Long-term risk of incident type 2 diabetes and measures of overall and regional obesity: the EPIC-InterAct case-cohort study

Abstract

Background: Waist circumference (WC) is a simple and reliable measure of fat distribution that may add to the prediction of type 2 diabetes (T2D), but previous studies have been too small to reliably quantify the relative and absolute risk of future diabetes by WC at different levels of body mass index (BMI).

Methods and findings: The prospective InterAct case-cohort study was conducted in 26 centres in eight European countries and consists of 12,403 incident T2D cases and a stratified subcohort of 16,154 individuals from a total cohort of 340,234 participants with 3.99 million person-years of follow-up. We used Prentice-weighted Cox regression and random effects meta-analysis methods to estimate hazard ratios for T2D. Kaplan-Meier estimates of the cumulative incidence of T2D were calculated. BMI and WC were each independently associated with T2D, with WC being a stronger risk factor in women than in men. Risk increased across groups defined by BMI and WC; compared to low normal weight individuals (BMI 18.5-22.4 kg/m(2)) with a low WC (<94/80 cm in men/women), the hazard ratio of T2D was 22.0 (95% confidence interval 14.3; 33.8) in men and 31.8 (25.2; 40.2) in women with grade 2 obesity (BMI≥35 kg/m(2)) and a high WC (>102/88 cm). Among the large group of overweight individuals, WC measurement was highly informative and facilitated the identification of a subgroup of overweight people with high WC whose 10-y T2D cumulative incidence (men, 70 per 1,000 person-years; women, 44 per 1,000 person-years) was comparable to that of the obese group (50-103 per 1,000 person-years in men and 28-74 per 1,000 person-years in women).

Conclusions: WC is independently and strongly associated with T2D, particularly in women, and should be more widely measured for risk stratification. If targeted measurement is necessary for reasons of resource scarcity, measuring WC in overweight individuals may be an effective strategy, since it identifies a high-risk subgroup of individuals who could benefit from individualised preventive action.

Figure 1. Hazard ratios for type 2 diabetes per 1 SD increase in BMI (SD = 3.6 kg/m²) in men.

Heterogeneity between centres: I ² = 48% (p = 0.012). HRs estimated from modified Cox regression with age as the underlying time scale, using Prentice weights. Centre-specific estimates are combined using random effects meta-analysis.

Figure 2. Hazard ratios for type 2 diabetes per 1 SD increase in BMI (SD = 4.4 kg/m²) in women.

Heterogeneity between centres: I ² = 59% (p = 0.012). HRs estimated from modified Cox regression with age as the underlying time scale, using Prentice weights. Centre-specific estimates are combined using random effects meta-analysis.

Figure 3. Hazard ratios for type 2 diabetes per 1 SD increase in WC (SD = 10.0 cm) in men.

Heterogeneity between centres: I ² = 31% (p = 0.11). HRs estimated from modified Cox regression with age as the underlying time scale, using Prentice weights. Centre-specific estimates are combined using random effects meta-analysis.

Figure 4. Hazard ratios for type 2 diabetes per 1 SD increase in WC (SD = 11.2 cm) in women.

Heterogeneity between centres: I ² = 69% (p<0.001). HRs estimated from modified Cox regression with age as the underlying time scale, using Prentice weights. Centre-specific estimates are combined using random effects meta-analysis.

Figure 5. Cumulative incidence of type 2 diabetes over 10 y by BMI and waist circumference groups in men.

Red line, WC<94 cm; blue line, WC≥94–101.9 cm; black line, WC≥102 cm.

Figure 6. Cumulative incidence of type 2 diabetes over 10 y by BMI and waist circumference groups in women.

Red line, WC<80 cm; blue line, WC≥80–87.9 cm; black line, WC≥88 cm.