Objectives: The aim of this study was to compare novel and established anthropometrical measures in their ability to predict cardiovascular disease (CVD), and to determine whether they improve risk prediction beyond classical risk factors in a cohort study of 60-year-old men and women. We also stratified the results according to gender to identify possible differences between men and women. Furthermore, we aimed to replicate our findings in a large independent cohort (The Malmö Diet and Cancer study-cardiovascular cohort).
Methods: This was a population-based study of 1751 men and 1990 women, aged 60 years and without CVD at baseline, with 375 incident cases of CVD during 11 years of follow-up. Weight, height, waist circumference (WC), hip circumference and sagittal abdominal diameter (SAD) were measured at baseline. Body mass index (BMI), waist-hip ratio (WHR), waist-hip-height ratio (WHHR), WC-to-height ratio (WCHR) and SAD-to-height ratio (SADHR) were calculated.
Results: All anthropometric measures predicted CVD in unadjusted Cox regression models per s.d. increment (hazard ratios, 95% confidence interval), while significant associations after adjustments for established risk CVD factors were noted for WHHR 1.20 (1.08-1.33), WHR 1.14 (1.02-1.28), SAD 1.13 (1.02-1.25) and SADHR 1.17 (1.06-1.28). WHHR had higher increases in C-statistics, and model improvements (likelihood ratio tests (P<0.001)). In the replication study (MDC-CC, n=5180), WHHR was the only measure that improved Cox regression models in men (P=0.01).
Conclusion: WHHR, a new measure reflecting body fat distribution, showed the highest risk estimates after adjustments for established CVD risk factors. These findings were verified in men but not women in an independent cohort.