Abstract
BACKGROUND: Visceral adiposity index (VAI) has recently been suggested to be used as a surrogate of visceral adiposity. We examined if VAI could improve predictive performances for CVD of the Framingham's general CVD algorithm (a multivariate model incorporating established CVD risk factors). We compared the predictive abilities of the VAI with those of simple anthropometric measures i.e. BMI, waist-to-height ratio (WHtR) or waist-to-hip ratio (WHpR). DESIGN AND METHODS: In a nine-year population-based follow-up, 6,407 (2,778 men) participants, free of CVD at baseline, aged≥30 years were eligible for the current analysis. The risk of CVD was estimated by incorporating VAI, BMI, WHpR, and WHtR, one at a time, into multivariate accelerated failure time models. RESULTS: We documented 534 CVD events with the annual incidence rate (95%CIs) being 7.3 (6.4-8.3) among women and 13.0 (11.7-14.6) among men. Risk of future CVD increased with increasing levels of VAI among both men and women. VAI was associated with multivariate-adjusted increased risk of incident CVD among women. However, the magnitude of risk conferred by VAI was not significantly higher than those conferred by BMI, WHpR, or WHtR. Among men, after adjustment for established CVD risk factors, VAI was no longer associated with increased risk of CVD. VAI failed to add to the predictive ability of the Framingham general CVD algorithm. CONCLUSIONS: Using VAI instead of simple anthropometric measures may lead to loss of much information needed for predicting incident CVD.