex stratified incidence rate of fatal and non-fatal CVD events across quartiles of estimated visceral adipose tissue. Curves were estimated using the Kaplan–Meier method. The number at risk at each year of follow-up is shown below figure. P-value corresponds to the log-rank test comparing survival distributions across quartiles. Abbreviations: eVAT, Estimated visceral adipose tissue; CVD, Cardiovascular disease; CI, Confidence interval; METS-VF, Metabolic Score for Visceral Fat.

• 

  Fig. 3. Association of eVAT with combined incident fatal and non-fatal CVD events. Panel a: Simulation-based plot with eVAT and outcomes. Panel b: Multivariable Cox proportional hazards models displaying the effect of eVAT quartiles on outcomes. Multivariate models were stratified by age group (≤44, 45–54, 55–65, ≥65 years) and sex, adjusted for urban/rural residence and smoking status and standard errors were clustered by cohort of origin. Abbreviations: eVAT, Estimated visceral adipose tissue; HR, Hazard ratio; CI, Confidence interval.

• 

  Fig. 4. Cox models stratified by potential effect modifiers evaluating the relationship between 100-g increments of eVAT and incident fatal and non-fatal CVD events. Models were stratified by age group (≤44, 45–54, 55–65, ≥65 years) and sex, adjusted for urban/rural residence and smoking status, and standard errors were clustered by cohort of origin. p-for-interaction derived from a likelihood-ratio test comparing models with and without the interaction term. Abbreviations: BMI, Body Mass Index; CI, Confidence Interval; HR, Hazard Ratio.

• 

  Fig. 5. Population-attributable fraction (PAF, black bars) with 95% confidence intervals of 10-year fatal and non-fatal CVD events under three hypothetical eVAT-reduction scenarios. The grey bars represent the residual risk remaining after each intervention. Scenario 1 (Q4 to Q1-Q3, equal shares): individuals in the highest eVAT quartile (≥1441 g) are redistributed equally across the three lower quartiles, representing a moderate reduction in visceral fat burden; PAF = 8.8% (95% CI: 2.8%–14.7%). Scenario 2 (Q3 & Q4 to Q1-Q2, equal shares): individuals in the two highest eVAT quartiles (≥1069 g) are redistributed equally across the two lowest quartiles, representing a more substantial population-level reduction; PAF = 22.6% (95% CI: 12.2%–31.8%). Scenario 3 (Q2, Q3 & Q4 to Q1): all individuals above the lowest eVAT quartile (≥735 g) are shifted to Q1 levels, representing a near-complete normalization of visceral fat across the population; PAF = 33.6% (95% CI: 16.6%–51.1%). PAFs were derived via model-based standardization of Cox models stratified by age group (≤44, 45–54, 55–65, ≥65 years) and sex, adjusted for urban/rural residence and smoking status. We obtained 95% CIs by bootstrap…