Abstract
Objectives
The association between body mass index (BMI) and total body adiposity differs across ethnic groups. For instance, South Asians (SA) and East Asians (EA) have lower body fat for a given BMI level than Europeans, while the opposite is true for African-Caribbeans (AC). This suggests that the relationship between BMI and type 2 diabetes (T2D) risk may also vary depending on ethnicity. We conducted a meta-analysis to investigate whether the association between BMI and the risk of T2D differs across ethnic groups.
Methods
MEDLINE, EMBASE and Web of Science were searched up to July 2020. We included prospective cohort studies of >2 years, which investigated the association between BMI and T2D incidence among adults of a specified ethnicity. Linear and non-linear dose-response meta-analyses were performed using random effects models, with subgroup analyses by ethnicity. The heterogeneity among studies was estimated using the Cochran Q test and I2 statistic. Study quality was assessed with the Newcastle-Ottawa Scale.
Results
54 studies were included. Cohorts were stratified into the following ethnic subgroups: AC (N = 67,453), EA (N = 1,012,135), European (N = 206,424), Indigenous (N = 10,533), Latin American (LA) (N = 4,669), SA (N = 9,395), and Southeast Asian (SEA) (N = 51,129). Linear dose-response associations between 1 kg/m2 increase in BMI and T2D were observed for the SEA (RR = 1.26; 95% CI, 1.10, 1.30) and SA (RR = 1.11; 95% CI: 1.04, 1.19) subgroups with no evidence of departure from linearity. Associations departed from linearity for all other subgroups. At a BMI level of 30 kg/m2, the non-linear dose-response curves for each of the other subgroups displayed the following risk ratios; AC: RR = 3.13 (95% CI, 1.95, 5.02), EA: RR = 2.39 (95% CI, 1.96, 2.92), European: RR = 7.41 (95% CI, 3.88, 14.18), Indigenous: RR = 8.15 (95% CI, 6.07; 10.95), and LA: RR = 12.82 (95% CI, 5.50, 29.92). For all subgroups, there was a high degree of interstudy heterogeneity (I2 > 75%).
Conclusions
Our findings indicated that the association between BMI and the risk of T2D differs across ethnic groups, suggesting that ethnic-specific BMI cut-offs could be helpful in identifying cardiometabolic risk profiles across different populations.
Funding Sources
Canadian Institutes for Health Research.